A novel ER-derived compartment, the ER body, selectively accumulates a β-glucosidase with an ER-retention signal in Arabidopsis

Authors


* For correspondence (fax +81 75 753 4142; e-mail ijhnishi@gr.bot.kyoto-u.ac.jp).

Summary

The ER body is a novel compartment that is derived from endoplasmic reticulum (ER) in Arabidopsis. In contrast to whole seedlings which have a wide distribution of the ER bodies, rosette leaves have no ER bodies. Recently, we reported that wound stress induces the formation of many ER bodies in rosette leaves, suggesting that the ER body plays a role in the defense system of plants. ER bodies were visualized in transgenic plants (GFP-h) expressing green fluorescent protein (GFP) with an ER-retention signal, HDEL. These were concentrated in a 1000-g pellet (P1) of GFP-h plants. We isolated an Arabidopsis mutant, nai1, in which fluorescent ER bodies were hardly detected in whole plants. We found that a 65-kDa protein was specifically accumulated in the P1 fraction of GFP-h plants, but not in the P1 fraction of nai1 plants. N-terminal peptide sequencing revealed that the 65-kDa protein was a β-glucosidase, PYK10, with an ER-retention signal, KDEL. Immunocytochemistry showed that PYK10 was localized in the ER bodies. Compared with the accumulation of GFP-HDEL, which was associated with both cisternal ER and ER bodies, the accumulation of PYK10 was much more specific to ER bodies. PYK10 was one of the major proteins in cotyledons, hypocotyls and roots of Arabidopsis seedlings, while PYK10 was not detected in rosette leaves that have no ER bodies. These findings indicated that PYK10 is the main component of ER bodies. It is possible that PYK10 produces defense compounds when plants are damaged by insects or wounding.

Introduction

Many ER-derived compartments with specific functions have been identified in plant cells (Chrispeels and Herman, 2000). Precursor accumulating (PAC) vesicles that are found in the maturing seeds of pumpkin mediate the direct transport of the precursors of storage proteins from the ER into protein-storage vacuoles (Hara-Nishimura et al., 1998). Two other types of ER-derived compartments, KDEL vesicles (KVs) and ricinosomes, have been shown to accumulate a cysteine proteinase with an ER-retention signal, KDEL (Schmid et al., 1998; Toyooka et al., 2000). KVs accumulate a vacuolar proteinase, SH-EP, which is responsible for the breakdown of the seed storage proteins of mung bean (Vigna mungo) (Okamoto and Minamikawa, 1998). KVs have been proposed to mediate protein mobilization in cotyledon cells of germinated seeds (Toyooka et al., 2000). Ricinosomes accumulate the same type of proteinase, Cys-EP, which is activated during senescence of castor bean (Ricinus communis) endosperm (Schmid et al., 1999). Ricinosomes have been suggested to be involved in programmed cell death in plant cells (Gietl and Schmid, 2001; Gietl et al., 2000).

Arabidopsis also develops ER-derived structures in its epidermal cells. Electron microscopic studies revealed that the structures have a characteristic shape and size (approximately 10 µm long and 0.5 µm wide) and that they are surrounded with ribosomes (Hayashi et al., 2001). The structures can be visualized by green fluorescent protein (GFP) with an ER-retention signal (GFP-HDEL) (Haseloff et al., 1997; Hawes et al., 2001; Ridge et al., 1999). We proposed to call them ‘ER bodies’ (Hayashi et al., 2001). Similar structures have been reported in the cells of various organs of Brassicaceae plants (Bonnett and Newcomb, 1965; Iversen, 1970b) which are related to Arabidopsis thaliana. However, their biological function has not been determined (Gunning, 1998). ER bodies have a characteristic shape and size which make them easy to distinguish from other ER-derived structures, which have spherical shapes (0.2–0.5 µm diameter). Therefore, the ER bodies appear to have a novel biological function.

The distribution of ER bodies in Arabidopsis plants varies with the tissue and the growth stage (Matsushima et al., 2002). The cotyledons, especially their epidermal cells, have a large number of ER bodies, but the number decreases during senescence. ER bodies are well developed in seedlings but are not detected in rosette leaves. However, ER bodies can be induced in rosette leaves by wounding them or treating them with methyl jasmonate (MeJA), a plant hormone involved in defense against wounding and chewing by insects. These observations suggest that ER bodies play a role in the defense system when plants are damaged by insects or wounding. However, direct evidence to clarify the biological function of ER bodies is still lacking because the main protein in the ER bodies has not been identified.

In this study, we isolated an Arabidopsis mutant that had no ER bodies. Using this mutant, we showed that a β-glucosidase, designated PYK10, is the main component of the ER bodies. ER bodies accumulate large amounts of PYK10 and protect the plants when damaged under stress conditions.

Results

Isolation of a mutant that has no ER bodies

Seeds of transgenic Arabidopsis (GFP-h) expressing ER-targeted GFP were mutagenized with ethyl methanesulfonate. Six- or 7-day-old GFP-h seedlings exhibited many rod-shaped fluorescences of ER bodies in cotyledons, hypocotyls and roots, as shown in Figure 1(a,c,e). Of the 239 M2 lines, we isolated a mutant that had no ER bodies in whole seedlings (Figure 1b,d,f). We named the mutant nai1. Previously, we showed that rosette leaves have no ER bodies, but roots at the same stage have many ER bodies (Matsushima et al., 2002) (Figure 1g,i). In the case of nai1, neither rosette leaves nor roots developed ER bodies (Figure 1h,j). nai1 had ER networks that were similar to those in GFP-h (Figure 1). These phenotypes mean that the nai1 mutation affects the formation of all ER bodies in various organs, but not the formation of the ER networks.

Figure 1.

Fluorescent images of a transgenic plant (GFP-h) and the nai1 mutant.

Five-day-old seedlings and 15-day-old plants were inspected with a fluorescence microscope. The GFP-h seedlings had many ER bodies in cotyledons (a), hypocotyls (c) and roots (e). On the contrary, the nai1 seedlings had no ER bodies in cotyledon (b), hypocotyl (d), or root (f). No ER bodies were detected in the rosette leaves of GFP-h(g) or nai1 (h), while many ER bodies were found in the roots of GFP-h(i), but not in that of nai1(j). Bars = 10 µm.

nai1 plants exhibited normal growth. No visual phenotypes were displayed under normal conditions. nai1 was backcrossed to the parental line, and 26 of 96 F2 progeny tested had no ER bodies, indicating that nai1 segregated as a single recessive allele (χ2 = 0.22, P > 0.5). nai1 was mapped genetically to the middle of chromosome 2 between PHYB and ERECTA. Therefore, nai1 is a recessive mutation that prevents the formation of ER bodies.

nai1 mutant does not accumulate β-glucosidase PYK10

Total homogenates from 8-day-old seedlings of GFP-h and nai1 were subjected to subcellular fractionation. ER bodies were observed to be concentrated in the P1 fraction of GFP-h (Figure 2c), but not in the P1 fraction of nai1 (Figure 2d). Small ER bodies were recovered in the P8 fraction (Figure 2e). The microsomes derived from the ER networks were concentrated in the P100 fraction (Figure 2g,h). We detected fluorescent dot-like structures in the total homogenate and P1 fraction of nai1 (Figure 2b,d) at low frequency. These signals may have been derived from the nuclear envelope.

Figure 2.

Fluorescent images of subcellular fractions from GFP-h and nai1.

Eight-day-old seedlings of GFP-h(a, c, e, g and i) and nai1 (b, d, f, h and j) were subjected to differential centrifugation to separate the total homogenate (a, b) into the P1 (c, d), P8 (e, f), P100 (g, h), and S100 (i, j) fractions. Each fraction was inspected with a fluorescence microscope. Note that the ER bodies were concentrated in the P1 fraction of GFP-h(c), while no ER bodies were found in the P1 fraction of nai1(d). Bars = 20 µm.

To identify proteins specific to the ER bodies, each fraction from GFP-h and nai1 was subjected to SDS–PAGE with CBB staining. A 65-kDa protein was concentrated in the P1 fraction from GFP-h but it was not detected in the P1 fraction from nai1 (indicated by arrowheads in Figure 3a). This result suggested that the 65-kDa protein was accumulated in the ER bodies. We detected no other proteins that were specific to the ER body-rich P1 fraction on the gel.

Figure 3.

A 65-kDa protein specific to the ER body-rich P1 fraction is PYK10.

(a) Each fraction obtained in Figure 2 was subjected to SDS–PAGE and stained with CBB. Note that the 65-kDa protein (arrowheads) was concentrated in the P1 fraction of GFP-h, but not in the P1 fraction of nai1. Molecular masses of marker proteins (M) are given on the left in kDa. The molecular mass markers were subjected in both sides of the gel. G, GFP-h.

(b) The N-terminal amino acid sequence of the 65-kDa proteins was determined (shaded amino acids). The 65-kDa protein was β-glucosidase, PYK10 (GenBank CAB50792) (Nitz et al., 2001). A cleavage site of a signal peptide is indicated by an arrow. The ER-retention signal KDEL is shown in boldface. Three possible glycosylation sites are indicated by ψ. Two peptide sequences (IM and CM) used for the preparation of anti-PYK10 antibodies are underlined and double underlined, respectively.

We determined the N-terminal amino acid sequence of the 65-kDa protein to be DGPVXPPSNKLARASFP (X, not determined). A database search showed that the 65-kDa protein was a β-glucosidase named PYK10 (GenBank accession number CAB50792) (Figure 3b). The N-terminal sequence of PYK10 revealed that the signal peptide was cleaved off on the carbonyl side of Ala-24 (indicated by an arrow in Figure 3b). The cleavage site was consistent with that predicted by application of the rules described in (Nielsen et al., 1997). Notably, PYK10 has an ER-retention signal sequence (KDEL) at its C terminus (Denecke et al., 1992).

The PYK10 gene is located in the top of chromosome 3 according to an Arabidopsis genome sequence search. This indicates that the PYK10 gene was not identical to the NAI1 gene, which was mapped to the middle of chromosome 2 as described above.

PYK10 is a major protein in seedlings, but not in rosette leaves

We raised anti-PYK10(IM) antibodies that are specific to an internal amino acid sequence (underlined in Figure 3b) and anti-PYK10(CM) antibodies that are specific to the C-terminal sequence (double-underlined in Figure 3b). Immunoblot analysis of Arabidopsis seedlings showed that anti-PYK10(IM) and anti-PYK10(CM) antibodies specifically recognized a 65-kDa protein in the wild type (Figure 4, lanes 1, 4) and GFP-h (Figure 4, lanes 2, 5). PYK10 was hardly detected in nai1 seedlings (Figure 4, lanes 3, 6). Longer exposure did not give any signals on the blot of nai1 seedlings (data not shown). This indicated that PYK10 is not expressed in nai1 seedlings. These results confirmed that the 65-kDa protein is PYK10.

Figure 4.

Immunoblot analysis showing that anti-PYK10(IM) and anti-PYK10(CM) antibodies specifically recognize the 65-kDa protein.

Extracts from 6-day-old seedlings of wild type (Wt; lanes 1, 4, 9), GFP-h (G; lanes 2, 5, 7, 10) and nai1 (lanes 3, 6, 8, 11) were subjected to an immunoblot analysis with anti-PYK10(IM) (lanes 1–3), anti-PYK10(CM) (lanes 4–6), anti-GFP (lanes 7, 8) and anti-BiP (lanes 9–11) antibodies. The molecular masses are given on the left in kDa.

The expression levels of GFP-HDEL (Figure 4, lanes 7, 8) or BiP, a member of the Hsp70 family in ER (Figure 4, lanes 9–11) were almost the same among wild-type, GFP-h and nail seedlings. This indicates that GFP-HDEL or BiP is not involved in the formation of ER bodies.

The organ specificity of PYK10 expression is shown in Figure 5. In wild-type and GFP-h seedlings, PYK10 was a major protein in cotyledons (Figure 5a, lanes 1, 2, 10, 11), hypocotyls (Figure 5a, lanes 4, 5, 13, 14) and roots (Figure 5a, lanes 7, 8, 16, 17). Especially, PYK10 was most abundant in hypocotyls and roots (indicated by an arrowhead in Figure 5a, lanes 13, 14, 16, 17). PYK10 was not detected in nai1 seedlings (Figure 5a, lanes 3, 6, 9, 12, 15, 18). In contrast to the seedlings, rosette leaves had no PYK10 (Figure 5b, lanes 1–3, 7–9). However, PYK10 was accumulated in roots of 15-day-old wild-type and GFP-h plants (Figure 5b, lanes 4, 5, 10, 11) but not in nai1 roots (Figure 5b, lanes 6, 12). It is noteworthy that PYK10 expression was completely consistent with the development of ER bodies (Figure 1). This result implied that PYK10 was the main component of the ER bodies.

Figure 5.

PYK10 is a major protein in Arabidopsis seedlings, but scarcely detected in rosette leaves.

(a) Six-day-old seedlings of wild type (Wt; lanes 1, 4, 7, 10, 13, 16), GFP-h (G; lanes 2, 5, 8, 11, 14, 17) and nai1 (lanes 3, 6, 9, 12, 15, 18) were separated into cotyledons (lanes 1–3, 10–12), hypocotyls (lanes 4–6, 13–15) and roots (lanes 7–9, 16–18). Extracts from each organ were subjected to SDS–PAGE followed by an immunoblot analysis with anti-PYK10 (IM) (lanes 1–9) antibodies or CBB staining (lanes 10–18).

(b) Extracts were prepared from rosette leaves (lanes 1–3, 7–9) and roots (lanes 4–6, 10–12) of 15-day-old wild-type (Wt; lanes 1, 4, 7, 10), GFP-h (G; lanes 2, 5, 8, 11) and nai1 (lanes 3, 6, 9, 12) plants. Extracts were subjected to SDS–PAGE followed by an immunoblot analysis with anti-PYK10(IM) (lanes 1–6) or CBB staining (lanes 7–12). Arrowheads indicate PYK10. The molecular masses are given on the left in kDa.

PYK10 is specifically localized in ER bodies

Figure 6(a) shows immunofluorescent staining of ER bodies in the P1 fraction with anti-PYK10(IM) antibodies. After mild fixation, ER bodies exhibited GFP fluorescence (Figure 6b,e). The immunofluorescence with anti-PYK10(IM) antibodies perfectly corresponded to the GFP fluorescence (Figure 6c). Almost all ER bodies were stained with anti-PYK10(IM) antibodies. This means that the ER bodies in the P1 fraction were uniform with respect to the accumulation of PYK10. Pre-immune serum gave no signal (Figure 6d) indicating that the labeling was fully specific to the PYK10 localization.

Figure 6.

PYK10 is specifically localized in the ER bodies.

The P1 fraction from 8-day-old GFP-h seedlings was fixed and stained with anti-PYK10(IM) antibodies (a) or pre-immune serum (d). GFP fluorescence of the same fields was observed (b, e). (a) and (b) are merged in (c). (d) and (e) are merged in (f). Note that all the ER bodies were stained with anti-PYK10 antibodies. Bars = 10 µm. The epidermal cells of 3- (g) and 5-day-old (h) cotyledons of wild type were subjected to immunocytochemistry with anti-PYK10(CM) (g) and anti-PYK10(IM) (h) antibodies. ER bodies were specifically stained with both antibodies. Bars = 1 µm.

Immunocytochemistry with cotyledons of wild-type plants showed that rod-shaped ER bodies were densely labeled with anti-PYK10(CM) (Figure 6g) and anti-PYK10(IM) (Figure 6h) antibodies. No signals were detected in other compartments in the cells. The signals were limited to the epidermal cells of cotyledons (data not shown). These observations indicate that PYK10 was specifically localized in the ER bodies.

Selective concentration of PYK10 in ER bodies

PYK10 was not detected in nai1 seedlings, while GFP-HDEL was accumulated at the same level in GPF-h and nai1 seedlings (Figure 4). This result implied that PYK10 was more specific to ER bodies than GFP-HDEL. To clarify whether PYK10 was selectively accumulated in the ER bodies, we compared the subcellular distribution of PYK10 and GFP-HDEL, both of which have ER-retention signals.

Figure 7 shows that PYK10 was most concentrated in the P1 fraction from GFP-h (lane 2), although PYK10 was also distributed in the P8 and P100 fractions (lanes 3, 4). In nai1, no PYK10 signal was detected (Figure 7, lanes 6–10). GFP-HDEL was distributed in all fractions from both GFP-h (Figure 7, lanes 12–15) and nai1 (Figure 7, lanes 17–20). PYK10, which was detected in the P1 fraction, was derived from the ER bodies (Figures 2c and 6a). It appears that PYK10 in the P8 fraction was derived from the smaller ER bodies (Figure 2e), and the lower amount of PYK10 was from microsomes (Figure 2g).

Figure 7.

PYK10 is concentrated more selectively in the ER bodies than GFP-HDEL.

Eight-day-old seedlings of GFP-h (lanes 1–5, 11–15) and nai1 (lanes 6–10, 16–20) were subjected to differential centrifugation. Equal volumes of each fraction were subjected to SDS–PAGE followed by an immunoblot analysis with anti-PYK10(IM) (lanes 1–10) and anti-GFP (lanes 11–20) antibodies: lanes 1, 6, 11, 16 for total homogenates; lanes 2, 7, 12, 17 for the P1 fraction; lanes 3, 8, 13, 18 for the P8 fraction; lanes 4, 9, 14, 19 for the P100 fraction, lanes 5, 10, 15, 20 for the S100 fraction. Molecular masses are indicated on the left.

PYK10 was more concentrated in the P1 (ER body) fraction than the P100 (microsome) fraction (Figure 7, lanes 2, 4) while GFP-HDEL was less concentrated in the P1 fraction than in the P100 fraction (Figure 7, lanes 12, 14). This result indicates that the concentration of PYK10 in the ER bodies was more selective than that of GFP-HDEL. Subcellular fractionation of wild-type plants gave the same distribution pattern of PYK10 (data not shown). Thus, over-expression of GFP-HDEL did not affect the subcellular distribution of PYK10.

The amount of GFP-HDEL in the P1 fraction of nai1 (Figure 7, lane 17) was less than the amount in the P1 fraction of GFP-h (Figure 7, lane 12). This difference may result from the loss of ER bodies in the P1 fraction of nai1. GFP-HDEL was detected as two bands with molecular masses of 30 and 28 kDa (Figure 7, lanes 11, 16). The 30-kDa form was most concentrated in the P100 (microsome) fraction (Figure 7, lanes 14, 19). The 28-kDa form was detected only in the S100 fraction (Figure 7, lanes 15, 20) which contained soluble vacuolar proteins (Hayashi et al., 2001). The 30-kDa form, GFP-HDEL with an HHHHHHDEL sequence at the C terminus, was recognized by anti-his-tag antibodies, while the 28-kDa form was not (data not shown). GFP-HDEL may be converted into the 28-kDa GFP by removal of the C-terminal sequence within the vacuoles.

PYK10 has three high-mannose oligosaccharides but no complex one

Three potential N-glycosylation sites were found at positions 60, 461, and 494 of the PYK10 polypeptide (indicated by ψ in Figure 3b). To determine whether PYK10 is glycosylated or not, total homogenates from seedlings were treated with glycosidases. N-glycosidase F (PNGase F) treatment increased the mobility of PYK10 on an SDS-gel, indicating that PYK10 had N-linked oligosaccharides (Figure 8a, lanes 1, 3). Endoglycosidase H (Endo H) treatment also increased the mobility of PYK10 to approximately the same extent as PNGase F, indicating that all of the N-linked oligosaccharides were sensitive to Endo H (Figure 8a, lanes 2, 4). These results show that PYK10 has high-mannose oligosaccharides. The small difference in mobility between PNGase F-treated and Endo H-treated PYK10 may be due to the fact that Endo H does not remove the asparagine-linked N-acetylglucosamine whereas PNGase F does.

Figure 8.

PYK10 has three high-mannose oligosaccharides, but not a complex one.

(a) Extract prepared from 8-day-old seedlings was incubated in the absence (–) or presence (+) of N-glycosidase F (PNGase F, 10 unit ml−1) at pH 7.5 (lanes 1, 3) and endoglycosidase H (Endo H, 0.05 unit ml−1) at pH 5.5 (lanes 2, 4). Reaction mixtures were subjected to an immunoblot analysis with anti-PYK10(IM) antibodies.

(b) The extract was incubated with 1 unit ml−1 PNGase F (lanes 1–7) or 0.005 unit ml−1 Endo H (lanes 8–14) for the indicated periods. ψ3, PYK10 with three oligosaccharides; ψ2, PYK10 with two oligosaccharides; ψ1, PYK10 with one oligosaccharide; ψ0, PYK10 without oligosaccharides. Molecular masses are indicated on the left.

By varying the incubation time with both PNGase F and Endo H, four separate bands were obtained for PYK10 (bands ψ3, ψ2, ψ1 and ψ0 in Figure 8b). The ψ0 form appeared to be free of oligosaccharides and had a mass of 57 kDa, which is consistent with the deduced mass for PYK10 itself, 57261 Da. These results indicated that PYK10 had three high-mannose oligosaccharides. The presence of high-mannose oligosaccharides in PYK10 showed that PYK10 did not pass through the Golgi apparatus. This provides further evidence that the ER bodies were directly derived from the ER but not the post-Golgi compartment.

Discussion

Mechanism of selective concentration of PYK10 in ER bodies

We showed that the β-glucosidase, PYK10, was the main component of the ER bodies. The localization of PYK10 in the ER bodies was more selective than that of GFP-HDEL, which may be passively drawn into the ER bodies. What is the mechanism responsible for the selective concentration of PYK10 in the ER bodies? PYK10 was abundant in the seedlings (Figure 5a) and is one of the proteins whose levels increase the most after seed germination (Gallardo et al., 2001). It is possible that PYK10 molecules, by forming aggregations in the ER cisternae, contribute to the formation of the ER bodies. Some examples of β-glucosidase aggregations in plant cells have been reported. In a certain maize genotype, β-glucosidase occurs as a large and insoluble aggregate (Blanchard et al., 2001; Esen and Blanchard, 2000). β-Glucosidases of flax (Fieldes and Gerhardt, 1994) and oat (Kim et al., 2000; Nisius, 1988) occur in high molecular mass aggregations ranging from 245 to 1200 kDa.

In many plant species, various ER-derived structures are formed by aggregation of proteins in ER. Rice and maize develop ER-derived protein bodies in seed cells for selective accumulation of insoluble prolamin and zein, respectively (Herman and Larkins, 1999; Okita and Rogers, 1996). The protein bodies have unique functions independent of the ER cisternae. PAC vesicles are formed by aggregation of storage protein precursors within the ER cisternae in maturing seeds of pumpkin (Hara-Nishimura et al., 1998). PYK10, which has a KDEL sequence at the C-terminus, was the main component of the ER bodies. Another example of a KDEL-tailed protein being the major component of an ER-derived compartment is cysteine endopeptidase, which is almost the only component of ricinosomes in the endosperm of castor bean (Schmid et al., 2001).

It is also possible that PYK10 mRNA is segregated on the ER membrane. The mRNA of rice prolamin has been shown to target the prolamin-protein body membrane (Choi et al., 2000; Li et al., 1993), which leads to the efficient segregation of proteins in the ER. Ultrastructural studies showed that the ribosomes are present on the membrane of ER bodies (Hayashi et al., 2001). This means that translation and translocation of proteins occur on the membrane of the ER bodies. Thus, PYK10 mRNA might target the ER-body membrane.

Possible biological function of PYK10 and ER bodies

We showed that PYK10 is the main component of ER bodies. Therefore, the biological function of ER bodies is probably linked with the function of PYK10. ER bodies are induced in rosette leaves when treated with MeJA or wound stress (Matsushima et al., 2002). We suggest that ER bodies have a role in defending against chewing insects and in responding in some way to other wound stresses. PYK10 may also have some defense function.

ER bodies did not develop in rosette leaves (Figure 1g). This means that ER bodies are not essential for fundamental growth. Consistent with this, the nai1 mutant had no visual phenotype when grown under normal conditions. ER bodies appear to be needed under stress conditions. To understand the function of ER bodies at the molecular level, it is necessary to identify the function of PYK10.

PYK10 is a β-glucosidase (β-d-glucoside glucohydrolase, EC 3.2.1.21). β-Glucosidase catalyzes the hydrolysis of aryl and alkyl β-d-glucosides as well as glucosides with a carbohydrate moiety such as cellobiose and other β-linked oligosaccharides (Esen, 1993). Three physiological roles for β-glucosidase have been reported: involvement in plant defense against pests (Bones and Rossiter, 1996; Poulton, 1990), phytohormone activation (Brzobohaty et al., 1993; Schliemann, 1984; Smith and van Staden, 1978), and cell wall catabolism (Dharmawardhana et al., 1995; Leah et al., 1995). The defense function of β-glucosidase against pathogens is mostly accomplished by producing toxic compounds as a result of cleavage of glucosinolates by thioglucosidases (Rask et al., 2000) or hydrolysis of cyanoglucosides (e.g. linamarin or amygdalin) by cyanogenic glucosidases (Poulton, 1990).

One of the homologs of PYK10 in Arabidopsis, BGL1 (GenBank CAC19786), has been suggested to be involved in the defense responses against herbivorous insects (Stotz et al., 2000). BGL1 has a putative ER-retention signal (REEL) and 70% identity with PYK10. The expression level of BGL1 in rosette leaves increased more than 10-fold after feeding by a herbivore insect, the diamondback moth, or by MeJA treatment (Stotz et al., 2000). Another PYK10 homolog (68% identity) is known to function as a zeatin-O-glucoside-degrading β-glucosidase in Brassica napus (GenBank CAA57913). This β-glucosidase is suggested to provide young tissues with active cytokinin from inactive cytokinin-O-glucosides (Falk and Rask, 1995). However, this β-glucosidase has no ER-retention signal and may function in a subcellular site different from PYK10.

Myrosinase is a thioglucosidase that catalyzes the hydrolysis of glucosinolates (Bones and Rossiter, 1996; Rask et al., 2000). ER body-like structures, ‘dilated cisternae’, have been reported to exist, mainly in the Brassicaceae family (Behnke and Eschlbeck, 1978; Iversen, 1970b). Several attempts have been made to correlate the presence of dilated cisternae with myrosinase (Iversen, 1970a). However, no direct evidence for such a correlation has been presented (Thangstad et al., 1990, 1991). Some researchers classified PYK10 as a myrosinase (Nitz et al., 2001), though PYK10 is distantly related to myrosinases (approximately 45% identity). Myrosinases identified so far have no ER-retention signals. It is unknown whether PYK10 has myrosinase activity or not. The aglycon-binding residues conserved in myrosinases are not conserved in PYK10 (Rask et al., 2000).

Many β-glucosidases are linked with plant defense systems (Babcock and Esen, 1994; Cicek and Esen, 1998; Gus-Mayer et al., 1994; Hughes et al., 1992; Oxtoby et al., 1991; Rask et al., 2000). In an undamaged oat cell, β-glucosidases and their specific substrate, avenocoside, are stored in different compartments of the cells (plastids and vacuoles, respectively), and the enzyme and the substrate are separated by intact membranes. However, when plant tissues and membranes are injured, the enzyme and substrate come into contact and toxic compounds (aglycons) are produced (Nisius, 1988). We previously reported that ER bodies start to fuse with lytic vacuoles when the cells are stressed with a concentrated salt solution (Hayashi et al., 2001). This fusion was also observed when epidermal cells were injured (data not shown), perhaps as a result of osmotic stresses. This means that ER bodies release their contents into vacuoles under stressed conditions. In this way, PYK10 in the ER bodies will come into contact with its natural substrates to produce the toxic compounds.

The nai1 mutant can be an ideal tool to clarify the physiological function of PYK10 and ER bodies. Comparison of HPLC chromatograms of homogenates of plant tissues of wild type and nai1 would help to identify the natural substrates of PYK10.

Experimental procedures

Plant materials and growth conditions

Previously, we transformed Arabidopsis plants (ecotype Columbia) with a p35s::sp-gfp-hdel gene encoding SP-GFP-HDEL composed of the signal peptide of pumpkin 2S albumin and GFP followed by an ER-retention signal, HDEL (Hayashi et al., 2001; Mitsuhashi et al., 2000). We designated this transgenic plant as GFP-h and used it for the experiments. Seeds were sown in soil or onto 0.5% Gellan Gum (Wako, Tokyo, Japan) with Murashige–Skoog's medium (Murashige and Skoog, 1962) and were grown at 22°C under continuous light conditions.

Isolation of nai1 mutant

GFP-h seeds were mutagenized by soaking them for 16 h in 0.1 or 0.2% (v/v) methanesulfonic acid ethyl ester (Sigma, Tokyo, Japan) and then washed for 11 h in running water. To obtain the M2 lines, each of which consisted of seeds derived from a single M1 plant, M2 seeds were collected from individual M1 plants after self-fertilization. Screening was carried out with some of the seeds from each M2 line. Six- or 7-day-old seedlings derived from M2 seeds were examined with a fluorescence microscope and an M2 line that had no ER bodies was selected. We named the mutant nai1. nai1 was crossed with Arabidopsis ecotype Landsberg erecta, and the genotype of nai1 in the F2 progeny were analyzed using a combination of cleaved amplified polymorphic sequence and simple sequence length polymorphism markers (Bell and Ecker, 1994; Konieczny and Ausubel, 1993) with data obtained from TAIR (http://www.arabidopsis.org).

Subcellular fractionation

The seedlings (1.2–2.0 g) were chopped with a razor blade in a Petri dish on ice in 3.6–6.0 ml of chopping buffer that contained 50 mm Hepes-NaOH (pH 7.5), 5 mm EDTA, 0.4 m sucrose and protease inhibitor cocktail (one tablet per 50 ml, Boehringer Mannheim, Tokyo, Japan). The homogenate was filtered through cheesecloth. An aliquot of the filtrate was used as the total fraction. Two milliliters of the filtrate was centrifuged at 1000 g at 4°C for 20 min. The pellet was re-suspended in 500 µl of chopping buffer and was designated as the P1 fraction. The supernatant was centrifuged again at 8000 g at 4°C for 20 min. The pellet was re-suspended in 500 µl of chopping buffer and was designated as the P8 fraction. The supernatant was ultracentrifuged at 100 000 g at 4°C for 1 h. The pellet was re-suspended in 500 µl of chopping buffer and was designated as the P100 fraction. The supernatant was designated as the S100 fraction and the volume of the fraction was measured. Each fraction was subjected to SDS–PAGE with CBB staining for Figure 3(a). To compare the relative distribution of PYK10 and GFP-HDEL, 0.8% (v/v) of each fraction of T, P1, P8, P100 and S100 was also subjected to SDS–PAGE followed by an immunoblot analysis with specific antibodies against PYK10 and GFP for Figure 7.

Determination of an N-terminal amino acid sequence

Each subcellular fraction was subjected to SDS–PAGE and then transferred electrophoretically to a GVHP membrane (0.22 µm; Nihon Millipore, Tokyo, Japan). After CBB staining, a 65-kDa protein was cut out from the blot and subjected to automatic Edman degradation on a peptide sequencer (model 492, Applied Biosystems, Foster City, CA).

Specific antibodies

Two peptides derived from PYK10, CSNHLEKPDPSKPRWMQDS (IM) and CDGYKNRFGLYYVDFKNNLTRYEKESGKYY (CM), were chemically synthesized with a peptide synthesizer (model 431 A; Applied Biosystems, Tokyo, Japan). These peptides were cross-linked to BSA by 3-maleimidobenzoic acid N-hydroxysuccinimide ester (Sigma). The peptide-BSA conjugates were injected into a rabbit subcutaneously with complete Freund's adjuvant. After 3 weeks, two booster injections with incomplete adjuvant were given at 7-day intervals. One week after the booster injections, blood was drawn and the antibodies were prepared. We also used antibodies against GFP (Clontech, Palo Alto, CA, USA) and BiP (Hatano et al., 1997).

Extraction of proteins from plants

Extracts were prepared from 6-day-old seedlings of wild type, GFP-h and nai1 for the immunoblot analysis shown in Figure 4. A whole seedling was homogenized in 100 µl extraction buffer; 100 mm Tris–HCl (pH 6.8), 4% (w/v) SDS, 20% (v/v) glycerol and 10% (v/v) 2-mercaptoethnol. The extracts (7 µl) were subjected to SDS–PAGE followed by the immunoblot analysis.

To study the organ-specific expression of PYK10, we harvested three pairs of cotyledons, 20 hypocotyls and 15 roots from 6-day-old seedlings and homogenized them in 100 µl of the extraction buffer for the immunoblot analysis shown in Figure 5(a). We also harvested the third and fourth rosette leaves and roots from one whole 15-day-old plant and homogenized them in 100 µl of the extraction buffer for the immunoblot analysis shown in Figure 5(b). Twenty-microliter extracts were subjected to SDS–PAGE followed by the immunoblot analysis.

Immunoblot analysis

After SDS–PAGE, the proteins were transferred electrophoretically to a GVHP membrane. The membrane was thoroughly dried for blocking and then incubated in Tris-buffered saline (pH 7.5) plus 0.05% (v/v)Tween 20 with the antibodies for 1 h. Dilutions of the antibodies were as follows; anti-PYK10(IM) 1 : 30 000 (v/v), anti-PYK10(CM) 1 : 30 000, anti-GFP 1 : 10 000, and anti-BiP 1 : 10 000. Horseradish peroxidase-conjugated goat antibodies against rabbit IgG (Amersham Pharmacia Biotech, Tokyo, Japan) were diluted (1 : 5000) to be used as second antibodies. Immunodetection was performed with an ECL kit (Amersham Pharmacia Biotech).

Fluorescence microscopy

Plants in water on glass slides were examined with a fluorescence microscope (Axiophot, Carl Zeiss, Jena, Germany). We used filter sets for GFP fluorescence (an excitation filter BP450-490, a dichroic mirror FT510 and a barrier filter BP515-565; Carl Zeiss) and the sets for rhodamine fluorescence (an excitation filter BP546, a dichroic mirror FT580 and a barrier filter BP590; Carl Zeiss). The fluorescent images were analyzed with Adobe Photoshop (Adobe Systems, Tokyo, Japan).

Immunofluorescent staining

The P1 pellet obtained from 8-day-old GFP-h seedlings was re-suspended in fixing buffer containing 1.0% (w/v) paraformaldehyde, 50 mm Hepes-NaOH (pH 7.5), 5 mm EDTA and 0.4 m sucrose. The P1 fraction in the buffer was incubated at 4°C on poly l-lysine-coated slides (POLY-PREP SLIDES; Sigma Diagnostics, St. Louis, USA). After fixing, the slides were incubated in PBS (137 mm NaCl, 2.7 mm KCl, 10 mm Na2HPO4 and 1.8 mm KH2PO4) plus 0.5% (v/v) Triton X-100 for 6 h. The slides were incubated in blocking buffer of PBS containing 0.1% Triton X-100 and 5% sheep serum (CHEMICON, Temecula, CA, USA) for 10 h. The slides were incubated with the anti PYK10 (IM) antibodies or the pre-immune serum (diluted 1 : 1000 in the blocking buffer) for 6 h at 4°C. After washing with blocking buffer for 15 min, the slides were incubated with secondary antibodies, Alexa FluorTM 546 goat antirabbit IgG (H + L) (Molecular Probes, Eugene, USA) for 3 h. Dilution of the secondary antibodies was 1 : 1000. After washing for 15 min, the slides were examined with a fluorescence microscope.

Immunocytochemistry

Three- or 5-day-old cotyledons were frozen with a high-pressure freezing machine (HPM010S, Bal-Tec, Balzers, Liechtenstein), as described (Craig and Staehelin, 1998). The frozen samples were dehydrated for 2 days and at −85°C with acetone and embedded in LR white resin. Immunogold labeling procedures were essentially the same as those described previously (Nishimura et al., 1993), except for the use of anti-PYK10 antibodies. Ultrathin sections were mounted on Formvar-coated nickel grids. The sections were treated with blocking solution of PBS containing 2% normal goat serum (Vector Laboratories, Burlingame, CA, USA) for 30 min and then incubated with anti-PYK10 antibodies (diluted 1 : 1000) for 12 h at 4°C. After washing with PBS, sections were incubated with AuroProbe EM Anti-Rabbit IgG (H + L) (diluted 1 : 30; 15 nm; Amersham Pharmacia Biotech) for 30 min. After washing, sections were stained with a solution of 1% (w/v) uranyl acetate and 0.4% (w/v) lead citrate. All sections were examined with a transmission electron microscope (model 1200EX; JOEL, Tokyo, Japan) at 80 kV.

Deglycosylation with N-glycosidase F and endoglycosidase H

Extracts from 8-day-old young seedlings were boiled in the presence of 0.1% SDS and 100 mm 2-mercaptoethanol for 5 min for denaturation. The extract (100–200 µl) was mixed with 0.1–1.0 unit of N-glycosidase F (Boehringer Mannheim) in 50 mm Tris–HCl (pH 7.5), 0.5% Triton-X-100 and 100 mm 2-mercaptoethanol. The extract was also mixed with 0.005–0.0005 unit of endoglycosidase H (Boehringer Mannheim) in 100 mm sodium acetate (pH 5.5), 0.5% Triton-X-100 and 100 mm 2-mercaptoethanol. The reaction mixtures were incubated for 11, 23, 45, 90, 180 min or overnight (14–20 h) at 37°C. Deglycosylated proteins were subjected to SDS–PAGE followed by an immunoblot analysis with anti-PYK10(IM) antibodies.

Acknowledgements

We thank Ms. Y. Makino and Ms. S. Ohsawa of NIBB for their helpful support with peptide sequencing and synthesis, respectively. This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (nos. 10182102, 12138205 and 12304049) and for a postdoctoral fellowship to R.M. from the Japan Society for the Promotion of Science (no. 14001468).

Ancillary