The composition of newly synthesized proteins in the endoplasmic reticulum determines the transport pathways of soybean seed storage proteins

Authors


(fax +81 774 38 3761; e-mail sutsumi@kais.kyoto-u.ac.jp).

Summary

Glycinin (11S) and β-conglycinin (7S) are major storage proteins in soybean (Glycine max L.) seeds and accumulate in the protein storage vacuole (PSV). These proteins are synthesized in the endoplasmic reticulum (ER) and transported to the PSV by vesicles. Electron microscopic analysis of developing soybean cotyledons of the wild type and mutants with storage protein composition different from that of the wild type showed that there are two transport pathways: one is via the Golgi and the other bypasses it. Golgi-derived vesicles were observed in all lines used in this study and formed smooth dense bodies with a diameter of 0.5 to several micrometers. ER-derived protein bodies (PBs) with a diameter of 0.3–0.5 μm were observed at high frequency in the mutants containing higher amount of 11S group I subunit than the wild type, whereas they were hardly observed in the mutants lacking 11S group I subunit. These indicate that pro11S group I may affect the formation of PBs. Thus, the composition of newly synthesized proteins in the ER is important in the selection of the transport pathways.

Introduction

Soybean is an important crop as a source of protein because the seed contains nutritious proteins. Storage proteins comprise about 35% of dry seed weight and are essential in germination as a source of nitrogen. Soybean storage proteins are classified into two major globulins, glycinin (11S) and β-conglycinin (7S), which consist of five kinds (group I: A1aB1b, A1bB2 and A2B1a. group II: A3B4 and A5A4B3) and three kinds (α, α′, and β) of subunits respectively (Utsumi et al., 1997). These proteins are synthesized in the endoplasmic reticulum (ER) and transported to the protein storage vacuole (PSV), where most of them are processed (reviewed in Müntz, 1996). Recently, a sorting signal of the α′ subunit essential for transport to the PSV was found to be present in its C-terminal 10 amino acids (Nishizawa et al., 2003).

Storage proteins are transported to the PSV by two pathways: direct ER-to-PSV and via the Golgi (Vitale and Raikhel, 1999). 7S and 11S in maturing pea cotyledons are transported to the PSV through the Golgi by dense vesicles (DVs) with a diameter of approximately 100 nm (Hohl et al., 1996). After DVs bud from the Golgi, they form multivesicular bodies (MVBs) which mediate the transport of storage proteins to the PSV (Robinson et al., 1998). On the contrary, 2S albumin and 11S in maturing pumpkin seeds are transported to the PSV bypassing the Golgi by precursor-accumulating (PAC) vesicles with a diameter of 300–500 nm (Hara-Nishimura et al., 1998). However, PAC vesicles in castor bean cotyledons contain complex glycans, suggesting that Golgi-derived vesicles are also involved in the formation of PAC vesicles. In transgenic soybean cotyledons where the expression of α and α′ subunits of 7S was suppressed, protein bodies (PBs) were observed (Kinney et al., 2001). These reports suggest that the composition of newly synthesized proteins in the ER is an important factor for selecting the transport pathways of seed storage proteins. Recently, soybean mutants with storage protein composition different from that of the wild type have been developed (Takahashi et al., 2003; Teraishi et al., 2001). To elucidate the transport mechanism of soybean 7S and 11S in detail, we analyzed the transport pathways of soybean storage proteins using soybean cotyledons of the wild type and mutants (7S null, 11S null, 7S α′ & α null, 7S α′ null, 7S α null, 7Sβ, 11S group I null, 11S group II null, 11S A5A4B3 null, 7S+11S A3B4 and 7S+11S A5A4B3).

Results

Vesicular transport of storage proteins to the PSV in the wild-type soybean cotyledons

To analyze the transport pathway of soybean storage proteins, at first we performed ultrastructural analysis of the wild-type soybean developing seeds. At an early stage, a single large vacuole was present and occupied a considerable portion of the space in a soybean cotyledon cell. During seed growth, the large vacuole was replaced by small PSVs where storage proteins accumulate. On the basis of differences in seed size and amount of accumulated proteins in the PSV, we classified developing soybean cotyledon cells into four stages (I–IV) (Figure 1). Proteins accumulated at the periphery (stage I), near the periphery (stage II), roughly inside (stage III), and finally filled the PSV (stage IV). The developmental changes of PSVs were consistent with previous reports (Craig et al., 1979; Melroy and Herman, 1991).

Figure 1.

Developmental changes of protein accumulation in developing wild-type soybean cotyledon PSVs (I–IV).
On the basis of the differences in seed size and amount of accumulated proteins in the PSV, soybean cotyledon cells were classified into four stages (I–IV). Bars, 1 μm.

Electron microscopic analysis of the wild type showed vesicles ranging in size from 0.5 to several micrometers at stages I–III (Figure 2a,b). As the vesicles were uniformly dense, we named them smooth dense bodies (SDBs). SDBs were frequently observed and seemed to fuse with each other at stages I–II (Figure 2c). The number and size of SDBs were less and smaller (approximately 0.5 μm in diameter), respectively, at stage III than those (several micrometers) at stages I–II.

Figure 2.

Smooth dense bodies (SDBs) in the wild-type soybean.
SDBs were labeled with anti-pro11S (a, d) and anti-7S (b) sera in resin embedded samples. It seemed that SDBs fused with each other (c) and formed larger ones before they fused with the PSV. Bars: (a and b) 500 nm; (c and d) 1 μm.

To confirm whether SDBs are involved in the transport of storage proteins, we performed immunologic analysis using anti-pro11S and anti-7S sera. Immunoelectron microscopic analysis showed that SDBs were labeled with anti-pro11S (Figure 2a,d) and anti-7S sera (Figure 2b). It seemed that SDBs fused with the PSV (Figure 2d). These indicate that SDBs mediated the transport of 11S and 7S to the PSV. As SDBs mediated the transport of both 11S and 7S and ranged in diameter from 0.5 to several micrometers, they were similar to MVBs observed in developing pea cotyledons (Robinson et al., 1998). However, most of the SDBs did not contain internal vesicles that are observed in MVBs. Thus, we do not use the term MVBs. It has been suggested that MVBs are formed by DVs in pea cotyledons (Robinson et al., 1998). Similarly, DVs with a diameter of approximately 100 nm mediated the transport of 11S and 7S to the PSV in soybean cotyledons at stages I–III (Figure 3a,b), suggesting that DVs are involved in the formation of SDBs.

Figure 3.

Golgi-mediated transport of soybean seed storage proteins.
DVs and the Golgi were labeled with anti-pro11S (a) and anti-7S (b) sera in resin-embedded samples. Bars, 500 nm.

Besides DVs and SDBs, PBs that contained an electron-dense core and were surrounded by translucent layer and ribosome-like structures were observed at stages I–III (Figure 4). The PBs were morphologically similar to PBs observed in transgenic soybean cotyledons (Kinney et al., 2001) and PAC vesicles in developing pumpkin cotyledons (Hara-Nishimura et al., 1998). PBs were morphologically distinguished from SDBs by their association with ribosomes (Figure 12a). PBs were observed at stages I–III, but not at stage IV. The frequency of observation of PBs was higher at stage III than at stages I–II. PAC vesicles in developing pumpkin cotyledons mediate the transport of 2S albumin and 11S. To determine whether PBs are involved in the transport of storage proteins, we performed immunologic analysis using anti-pro11S and anti-7S sera. Immunoelectron microscopic analysis showed that PBs were labeled with anti-pro11S serum (Figure 4a), but not with anti-7S serum (Figure 4b). Thus, 11S is preferentially incorporated into PBs. We discuss below the possible role of these structures in transporting 11S to the PSV.

Figure 4.

Protein bodies (PBs) in the wild type.
PBs were labeled with anti-pro11S (a) but not with anti-7S (b) sera in resin-embedded samples. Bars, 500 nm.

Figure 12.

Immunoelectron microscopy of developing soybean cotyledons with anti-BiP serum.
Protein bodies are observed within and around the ER and labeled with the anti-pro11S in the resin-embedded samples (a) and anti-BiP sera in the cryosectioned samples (b) respectively. Bars, 500 nm.

Transport vesicles in mutant soybean cultivars

To determine whether the mechanism of transport of storage proteins in mutant soybean cultivars is similar to that in the wild type, we performed ultrastructural analysis of the seeds of mutant cultivars (7S null, 11S null, 7S α′ null, 7S α null, 7S α′ & α null, 11S group I null, 11S group II null, 11S A5A4B3 null, 7S+A3B4, 7S+A5A4B3 and 7Sβ, see Table 1). SDS-PAGE of seed extracts showed the seed storage protein and subunit compositions of these cultivars (Figure 5). The processes of protein accumulation in the PSV of all mutants were almost similar to that of the wild type. At first, we analyzed 7S null and 11S null mutants. SDBs were observed at stages I–III, especially at stages I–II, in 11S null mutant (Figure 6). On the contrary, the 7S null mutant did not exhibit such a tendency. SDBs in the 7S null and 11S null mutants were labeled with anti-pro11S and anti-7S sera respectively (Figure 6a,b). A small number of SDBs in the 11S null mutant contained internal vesicles (Figure 6c).

Table 1.  Subunit composition (+, presence; −, absence) of soybean cultivars
 7S subunit composition11S subunit composition
ααβIII (A3B4)II (A5A4B3)
Fukuyutaka (wild type)++++++
EnB1 (11S null)+++
QY2 (7S null)+++
QF2 (double null)
Yumeminori (α′&α null)++++
7S α′ null+++++
7S α null+++++
Enrei (A5A4B3 null)+++++
EnF2 (7S β)+
11S group I null+++++
11S group II null++++
7S+A3B4++++
7S+A5A4B3++++
Figure 5.

Comparison of protein patterns of mature seeds of wild-type and mutant cultivars.
SDS-PAGE was performed in the presence of 2ME using 11% polyacrylamide gel. Lane 1, wild type; lane 2, 7S null; lane 3, 11S null; lane 4, 7S α′&α null; lane 5, 7S α′ null; lane 6, 7S α null; lane 7, 11S group I null; lane 8, 11S group II null; lane 9, 11S A5A4B3 null; lane 10, 7S+A3B4; lane 11, 7S+A5A4B3; lane 12, 7S β. A: 11S acidic chain; B: 11S basic chain.

Figure 6.

Smooth dense bodies (SDBs) in mutant cultivars.
SDBs were observed in these mutants and labeled with anti-pro11S (a) or anti-7S (b) serum. 7S null (a), 11S null (b and c). Bars, 500 nm.

Protein bodies were hardly observed in the 11S null mutant, whereas vesicles were observed at a higher frequency in the 7S null mutant than in the wild type (Figure 7a). Similar to those of the wild type, PBs in the 7S null mutant were labeled with the anti-pro11S serum. In the 7S null mutant, PBs were easily observed at stages I–III, and were observed even at stage IV and in mature seeds where the wild type did not have the vesicles (Figure 7c). Furthermore, we performed Western blot analysis of proteins extracted from developing (stage IV) and mature wild type and 7S null mutant seeds using the anti-pro11S serum (Figure 8). The wild type hardly gave bands of pro11S, unprocessed form of 11S subunit (Figure 8, lanes 2 and 3), while the 7S null mutant gave intense bands (Figure 8, lanes 4 and 5).

Figure 7.

Protein bodies (PBs) in the 7S null mutant.
PBs were labeled with anti-pro11S serum (a) and remained in the cytoplasm at stage IV (b) and in mature seeds (c). Bars, 500 nm.

Figure 8.

Pro11S remains in the 7S null mutant.
Seed extracts were subjected to SDS-PAGE and subsequent immunologic analysis with anti-pro11S serum. Lane 1, purified recombinant proA1aB1b; lanes 2 and 4, protein extracts from developing seeds at stage IV; lanes 3 and 5, protein extracts from mature seeds. P, pro11S; A, 11S acidic chain; B, 11S basic chain.

To analyze the relationship between the composition of storage proteins and the formation of transport vesicles in detail, we performed ultrastructural analysis of other mutants (7S α′ null, 7S α null, 7S α′ & α null, 11S group I null, 11S group II null, 11S A5A4B3 null, 7S+A3B4, 7S+A5A4B3 and 7Sβ, see Table 1). In the mutants lacking α′ and/or α subunits of 7S (7S α′ null, 7S α null and 7S α′ & α null), SDBs and PBs were observed. PBs and pro11S remained in mature seeds of 7S α′ & α null mutant (data not shown). In the 7Sβ mutant, SDBs were observed, but PBs were not observed similar to the 11S null mutant. In all the mutants lacking some subunits of 11S (11S group I null, 11S group II null, 11S A5A4B3 null, 7S+A3B4 and 7S+A5A4B3), SDBs were observed. However, PBs were observed mainly in the mutants containing 11S group I subunits (11S group II null and 11S A5A4B3 null) (Figure 9a). Furthermore, pro11S remained in mature seeds of 11S group II null mutant (Figure 9b, lane 2). These suggest that the 11S group I subunit plays an important role in the formation of PBs. To elucidate the developmental changes of the number of PBs, we scored the vesicles (Table 2). PBs were observed exclusively in the wild type and mutants containing 11S group I subunits (7S α′ null, 7S α null, 7S null, 7S α′ & α null, 11S group II null and 11S A5A4B3 null). The number of PBs observed in the mutants (7S null, 7S α′ & α null and 11S group II null) was higher than in the wild type and some of them remained even in mature seeds. Importantly, in the wild-type seeds and in seeds from all mutants where PBs were observed, the number of PBs present decreased substantially from developmental stage III to stage IV. One explanation for this decrease in PB number would be their transport to, and incorporation in, PSVs.

Figure 9.

Protein bodies (PBs) formed in the mutant containing 11S group I subunits.
Electron microscopy of mutant cultivars showed that PBs were frequently observed in 11S group II null mutant (a). Seed extracts were subjected to SDS-PAGE and subsequent immunologic analysis with anti-pro11S serum (b). For (a) and (b), seeds at stages II–III were used. Lane 1, purified recombinant proA1aB1b; lane 2, 11S group II null; lane 3, 11S group I null; P, pro11S; A, 11S acidic chain; Bars, 1 μm.

Table 2.  Quantification of protein bodies (PBs) in developing soybean cotyledons
 Developmental stages
IIIIIIIV
  1. Twenty cells were chosen by chance in five sections of each stage of developing soybean cotyledons and compared with the developmental changes of the number of PBs per 1000 square micrometers.

Fukuyutaka (Wild)<0.1<0.122.3<0.1
7S null6.436.677.226.4
7S α′&α null11.447.883.531.7
7S α′ null<0.1<0.120.21.7
7S α null<0.1<0.124.51.4
Group II null<0.1<0.156.713.8
Enrei (A5A4B3 null)2.43.4254.1
Others (7S β, 11S null, group I null, 7S+A3B4, 7S+A5A4B3)<0.1<0.1<0.1<0.1

To examine the relationship between the solubility and transport pathway of 11S, we purified the 11S species (11S, 11S group I, 11S group II, A3B4 and A5A4B3) from the wild-type and mutant cultivars. The ionic strength in the ER is not exactly known, but it was supposed to be more or less 0.06–0.15 (Maruyama et al., 1998; Shimoni and Galili, 1996). Thus, we measured the solubility of these 11S species at various pHs under ionic strength 0.08. 11S group II and A5A4B3 were completely soluble and 11S group I was almost insoluble in the vicinity of pH 7.0 (Figure 10). The solubility of 11S was in the middle of 11S groups I and II, and similar to that of A3B4. The solubility of 11S group II was attributed to that of A5A4B3. These indicate that group I-rich 11S species was insoluble in the ER.

Figure 10.

Dependency of the solubility of 11S species on pH at ionic strength 0.08.
Solubilities of 11S, 11S group I, 11S group II, A3B4 and A5A4B3 are shown by solid line with closed circles, dotted line with open squares, dashed and double-dotted line with closed squares, dashed line with inverted triangle and dashed and single-dotted line with open triangles respectively.

Origins of PBs and SDBs

To confirm the origins of PBs and SDBs, we prepared anti-soybean binding protein (BiP) serum and anti-complex glycan antibody. Although BiP is a molecular chaperon involved in protein folding and localizes in the ER, a part of the proteins is taken in ER-derived vesicles and transported to the PSV (Hara-Nishimura et al., 1998). Complex glycans are formed in the Golgi, and proteins containing them are taken in Golgi-derived vesicles. Anti-horseradish peroxidase (HRP) serum contains antibodies specific for fucose, one of the components of complex glycans (Faye et al., 1993). Thus, we purified complex glycan-specific antibody from anti-HRP serum. To confirm the specificity of these antibodies, we performed Western blot analysis using soybean protein extracts, purified phytohemagglutinin (PHA, Wako, Osaka, Japan) with complex glycan containing fucose and purified 7S. Anti-BiP serum gave a single band, corresponding to BiP, in soybean seed extracts (Figure 11a), and the anti-complex glycan antibody clearly interacted with PHA but not purified 7S which has high mannose type glycans (Figure 11b). As PHA is a lectin, it is possible to interact with glycans of antibodies. However, PHA separated on the SDS-PAGE did not interact with anti-pro11S serum (data not shown). These indicate that anti-complex glycan antibody specifically interacted with the complex glycans of PHA. Accordingly, we examined the origins of PBs and SDBs by using anti-BiP serum and anti-complex glycan antibody.

Figure 11.

Analysis of the specificities of anti-BiP serum (a) and anti-complex glycan antibody (b).
(a) Lane 1, purified recombinant BiP; lane 2, seed extracts from the wild type at stage III.
(b) Lane 1, purified 7S; lane 2, PHA. Single band was detected by anti-BiP serum in seed extracts (lane 2a), corresponding to purified recombinant BiP (lane 1a). A band, corresponding to PHA, was intensely detected by anti-complex glycan antibody (lane 2b).

Protein bodies were frequently observed within and around the ER (Figure 12a), suggesting that the vesicles may also be derived from the ER. Immunoelectron microscopic analysis of cryosectioned samples of the 7S null mutant showed that PBs were labeled with anti-BiP serum (Figure 12b). This indicates that the vesicles are derived from the ER. However, SDBs were not labeled with anti-BiP serum (data not shown), suggesting that PBs are not involved in the formation of SDBs.

Small SDBs (approximately 0.5 μm in diameter) were frequently observed near the Golgi (Figure 13a). Immunoelectron microscopic analysis showed that SDBs were labeled with anti-complex glycan antibody (Figure 13b). On the contrary, PBs were not labeled with the anti-complex glycan antibody (data not shown).

Figure 13.

Immunoelectron microscopy of developing soybean cotyledons with the anti-complex glycan antibody.
Small smooth dense bodies were observed near the Golgi at a high frequency (a) and labeled with the anti-complex glycan antibody (b) in resin-embedded samples. Bars, 500 nm.

Discussion

There are two transport pathways of storage proteins in seeds: via the Golgi and direct from the ER to the PSV (Vitale and Raikhel, 1999). Pea legumin (11S) and vicilin (7S), and rice glutelin are transported via the Golgi (Hohl et al., 1996; Krishnan et al., 1986), whereas pumpkin 2S albumin and 11S (Hara-Nishimura et al., 1998), and wheat gliadin are transported bypassing the Golgi (Levanony et al., 1992). The transport of storage proteins via the Golgi and bypassing the Golgi are mediated by DVs and ER-derived vesicles (pumpkin PAC vesicles and wheat provacuoles) respectively (Hara-Nishimura et al., 1998; Levanony et al., 1992). In pea cotyledons, DVs form MVBs (Robinson et al., 1998). Our results suggest that in soybean cotyledons two types of vesicles are involved in the transport of 11S and 7S to the PSV: ER-derived PBs and Golgi-derived DVs which form SDBs (Figure 13). PAC vesicles in castor bean cotyledons are labeled with the anti-complex glycan antibody (Hara-Nishimura et al., 1998). PV72 on the periphery of PAC vesicles contains complex glycans (Shimada et al., 2002). These indicate that Golgi-derived proteins are involved in the formation of PAC vesicles (Hara-Nishimura et al., 1998). However, PBs in soybean cotyledons were not labeled with anti-complex glycan antibody (data not shown). PBs in transgenic soybean cotyledons also do not contain complex glycans (Kinney et al., 2001). Thus, the mechanisms of the formation of PBs in soybean may be different from that of PAC vesicles in castor bean. However, we cannot exclude the possibility that the amount of proteins containing complex glycans are too small to be detected in this method, although there is a route from the Golgi to PBs.

Smooth dense bodies transported both 11S and 7S to the PSV. SDBs ranged in size. The formation of small SDBs (approximately 0.5 μm in diameter) was observed near the Golgi. At stages I–II, large SDBs were major, whereas at stages III–IV only small SDBs were evident. PSVs occupy a considerable portion of the space in the cell at stages I–II. Thus, it is possible that SDBs are likely to get in contact with each other. These suggest that at stages I–II when protein synthesis is active, small SDBs formed by DVs fuse with each other to form larger ones, and finally fuse with PSVs (Figure 2). Alternatively, these two types of SDBs may be different bodies, while they are morphologically and functionally similar to each other.

Unlike MVBs, most of the SDBs did not contain internal vesicles. However, a small number of SDBs observed at stages III–IV contained vesicle-like structures (Figure 6c). The PSV is generally composed of matrix, crystalloid and globoid (Jiang et al., 2000, 2001). Jiang et al. (2001) suggested that globoid is a lytic compartment in the PSV because the globoid membrane contains γ-TiP, a marker for a lytic vacuole. Jiang et al. (2002) proposed that MVBs are also composed of the storage compartment and the lytic compartment (internal vesicles), and are involved in the transport of the components to these compartments in the PSV. Lott and Buttrose (1978) reported that soybean PSVs contain the matrix and globoid. We also observed such globoids in our samples (Figure 1IV). MVBs were identified as prevacuolar compartments (PVCs) in BY-2 cells (Tse et al., 2004). Thus, SDBs may be PVCs similar to MVBs and may be involved in the transport of the components of the matrix and globoid. This suggests that the frequency of observation of SDBs that contain internal vesicles may depend on the amount of the components of globoids.

Protein bodies were frequently observed at stage III, but not at stage IV in the wild type (Table 2). Furthermore, PBs were labeled with anti-11S antibody (Figure 4a). These suggest that PBs are involved in the transport of 11S to the PSV. However, some of the PBs remained in the cytoplasm (Figure 7b) (Table 2). Thus, it is possible that PBs are not only stable compartments in the cytoplasm but they can also be incorporated into the PSV. Another possibility is that PBs are within the ER lumen because PBs are surrounded by ribosomes. Their absence indicates that they were delivered to PSVs, although we cannot exclude the possibility that they were instead degraded with the ER.

The stages where PBs were observed differed between the wild type and the 7S null mutant, suggesting that the composition of synthesized proteins in the ER is a strict factor in the formation of PBs. SDS-PAGE of the wild type and the 7S null seed extracts showed that α and α′ subunits of 7S are synthesized at stages I–III, 11S at stages I–IV, mainly stages II–IV, and the β subunit of 7S at stages III–IV (data not shown). A previous analysis of the 7S null mutant revealed that the loss of 7S was compensated by the increase in 11S (Ogawa et al., 1989). SDS-PAGE of seed extracts showed that the ratio of 11S to total proteins in the 7S null mutant was higher than that in the wild type. Furthermore, PBs were hardly observed in the mutants lacking 11S group I subunit (11S group I null, 7S+A3B4, and 7S+A5A4B3) (Table 2). These suggest that the increase in newly synthesized 11S group I subunit in the ER induced the formation of PBs. Thus, in the mutant containing a higher amount of 11S group I subunit than the wild type, many PBs could be formed and some of them may remain in the cytoplasm. These results clearly indicate the relationship between the formation of PBs and 11S species. There is no discrepancy between our results and the preliminary observation by Kinney et al. (2001).

Proteins with low solubility interact with each other and form aggregates in the ER. Wheat storage proteins aggregate in the ER and are transported to the PSV (Levanony et al., 1992). Synthesis of water-insoluble maize zein induces ER-derived PBs in transgenic tobacco endosperm (Coleman et al., 1996). Thus, physicochemical properties such as the solubility of storage proteins affect the formation of ER-derived vesicles. At μ = 0.08, the solubility is different between 11S groups I and II. 11S group I is almost insoluble and 11S group II is completely soluble in the vicinity of pH 7.0 (Figure 10). A3B4 and A5A4B3, the 11S group II subunits, are larger than the 11S group I subunits in molecular size, because they include longer hypervariable regions, which are rich in negatively charged amino acid residues. These regions may account for the difference in their solubilities. Thus, the solubility of the pro11S group I may be lower than that of pro11S group II in the ER. Furthermore, PBs were observed at a high frequency in the 11S A5A4B3 null mutant. 11S A5A4B3 was most soluble subunit in the vicinity of pH 7.0. These indicate that the low solubility of 11S group I may affect the formation of PBs. Although the solubility of pro7S has not been studied, at μ = 0.08 7S heterotrimers (α2β1, α′2β1, α1β2, α′1β2) and 7S homotrimers (α3, α′3), molecular species composed of mature subunits of 7S, are completely soluble in the vicinity of pH 7.0 (Maruyama et al., 2002a,b). It is likely that these species of pro7S are soluble in the ER, because their pro-region is rich in charged residues. Thus, soluble proteins (7S and group II-rich 11S species) and insoluble proteins (group I-rich 11S species) in the ER may be transported by DVs and by PBs respectively. However, at μ = 0.08, 7Sβ homotrimer is insoluble at pH 4.8–8.5 (Maruyama et al., 2002b). Although it is expected that 7Sβ homotorimer aggregates and is taken in PBs in a similar manner as the pro11S group I, PBs were not observed in the 7Sβ mutant. If PBs are formed by the aggregation of proteins, it is possible that there is a mechanism that inhibits the aggregation of 7Sβ homotrimer in the ER and transports it from the ER to the Golgi. One possibility is that aggregation of 7Sβ is inhibited by chaperons such as calnexin in the ER as the subunit is a glycoprotein (Utsumi et al., 1997).

Experimental procedures

Plant materials

Soybean plants (Glycine max L.) were grown in a greenhouse. Fukuyutaka was used as the wild-type cultivar and the other mutants were EnB1 (11S null), QY2 (7S null), Yumeminori (7S α′ & α null), 7S α′ null, 7S α null, Enrei (A5A4B3 null), EnF2 (7Sβ), 11S group I null, 11S group II null, 7S+11S A3B4, 7S+11S A5A4B3 (Takahashi et al., 2003; Yagasaki et al., 1997). Globulin and subunit compositions of the wild type and mutants are summarized in Table 1. Seeds were harvested and the cotyledons were used for all experiments.

Preparation of antibodies

A polyclonal antibody against complex glycans was purified from anti-HRP (Sigma, Tokyo, Japan). Phospholipase A2 that contains complex glycans was coupled to a HiTrap NHS-activated HP (Amersham Biosciences, Uppsala, Sweden) according to the manufacturer's instructions. Serum (0.5 ml) was applied on the column equilibrated with 20 mm phosphate buffer (pH 7.2). After washing the column, bound antibodies were eluted with the buffer (0.1 m glycine-HCl, pH 3.0, 0.5 m NaCl). Eluates were neutralized by 1 m phosphate buffer (pH 7.2).

To prepare an antiserum against soybean BiP, the mRNA was isolated from soybean seeds as described previously (Maruyama et al., 2003), and the cDNA was obtained from soybean mRNA by RT-PCR amplification using the following primers: BiPN primer (5′-AAGGAGGAAGCCACCAAGTTGGGG-3′) corresponded to the N-terminal sequence of BiP, and BiPC primer (5′-CGCGGATCCACATCTCTAGAGCTCGTCATGAG-3′) containing BamHI restriction site (underlined) complementary to the C-terminal sequence of BiP. The amplified product was blunted, and then digested with BamHI. The resultant DNA fragment was inserted into the NcoI (filled-in) and BamHI sites of pET-21d vector to construct pEBiP. The expression plasmid pEBiP was transformed into Escherichia coli BL21(DE3). Cells were disrupted by sonication and the proteins were initially fractionated using ammonium sulfate (25–75% saturation). The precipitate (75% saturation) was dissolved in buffer A [20 mm Tris-HCl (pH 7.5), 1 mm EDTA, 0.1 mm (p-aminophenyl) methanesulfonylfluoride hydrochloride (p-APMSF), 1.2 μm leupeptin, 0.2 μm pepstatinA], and dialyzed against the same buffer overnight. The dialysate was applied to a MonoQ column (Amersham Biosciences) equilibrated with buffer A. The bound proteins were eluted by linear gradient 0–0.5 m NaCl in buffer A. The N-terminal sequence of the purified protein was determined to be KEEATK. The sequence was consistent with that of soybean BiP (Kalinski et al., 1995). Anti-BiP serum was raised against the purified recombinant BiP in rabbits.

To prepare an antiserum against the β subunit of 7S, we purified the recombinant β subunit of 7S (Maruyama et al., 1998). Anti-7S serum was raised against the purified recombinant β subunit of 7S in rabbits.

Resin embedding

Soybean cotyledons were vacuum infiltrated in 1.5% glutaraldehyde, and 50 mm phosphate buffer (pH 7.2) for 1.5 h at room temperature. The tissues were sliced into 1 mm thickness and fixed with the fixative for 4 h at room temperature. After washing in 50 mm phosphate buffer (pH 7.2), the tissues were dehydrated with a graded ethanol series and embedded in epoxy resin (Nisshin EM, Tokyo, Japan). Ultrathin sections were cut and mounted on copper grids with carbon-coated formvar films.

Cryosectioning

Soybean cotyledons were fixed in the same procedures described above. The tissues were washed in 50 mm phosphate buffer (pH 7.2) and transferred to a 2.3 m sucrose solution overnight at 4°C (Tokuyasu, 1980). The tissues were frozen in liquid nitrogen. Ultrathin sections were cut and mounted on copper grids with carbon-coated formvar films.

Immunoelectronmicroscopy

Thin sections were immersed in blocking solution [10 mm PBS, pH 7.4, 0.25% (v/v) Tween 20, 5% (w/v) BSA] for 30 min, and then washed in diluting solution [10 mm PBS, pH 7.4, 0.25% (v/v) Tween 20, 1% (w/v) BSA]. The sections were incubated with antisera specific for the proA1aB1b subunit of pro11S (Utsumi et al., 1994), the β subunit of 7S, BiP, and antibody specific for complex glycans in diluting solution. After washing in diluting solution, the sections were incubated with a secondary antibody conjugated to 15 nm gold in diluting solution (Amersham Biosciences). Thin sections of the resin-embedded tissues were stained in uranyl acetate solution and lead citrate solution. Cryosectioned samples were stained in the solution containing polyvinyl alcohol and uranyl acetate. These samples were examined with an electron microscope (Hitachi, H-700H, Tokyo, Japan).

SDS-PAGE and Western blot analysis

Soybean cotyledons were homogenized in an extraction buffer [50 mm Tris-HCl, pH 6.8, 4% SDS (w/v), 10% (v/v) glycerol, 0.5 μm leupeptin, 0.2 μm pepstatinA, 0.1 mmp-APMSF]. Homogenates were centrifuged and the supernatants were collected. Protein concentrations were estimated according to the method of Bradford (1976) with BSA as a standard. These samples were separated by SDS-PAGE on 11% polyacrylamide gel. Separated proteins were stained with Coomassie brilliant blue R-250 or transferred to nitrocellulose membranes (0.45 μm; Schleicher & Schuell Inc., Dassel, Germany). Immunodetection was performed as described (Utsumi et al., 1988). The polyclonal antibody and antisera were as follows: anti-complex glycan, anti-pro11S, and anti-BiP.

Solubility

11S fraction was prepared from the wild-type and mutant cultivars (11S group I null, 11S group II null, 7S+A3B4 and 7S+A5A4B3) as described previously (Nagano et al., 1992). After centrifugation, the precipitate was dissolved in 35 mm sodium phosphate buffer containing 0.4 m NaCl, 1 mm EDTA, 10 mm 2-mercaptoethanol, 0.1mmp-APMSF and 50% ammonium sulfate. The solution was centrifuged at 20 000 g for 20 min at 20°C. The protein was fractionated by ammonium sulfate 65% (A3B4), 70% (11S group II and A5A4B3) and 75% (11S group I) saturation, and then used for the measurement of their solubilities at μ = 0.08 as described previously (Maruyama et al., 1999).

Ancillary