- © 1999 American Society of Plant Physiologists
Abstract
In the parenchyma cells of developing legume cotyledons, storage proteins are deposited in a special type of vacuole, known as the protein storage vacuole (PSV). Storage proteins are synthesized at the endoplasmic reticulum and pass through the Golgi apparatus. In contrast to lysosomal acid hydrolases, storage proteins exit the Golgi apparatus in 130-nm-diameter electron-dense vesicles rather than in clathrin-coated vesicles. By combining isopycnic and rate zonal sucrose density gradient centrifugation with phase partitioning, we obtained a highly enriched dense vesicle fraction. This fraction contained prolegumin, which is the precursor of one of the major storage proteins. In dense vesicles, prolegumin occurred in a more aggregated form than it did in the endoplasmic reticulum. The putative vacuolar sorting receptor BP-80 was highly enriched in purified clathrin-coated vesicles, which, in turn, did not contain prolegumin. The amount of BP-80 was markedly reduced in the dense vesicle fraction. This result was confirmed by quantitative immunogold labeling of cryosections of pea cotyledons: whereas antibodies raised against BP-80 significantly labeled the Golgi stacks, labeling of the dense vesicles could not be detected. In contrast, 90% of the dense vesicles were labeled with antibodies raised against α-TIP (for tonoplast intrinsic protein), which is the aquaporin specific for the membrane of the PSV. These results lead to the conclusions that storage proteins and α-TIP are delivered via the same vesicular pathway into the PSVs and that the dense vesicles that carry these proteins in turn do not contain BP-80.
INTRODUCTION
Plant vacuoles perform various functions, of which the most important are the hydrolysis and storage of molecules and the maintenance of cell turgor (Wink, 1993). As we now know, not only is vacuolar function dependent on cell type and developmental state, but functionally different vacuoles can exist in the same plant cell. For example, Paris et al. (1996) have shown that root tip cells contain two types of vacuoles that can be distinguished on the basis of their lumenal as well as their membrane proteins. Moreover, by using a fusion protein comprising the C-terminal vacuolar targeting motif of chitinase A and the green fluorescent protein, Di Sansebastiano et al. (1998) have recently demonstrated that tobacco leaf cells possess distinct sets of vacuoles. The same is true for the parenchyma cells of maturing pea cotyledons (Hoh et al., 1995). During development, these cells are characterized by the presence of an additional type of vacuole, the protein storage vacuole (PSV), the function of which is to accumulate newly synthesized storage proteins that later supply the growing seedling with carbon and nitrogen as it germinates (Müntz, 1998). This organelle arises de novo and can easily be distinguished from lytic vacuoles in the same cells by the presence of osmiophilic storage proteins in its lumen and by the presence of the unique aquaporin α-TIP (for tonoplast intrinsic protein) in its membrane (Hoh et al., 1995).
Pea cotyledons contain two major types of storage proteins, vicilin and legumin. They are synthesized as prepropolypeptides at the rough endoplasmic reticulum (ER) and, after cleavage of the presequence and oligomerization into transport-competent trimers, are exported to the Golgi apparatus (Chrispeels et al., 1982a). At the trans Golgi network (TGN), they are then segregated from secretory proteins and polysaccharides and transferred to the PSVs (Chrispeels et al., 1982b; Chrispeels, 1983), where they are proteolytically processed into mature proteins (reviewed in Müntz, 1996). In addition, they must also be separated from proteins destined for the lytic vacuole. To cope with these different vacuolar compartments, these cells must possess an additional vacuolar sorting machinery.
Soluble acid hydrolases of the mammalian lysosome possess phosphorylated mannose residues as positive sorting information (reviewed in Le Borgne and Hoflack, 1998). These motifs are recognized by a specific receptor—the mannose-6-phosphate receptor—located at the TGN. Binding of the precursor proteins results in receptor and ligand being packaged into clathrin-coated vesicles, which are then delivered to endosomes. After releasing the ligand into the endosome, the receptor recycles back to the TGN, and the endosome finally fuses with the lysosome (Futter et al., 1996). A similar mechanism has also been described for the transport of carboxypeptidase Y (CPY) into the vacuole of yeast cells (reviewed in Horazdovsky et al., 1995), in which a four–amino acid sequence, QRPL, at the N terminus of CPY, is sufficient to ensure its correct sorting to the vacuole (Valls et al., 1990). This tetrapeptide is recognized by a 100-kD type I membrane protein called Vps10p at the TGN (Marcusson et al., 1994). As in mammalian cells, CPY reaches the vacuole via an endosomal compartment, and the receptor recycles back into the TGN (Stack et al., 1995; Cooper and Stevens, 1996).
Several different vacuolar targeting signals have been identified in the primary sequence of soluble vacuolar proteins (Neuhaus and Rogers, 1998). A transmembrane receptor protein with a molecular mass of 80 kD, called BP-80, which recognizes one of these motifs, has also been purified and cloned from a clathrin-coated vesicle fraction isolated from pea cotyledons (Kirsch et al., 1994, 1996). Homologous proteins are also present in Arabidopsis (Ahmed et al., 1997), pumpkin (Shimada et al., 1997), rice, and maize (Paris et al., 1997). Because of the presence of a distinct set of glycoproteins in a highly purified clathrin-coated vesicle fraction, which is absent from the PSV, Hohl et al. (1996) have discussed the possibility that clathrin-coated vesicles may carry soluble acid hydrolases into the second (lytic) vacuolar compartment of these cells.
In yeast cells, at least two independent vacuolar transport pathways exist (reviewed in Conibear and Stevens, 1998). One is responsible for the transport of the soluble hydrolase CPY and the membrane protein carboxypeptidase yscC from the Golgi apparatus to the endosome via clathrin-coated vesicles containing adapter complex 1 (AP-1) or 2 (AP-2), and a second transports alkaline phosphatase from the Golgi apparatus to the vacuole. The second vacuolar transport pathway uses unidentified transport vesicles containing adapter complex 3 (AP-3). These AP-3–containing transport vesicles seem to bypass the endosome because mutations that delete Vps4p, a protein required for transport from the late endosome to the vacuole, do not disturb the correct import of alkaline phosphatase into the vacuole.
In plants, precursor polypeptides of vacuolar storage proteins are not carried by clathrin-coated vesicles but rather by an additional type of Golgi apparatus–derived vesicles, the so-called dense vesicles (Hohl et al., 1996). Dense vesicles have an average diameter of 130 nm, are characterized by an electron-dense core, and are morphologically very homogeneous. After release from the Golgi stack, they have no visible coat proteins on their cytosolic surface. These observations raise the question of whether BP-80 is responsible for sorting proteins into both types of transport vesicles or whether the transport of storage proteins in dense vesicles is independent of BP-80 and employs another sorting mechanism.
Pure fractions of transport vesicles are needed for a better understanding of the complex vesicular sorting processes in the cells of plant storage tissues. Techniques for the isolation of PSVs and of clathrin-coated vesicles from pea cotyledons have already been established (Hohl et al., 1996; Hinz et al., 1997). By developing a method for the isolation of dense vesicles from this tissue, we are now able to demonstrate that α-TIP is transported together with the storage proteins in dense vesicles into the PSV and that BP-80 does not seem to be present in dense vesicles.
RESULTS
Isolation and Characterization of Dense Vesicles
A method that combines differential centrifugation with isopycnic and rate zonal gradient centrifugation steps has been developed for the isolation of dense vesicles from developing pea cotyledons. Because proteolytic processing of prolegumin into the mature α and β chains occurs in the PSV (Müntz, 1996), the relative distribution of prolegumin and mature legumin together with the distribution of ER and Golgi apparatus marker proteins was used to monitor the purification of the dense vesicles as a post-Golgi prevacuolar compartment. Because several members of the legumin multigene family are expressed in pea (Casey, 1979), a number of prolegumin precursor polypeptides as well as mature α and β chains can be detected in protein blots by using polyclonal antisera raised against legumin (Chrispeels et al., 1982a; Hinz et al., 1997).
Figure 1 shows a flow chart of the purification scheme. After a short, low-speed centrifugation of the homogenate (Figure 1, step 2), prolegumin was enriched in the supernatant, whereas most of the mature legumin, which is the marker for the PSV, was sedimented, as shown in Figure 2 (lanes 1 and 2). After further removal of mitochondria and plastids (Figure 1, step 3), the supernatant was fractionated on a sucrose step gradient for 2.5 hr at 80,000g under low Mg2+ conditions (Figure 1, step 4), thereby achieving isopycnic conditions for the dense vesicles (data not shown). Prolegumin was found to be enriched in the 41/55% sucrose interphase (Figure 2, lane 3). After rate zonal recentrifugation of this fraction on a linear sucrose gradient for 20 min at 25,000g, two pools were collected: a low-density fraction from 22 to 26% sucrose and a high-density fraction from 29 to 33% sucrose (Figure 1, step 5). Whereas both fractions contained prolegumin, mature legumin was enriched in the high-density fraction (Figure 2, lanes 4 and 5).
Both rate zonal fractions were fixed and embedded for electron microscopy. As shown in Figure 3, dense vesicles were enriched in the low-density pool (Figure 3A), whereas the high-density fraction, containing significantly fewer dense vesicles, was enriched in larger organelles also containing electron-dense protein aggregates (Figure 3B). This fraction was further characterized by the presence of multivesicular body–like structures and mitochondria. The low-density fraction was estimated to have a purity of ∼60% after particle counting on micrographs of thin sections. Approximately 0.5 to 0.7 mg of protein was isolated from 70 g of pea cotyledons.
Flow Chart Diagram of Dense Vesicle Purification.
This flow chart diagram provides a brief overview of the purification of the dense vesicles (DV), as given in Methods. The dense vesicle–enriched fraction is marked by x’s.
Because of its high density (Robinson et al., 1994), the plasma membrane was enriched in the 41/55% sucrose interphase of the isopycnic step gradient. Both rate zonal fractions were contaminated with plasma membrane as well, as shown by the distribution of the plasma membrane ATPase in Figure 2. Contamination of the dense vesicle fraction by the ER was low, as shown in Figure 2 by the distribution of the binding protein BiP, which is a marker for the ER (Denecke et al., 1991). Relative to the dense vesicle fraction, BiP was significantly enriched in the 20/35% and 35/41% sucrose interphases of the isopycnic step gradient, which, in turn, contained less prolegumin than did the dense vesicle fraction (Figure 2, lanes 4, 6, and 7).
To determine the distribution of the Golgi apparatus, we used two marker proteins. The reversibly glycosylated protein (RGP; Dhugga et al., 1997) was specifically enriched in the supernatant/20% and 20/35% sucrose interphase of the isopycnic step gradient (Figure 2, lanes 6 and 8) and was no longer detectable in the dense vesicle fraction (Figure 2, lane 4). Because the RGP is not an integral membrane protein, we considered the high amount of RGP in the supernatant to be due to the release of this protein during membrane isolation. To confirm the distribution of the Golgi apparatus, we determined a second marker (latent inosine diphosphatase [IDPase]; Robinson et al., 1994). As shown in Table 1, the highest specific activity (6.4 nanokatals [nkat] mg-1 protein) was found in the 20/35% sucrose interphase. No activity was detected in the supernatant/20% sucrose interphase. Therefore, Golgi apparatus and ER membranes cosediment in the 20/35% sucrose fraction under the low Mg2+ conditions used for dense vesicle purification. However, 1.5 nkat mg-1 protein of IDPase activity was still detectable in the 41/55% sucrose interphase, corresponding to 25% of the specific activity of the 20/35% fraction. In both rate zonal fractions, the activity of the latent IDPase was below the limit of detection.
Distribution of Marker Proteins during Dense Vesicle Isolation.
Legumin (Leg), mature legumin (mL), and prolegumin (pL) are marker polypeptides for the PSV and for prevacuolar compartments, respectively; BiP is a marker for the ER; RGP is a marker for the Golgi apparatus; β-fructosidase (βF1) is an antibody raised against complex glycoproteins as a marker for late Golgi and post-Golgi compartments; plasma membrane H+-ATPase (PM-ATPase) is a marker for the plasma membrane. Lanes 1 to 5 show dense vesicle (DV) purification. Lane 1 contains the crude homogenate (step 1); lane 2, the 200g supernatant (step 2); lane 3, the 41/55% interphase of the isopycnic step gradient (step 4); lane 4, the low-density fraction of the rate zonal gradient (highly enriched dense vesicles; step 5); and lane 5, the high-density fraction of the rate zonal gradient. Lanes 6 to 8 show the distribution of the marker proteins in the isopycnic step gradient. Lane 6 contains the 20/35% interphase; lane 7, the 35/41% interphase; and lane 8, the supernatant/20% interphase. Each lane contains 15 μg of protein. Glycoproteins enriched in the 41/55% fraction and in the rate zonal fractions are marked with asterisks.
Electron Microscopy of the Rate Zonal Fractions.
(A) Pool 1 shows the low-density fraction of the rate zonal gradient.
(B) Pool 2 shows the high-density fraction of the rate zonal gradient.
Some of the dense vesicles are marked with arrowheads; putative multivesicular bodies are labeled with stars; mitochondria are marked with M; small protein bodies are marked with a P. Bars in (A) and (B) = 450 nm.
A latent IDPase activity has been reported to be present in the plasma membrane (M’Voula-Tsieri et al., 1981). However, as discussed by Widell and Larsson (1990), this activity is most likely not a plasma membrane IDPase but reflects the unspecific cleavage of IDP by other attached nucleotide triphosphatases or diphosphatases. To test whether the latent IDPase activity in the 41/55% fraction was due to contaminating plasma membrane, we subjected both the 20/35% and the 41/55% sucrose fractions to aqueous two-phase partitioning (Robinson et al., 1994). The plasma membrane ATPase was completely partitioned into the upper phase (Figure 4B). As shown in Table 2, latent IDPase activity was highly enriched in the upper phase of the 41/55% fraction, whereas no such activity could be measured in the upper phase of the 20/35% fraction, although similar amounts of protein were partitioned. These results clearly indicate that although there was still Golgi contamination, a part of the latent IDPase activity in the high-density fraction was indeed due to contaminating plasma membrane.
The fraction enriched in dense vesicles contained complex glycans, as determined by the distribution of proteins cross-reacting with the β-fructosidase antibody, which recognizes xylose residues of processed glycoproteins (Laurière et al., 1989). At least two such proteins with molecular masses of 70 and 28 kD, respectively, were highly enriched in the 41/55% fraction of the isopycnic step gradient (Figure 2, lane 3; the proteins are marked with asterisks). Both proteins could also be detected in both rate zonal fractions (Figure 2, lanes 4 and 5).
Transport of α-TIP
The results of Gomez and Chrispeels (1993) suggested that the aquaporin α-TIP, which is the major integral membrane protein of the PSV, and phytohemagglutinin, which is a lectin from bean seeds, may use different vesicular pathways as they move from the Golgi apparatus to the vacuole when they are coexpressed in leaves of transgenic tobacco plants. Previously, we were able to demonstrate that neither α-TIP nor the storage proteins legumin and vicilin are transported by clathrin-coated vesicles in pea cotyledons (Hohl et al., 1996). By contrast, dense vesicles were shown to contain storage proteins; however the question remained as to whether or not they might also transport α-TIP.
To determine whether α-TIP is transported in dense vesicles together with prolegumin, we monitored the distribution of α-TIP during isolation of dense vesicles. As shown in Figure 4A, most of the α-TIP sedimented together with the mature legumin after the initial centrifugation step (lanes 1 and 2). α-TIP was enriched in the ER/Golgi apparatus (20/35%) fraction of the isopycnic step gradient (Figure 4A, lane 6).
The 41/55% interphase still contained significant amounts of α-TIP (Figure 4A, lane 3). Nevertheless, the dense vesicle fraction still contained significant levels of α-TIP (Figure 4A, lanes 4 and 5).
Distribution of the Latent IDPase, a Marker for the Golgi Apparatus, in the Isopycnic Sucrose Step Gradient and the Rate Zonal Gradient
Distribution of α-TIP in Dense Vesicles, the High-Density Fraction, and the Plasma Membrane.
(A) Distribution of α-TIP during dense vesicle (DV) purification. Lanes 1 to 5 show dense vesicle purification. Lane 1 contains the crude homogenate (step 1); lane 2, the 200g supernatant (step 2); lane 3, the 41/55% interphase of the isopycnic step gradient (step 4); lane 4, the low-density fraction of the rate zonal gradient (highly enriched dense vesicles; step 5); and lane 5, the high-density fraction of the rate zonal gradient. Lanes 6 to 8 show the distribution of the marker proteins in the isopycnic step gradient. Lane 6 contains the 20/35% interphase; lane 7, the 35/41% interphase; and lane 8, the supernatant/20% interphase.
(B) Distribution of α-TIP and of the plasma membrane H+-ATPase (PM-ATPase) between the high-density fraction and the plasma membrane. Lane 1 contains the 41/55% interphase (step 4); lane 2, the high-density fraction of the rate zonal gradient (step 5); lane 3, the lower phase of the 41/55% interphase; lane 4, the lower phase of the high-density fraction of the rate zonal gradient; and lane 5, the upper phase of the high-density fraction of the rate zonal gradient.
(C) Distribution of α-TIP during dense vesicle preparation after stripping the membranes with 1 M KI. Lane 1 contains the stripped membranes of the 41/55% interphase of the isopycnic step gradient; lane 2, the membranes of the low-density fraction of the rate zonal gradient (dense vesicle fraction); and lane 3, the membranes of the high-density fraction of the rate zonal gradient.
In (A) to (C), each lane contains 15 μg of protein.
Distribution of Latent IDPase after Aqueous Two-Phase Partitioning of ER/Golgi and Dense Vesicle–Enriched Fractions of the Isopycnic Sucrose Step Gradient
Robinson et al. (1996) had previously demonstrated the presence of vacuolar membrane proteins (γ-TIP, V-type ATPase, and pyrophosphatase) in the plasma membrane of pea cotyledons. Therefore, the possibility remained that the presence of α-TIP in these fractions was due to plasma membrane contamination. To test this possibility, we subjected both the 41/55% sucrose fraction (Figure 1, step 4) and the high-density fraction of the rate zonal gradient to aqueous two-phase partitioning. α-TIP remained in the lower phase and was not detectable in the upper phase (Figure 4B, lanes 3 to 5), whereas the plasma membrane ATPase was completely partitioned into the upper phase (lanes 3 to 5).
Immunogold Staining of Cryosectioned Cotyledon Tissue with Antibodies Raised against α-TIP.
(A) Labeling of the membranes of the PSV (arrows), the Golgi stack, and dense vesicles (arrowheads) with the α-TIP antiserum. The background label is very low; however, the membranes of the PSV and the dense vesicles are significantly labeled.
(B) Labeling of a single dictyosome and three connected dense vesicles (arrowheads) with the α-TIP antiserum. The dense vesicles are significantly labeled.
Bars in (A) and (B) = 100 nm.
The presence of high amounts of soluble protein in these fractions may interfere with the relative distribution of the membrane protein α-TIP. Therefore, the soluble proteins of the 41/55% interphase of the step gradient and of both rate zonal fractions were removed by repeatedly washing the membranes with 1 M KI, a treatment that can dissociate mature storage proteins from PSV membranes (Hinz et al., 1997). More than 90% of the protein was removed from the membrane pellet after three wash cycles with KI (data not shown). As shown in Figure 4C, even after removing the soluble proteins α-TIP was still equally distributed between the membranes of the two rate zonal fractions, indicating the presence of α-TIP in different compartments in each fraction.
To confirm the results obtained by subcellular fractionation, we examined the in situ distribution of α-TIP by using immunoelectron microscopy. The amount of antigen in the membranes of the dense vesicles was too low to be detected by on-grid immunolabeling of resin sections, as used previously (Hohl et al., 1996). Therefore, we used the cryosectioning technique of Tokuyasu (1980) to increase the sensitivity of the immunolabeling. Figures 5A and 5B show that the membrane of the PSV was heavily labeled with the α-TIP antibody, whereas background and ER labeling were low. Golgi cisternae were labeled with approximately five to six gold particles per stack (Figure 5 and Table 3). As shown in Table 3, up to 90% of the dense vesicles were labeled with an average of two to three gold particles per dense vesicle, even though a lower concentration of primary antibody was employed, as by Hohl et al. (1996). Low random labeling was observed with a nonimmune rabbit serum (data not shown).
Distribution of BP-80 and α-TIP between Golgi Cisternae and Dense Vesicles: Quantitative Analysis of Cryoimmunoelectron Microscopy
Clathrin-Coated Vesicles Carry BP-80 but Dense Vesicles Do Not
BP-80 was the first putative vacuolar sorting receptor to be identified in plants (Kirsch et al., 1994). The distribution of BP-80 between dense vesicles and clathrin-coated vesicles has been investigated by comparing the relative distribution of this protein during the isolation of these organelles and by cryoimmunoelectron microscopy.
As shown in Figure 6A, BP-80 was enriched severalfold in the purest clathrin-coated vesicle fraction compared with a crude microsomal pellet (Figure 6A, lanes 2 and 7), but it was not detectable in the pure PSV fraction (Figure 6A, lane 8). Even after the storage protein was stripped off with KI, BP-80 was not detected in the PSV membrane (data not shown). In comparison, the clathrin-coated vesicle fraction did not contain detectable amounts of prolegumin (Figure 6A, lane 7).
During isolation of the dense vesicles, BP-80 was enriched in the 41/55% interphase of the isopycnic step gradient (Figure 6B, lanes 1 to 3). After the rate zonal gradient, BP-80 was enriched in the high-density fraction (Figure 6B, lanes 3 and 5), whereas the relative amount of BP-80 in the dense vesicle–containing fraction was strongly reduced (Figure 6B, lanes 3 and 4). A significant amount of BP-80 was also to be found in the ER- and Golgi apparatus–derived fractions of the isopycnic sucrose step gradient (Figure 6B, lane 6). As shown in Figure 6A, clathrin-coated vesicles were highly enriched in BP-80. Therefore, even a low degree of contamination with either clathrin-coated vesicles or clathrin-coated TGN membranes results in a significant BP-80 signal in the isolated dense vesicle fraction.
Because clathrin-coated vesicles could not be detected by electron microscopy in the dense vesicle fractions (Figure 3A), the distribution of β-adaptin during the isolation of dense vesicles was monitored for the presence of clathrin-coated TGN elements or clathrin-coated immature dense vesicles, which are also seen at the TGN (Robinson et al., 1997). β-Adaptin is part of the adapter complex in clathrin-coated vesicles connecting the lysosomal sorting receptor with the clathrin coat of the vesicles (Robinson, 1996; Le Borgne and Hoflack, 1998). A monoclonal antibody raised against the mammalian protein was used because it also recognizes the plant β-adaptin homolog in pea clathrin-coated vesicles (Holstein et al., 1994). As shown in Figure 6B, β-adaptin was highly enriched in the ER- and Golgi apparatus–derived fractions of the step gradient (lane 6). β-Adaptin was enriched in both rate zonal fractions relative to the 41/55% interphase of the isopycnic step gradient, thus indicating the presence of clathrin-coated and therefore BP-80–carrying membranes in the dense vesicle fraction (Figure 6B, lanes 3 to 5). Because it was demonstrated (Figure 4C) that soluble cargo proteins may interfere with the estimation of the relative distribution of membrane proteins between two organelle fractions, the distribution of BP-80 was also determined on KI-stripped membranes. However, in contrast to α-TIP, no change in the distribution pattern of BP-80 between the two fractions was found (Figure 6C).
These results were also subjected to confirmation by immunoelectron microscopy. Before fixation, the sensitivity of the antibody to antigen interaction toward fixatives was tested using a dot blot assay according to Riederer (1989). Concentrations of >2% of paraformaldehyde (as used by Paris et al. [1997] in root tissue) were shown to be deleterious in pea cotyledons. Even when a mild fixative was used, embedding in London resin prevented the detection of an immunogold signal. Cryosectioning was therefore used for these experiments as well. As shown in Figure 7A, background labeling with the antibody against BP-80 was low, and neither the PSV nor the ER membranes were labeled. In contrast, the Golgi cisternae were labeled, but without a preference between the cis or trans pole of the Golgi stack (Figure 7B). Dense vesicles were not labeled.
The distribution of BP-80 was analyzed quantitatively (Table 3). The Golgi cisternae were labeled with approximately four gold particles per stack, but there was no labeling of the dense vesicles. The gold label over the Golgi stacks with BP-80 antiserum was only 30% less than the labeling of Golgi stacks with α-TIP. However, under the same conditions, dense vesicles were significantly (approximately three gold particles per dense vesicle) labeled with α-TIP antibodies. Thus, if BP-80 had the same distribution as α-TIP, then dense vesicles carrying BP-80 should also have been detectable. In addition, when comparing the labeling density of dense vesicles with α-TIP and BP-80 antibodies, respectively, one should keep in mind that comparable dilutions of rabbit polyclonal antibodies were used in both cases and that sections were probed under the same conditions with identical gold-conjugated secondary antibodies (see Methods). Examination of sections with nonimmune rabbit serum showed only a very low background labeling (data not shown).
Distribution of the Vacuolar Sorting Receptor BP-80 between Clathrin-Coated Vesicles and Dense Vesicles.
Legumin (Leg), mature legumin (mL), and prolegumin (pL) are marker polypeptides for the PSV and prevacuolar compartments, respectively. Clathrin heavy chain (CHC) is a marker for clathrin-coated vesicles. β-Adaptin is an additional marker for clathrin-coated vesicle–mediated transport.
(A) Distribution of BP-80 during the purification of clathrin-coated vesicles. Lane 1 contains the homogenate; lane 2, the crude postmicrosomal pellet; lane 3, the supernatant from purification step 1; lane 4, the pellet from purification step 2; lane 5, the pellet from purification step 3; lane 6, the supernatant from purification step 4; lane 7, the supernatant from purification step 5 (highly enriched clathrin-coated vesicles); and lane 8, isolated protein bodies.
(B) Distribution of BP-80 and β-adaptin during dense vesicle (DV) purification. Lanes 1 to 5 show dense vesicle purification. Lane 1 contains the crude homogenate (step 1); lane 2, the 200g supernatant (step 2); lane 3, the 41/55% interphase of the isopycnic step gradient (step 4); lane 4, the low-density fraction of the rate zonal gradient (highly enriched dense vesicles; step 5); and lane 5, the high-density fraction of the rate zonal gradient. Lanes 6 to 8 show the distribution of the marker proteins in the isopycnic step gradient. Lane 6 contains the 20/35% interphase; lane 7, the 35/41% interphase; and lane 8, the supernatant/20% interphase.
(C) Distribution of BP-80 after stripping the membranes with 1 M KI. Lane 1 contains the stripped membranes of the 41/55% interphase of the isopycnic step gradient; lane 2, the membranes of the low-density fraction of the rate zonal gradient (dense vesicle fraction); and lane 3, the membranes of the high-density fraction of the rate zonal gradient.
In (A) to (C), each lane contains 15 μg of protein.
Oligomerization Status of Prolegumin in an Enriched Dense Vesicle Fraction
Dense vesicles and the PSV, unlike the lumen of the ER and of the Golgi cisternae, are characterized by their osmiophilic, electron-opaque content (Hohl et al., 1996), which is due to the aggregation of storage globulins (Craig et al., 1979). Therefore, the degree of oligomerization of prolegumin was compared between dense vesicles and ER fractions by using a rate zonal sucrose density centrifugation. The 20/35% interphase of the isopycnic sucrose step gradient (Figure 2, lane 6) was used as an ER/Golgi apparatus fraction. As shown in Figure 8 and Table 4, >50% of the precursor sedimented to a density of ⩽26% in the ER/Golgi apparatus fraction, but >50% of the precursor was found at densities >29% in the dense vesicle fraction. These data indicate that upon arrival in the dense vesicles, prolegumin condensed to higher aggregated oligomers.
DISCUSSION
Transport vesicles carrying vacuolar precursor proteins from developing castor bean and pumpkin cotyledons have previously been enriched and characterized by Hara-Nishimura and co-workers (1991, 1993, 1998). The membrane of these vesicles is characterized by the presence of integral proteins homologous to vacuolar sorting receptors (Shimada et al., 1997). However, these vesicles, with a diameter of ∼300 nm, are significantly larger than the dense vesicles from pea seeds, and they originate directly from the ER rather than from the Golgi apparatus of these cells (Hara-Nishimura et al., 1998). Thus, the method presented here allows access to TGN-derived transport vesicles carrying precursor polypeptides of vacuolar storage proteins. In combination with the application of cryosectioning methods, we were thus able to demonstrate, both in vitro and in situ, the presence of two functionally different vacuolar sorting vesicles at the TGN of developing pea cotyledons.
Immunogold Staining of Cryosectioned Cotyledon Tissue Using Antibodies Raised against BP-80.
(A) Distribution of BP-80 between the membranes of PSV (arrow), the Golgi stack, and dense vesicles (arrowhead). The background label is very low. The membrane of the PSV is not labeled. There is low labeling of the Golgi stack but no label on the dense vesicle.
(B) Distribution of BP-80 in a single dictyosome. Four gold particles are visible in the cisternae. By contrast, the two connected dense vesicles (arrowhead) are not labeled.
The physiological function of vacuoles can be correlated with the appearance of distinct aquaporins in their membranes, as has been demonstrated recently by immunocytochemical examination of the spatial and temporal distribution of two aquaporins in the tonoplast of vegetative storage vacuoles in soybean leaves (∂- and γ-TIP; Jauh et al., 1998). Sorting of these proteins to vacuoles is still not understood, especially in cells that contain two different types of vacuoles, such as those of pea cotyledons. Transport of α-TIP to the tonoplast of the PSV was originally considered to be mediated by the ER and the Golgi apparatus (Melroy and Herman, 1991), and the sorting information seemed to be located in the transmembrane domain of the protein (Höfte and Chrispeels, 1992). However, α-TIP and storage proteins apparently do not share the same vesicular transport pathway to the vacuole in transgenic tobacco leaves (Gomez and Chrispeels, 1993).
These results are consistent with findings obtained with yeast and mammalian cells. In these organisms, vacuolar membrane proteins do not appear to share the same vesicular pathway to the endosome as the soluble proteins. In mammalian cells, type I lysosomal integral membrane proteins and lysosomal acid phosphatase are first transported to the cell surface, where they are subsequently internalized by receptor-mediated endocytosis, before reaching the endosome (Chao et al., 1990; Sandoval and Brakke, 1994). Transport of these proteins is independent of phosphorylated mannose residues and the mannose-6-phosphate receptor. Instead, a tyrosine residue, which is located in a cytoplasmic domain containing a β tight turn configuration, is required for the correct targeting (Trowbridge et al., 1993). Similarly, transport of membrane proteins in yeast cells is not affected by mutations in genes involved in the clathrin-coated vesicle–mediated sorting of CPY (Marcusson et al., 1994).
Aggregation Shift of Prolegumin between the ER/Golgi Apparatus and Dense Vesicles.
The results of a sucrose density gradient separation of prolegumin oligomers released from lysed membrane fractions are shown. The protein gel blot shows the distribution of the two major prolegumin polypeptides with molecular masses of 75 and 60 kD, respectively. The 20/35% interphase of the isopycnic step gradient (Figure 2, lane 6) is the ER/Golgi apparatus fraction. The low-density pool of the rate zonal gradient (Figure 2, lane 4) is the dense vesicle (DV) fraction. Protein was precipitated in equal volumes.
Aggregation Shift of Prolegumin between ER/Golgi and Dense Vesiclesa
Two pathways for the transport of integral membrane proteins appear to exist in protoplasts of suspension-cultured tobacco cells, as Jiang and Rogers (1998) have recently demonstrated by using chimeric integral membrane reporter proteins. Whereas γ-TIP and BP-80 seemed to share the same vesicular pathway to the prevacuolar compartment of these cells, α-TIP did not colocalize in the same transport compartments. Instead, after leaving the ER, α-TIP apparently reached an undefined compartment, bypassing the Golgi apparatus. In contrast to these results, α-TIP does indeed pass through the Golgi apparatus of developing pea cotyledons. Furthermore, because >90% of the dense vesicles were significantly labeled with α-TIP antibodies in situ and because nearly all of the dense vesicles were also labeled with antibodies raised against the two storage proteins vicilin and legumin (Hohl et al., 1996), α-TIP seems to be cotransported with the storage protein precursor polypeptides along the same vesicular pathway into the PSV. A further divergence of α-TIP and storage proteins into two distinct classes of dense vesicles seems unlikely.
Although brefeldin A also interferes with dense vesicle production in the pea cotyledon (Robinson et al., 1997), a result that is in agreement with the finding that phytohemagglutinin transport to the vacuole in transgenic tobacco cells is blocked by this inhibitor (Gomez and Chrispeels, 1993), the fact that the latter authors did not observe inhibition of α-TIP transport with this inhibitor may be a function of the heterologous nature of the expression system used. The same applies to the results of Jiang and Rogers (1998). It could be that in mesophyll or suspension-cultured tobacco cells, the vesicular protein-sorting machinery is strictly adapted to the respective physiological conditions in each cell type, that is, it is geared up for a particular class of vacuoles. Moreover, dense vesicles are not produced in these two tobacco cell types. Thus, the constraints of packaging α-TIP and storage proteins into the same transport vesicle are no longer given, and the two proteins may be forced to take other and perhaps different routes to the (lytic) vacuole.
BP-80 is a type I integral membrane protein that binds with high specificity to the N-terminal vacuolar sorting signal NPIR (Kirsch et al., 1994, 1996) and shows structural similarities to some other eukaryotic vacuolar sorting receptors (Ahmed et al., 1997; Paris et al., 1997). It is a member of a novel protein family in plants, with representatives having been identified in pea, maize, rice, pumpkin, and Arabidopsis (Paris and Rogers, 1996; Ahmed et al., 1997; Paris et al., 1997; Shimada et al., 1997; Neuhaus and Rogers, 1998). Three different homologous BP-80 cDNAs have been identified in pea and also in Arabidopsis (Paris and Rogers, 1996). This finding may indicate either that there are functionally different vacuolar sorting receptors in the same cell or that these homologs may have redundant functions. BP-80 is present at the TGN in pea roots, as shown by immunogold labeling of chemically fixed and resin-embedded tissue (Paris et al., 1997).
In yeast as well as in mammalian cells, the vacuolar targeting receptors cycle between the TGN and a prevacuolar compartment, namely, the endosome, and are not detectable in the vacuolar/lysosomal membrane (Robinson and Hinz, 1997). Recent results suggest that the same may also be true for plant cells. In pea root tips, Paris et al. (1997) were able to localize BP-80 to small vacuole-like organelles in situ, but the protein was not detectable in the tonoplast of the central vacuole. Using subcellular fractionation and in situ immunocytochemistry, Raikhel and co-workers (Ahmed et al., 1997; da Silva Conceição et al., 1997; Sanderfoot et al., 1998) have identified a post-Golgi prevacuolar compartment in Arabidopsis. This compartment was characterized by the presence of AtPep12p. AtPep12p is a functional homolog of the yeast tSNARE Pep12p, which is a marker for the endosomal compartment in yeast (Bassham et al., 1995). In Arabidopsis, this protein is not present in the ER, the Golgi apparatus, or the tonoplast, but it is enriched in a post-Golgi prevacuolar compartment (da Silva Conceição et al., 1997). The Arabidopsis homolog of BP-80, AtELP, is located at the trans Golgi apparatus but not in the tonoplast of these cells. In addition, AtELP colocalizes with AtPep12p in the prevacuolar compartment, thus indicating that this compartment may have the same receptor-retrieval function as does the endosome in mammals or yeast (Ahmed et al., 1997; Sanderfoot et al., 1998).
In agreement with this result, BP-80 was found not to be present in the membranes of isolated protein bodies but instead was highly enriched in the high-density fraction of the rate zonal gradient. This fraction, as determined by the distribution of the marker proteins latent IDPase, RGP, and β-adaptin together with the distribution of processed xylose-containing glycoproteins, is likely to contain a prevacuolar compartment in addition to clathrin-coated vesicles and the TGN.
Matsuoka and co-workers (1995) were able to distinguish biochemically between two different populations of vacuolar transport vesicles in transgenic tobacco cells that were transformed with sporamin, a vegetative storage protein, and barley lectin, a protein of the PSV. Whereas the transport of barley lectin, which is a protein with a C-terminal vacuolar sorting signal, was blocked with wortmannin, the transport of sporamin and the possessing of its N-terminal NPIR signal remained undisturbed, thereby indicating that BP-80 is part of a wortmannin-insensitive pathway. However, the internally localized vacuolar sorting information of legumin shares no sequence or structural homologies with the NPIR signal, which is recognized by BP-80, or with any other known C-terminal signal (Saalbach et al., 1991). Therefore, with multiple vacuolar sorting mechanisms, we presumably have multiple vacuolar sorting receptors existing together in the same cell (Neuhaus, 1996). However, the data presented here indicate that not only multiple vacuolar sorting receptors but also multiple vesicular vacuolar sorting mechanisms may exist in the same cell. Thus, in developing pea cotyledons, prolegumin and α-TIP but not BP-80 are present in dense vesicles. However, BP-80 is enriched in clathrin-coated vesicles, which in turn do not carry prolegumin (Hohl et al., 1996) or α-TIP (Hohl et al., 1996).
Because dense vesicles and clathrin-coated vesicles can be detected at the same dictyosome (Hohl et al., 1996; Robinson et al., 1997), the question remained as to how prolegumin gets sorted into the dense vesicles. Several lines of evidence suggest that an aggregation process may be involved. Inside the ER, prolegumin is rapidly assembled into trimers, which are then transported to the Golgi apparatus (Vitale et al., 1993). Although the lumen of the ER and Golgi cisternae contain high amounts of trimeric prolegumin (Chrispeels et al., 1982a; Hinz et al., 1997), this protein cannot normally be detected by using immunogold labeling (for exceptions, see Robinson and Hinz, 1996). In contrast, dense vesicles are characterized by osmiophilic, condensed aggregates of storage protein precursor polypeptides. Because these aggregates are first visible inside the dilated periphery of the Golgi cisternae (Hohl et al., 1996; Robinson et al., 1997), storage protein aggregation and vesicle formation must be coincidental events. In support of this hypothesis is the fact that prolegumin in the dense vesicles is aggregated to a higher degree relative to that in the ER.
The mechanism of prolegumin aggregation, however, remains obscure. By using in vitro oligomerization assays, Dickinson et al. (1989) demonstrated that trimeric prolegumin does not oligomerize into higher aggregates unless it is processed into mature legumin. Jung et al. (1998) have also shown that in transgenic tobacco seeds that express broad bean prolegumin, processing of the precursor is a prerequisite for hexamer formation. By contrast, dense vesicles isolated from pea cotyledons contain only unprocessed prolegumin. Moreover, prolegumin complexed in hexameric oligomers has also been demonstrated in PSVs isolated from developing pea cotyledons (Hinz et al., 1997). Therefore, condensation of prolegumin into higher molecular weight aggregates is not a property of the PSV. Whether such an aggregation is a chaperone-mediated process or whether it occurs spontaneously after exceeding a critical concentration remains to be elucidated. In favor of the second hypothesis are the results of Wandelt et al. (1992), who expressed a vicilin KDEL construct in tobacco leaves. This protein was also not proteolytically processed and, upon accumulation, made up to 2.5% of the total leaf protein. Due to the KDEL sequence, the polypeptides were retained in the lumen of the ER and finally aggregated into protein body–like structures. Vicilin constructs without the KDEL sequence are not retained in the ER and therefore do not aggregate in this compartment.
Whether the sorting of prolegumin into the dense vesicles is a receptor-mediated process or whether aggregation alone is responsible for proper sorting into the transport vesicles remains to be determined. In favor of receptor-mediated transport is the fact that prolegumin contains a vacuolar sorting domain that directs reporter constructs into the vacuole of transgenic tobacco leaves and seeds (Saalbach et al., 1991). In clathrin-coated vesicles of yeast or mammals, soluble vacuolar hydrolases bind with a stoichiometry of one or two to their receptors (Le Borgne and Hoflack, 1998). The highly aggregated state of proteins in the lumen of the dense vesicles makes it very unlikely that the binding of prolegumin to a putative receptor shows the same binding properties. Because prolegumin can be detected in forming dense vesicles attached to the periphery of the first cis cisternae of the Golgi stack (Hohl et al., 1996; Robinson et al., 1997), one might assume that newly arrived prolegumin is recognized by a receptor located at the cis cisternae and selectively transported laterally into the growing dense vesicles where, due to the high concentration of protein, it aggregates.
As has been discussed earlier (Hohl et al., 1996; Robinson et al., 1997), this process would be similar to the sorting of regulated secretory proteins in endocrine glands of mammals. At the TGN of these cells, so-called immature secretory granules (ISGs) are formed as the result of homophilic aggregation of the regulated secreted proteins. This aggregation process specifically excludes constitutively secreted proteins (Thiele et al., 1997; Tooze, 1998). However, these ISGs still contain lysosomal hydrolases. During the transformation of the ISG into the mature secretory granules, these proteins are removed by mannose-6-phosphate receptors that sort them into clathrin-coated vesicles, which in turn bud from the ISG (Kuliawat et al., 1997). Interestingly, clathrin-coated vesicles budding from dense vesicles can often be seen at the TGN of pea cotyledons (Figure 2 in Hinz et al., 1993; Figure 3 in Hohl et al., 1996; Figure 7 in Robinson and Hinz, 1997; Figures 1A and 1F in Robinson et al., 1997), thus indicating that the targeting of storage protein precursor polypeptides may follow a similar sorting mechanism in the secretory system of developing pea cotyledons. Two observations further support this hypothesis. Prolegumin, by contrast to mature legumin, is a hydrophobic protein (Hinz et al., 1997). In addition, prolegumin, unlike mature legumin, is tightly attached to the membranes of the secretory pathway (Hinz et al., 1997). As discussed by Thiele et al. (1997), sorting of regulated secretory proteins may also be due to the interaction of secretory proteins with a membrane-attached subpopulation of such cargo proteins. In this scenario, both membrane association and the homophilic aggregation are possibly the result of hydrophobic protein–protein interactions.
METHODS
Plant Material
Pea (Pisum sativum var Exzellent) plants were grown hydroponically. Cotyledons with 8- and 9-mm-long axis diameters were collected (20 to 22 days after flowering).
Isolation of Dense Vesicles
Dense vesicles were isolated from pea cotyledons by a combination of isopycnic and rate zonal sucrose density gradient centrifugation. Testa-free cotyledons were homogenized in a mixer with 2 mL g-1 fresh weight of a slushy frozen medium of 0.3 M sorbitol, 50 mM Mops-KOH, pH 6.5, 3 mM EDTA, 0.5 mM MgCl2, 1.5 mg mL-1 fatty acid–free BSA, 2 μg mL-1 leupeptin, 2 μg mL-1 aprotinin, 1 μg mL-1 trans-epoxysuccinyl-l-leucylamido-(4-guanido)-butan (E-64), 0.7 μg mL-1 pepstatin, and 1 mM o-phenanthroline. The homogenate was filtered through two layers of paper fleece (Schleicher and Schuell; step 1) and then centrifuged for 10 min at 200g (step 2). The supernatant from step 2 was centrifuged for 20 min at 16,000g (step 3). The supernatant from step 3 was then layered onto a sucrose step gradient (7 mL 20%, 7 mL 35%, 7 mL 41%, and 5 mL 55% sucrose [w/w] in medium A [50 mM Mops, pH 6.5, 3 mM EDTA, and 0.5 mM MgCl2]) and centrifuged for 150 min at 80,000g in a swing-out rotor (step 4). The 41/55% sucrose interphase was harvested, pooled, diluted with 2 volumes of medium A, and sedimented for 60 min at 100,000g in a swing-out rotor onto a 55% (w/w) sucrose cushion. The membranes were diluted in medium A to a sucrose concentration of ∼18% (w/w), layered onto a linear sucrose gradient (22 mL, 20 to 55% sucrose [w/w] in medium A), and centrifuged for 25 min at 24,000g in a swing-out rotor (model HB-4; Sorvall, Newton, MA). The gradient was fractionated into two pools: from 22 to 26% (pool 1) and from 29 to 33% (pool 2, step 5). Pool 1 contained highly enriched dense vesicles. All procedures were performed at 4°C.
Isolation of Plasma Membranes and Protein Bodies
To eliminate plasma membrane contamination from the dense vesicle fraction, the fractions were diluted twice with medium A and sedimented for 60 min at 100,000g. The membrane pellet was further purified by aqueous two-phase partitioning (Larsson et al., 1987), as described for pea cotyledons by Robinson et al. (1996). The three-time partitioned lower phase contained highly enriched dense vesicles.
Protein bodies from mature pea cotyledons were isolated according to the method of Mäder and Chrispeels (1984).
Isolation of Clathrin-Coated Vesicles
Clathrin-coated vesicles were isolated as described by Hohl et al. (1996). Pea cotyledons were homogenized in a medium containing 0.3 M sorbitol, 1 mM DTT, 3 mM EDTA, 1.5% (w/v) fatty acid–free BSA, 2 μg mL-1 aprotinin, 1 μg mL-1 E-64, 0.7 μg mL-1 pepstatin, and 1 mM o-phenanthroline dissolved in buffer B (0.5 mM MgCl2 and 1 mM EGTA in 0.1 M Mes-NaOH, pH 6.4); the microsomal membranes were removed by centrifuging at 40,000g for 30 min. Clathrin-coated vesicles were purified from a crude postmicrosomal fraction (133,000g pellet) in a five-step procedure. Step 1 included treatment with pancreatic RNase (4.7 Kunitz units per mg protein for 25 min at 27°C) followed by a centrifugation at 8500g for 30 min. In step 2, the supernatant from step 1 was loaded onto a sucrose step gradient (15 mL 5% [w/v] and 10 mL 30% [w/v] sucrose in buffer B) and centrifuged at 67,000g for 40 min. The 5% layer and the 5/30% interphase were removed, diluted in medium B, and pelleted (133,000g for 70 min). In step 3, the pellet from step 2 was resuspended in buffer B, layered onto a linear (5 to 35% [w/v] sucrose in buffer B) sucrose gradient, and centrifuged at 67,000g for 90 min. Fractions corresponding to 9 to 19% sucrose were pooled, diluted with buffer B, and pelleted again (133,000g for 70 min). In step 4, the pellets from step 3 were resuspended in buffer B and recentrifuged in an Eppendorf tube for 10 min at 9000g. In step 5, the supernatant from step 4 was layered onto a linear (25 to 55% [w/v] sucrose in buffer B) sucrose gradient and centrifuged at 167,000g for 150 min in a vertical rotor. Fractions from 40 to 45% sucrose were pooled, diluted in buffer B, and centrifuged for 70 min at 133,000g. The pellet was resuspended in buffer B and recentrifuged in an Eppendorf test tube for 10 min at 9000g. The resulting supernatant contained highly purified clathrin-coated vesicles.
Separation of Legumin Oligomers on Sucrose Gradients
The separation of legumin oligomers was performed as described previously (Hinz et al., 1997). Aliquots of the 20/35% sucrose interphase (Figure 1, step 4) and of the enriched dense vesicle fraction (22 to 26% sucrose; Figure 1, step 5) were diluted in homogenization medium and sedimented for 1 hr at 100,000g. The membranes were lysed on ice in buffer C containing 10 mM Tris-HCl, pH 7.5, 180 mM NaCl, and 0.05% Triton X-100 with the protease inhibitors 2 μg mL-1 leupeptin, 0.7 μg mL-1 pepstatin, 2 μg mL-1 aprotinin, and 1 mM o-phenanthroline for 1 hr and centrifuged for 1 hr at 100,000g. The supernatant (0.3 mL, ∼0.2 mg of protein) was separated on a 5-mL linear sucrose gradient (20 to 40% sucrose [w/w]) for 20 hr at 200,000g in a swing-out rotor (model AH 650; Sorvall). Fractions (0.5 mL) were removed, and the proteins were recovered by precipitation with methanol chloroform according to the method of Wessel and Flügge (1984).
Gel Electrophoresis, Protein Gel Blotting, Protein Determination, and IDPase Assay
Before SDS-PAGE under reducing conditions (Laemmli, 1970), proteins were precipitated with trichloracetic acid and neutralized by washing the sediments twice with ice-cold pure ethanol. Proteins were electroblotted onto nitrocellulose (Towbin et al., 1979) and probed with appropriate primary and secondary antibodies. Secondary antibodies were coupled to horseradish peroxidase (Sigma). Visualization of the bound antibodies was with an enhanced chemiluminescence (ECL) kit (Amersham Life Sciences, Braunschweig, Germany). Blots were quantitated by determining the optical density of the bands on the ECL film by using the BASYS gel analyzing system (Biotec Fischer, Reiskirchen, Germany). Protein was measured according to the method of Lowry et al. (1951). The activity of the latent IDPase was used as a marker for the Golgi apparatus and was measured as described by Robinson et al. (1994).
Antibodies and Their Dilution
A polyclonal antiserum was prepared against pea legumin. Legumin was purified from isolated protein bodies using a DEAE–Sephacel anion exchange column, as previously described (Hinz et al., 1997). The column was equilibrated with 50 mM Mops-KOH, pH 6.5, and 180 mM NaCl, and the protein was eluted by a linear salt gradient from 180 mM to 1 M NaCl in the same buffer. The legumin-containing fractions, which eluted at ∼450 mM NaCl, were pooled and dialyzed against 10 mM Tris-HCl, pH 7.5. The precipitated protein was separated on a SDS–polyacrylamide gel in the presence of β-mercaptoethanol. The α and β legumin chains were cut from the Coomassie Brilliant Blue R 250–stained (Neuhoff et al., 1985) gel, electroeluted for 36 hr at 100 V in SDS-PAGE buffer, dialyzed against 10 mM Tris-HCl, pH 7.5, for 48 hr, and lyophilized. The antiserum raised against this purified legumin was raised commercially in rabbits by Eurogentec Bel S.A. (Ougree, Belgium). The rabbits were immunized three times with ∼200 μg of legumin for each injection. The antiserum was used for protein gel blots at a dilution of 1:10,000.
Polyclonal antibodies were raised in rabbits against α-TIP (for tonoplast intrinsic protein; Johnson et al., 1989) (1:6000; dilution for immunocytochemistry on cryosectioned specimens was 1:400); polyclonal antibodies were raised in rabbits against tobacco BiP (Denecke et al., 1991) (1:8000); polyclonal antibodies were raised in rabbits against the trans Golgi resident protein reversibly glycosylated protein (RGP; Dhugga et al., 1997) (1:20,000); polyclonal antibodies were raised in rabbits against the complex glycoprotein β-fructosidase from Daucus carota (Laurière et al., 1989) (1:5000); monoclonal antibodies were raised in mouse against the plasma membrane H+-ATPase (Fichmann et al., 1989) (1:500); a polyclonal antiserum was raised in rabbits against the N-terminal 10 amino acids of BP-80 (Paris et al., 1997) (1:2000; dilution for immunocytochemistry on cryosectioned specimens was 1:200); a polyclonal antibody was raised in rabbits against the zucchini hypocotyl clathrin heavy chain (Drucker et al., 1995) (1:20,000); and a monoclonal antibody that also recognizes the plant β-adaptin (Holstein et al., 1994) was raised in mouse against the mammalian protein (B1M6) (1:1000).
Electron Microscopy and Cryosectioning
Fractions from the rate zonal linear sucrose gradient (Figure 1, step 5) were mixed in equal volumes of primary fixative containing 2% (w/v) glutaraldehyde and 50 mM potassium phosphate buffer, pH 7.0, and incubated for 1 hr on ice. The fixed organelles were layered onto sucrose step gradients (30 and 35% sucrose [w/v] in potassium phosphate buffer) containing 1% (w/v) OsO4 in the 35% sucrose step and centrifuged for 2 hr at 100,000g in a swing-out rotor. The fixed organelle sediments were embedded in 2% (w/v) low-melting-point agarose and, after washing in phosphate buffer and then water (twice each for 10 min), postfixed in 2% (w/v) uranyl acetate for 14 hr on ice. After washing in water (twice for 15 min), the specimens were dehydrated in a graded ethanol series. Specimens were embedded in London Resin White hard grade (Plano, Marburg, Germany), as given by Hoh et al. (1995). Lead citrate/uranyl acetate–poststained sections were examined in an electron microscope (model CM 10; Philips, Eindhoven, The Netherlands).
For cryosectioning, small blocks of cotyledon tissue close to the inner epidermis were excised and immersed in 100 mM potassium phosphate buffer, pH 7.0, containing 1.5% (w/v) paraformaldehyde and 0.2% (v/v) glutaraldehyde. To achieve better infiltration of the fixatives, we evacuated the samples for 15 min at room temperature before transferring them to 4°C for an additional 16 hr. Afterward, the blocks were washed in buffer and transferred to 2.3 M sucrose for 24 hr at 4°C (Tokuyasu, 1980). The blocks were then frozen in liquid nitrogen and sectioned at -120°C (Leica UCT; Leica, Bensheim, Germany). Frozen sections were picked up from a diamond knife by using a mixture of sucrose and methylcellulose to prevent over-stretching (Liou et al., 1996) and transferred to carbon-coated nickel grids. The labeling conditions were identical to the procedures that have been described for sections with resin-embedded specimens. Secondary antibodies conjugated to 10-nm gold (goat anti–rabbit IgG) were obtained from Biocell (Cardiff, UK) and diluted 1:30 in Tris-buffered saline (50 mM Tris-HCl, pH 7.5, 0.9% [w/v]) containing 1% (w/v) BSA and 0.01% (w/v) acetylated BSA (BSA-C; Aurion, Wageningen, The Netherlands).
Acknowledgments
We thank David G. Robinson for critically reading the manuscript and Dorothee Dassbach and Birgit Zeike for technical assistance. We also thank Sybille Hourticolon for the photographic work and Bernd Raufeisen for preparing the drawings. We are indebted to Drs. Maarten J. Chrispeels, Jürgen Denecke, Kanwarpal S. Dhugga, Nadine Paris, Margaret S. Robinson, and David G. Robinson, who provided us with the necessary antibodies for this study. This research was supported by grants from the Deutsche Forschungsgemeinschaft (Nos. SFB 523 Teilprojekte A7, B2, and Z2).
- Received January 20, 1999.
- Accepted April 26, 1999.
- Published August 1, 1999.
REFERENCES
