Ultrastructure of the endocytotic pathway in glutaraldehyde-fixed and high-pressure frozen/freeze-substituted protoplasts of white spruce (Picea glauca)

Endocytosis via clathrin-coated pits and vesicles, and the subsequent fate of endocytosed proteins and other sub-stances, has been extensively studied in animal cells. In plant cells until recently, research on endocytosis was hindered by (a) the supposition that endocytosis was not ener-getically possible in turgid plant cells, and (b) the presence of cell walls, which limit access to plant plasma membranes (Fowke et al., 1991).

Clathrin-coated pits on the plasma membrane and clathrin-coated vesicles in the cytoplasm of plant cells are believed to function mainly in the uptake and recycling of plasma membrane and plasma membrane components (Coleman et al., 1988). However, many studies have indi-cated that plant cells can internalize a variety of large, exogenously supplied molecules (Nishizawa and Mori, 1977; Coleman et al., 1988, and references therein).

Recently, a number of ultrastructural studies using mem-brane-impermeant, electron-opaque tracers have confirmed that endocytosis via coated pits and coated vesicles is pos-sible in plant cells and protoplasts (reviewed by Fowke et al., 1991; see also Samuels and Bisalputra, 1990; Owen et al., 1991; Lazzaro and Thomson, 1992). The subsequent appearance of tracers in various organelles, vesicles and vacuoles has provided valuable information about organelle functions and the pathways of intracellular transport in plant cells. Two types of markers have been employed in ultra-structural studies: soluble tracers of fluid-phase uptake such as heavy metal salts and lucifer yellow (Hübner et al., 1985; Samuels and Bisalputra, 1990; Owen et al., 1991; Lazzaro and Thomson, 1992) and the insoluble plasma membrane-bound tracers, cationized ferritin (CF) and lectin-gold con-jugates (Tanchak et al., 1984; Joachim and Robinson, 1984; Hillmer et al., 1986; Domozych and Nimmons, 1992). Since the soluble fluid-phase markers can penetrate cell walls, they can be used to study endocytosis in intact cells and tissues at normal turgor pressure. However, they must be precipitated in situ during fixation to visualize them for electron microscopy. Drawbacks are that lucifer yellow may not be completely membrane-impermeant (Robinson and Hedrich, 1991), while the heavy metals are toxic and can have deleterious effects on cell structure and function (Wheeler et al., 1972; Romanenko et al., 1986; Lazzaro and Thomson, 1992).

Although endocytosis of insoluble membrane-bound tracers should reflect the normal processes of vesicle-medi-ated turnover of plasma membrane (Fowke et al., 1991), these tracers can’t penetrate the cell walls of higher plants and therefore have only been used to study endocytosis in protoplasts. Nevertheless, the most detailed information on the time course and pathway of tracer uptake has been obtained from studying the endocytosis of CF in protoplasts of suspension-cultured soybean cells (Tanchak et al., 1984; Tanchak, 1987; Tanchak and Fowke, 1987; Tanchak et al., 1988; Fowke et al., 1989). This is because the plasma mem-branes of many protoplasts can be rapidly and uniformly labelled at once, resulting in the endocytosis of large quantities of cationized ferritin and subsequent labelling of relatively large numbers of organelles.

We wished to determine whether the time course and pathway of CF endocytosis as elucidated in soybean and also in bean leaf protoplasts (Joachim and Robinson, 1984) differs in protoplasts of more distantly related species. For this purpose, we chose meristematic protoplasts prepared from embryogenic suspension cultures of the gymnosperm Picea glauca, or white spruce (Attree et al., 1989a,b; Attree and Fowke, 1993).

Another concern to us was that all previous ultrastructural studies of endocytosis (apart from one study of heavy-metal uptake) have employed slow-acting aldehyde fixatives, which can disrupt labile cellular membranes and membrane-bound compartments (Mersey and McCully, 1978; Wilson et al., 1990) and may therefore be particularly unsuitable for studying the ultrastructural cytology of endocytosis. To address the possibility of fixation artefacts, we employed both glutaraldehyde fixation and high-pres-sure freezing in a prototype Balzers HPM 010 high-pressure freezing apparatus (see Gilkey and Staehelin, 1986 for details), followed by freeze substitution.

The WS1 line of suspension-cultured embryogenic white spruce, originally established by Hakman and Fowke (1987), was main-tained as previously described (Hakman and Fowke 1987), except that subculturing was carried out every 7 days, by mixing 50 ml of fresh LP medium with each 50 ml of WS1 culture, and decanting half of the mixture into a new 250 ml flask.

The optimized method for a rapid, high yield of WS1 protoplasts was modified from Attree et al. (1989a) by substituting pectolyase (Sigma Chemical Company, St. Louis, MO, USA) for the mix-ture of rhozyme HP-150, pectinase and driselase. WS1 suspension-cultured cells (6-7 days old) were collected on Miracloth (Chicopee Mills, New York, USA), drained of excess liquid and weighed. They were then preplasmolyzed for 1 hour on a gyratory shaker at 50-75 rpm in a protoplast isolation buffer consisting of 10 mM 2[N-morpholino]ethanesulfonic acid (MES) at pH 5.6, 6 mM CaCl₂.2H₂O, 0.7 mM NaH₂PO₄ and 0.44 M sorbitol. The plasmolyzed cells were collected again by filtration and trans-ferred to isolation buffer containing 1% (w/v) Onozuka R-10 cellulase (Yakult Pharmaceutical Industry, Nishinomiya, Japan) and 0.2% (w/v) pectolyase. This enzyme solution was centrifuged before use for 5-10 minutes at 2500 g to precipitate insoluble solids. The preplasmolyzed cells were added to the enzyme solution at a ratio ranging between 0.2 and 0.6 g wet weight of WS1 cells per ml of solution, and incubated on a gyratory shaker at 100 rpm for 2-2.5 hours. The resultant protoplasts were collected by filtration through an 80 μm mesh nylon filter and collected in conical centrifuge tubes by centrifugation for 4 minutes at 100 g. Protoplasts were resuspended and washed in enzyme-free isolation buffer and then in an incubation buffer containing 10 mM N-[2-hydroxyethyl]piperazine-N′-2-ethanesulfonic acid (HEPES) (tissue culture grade) at pH 7, in place of MES. Protoplasts were allowed to recover for at least 1 hour prior to subsequent labelling with CF in the incubation buffer.

Incubation of protoplasts in CF was carried out as described by Tanchak et al. (1988). Protoplasts in incubation buffer were col-lected by centrifugation in 1.5 ml polypropylene microcentrifuge tubes to yield pellets of approximately 0.15 ml volume. Pellets were resuspended to 0.4 ml, then diluted to 0.8 ml in incubation buffer containing 1 mg/ml CF (stock contained 10 mg/ml in 0.125 M NaCl, 2 mM azide; Molecular Probes, Eugene, OR, USA), to give a final concentration of 0.5 mg/ml CF. Control protoplasts were incubated in CF-free incubation buffer containing 6.3 mM NaCl, 0.1 mM NaN₃ (approx. 0.006%). The small amounts of NaCl and NaN₃ had no detectable effects on protoplasts. Proto-plasts incubated for 30 minutes or less in CF were fixed by the addition of 0.4 ml of ice-chilled 3% w/v glutaraldehyde (in incubation buffer) to give a final concentration of 1% glutaraldehyde. For incubations of 60 minutes or more, protoplasts were first washed several times in succession by centrifugation and resus-pension in CF-free incubation buffer before fixation in ice-chilled 1% glutaraldehyde. All protoplasts were fixed in 1% glutaralde-hyde on ice for 1-2 hours, followed by fixation in 3% glutaralde-hyde in incubation buffer for 2-3 hours at room temperature. Fixed protoplasts were washed in 10 mM HEPES, then postfixed in 1% osmium tetroxide in 10 mM HEPES, dehydrated first to 100% ethanol, then to propylene oxide, followed by infiltration and embedding in Araldite resin for electron microscopy.

Protoplasts isolated as described were washed and resuspended in incubation buffer containing 0.5% agarose (Sigma Type IX, ultra-low gelling temperature, Sigma Chemical Co., St. Louis, MO, USA) to cushion protoplasts during freezing and hold them together during subsequent processing. Agarose increased the viscosity of the incubation medium, but did not interfere with coat-ing of protoplasts with CF since protoplasts with or without agarose in the incubation medium became rapidly coloured as the orange-brown CF adsorbed to their surfaces. Electron microscopy of freeze-fixed protoplasts incubated in CF with agarose revealed that the plasma membranes were extensively coated with CF, similar to protoplasts not exposed to agarose. The protoplasts were incubated in CF as described above, except that agarose was added and the final ratio of total volume to protoplast pellet volume during CF incubation was decreased from 5.3:1 to 2:1, so that the final concentration of CF used ranged from 0.5 to 1 mg/ml. This increase in protoplast density ensured that protoplast samples were sufficiently concentrated for subsequent sectioning and electron microscopy, since pelleting the CF-labelled protoplasts before freezing was not possible for short incubations in CF. Droplets of the protoplast suspensions were frozen in liquid nitrogen under high pressure in a prototype Balzers HPM 010 high-pressure freezing apparatus using the specimen cups and holder designed by Craig et al. (1987). Just before use, the upper part of each specimen cup was coated in a freshly prepared solution of vegetable lecithin in chloroform (100 mg/ml) to assist in separating the cups after freezing. Freezing took place within 30 seconds of loading the samples into the cups.

After freezing, the cups were transferred immediately to liquid nitrogen, the two halves were separated, and the specimens were transferred to vials containing a liquid nitrogen-cooled solution of 2% osmium tetroxide in acetone. Substitution was carried out in this solution at approximately −79°C in a dry ice/acetone bath for a minimum of 3 days. The samples were gradually warmed to room temperature over a 6 hour period, rinsed in acetone, and infiltrated with Spurr’s resin (which infiltrates freeze-substituted tissue better than Araldite) over 3 days before polymerization.

Ultra thin sections were cut on a diamond knife. Sections were mounted on uncoated or formvar-coated copper mesh grids, and were observed either unstained or briefly stained in lead citrate (for visualizing CF) or stained fully in uranyl acetate and lead cit-rate, and viewed with a 420 model Philips electron microscope.

The WS1 line of suspension-cultured, embryogenic white spruce consists of small clumps of undifferentiated cells and somatic embryos at various stages of development. These embryos, like zygotic embryos, are a mass of small densely cytoplasmic meristematic cells from which extend a ‘tail’ of larger, elongated and vacuolate suspensor cells (Fig. 1; see also Attree et al., 1991). Suspensor cells seldom released intact protoplasts. Nearly all protoplasts were small and cytoplasmic, like the meristematic cells of the embryos. Since the cells in the meristematic region are interconnected by numerous plasmodesmata (Hakman et al., 1987), the best yields were obtained by vigorous shaking during the enzyme treatment in order to separate the protoplasts, which otherwise remained connected by their plasmodesmata and/or fused to yield multinucleate protoplasts. Fluorescein diacetate staining confirmed that 80 to 90% of newly iso-lated protoplasts were viable (not shown).

Thin sections of single well-preserved, uninucleate proto-plasts were used to establish the time course of CF uptake, and to study protoplast ultrastructure (Figs 2 and 3). Much of the volume of each protoplast was occupied by the nucleus, which was surrounded by cytoplasm, organelles and large vacuolar profiles. Membranous material was usually attached to the outer surface of the plasma mem-branes in glutaraldehyde-fixed protoplasts (Fig. 3), and may have been extruded from the protoplast during fixation and/or released from damaged protoplasts during protoplast preparation.

In contrast, freeze-fixed protoplasts had little or no extraneous membrane at their surfaces, although the plasma membranes were quite ruffled (Fig. 2). The nuclear chromatin in some freeze-fixed protoplasts had a holey appear-ance (Fig. 2), indicating ice damage within the nucleus. However, the cytoplasm of these protoplasts was not excluded from observation if it was free of detectable ice damage.

More glutaraldehyde-fixed than freeze-fixed protoplasts were examined in this study because the small size of the specimen cups limited the number of protoplasts that could be frozen, and because only a subpopulation of protoplasts were well-preserved by high-pressure freezing. A survey of protoplasts fixed in five different freezing runs showed that there were between 2 and 6% uninucleate, well-preserved protoplasts (for 128 to 413 protoplasts sampled per run).

The ultrastructure of CF-labelled and unlabelled WS1 protoplasts was similar. However, significant differences were noted in the ultrastructure of freeze-fixed compared to glu-taraldehyde-fixed protoplasts. In general, the membranes of all organelles and vacuoles were smoother and the organelles appeared turgid in freeze-fixed protoplasts as compared to glutaraldehyde-fixed protoplasts (Figs 2-4). Nuclear profiles were full and round after freeze fixation, but convoluted after glutaraldehyde fixation (compare Figs 2 and 3). Membrane contrast was generally poorer after freeze fixation regardless of staining method employed.

Both freeze-fixed and glutaraldehyde-fixed WS1 protoplasts contained smooth invaginations of the plasma mem-brane, which were 4 times larger than the invaginations of coated pits. In the cytoplasm, smooth vesicles similar in size and appearance to these large smooth invaginations were only observed near the plasma membrane (Figs 2 and 3; see also below). These large smooth vesicles should not be confused with the small smooth or uncoated vesicles, similar in size to coated vesicles, which were distributed throughout the cytoplasm.

All glutaraldehyde-fixed protoplasts contained amor-phous dark deposits scattered over protoplast membranes, particularly plasma membrane, the bounding membranes of plastids, mitochondria, multivesicular bodies (MVBs) as well as Golgi bodies (e.g. Figs 3, 8, 10, 12, 24, 28) where they hindered observation of CF particles within the cis-ternae. These osmiophilic deposits were visible in both unstained and stained sections, and in unstained sections a linear/punctate substructure was detected, suggesting that the material consisted of closely packed rod-shaped ele-ments (Fig. 5). These deposits were absent from high-pres-sure frozen material and from intact WS1 embryos (not shown).

Golgi body ultrastructure differed somewhat between embryonic cells of the somatic embryos and the protoplasts derived from them (Figs 6-11). In the Golgi bodies of embryonic cells (studied by Hakman et al., 1987) coated or smooth vesicles were associated with the trans-Golgi faces (Fig. 6); reticulate networks of tubular membranes bearing clathrin-coated regions, the partially coated reticula (PCR), were often near by, although not detectably connected to the Golgi bodies (Fig. 6, see also Fig. 13,Hakman et al., 1987). In the protoplasts, similar Golgi bodies with adja-cent PCR were observed (Fig. 7), but in most Golgi bodies the structure of the trans faces was more complex (Figs 8-11). The ends of the transmost Golgi cisternae were often curled inward and associated with clusters of coated or smooth vesicles (Figs 8, 9, 11). Profiles of closed or almost closed membrane rings were also found adjacent to trans-Golgi cisternae (Figs 9, 10, 11). Some of these rings also enclosed coated and smooth vesicles (Fig. 10). Like embryonic cells, PCR was often located near Golgi bodies in the protoplasts, typically covering an area about two to three times that occupied by the Golgi bodies (Figs 4, 7, 12, 23). Microtubules lay just beneath the plasma membrane in both freeze-fixed and glutaraldehyde-fixed protoplasts (Figs 13 and 14). All microfilament bundles observed in glutaraldehyde-fixed protoplasts were straight and compact (Fig. 15), but one bundle consisting of wavy microfilaments was found in an otherwise well-preserved protoplast after freeze-fixation (Fig. 16).

Protoplasts were labelled continuously in CF at room temperature and not by pulse-labelling chilled protoplasts, then rewarming them as described by Joachim and Robinson (1984) and Hillmer et al. (1986) because fewer coated pits were found on the plasma membranes of rewarmed proto-plasts than on the membranes of comparable unchilled protoplasts, and CF formed discontinuous clumps at the plasma membranes of rewarmed protoplasts (unpublished results). These factors reduced CF uptake in rewarmed protoplasts compared to protoplasts held at room temperature, which in turn reduced CF labelling of organelles and made it more difficult to determine the timing of CF transit into different organelles.

No differences were detected in the time course and path-way of CF endocytosis in glutaraldehyde-fixed compared to freeze-fixed protoplasts, so examples have been selected from protoplasts fixed by both methods (Figs 17-29). Ultra-structural observations of CF uptake were made either from unstained thin sections of protoplasts or after 5 minutes staining in lead citrate only, since combined uranyl acetate/lead citrate staining obscured the less electron-opaque CF. However, this meant that the clathrin coats of coated pits, vesicles and membranes were often poorly stained, although still identifiable by a ribosome-free zone of similar thickness to clathrin coats (Figs 17-22).

Within 10 seconds of labelling, and thereafter, protoplasts were coated with a layer of CF, which was also present in coated pits (Figs 17 and 18), large smooth invaginations of the plasma membrane (Fig. 19), coated vesicles (Figs 20-22) and large smooth vesicles lying close to the plasma membrane (Fig. 22). The large smooth invaginations were connected by openings of various sizes to the surface of the protoplasts, and serial sectioning showed that some of the large smooth vesicles were also connected by narrow channels to the surface (not shown). Coated pits appeared on both the large smooth invaginations and vesicles just as on the plasma membrane proper, and coated vesicles were observed in their vicinity (Figs 14 and 19). When the mem-branes of these invaginations/vesicles were CF-labelled, the associated coated pits and vesicles were often labelled also (Fig. 19). CF in uncoated vesicles (Figs 17 and 19) first appeared in samples fixed 2 minutes after exposure to CF. Some CF was observed in Golgi bodies, trans-Golgi membranes (Fig. 8), and PCR after only 5 minutes of labelling in both freeze-fixed and glutaraldehyde-fixed samples, whereas in MVBs the first CF was not detected until 10 minutes after labelling. The number of CF-labelled coated and smooth vesicles, Golgi bodies, PCR (Fig. 23) and MVBs in the protoplasts increased rapidly within the first 30 minutes of labelling. CF within Golgi bodies was located principally at the periphery of medial and trans cisternae (Fig. 24). CF was also observed in clathrin-coated and smooth vesicles in the vicinity of Golgi cisternae, and in the trans-Golgi membranes and associated vesicles (Fig. 8). However, more CF appeared in PCR and MVBs than in Golgi bodies in protoplasts incubated in CF for 30 min-utes to 2 hours. In PCR, CF occurred throughout the tubular membranes and their coated buds (Figs 12, 23, 25).

Although some CF particles were observed at the internal surface of the bounding membranes of MVBs, most CF inside these organelles was located at the surfaces of the clustered small internal vesicles (Figs 25-27). Vesicles, sometimes containing CF, were occasionally seen fused to the outer surface of CF-labelled MVBs (Fig. 25). After incubation in CF for 1-2 hours, some MVBs contained a large number of CF particles (Fig. 26). In a few cases MVBs lay immediately adjacent to vacuoles (Figs 25, 27). Clusters of vesicles similar in size to those present in MVBs were also noted within some smaller vacuolar profiles (Fig. 28), and in one instance these vesicles were labelled with CF (Fig. 29).

To the best of our knowledge, this is the first ultrastructural study of rapidly frozen and freeze-substituted protoplasts. Freeze fixation provided excellent preservation of a small number of protoplasts. It should be possible to increase the proportion of protoplasts preserved free from detectable ice damage by modifying the freezing or freeze substitution method used here. The lack of direct contact between the 20 to 50 μm diameter protoplasts and the specimen cups, which act as heat sinks for larger specimens such as root tips (Moor, 1987) may have contributed to ice crystal formation in protoplasts during the initial freezing by slowing heat loss from the cells. The 0.44 M (8% w/v) sorbitol in the CF incubation buffer should have provided some cry-oprotection for the protoplasts during the initial freezing, by reducing extracellular ice crystal formation and there-fore reducing the release of heat, which accompanies this crystallization (Gilkey and Staehelin, 1986; Moor, 1987). However, cryoprotection could be increased further by adding inert non-penetrating cryoprotectants such as dextran or 1-hexadecene (Kiss et al., 1990) to the protoplasts during incubation. Large ice crystals can also form by secondary growth during specimen rewarming if ice removal during substitution is incomplete. Although the substitution and rewarming method used here has been successfully used to preserve seedling root tips after high-pressure freezing (Kiss et al., 1990), and the 3-day substitution period at −79°C was considered adequate due to the small sample volumes imposed by specimen cup sizes, it may be neces-sary to increase this substitution time to ensure complete removal of ice from protoplasts and the surrounding medium.

Ruffled plasma membranes have not previously been reported in well-preserved cells after high-pressure freez-ing, although rippled plasma membranes were reported in ice-damaged walled cells of seedling root tips (Craig and Staehelin, 1988). It is possible that the ruffling is an arte-fact of exposure to high pressure. Although samples are pressurized for only about 0.5 seconds, a number of such artefacts have been reported, including the bursting or breakage of large vesicles in root cap cells of Arabidopsis (Kiss et al., 1990) and large peripheral vesicles in fungal sporangia (Hyde et al., 1991a).

The appearance of loose wavy bundles of microfilaments in cells after high-pressure freezing/freeze substitution may be another artefact of pressurization, since Ding et al. (1992) noted such microfilament bundles in Nicotiana leaf tissue after high pressure freezing, but found only compact bundles of straight, regularly spaced microfilaments in well-preserved cells after chemical or freeze fixation at normal atmospheric pressure.

The ultrastructure of protoplasts freshly prepared from WS1 embryos has not previously been examined in detail. WS1 protoplasts evidently differ from the meristematic cells of the embryos since protoplasts but not embryo cells have large smooth invaginations of the plasma membranes and similar smooth vesicles, amorphous dark deposits associated with various intracellular membranes, and different arrangements of membranes at the trans face of Golgi bodies.

The large smooth plasma membrane invaginations and vesicles present in protoplasts may reflect disturbances of the plasma membranes during protoplast isolation, when, for example, the numerous plasmodesmata between the meristematic cells of the embryos become severed. It is not clear whether the large smooth vesicles near the plasma membranes of protoplasts are formed by internalization of large smooth plasma membrane invaginations. However, since these vesicles remained in the peripheral cytoplasm and did not fuse with other organelles or vacuoles, there is no evidence that they delivered CF directly to the endocytotic pathway. Indeed, the occurrence of CF-labelled coated pits on such smooth invaginations and vesicles and of CF-labelled coated vesicles nearby suggest that these structures continued to behave as plasma membrane.

The membrane-associated amorphous deposits detected in glutaraldehyde-fixed protoplasts are absent from intact somatic embryos and older white spruce protoplasts in cul-ture (Hakman et al., 1987, and unpublished results). They may have been induced by stress, caused for example by exposure to cell wall-degrading enzymes, or water loss upon plasmolysis, since very similar ‘membraglobuli’ form in Arabidopsis leaves exposed to low temperatures (Ristic and Ashworth, 1993). The absence of deposits in freeze-substituted protoplasts suggests that they were retained by glutaraldehyde and osmium tetroxide fixation but not by exposure to osmium tetroxide and acetone during freeze substitution. These deposits could be tannins, since tannin synthesis is induced in suspension-cultured white spruce by mild environmental changes (Durzan et al., 1973) and tannin synthesis is associated with the formation of amorphous dark deposits in the white spruce cells (Chafe and Durzan, 1973). Soluble phenols, including some tannins, are known to diffuse throughout the cytoplasm during fixation (McClure, 1979), perhaps accounting for the odd association of the amorphous deposits with cellular membranes. The elaboration of membranes at the trans face of Golgi bodies in WS1 protoplasts marks another difference between protoplasts and their source embryonic cells. The curled trans cisternae and associated vesicles observed in the protoplasts resemble the trans-Golgi membranes identified as trans-Golgi network (TGN) in root cap cells (Stae-helin et al., 1990 Fig. 6; Hillmer et al., 1988,Fig. 1b). We distinguished between PCR and TGN on the basis that both WS1 protoplasts and embryonic cells contained partially coated reticula, but only in protoplasts were TGN-like curled membranes and membrane rings associated with trans faces of Golgi bodies. The terms TGN and PCR have at times been used interchangeably to describe all partially coated membranes in plant cells (e.g. see Hillmer et al., 1988; Staehelin et al., 1990). Some of this uncertainty over the identity of PCR and TGN has been due to our incom-plete knowledge of the biogenesis and functions of these membranes (Griffing, 1991), and to variations in the promi-nence of TGN and PCR in different cell types (for examples see Pesacreta and Lucas, 1985). Fixation can also result in differential preservation of TGN and PCR. For example, Samuels and Bisalputra (1990) did not find any PCR in glu-taraldehyde-fixed root cells of Lobelia, but they found that rapid freezing and freeze substitution preserved this organelle and revealed that it accumulated a fluid-phase tracer of endocytosis. Griffing (1991) has summarized the physical and cytochemical evidence suggesting that PCR and TGN are indeed distinct membrane-bound compart-ments with different functions in plant cells. At present, PCR is believed to be an intermediate organelle of the lytic pathway, capable of fusing with endocytotic vesicles from the plasma membrane and perhaps (by analogy with simi-lar structures in animal cells) sorting the contents of these vesicles for recycling to the cell surface or delivery to Golgi bodies or MVBs (Tanchak et al., 1988; Fowke et al., 1991). The TGN membranes are thought to be sites for sorting vacuolar proteins from proteins and polysaccharides des-tined for secretion (Staehelin et al., 1990; Moore et al., 1991). The development of TGN-like membranes at the Golgi bodies of WS1 protoplasts therefore suggests there is an increase in the secretory activity of the protoplasts compared to embryonic cells, perhaps related to the initiation of cell wall regeneration in the protoplasts.

We were unable to detect any differences between the time course and pathway of internalized CF in WS1 protoplasts and that detailed for protoplasts from suspension-cultured soybean (Tanchak et al., 1984; Tanchak and Fowke, 1987; Tanchak et al., 1988). The presence of exogenously sup-plied CF in coated vesicles and in coated pits at all stages of invagination shows that CF enters the protoplasts via this pathway as in soybean protoplasts (Tanchak et al., 1984), bean leaf protoplasts (Joachim and Robinson, 1984) and wound-induced protoplasts of the green alga Boergesenia (O’Neil and La Claire, 1988). All coated pits are assumed to be invaginating, not fusing with the plasma membrane, because of evidence that the clathrin coat must disassem-ble before membrane fusion can occur (Altstiel and Bran-ton, 1983). Serial sectioning of similar coated vesicles in soybean protoplasts has revealed that coated vesicles located in the cytoplasm beyond the depth of coated pit invaginations on the plasma membrane are indeed free vesi-cles (Fowke et al., 1989).

The intracellular distribution of endocytosed CF changed and increased during labelling, allowing different stages in the uptake of this marker to be identified. Thus CF first appeared rapidly in coated and smooth vesicles near the plasma membrane then deeper in the cytoplasm. As in soy-bean protoplasts, Golgi bodies and PCR became CF-labelled within the first 5 minutes of exposure to CF (Tan-chak et al., 1984). Likewise, lectin-gold conjugates were endocytosed and transported to the Golgi bodies of carrot protoplasts within 7 minutes of labelling (Hillmer et al., 1986). These results and others confirm that both fluid-phase and membrane-bound tracers of endocytosis are delivered after uptake to PCR in protoplasts and walled cells of higher plants (Hübner et al., 1985; Hillmer et al., 1986; Tanchak et al., 1988; Samuels and Bisalputra, 1990; Owen et al., 1991). However, access of tracers to Golgi bodies differs between fluid-phase and insoluble mem-brane-bound tracers, suggesting that while tracers are trans-ported directly in plasma membrane-derived coated vesicles to PCR, transport to Golgi bodies is secondary or indirect (Fowke et al., 1991). Our results agree with previous studies that insoluble membrane-bound tracers pass into the peripheries of both cis or trans Golgi cisternae after endocytosis (Hillmer et al., 1986; Tanchak et al., 1988). The distribution of endocytosed lead nitrate in Golgi cis-ternae of seedling root cells was similar (Hübner et al., 1985), but lanthanum nitrate did not enter Golgi bodies at all in epidermal cells of Lobelia roots (Samuels and Bisalputra, 1990), while lucifer yellow was distributed throughout Golgi cisternae after endocytosis in absorptive trichomes of Brocchinia (Owen et al., 1991). The efficiency of transport of endocytosed tracers into Golgi bodies (from PCR?) probably varies according to tracer characteristics such as charge and solubility (Samuels and Bisalputra, 1990; Owen et al., 1991).

In WS1 protoplasts, the first lightly labelled MVB was observed after 10 minutes; MVB labelling clearly lagged behind that of the Golgi and PCR, a lag that was also noted in CF-labelled soybean protoplasts (Tanchak et al., 1984). This lag suggests that endocytosed CF is transported either from PCR or Golgi bodies to MVBs (Fowke et al., 1991). The fusion of CF-labelled smooth vesicles with MVBs as observed in WS1 protoplasts is consistent with a vesicle-mediated delivery of CF to the MVBs. Such fusion events would deliver CF to the interior surface of the bounding membrane; subsequent internal invagination of this mem-brane could account for the occurrence of internal vesicles with surface-bound CF, as proposed by Tanchak and Fowke (1987). MVBs with labelled internal vesicles were also observed in protoplasts exposed to gold-labelled lectins (Hillmer et al., 1986), and in lanthanum-labelled root epi-dermal cells (Samuels and Bisalputra, 1990).

The detection of MVB-type vesicle clusters, in one case CF-labelled, in vacuoles of the WS1 protoplasts suggests that CF in MVBs may ultimately be delivered to the vac-uoles for degradation. Similar evidence for MVB fusion with vacuoles was obtained in soybean protoplasts (Tan-chak and Fowke, 1987; Record and Griffing, 1988). The co-localization of CF and acid phosphatases (enzymes associated with the lysosomal or degradative pathway in cells) in some MVBs and the vacuoles of soybean protoplasts pro-vides further evidence that internalized CF is ultimately degraded (Record and Griffing, 1988).

In conclusion, our results confirm that there is a common time course and pathway for the uptake of CF by endocy-tosis in protoplasts derived from taxonomically distant species of vascular plants. This endocytotic pathway is probably similar to that in intact higher plant cells, judging by the available data on endocytosis of fluid-phase tracers in seedling roots and absorptive trichomes. However, phys-iologically-based differences in the fate of endocytosed tracers are evident in studies of CF taken up via contrac-tile vacuoles in fungal zoospores and flagellated green algae (Cerenius et al., 1988; Domozych and Nimmons, 1992) and in CF taken up via coated pits of wound-induced proto-plasts of the coenocytic green alga Boergesenia (O’Neil and LaClaire, 1988).

In contrast to Samuels and Bisalputra (1990) and Hyde et al. (1991b) we have found no significant differences in the ultrastructure of endomembranes after freeze fixation as compared to glutaraldehyde fixation, although high-pres-sure freezing and freeze substitution greatly improved the preservation of cell membranes, organelle structure and probably cytoplasmic organization. These results confirm the view of CF transport and distribution within the endomembrane system previously established from study of aldehyde-fixed protoplasts. Improvements to the freeze-fixation methods used here should enable larger samples of well-preserved protoplasts to be examined in future, which will assist in cytological and developmental studies of these fragile cells.

We thank Dr Andrew Staehelin for use of his Balzers high-pressure freezing apparatus, Dr Tom Giddings for his assistance and much helpful advice in the operation of the high-pressure freezer and in freeze substitution of the samples and Dr Stephen Attree for advice on culturing and preparing protoplasts from the WS1 culture. This research was supported by the Natural Sciences and Engineering Research Council of Canada, including an NSERC postdoctoral award to M.E.G.