• Open Access

Development of transgenic rice seed accumulating a major Japanese cedar pollen allergen (Cry j 1) structurally disrupted for oral immunotherapy

Authors


* Correspondence (fax +81 29 838 8379; e-mail: takaiwa@nias.affrc.go.jp)

Summary

Rice seed-based edible vaccines expressing T-cell epitope peptides derived from Japanese cedar major pollen allergens have been used to successfully suppress allergen-specific Th2-mediated immunoglobulin E (IgE) responses in mouse experiments. In order to further expand the application of seed-based allergen-specific immunotherapy for controlling Japanese cedar pollinosis, we generated transgenic rice plants that specifically express recombinant Cry j 1 allergens in seeds. Cry j 1 allergens give low specific IgE-binding activity but contain all of the T-cell epitopes. The allergens were expressed directly or as a protein fusion with the major rice storage protein glutelin. Fusion proteins expressed under the control of the strong rice endosperm-specific GluB-1 promoter accumulated in rice endosperm tissue up to 15% of total seed protein. The fusion proteins aggregated with cysteine-rich prolamin and were deposited in endoplasmic reticulum-derived protein body I. The production of transgenic rice expressing structurally disrupted Cry j 1 peptides with low IgE binding activity but spanning the entire Cry j1 region can be used as a universal, safe and effective tolerogen for rice seed-based oral immunotherapy for cedar pollen allergy in humans and other mammals.

Introduction

Japanese cedar (Cryptomeria japonica) pollen is a potent seasonal aeroallergen that is spread over most areas of Japan in the early spring. C. japonica pollen causes cedar pollinosis with rhinitis, asthma and conjunctivitis as clinical symptoms, which can result in severe health problems. About 20% of the Japanese population is currently afflicted (Kaneko et al., 2005; Takaiwa, 2007), and more than half of the Japanese population has specific circulating immunoglobulin E (IgE) for cedar pollen allergens. It is interesting to note that cedar pollinosis has also been reported in dogs with atopic dermatitis in Japan (Sasaki et al., 1995). Approximately 20% of atopic dogs have specific IgE to Japanese cedar pollen allergen (Masuda et al., 2000). The economic costs associated with this disease are already high and are expected to increase as healthcare costs increase and as environmental adjuvants, such as air pollutants, exacerbate symptoms. The development of an affordable and clinically effective vaccine for controlling the disease is thus a high priority for public health.

Two major allergens, designated Cry j 1 and Cry j 2, have been isolated from Japanese cedar pollen and characterized in detail (Yasueda et al., 1983; Sakaguchi et al., 1990; Komiyama et al., 1994; Namba et al., 1994; Sone et al., 1994, 1998). More than 90% of cedar pollinosis patients have specific IgE to both allergens (Hashimoto et al., 1995). Mature Cry j 1 is a basic glycoprotein with pectate lyase activity, consisting of 353 amino acid residues (Sone et al., 1994).

Allergen-specific immunotherapy is the only treatment that can provide a cure for cedar pollinosis. Conventional allergen-specific immunotherapy has been conducted by repeated subcutaneous administration of increasing doses of crude allergen preparations (intact allergen) throughout a period of 3–5 years. This treatment is plagued by high patient dropout rates and is associated with side-effects such as anaphylactic shock due to the presence of IgE-binding activity, and pain caused by inflammation. A safe, easy and convenient treatment would thus be a boon to public health. Peptide immunotherapy using dominant T-cell epitopes has been shown to be a safe and effective treatment for the control of IgE-mediated allergic diseases because of the absence of specific tertiary structures or B-cell epitopes recognized by specific IgE (Haselden et al., 2000; Crameri and Rhyner, 2006).

Furthermore, taking advantage of the mucosal immune system by oral administration as an alternative to subcutaneous injection may provide a safe and convenient treatment. Oral administration of antigens through the mucosal route has shown promise for inducing a systemic immune response by immune tolerance, in addition to the mucosal response (Walker et al., 1995). Mucosal tolerance is believed to be achieved by reduction of the allergen-specific T-cell response via anergy or apoptosis of specific T cells, or active suppression of regulatory T cells, resulting in suppression of allergen-specific IgE production (Faria and Weiner, 2005).

A rice seed-based edible vaccine expressing mouse T-cell epitopes derived from Cry j 1 and Cry j 2 has successfully inhibited allergen-specific Th2-mediated IgE responses in mouse models (Takagi et al., 2005a). This result strongly supports the clinical feasibility of using allergen-derived peptide expressed in rice seed in other mammals, including humans, and indicates that rice seed-based peptide vaccines can be used as a new allergen-specific immunotherapy for treatment of airway allergies (Hiroi and Takaiwa, 2006).

Transgenic plants are an emerging technology for the production of pharmaceuticals, vaccines and antibodies with many advantages over cell culture, microbial or animal production systems, including direct delivery, extremely low risk of contamination with animal or human pathogens, and there is no need for extraction or processing, thus providing a cost-effective production platform (Giddings et al., 2000; Walmsley and Arntzen, 2000; Stoger et al., 2005). The use of rice seed production as a bioreactor has several advantages over other cereals. Rice is consumed as a staple food by more than half the world's population, and recombinant proteins expressed in rice seeds are highly stable at room temperature (Stoger et al., 2000). Furthermore, it has been recently demonstrated that the edible rice endosperm is a good production platform for foreign products (Takaiwa et al., 2007).

A critical factor for determining the economic viability of a plant production system is the yield of recombinant proteins in transgenic plants. The general approach for enhancing recombinant protein expression is to combine the effects of strong tissue-specific promoters, optimization of codons and protein stability, targeting to a suitable subcellular compartment, and making translational fusions with an appropriate carrier protein (Twyman et al., 2003; Streatfield, 2007).

Glutelin, a major seed-storage protein in rice, accounts for 60–70% of total seed protein and is encoded by a multigene family (Takaiwa et al., 1999). Rice glutelin is synthesized as a 57-kDa precursor that is subsequently processed into 37-kDa acidic and 20-kDa basic subunits at the conserved Asn-Gly site. Cleavage takes place within protein storage vacuoles known as protein body II (PB-II) (Krishnan and White, 1995). We have recently reported that under the control of the strong rice endosperm-specific GluB-1 promoter (Qu and Takaiwa, 2004), a hypocholesterolemic peptide, lactostatin, and an antihypertension peptide, novokinin, were highly accumulated in transgenic rice seed as glutelin A2 (GluA-2) fusion proteins (Wakasa et al., 2006; Yang et al., 2006). These fusion proteins are efficiently processed into mature acidic and basic subunits, indicating that glutelin is a potential carrier for heterogeneous bioactive peptide(s).

The clinical use of a rice seed-based edible vaccine with low IgE binding activity but that retains immunogenicity for humans and other mammals affected by cedar pollinosis (Masuda et al., 2004), will depend on the accumulation of allergen in rice seed at a pharmacologically appropriate level. In this study, we expressed Cry j 1 either as an independent gene cassette or as a fusion molecule as an alternative to T-cell epitope peptides. Our preliminary results showed that higher levels of accumulation were achieved by expressing recombinant Cry j 1 as a fusion protein with rice glutelin. The highest accumulation level of the fusion protein reached about 15% of total seed protein, but fusion protein precursors containing Cry j 1 with an altered structure were not post-translationally processed into mature forms and, thus, aggregated with cysteine-rich prolamins in protein body I (PB-I) of seed endosperm tissues.

Results

Construction of expression vectors and generation of transgenic rice plants

The coding sequence of the mature full-length Cry j 1 gene was optimized for translation according to the codon bias of rice seed storage protein genes. The gene was ligated downstream of the 2.3 kb GluB-1 promoter (Qu and Takaiwa, 2004) containing a signal peptide sequence (Figure 1a, pJ1full). To disrupt the native conformational structure of Cry j 1 required for recognition by Cry j 1-specific IgE (Sakaguchi et al., 2001), the coding region was divided into two parts to reduce IgE binding activity. Nucleotide sequences encoding the N-terminal (amino acid residues 1–195) and C-terminal halves (amino acid residues 135–353) of Cry j 1 were linked to the 2.3 kb GluB-1 promoter containing a signal sequence (Figure 1a, pJ1N-half and pJ1C-half). To stabilize and increase transgene products in endoplasmic reticulum (ER) (Takagi et al., 2005b), an ER retention KDEL signal was attached to the C-termini of all three constructs, followed by 0.65 kb of the GluB-1 terminator. The resulting constructs were cloned into binary vector pGPTV-35S-HPT (Figure 1a). Construct pJ1full has native conformational structure was thus used as a control for comparison with fragmented Cry j 1 peptide and peptide fusions with disrupted structures.

Figure 1.

Construction of expression vectors for rice transformation. (a, c) Seven expression vectors used for rice transformation. hpt, hygromycin phosphotransferase gene; pAg7, 3′ terminal region of agropine synthase gene; RB, right border; LB, left border; CaMV 35S P, cauliflower mosaic virus 35S promoter; GluB-1 P, 2.3 kb glutelin B1 promoter; GluB-1 T, glutelin B1 terminator; SP, glutelin B1 signal peptide; Cry j 1 full, full length mature Cry j 1; Cry j 1 N-half, N-terminal 195 amino acid residues of mature Cry j 1; Cry j 1 C-half, C-terminal 218 amino acid residues of mature Cry j 1; GluA2, glutelin A2; 5′UTR, 5′untranslated region of GluA2; Cry j 1 F1, 1–144 amino acid residues of mature Cry j 1 protein; Cry j 1 F2, 126–257 amino acid residues of mature Cry j 1 protein; Cry j 1 F3, 231–353 amino acid residues of mature Cry j 1 protein. Numbers indicate the position of the amino acid residues of GluA2 and Cry j 1. (b) Partial nucleotide and amino acid sequences at the modified region of GluA2 Version 1 (V1) and Version (V2). Arrow indicates the processing site for release of acidic and basic subunits; the SmaI insertion site is underlined. Numbers represent the position corresponding to the amino acid residues of GluA2.

In a second set of constructs, Cry j 1 protein was divided into three overlapping subregions (for details, see the Experimental procedures section), and the DNA sequences encoding these three regions were inserted into two modified GluA2 constructs (GluA2-V1 and GluA2-V2). GluA2-V1 has a 27 amino acid residue deletion in the C-terminal highly variable region of the acidic subunit. GluA2-V2 has a 93 amino acid residue deletion, including 66 amino acids upstream of the 27 amino acid variable region. The conserved post-translational processing site Asn282-Gly283 between the acidic and basic subunits was retained in both cases (Figure 1b). The 2.3 kb GluB-1 promoter was ligated upstream of these chimeric genes, which were followed by the GluB1 terminator (Figure 1c). At least 30 independently transformed transgenic plants were regenerated for each construct.

Expression of Cry j 1 protein in rice seed

Accumulation of Cry j 1 was measured by Coomassie Brilliant Blue (CBB) staining intensity or Western blots of 12% SDS-PAGE of proteins extracted from four seeds of individual transgenic plants for each construct. As shown in Figure 2(a), there were no Cry j 1 CBB-visible bands corresponding to the J1-full, J1N-half and J1C-half constructs, but Western blots with anti-Cry j 1 antibody showed some faint bands with molecular masses around 40, 21 and 26 kDa, which are close to the sizes predicted from construct sequences (Figure 2b). The average accumulation level from all of the transgenic plants carrying these constructs was less than 1 µg/seed (data not shown).

Figure 2.

Expression of pJ1full, pJ1N-half and pJ1C-half transgenic rice seeds. (a) SDS-PAGE analysis of total proteins extracted from non-transgenic Kita-ake (Con) control and transgenic rice seeds of pJ1full, pJ1N-half and pJ1C-half. Four seeds from independent transgenic lines for each construct were analysed by 12% SDS-PAGE. (b) Immuno-detection of Cry j 1 with anti Cry j 1 antibody. The samples are the same as (a). Arrows indicate the expressed transgene products and molecular markers shown on the right. Note that the extra faint bands around molecular masses of 18 kDa and 25 kDa are nonspecific ones present in both non-transgenic (Con) and transgenic samples.

To improve the expression of Cry j 1 in rice seed, Cry j 1 was divided into three overlapping peptide fragments (F1–F3). F1 was inserted into the C-terminal regions of the acidic subunit of modified glutelin GluA-2 constructs GluA2-V1 and GluA2-V2, and F2 and F3 were inserted into GluA2-V2 (Figure 1b,c). Analysis of representative plants containing pV1-F1, pV2-F1, pV2-F2 and pV2-F3 on 12% SDS-PAGE showed bands with molecular masses of ca. 58 kDa to 65 kDa (Figure 3a). Anti-Cry j 1 antibody reacted strongly to bands of V1-F1 and V2-F1, and weakly to V2-F3 on Western blots, but not to V2-F2 or non-transgenic rice controls (Figure 3b). The absence of an immunoreaction to V2-F2, which clearly has a 60-kDa band on the CBB stained gel, is probably because the immunogenic domain of Cry j 1 is mainly restricted to the N-terminal region. To further characterize the ~58-kDa to 65-kDa bands, we carried out Western blot analysis using anti-GluA full-length antibody. Except for native GluA, the ~58-kDa to 65-kDa bands specifically reacted to anti-GluA, confirming that they were the fusion protein precursors of V1-F1, V2-F1, V2-F2 and V2-F3 (Figure 3c). In addition, there were no apparent extra bands observed around the acidic and basic subunit regions as compared with non-transgenic rice, indicating that the Cry j 1 fusion proteins were only detected as precursors that are not subsequently processed into mature subunits.

Figure 3.

Expression of Cry j 1 peptides fused with modified GluA2. (a) SDS-PAGE analysis of total proteins extracted from non-transgenic Kita-ake (lane 1) and transgenic rice seeds of pV1-F1 (lanes 2 and 3), pV2-F1 (lanes 4 and 5), pV2-F2 (lanes 6 and 7) and pV2-F3 (lanes 8 and 9). Two representative transgenic lines from each construct were analysed by 12% SDS-PAGE. Arrows indicate the expressed fusion protein bands, and molecular markers are shown on the right. (b, c) Immuno-detection of fusion proteins with anti-Cry j 1 and GluA full-length antibodies. The samples are the same as (a).

Total seed proteins were extracted from 30 independent transgenic lines for each construct and separated by SDS-PAGE. Six to eight grains from each of the six selected independent highly expressing lines were extracted for each construct, and Cry j 1 fusion protein yield was measured using BSA as a standard. The yield for most lines ranged from 40 to 110 µg fusion protein/grain. The highest expression lines for each construct were no. 10 of pV1-F1, no. 16 of pV2-F1, no. 30 of pV2-F2 and no. 8 of pV2-F3, which accumulated up to 110 µg, 220 µg, 80 µg and 100 µg fusion protein/grain, respectively, which is equivalent to 35 µg, 85 µg, 27 µg and 40 µg allergen/grain or about 8%, 15%, 6% and 7% of total seed proteins (Figure 4 and Table 1). It should be noted that the variation in fusion protein contents within a fusion gene construct might be attributed to the results of the transgene copy number (gene dosage effect) as well as the site where the transgenes were inserted (position effect).

Figure 4.

Accumulation of fusion proteins in the mature seeds of transgenic rice plants. Six higher expression lines for each construct were selected for determining accumulation levels. Six to eight independent seeds from each line were analysed on 7% SDS-PAGE and CBB stained. BSA was used as a standard. (a) pV1-F1, (b) pV2-F1, (c) pV2-F2, (d) pV2-F3.

Figure 1.

Summary of the development of edible rice vaccines for Japanese cedar pollinosis in our laboratory

Southern blot analysis of transgenic rice plants

Transgene copy numbers for each independent transgenic line were determined by genomic Southern blot analysis. Genomic DNA was digested either with HindIII or XbaI. Since these restriction enzymes cut only once within the expression vector constructs used for transformation of rice, the number of hybridizing bands represents the transgene copy number. Two lines for each construct were chosen for the analysis. One to three copies of pV2-F1 and pV2-F3 were integrated into the rice genome, and four to six copies of pV1-F1 and pV2-F2 were integrated (Figure 5). Generally, there is no clear correlation between copy number and accumulation level, except for the pV2-F1 construct, in which a higher level of accumulation (220 µg fusion protein/grain, line no. 16) might be the result of random insertion of three transgene copies. In comparison, transgenic line no. 11 in the pV2-F1 had a level of only 84 µg fusion protein/grain with one copy (Figures 4b and 5). This result suggests that accumulation level may be partially dependent on transgene copy number.

Figure 5.

Southern blot analysis of transgenic rice plants. 5 µg genomic DNA isolated from non-transgenic Kita-ake and transgenic rice plants were digested with HindIII (a) or XbaI (b) and fractionated on a 0.8% agarose gel. hpt was used as probe. Non-transgenic Kita-ake (lane 1), transgenic rice lines 11 and 16 of pV1-F1 (lanes 2 and 3), lines 11 and 16 of pV2-F1 (lanes 4 and 5), lines 24 and 30 of pV2-F2 (lanes 6 and 7), lines 7 and 8 of pV2-F3 (lanes 8 and 9). Numbers on the right indicate DNA fragment sizes.

Analysis of endogenous glutelin precursors

Glutelin precursors increased significantly in all of the transgenic lines analysed (Figure 3a and 6a) compared with non-transgenic rice controls. Western blot analysis was performed using specific anti-GluA, anti-GluB and anti-GluC peptide antibodies to identify what types of glutelin precursors were increased. GluA and GluC precursors were increased about 4- and 15-fold, respectively, whereas GluB precursor was hardly detected (Figure 6). Although the reason for this observation is not clear at present, preliminary experiments suggest that precursors of GluA1 and GluA2 aggregate to form polymeric structures in rice seed due to a higher number of free cysteine residues than in the GluB family. GluA2 fusion protein precursors introduced into transgenic rice seed may cause an increase in inherent GluA production (most likely GluA1) to stabilize the increased fusion protein precursor by formation of polymeric aggregates. Further work will be required to clarify the plant's response to elevated levels of storage proteins.

Figure 6.

Analysis of internal glutelin precursors. Non-transgenic Kita-ake controls (lane 1), transgenic rice plant pV1-F1 (lane 2), pV2-F1 (lane 3), pV2-F2 (lane 4) and pV2-F3 (lane 5); Total proteins were extracted from seeds and separated by 12% SDS-PAGE. After blotting on to a membrane, the proteins were detected with corresponding antibodies. (a) CBB stained gel; (b–d) Detection of internal glutelins with anti-GluA, anti-GluB and anti-GluC peptide antibodies, respectively. Note that the anti-GluA peptide antibody (corresponding to the amino acid residues from 188 to 203 of GluA) reacts with internal GluA precursor and acidic subunit, as well as V1-F1 precursor (indicated by an arrow), but had no reaction to V2-related V2-F1, V2-F2 and V2-F3 precursors.

Fusion protein precursor aggregated with cysteine-rich prolamin

The fusion protein precursor was not processed into acidic and basic subunits (Figures 3 and 6), suggesting that the chimeric precursor is not transported to protein storage vacuoles, or may not be recognized by processing enzymes due to the insertion length of the Cry j 1 fragments. To determine the intracellular localization of the fusion protein precursor, we first examined the extraction conditions of V2-F1 precursor from pV2-F1 transgenic seeds. Glutelins were extracted only after the removal of globulins from endosperm (Figure 7, lanes 1 and 2), and removal of cysteine-poor prolamins after globulin extraction had no effect on V2-F1 precursor extraction (Figure 7, lane 3). The 63-kDa V2-F1 precursor was efficiently extracted by 1% (v/v) lactic acid only after cysteine-rich prolamins were completely removed following globulin extraction (Figure 7, lane 4). The necessity for preremoval of cysteine-rich prolamins for glutelin extraction suggests the possibility that the V2-F1 precursor is aggregated with cysteine-rich prolamin by disulphide bonds.

Figure 7.

SDS-PAGE analysis of fusion proteins extracted with 1% lactic acid after pre-extraction with different solvents. (a) Non-transgenic Kita-ake; (b) transgenic pV2-F1; Total, total seed proteins; +, with pre-extraction; –, without pre-extraction. Arrows indicate the V2-F1 fusion protein precursor.

Fusion protein precursor is localized in PB-I of transgenic rice seed

There are two types of storage protein bodies present in rice seed endosperm tissues (Krishnan and White, 1995). Prolamin is localized in roughly spherical, ER-derived protein bodies called PB-I, whereas glutelin and globulin are packaged in higher density and irregularly shaped protein storage vacuoles designated PB-II. To determine whether the V2-F1 fusion protein precursor was targeted to protein bodies in the endosperm, we investigated its subcellular localization by immunocytochemical electron microscopy using anti-Cry j 1 antibody followed by secondary antibody conjugated to 20 nm gold particles.

V2-F1 accumulated inside spherical protein bodies delimited by ribosome-associated membranes resembling PB-I (Figure 8a). However, some invaginations or folds were observed inside these storage bodies that are not present in the PB-I structures of non-transgenic rice controls. Protein bodies in which V2-F1 precursor was located reacted preferentially with anti-13-kDa cysteine-rich prolamin antibody (Figure 8b), indicating that V2-F1 precursor was specifically translocalized into PB-I.

Figure 8.

Intracellular localization of V2-F1 in developing rice endosperm cells observed by immuno-electron microscopy. (a) Immuno-electron microscopy with anti-Cry j 1 antibody. (b) Immuno-electron microscopy with anti 13 kDa prolamin antibody. In both (a) and (b), 20 nm gold-labelled particles are specifically present inside PB-I. PB-I, protein body I; PB-II, protein body II; bar = 1 µm.

Discussion

It has been reported that the tertiary structure of Cry j 1 plays a critical role for specific IgE binding activity (Sakaguchi et al., 2001). An ideal safe and effective edible vaccine should have high immunogenicity and low allergenicity. Peptide vaccines using T-cell epitopes would be an ideal allergen-specific immunotherapy, since T-cell epitopes do not have IgE binding activity or B cell epitopes. Therefore, identification of T-cell epitopes for use as tolerogens is necessary for its development, because there is a wide variation in T-cell epitopes. It should be noted that T-cell epitopes derived from Cry j 1 are variable even among same Japanese cedar pollinosis patients and dogs (Sone et al., 1998; Masuda et al., 2004). Development of safe tolerogens with low IgE binding activity can be achieved by cleavage of the Cry j 1 molecule into several fragments and co-expression these fragments in the same transgenic rice seed (Table 1, construct multiple). Fragmented vaccines spanning the whole allergen are applicable not only to humans, but also to other mammals, due to their low allergenic activities, despite the presence of the same B-cell and T-cell epitopes as the native molecule.

In this study, we described the expression and production of the Japanese cedar pollen allergen Cry j 1 in transgenic rice seed using two approaches. First, DNA sequences encoding full-length, N-terminal or C-terminal halves of Cry j 1 were expressed independently under the control of the strong rice endosperm-specific GluB-1 promoter. Although the transgene products were detected by Western blot analysis, accumulation levels were estimated to be less than 1 µg/grain (data not shown). This low level of accumulation may be accounted for by the pectate lyase activity of Cry j 1 (Sone et al., 1994). Okada et al. (2003) also expressed Cry j 1 as a fusion protein with the GFP reporter gene, but the expression level of mature Cry j 1 in transgenic rice seed was only about 0.15 µg/mg of total seed protein. Use of GFP as a carrier may limit the commercial application of any fusion with it as an edible vaccine, even if it proves to be an excellent carrier protein.

An alternative approach has been exploited in this study to increase the accumulation of Cry j 1 in transgenic rice seed. Cry j 1 peptide fragments were expressed as translational fusions with the rice major storage protein glutelin by insertion into the C-terminal variable region of the acidic subunit of the modified glutelins (Figure 1b,c). Overall expression levels were significantly increased for all four constructs tested and the highest accumulation level reached about 15% of total seed protein (Figure 4, Table 1). We showed that modified GluA2-V1 and GluA2-V2 could both be highly expressed as fusion proteins. It is interesting to note that the shorter GluA-V2 plays a role as a fusion protein even though it retains only 80% of its original length. However, such high accumulation phenomena obtained by expressing as fusion protein might be result of the first-generation transgenic rice. From viewpoint of practical use, further work on genetically fixed or stable transgenic rice lines will be required to examine whether high expression levels in mature seed are inherited to their progenies.

It should be noted that higher level of Cry j 1 accumulation in transgenic rice seed could be achieved by using a strong endosperm-specific GluB-1 promoter and expressing the protein fragment as a fusion with GluA2 storage protein. De Jaeger et al. (2002) reported that expression under the control of the common bean seed storage protein arcelin 5-1 and β-phaseolin promoters increased accumulation levels of functional murine ScFv to 12.5% and 36.5% of total soluble proteins, respectively, in transgenic dicotyledonous seeds. Taken together, seed storage protein promoters are a useful tool for large-scale production of economically important recombinant proteins.

It is particularly notable that the Cry j 1 peptide fusion proteins have an altered structure and, thus, would be expected to reduce specific IgE binding activity. This would make them excellent candidates for rice seed-based oral immunotherapy for cedar pollen allergy in humans and the other mammals.

In contrast with the GluA2-V1 construct used in this study, the tandem dodeca-repeat bioactive lactostatin peptide (IIAEK), having a total of 92 amino acids, was highly expressed and correctly processed in transgenic rice seeds, even if it was expressed as a fusion protein with GluA2-V1 (Wakasa et al., 2006). However, none of the precursors of V1-F1, V2-F1, V2-F2 or V2-F3 were processed into mature forms (Figure 3). These results suggest that there are certain size limitations for inserts in the GluA2-V1 and GluA2-V2 constructs for post-translational processing into mature acidic and basic subunits. When the size of an insert reaches or surpasses some as-yet undefined limitation, it may not be possible for the fusion protein to assume a conformation that allows translocation and subsequent processing. The insert lengths of Cry j 1-F1, -F2 and -F3 are 144, 132 and 123 amino acid residues, respectively, which may be beyond the translocation or processing size limitation. In addition, at least for the GluA2-V2 construct, the N-terminal 66 amino acid residues of the deleted region may contain some critical but unidentified recognition domain(s) or residue(s) that is (are) necessary for transporting fusion proteins to a protein storage vacuole. Additional research is in progress for examining this region for its possible role in transport and processing.

Intracellular targeting plays an important role in determining the yield of recombinant proteins because the deposition site of a recombinant protein strongly influences the interrelated processes of folding, assembly and post-translational modification (Streatfield, 2007). We examined the subcellular deposition and localization of V2-F1 precursor in transgenic rice seed endosperms by stepwise extraction and immunocytochemical electron microscopic analysis. Our results showed that V2-F1 precursor aggregated with cysteine-rich prolamin and localized in ER-derived PB-I (Figures 7 and 8). It is conceivable that ‘transport-defective’ V2-F1 precursor was retained by passive trapping as aggregates with cysteine-rich prolamins by disulphide bonds. These observations show that the structural features of protein and protein–protein interactions can play key roles in intracellular transport. Our results also suggest that interaction with cysteine-rich prolamins and the colocalization with prolamins within PB-I may be involved in stabilizing the fusion protein precursor, even though the V2-F1 precursor is not processed into a stable, mature form. These precursors may thus be protected from degradation caused by ER stress, allowing increased accumulation.

The development of plant-derived biopharmaceuticals is a promising field for vaccine, antibody and bioactive peptide production. However, the human and ecological risks associated with cultivating biopharmaceutical-producing transgenic plants must be assessed in advance of the commercialization process. Further developments in physical isolation, biological or genetic containment and a marker-free selection system may be required for public acceptance of transgenic plants (Zuo et al., 2002; Mascia and Flavell, 2004). Nevertheless, the use of edible plant parts for the production and delivery of useful pharmaceuticals has reached an exciting stage (Walmsley and Arntzen, 2000).

Table 1 is a summary of the generation of rice-based cedar pollinosis vaccines in our laboratory. Except for candidate vaccines that have been tested in mouse experiments (Takagi et al., 2005a), transgenic tolerogenic plants generated in this study will be tested for efficacy and safety in animal model experiments in the near future. Meanwhile, the production of host plants that co-express V2-F1, V2-F2 and V2-F3 fusion proteins is in progress with the goal of generating a universal edible vaccine which is suitable for all affected mammals. Seed-based vaccines that we have developed or that are under development for cedar pollinosis immunotherapy have great potential for clinical prophylactic and remedial treatment of the disease (Takaiwa, 2007).

Experimental procedures

Plasmid construction and rice transformation

The full-length Cry j 1 gene (accession number: D34639) was synthesized by GenScript Corporation (Piscataway, NJ, USA). The DNA sequence encoding full-length, N-terminal (amino acid residues 1–195) and C-terminal (amino acid residues 135–353) halves of Cry j 1 were ligated downstream of the 2.3 kb GluB-1 promoter (Qu and Takaiwa, 2004; accession number: AY427569) containing a signal peptide sequence (Figure 1a). A KDEL ER retention signal was attached to the C-termini of all three constructs, followed by 0.65 kb of the GluB-1 terminator (accession number: X54314). The resulting constructs were cloned into HindIII and EcoRI sites of binary vector pGPTV-35S-HPT (Figure 1a, Goto et al., 1999).

Fragmented Cry j 1 was inserted into rice storage protein glutelin GluA-2 as a fusion protein. The DNA sequences corresponding to GluA-2 (accession number: X05664) between amino acid residues 254–280 or 188–280 in the highly variable C-terminal region of the acidic subunit were deleted, and a SmaI site was introduced ahead of Pro281 for subsequent insertion of Cry j 1 sequences (Figure 1b, GluA2-V1 and GluA2-V2). The conserved processing site of Asn282-Gly283 between the acidic and basic subunits was retained. The deletion and modification of GluA2 was accomplished using a method with two stages of PCR amplification as previously described (Takagi et al., 2005a). The DNA sequence encoding Cry j 1 F1 (amino acid residues 1–144) was cloned into GluA2-V1 and GluA2-V2 at a SmaI site. Cry j 1 F2 (residues 126–257) and Cry j 1 F3 (residues 231–353) were inserted into GluA2-V2 at the same site (Figure 1c). The orientation of inserts and the fidelity of all DNA constructs were confirmed by DNA sequencing. These chimeric constructs were ligated downstream of the 2.3 kb GluB-1 promoter and signal peptide sequence, and upstream of the 0.65 kb GluB-1 terminator. The resulting plasmids were then cloned into the binary vector pGPTV-35S-HPT.

The expression plasmids were introduced into the rice genome (Oryza sativa cv. Kita-ake) by Agrobacterium tumefaciens-mediated transformation, and transgenic plants were selected by resistance to hygromycin as described (Goto et al., 1999). Transgenic plants were grown in a controlled greenhouse (28 °C, 12-h light/dark cycle).

Antibodies

Anti-Cry j 1 full-length antibody was purchased from Hayashibara Biochemical Laboratories (Okayama, Japan). Anti-13 KDa Cys-rich prolamin, anti-GluA, anti glutelin B (GluB) and anti glutelin C (GluC) peptide antibodies were prepared as described previously (Takagi et al., 2006). Anti-GluA full-length antibody was kindly provided by Dr T. Utsumi (Kyoto University, Japan).

SDS-PAGE and Western blot analysis

At least four mature seed grains from independent transgenic rice plants were harvested and ground separately into a fine powder with a multibeads shocker (Yasui Kikai, Tokyo, Japan). Total protein was extracted from the powder with 600 µL urea-SDS buffer (50 mm Tris-Cl, pH 6.8, 8 m urea, 4% SDS, 5% 2-mercaptoethanol, 20% glycerol) as previously described (Tada et al., 2003). After separation by 12% SDS-PAGE, the proteins were visualized by CBB-R250 staining or transferred to PVDF membranes (Millipore, Billerica, MA, USA) for immuno-detection. The proteins were detected with anti Cry j 1, anti-GluA, anti-GluB and anti-GluC antibodies followed by a goat anti-rabbit IgG secondary antibody (Cell Signalling Technology, Danvers, MA, USA).

The fusion protein yield as a percentage of total seed protein of transgenic rice was estimated on the basis of the intensity of bands stained with CBB using Bio-Rad Protein Assay Standard II–BSA as a calibration control (Bio-Rad Laboratories, Hercules, CA, USA). Gel images were scanned into a computer, and the corresponding bands were quantified with NIH image software (US National Institutes of Health). Extracts of eight independent seeds from six higher expression lines for each construct were used for the estimation.

Sequential protein extraction

Sequential extraction of proteins was performed according to Tada et al. (2003). Briefly, glutelin and proglutelin were extracted with 1% (v/v) lactic acid from 20 mg seed powder after the stepwise removal of albumins and globulins with 500 µL saline buffer (0.5 m NaCl, 10 mm Tris-Cl, pH 7.5) followed by removal of cysteine-poor prolamins with 500 µL 60% (v/v) n-propanol solution and cysteine-rich prolamins with the same solution but containing 5% (v/v) 2-mercaptoethanol. Each extraction step was accomplished by re-suspending seed powder in the solution and sonicating on ice for 1 min. After centrifugation (9100 g, 10 min) the residues were extracted twice more with the same solution.

DNA extraction and Southern blot analysis

Genomic DNA was prepared from young leaves of non-transgenic and transgenic rice plants according to cetyltrimethylammonium bromide (CTAB) method (Milligan, 1989). Five micrograms of DNA were digested with HindIII or XbaI, fractioned by electrophoresis on 0.8% agarose gels and transferred on to HybondN+ membranes (GE Healthcare Bio-Sciences, Buckinghamshire, UK). Southern blot analysis was carried out with a Gene Images AlkPhos Direct Labelling and Detection System kit (GE Healthcare Bio-Sciences). Hybridization was performed at 62 °C using a full-length hpt probe (accession number: K01193).

Immunocytochemical electron microscopy

Immature transgenic rice seeds were collected (18–20 DAF) and used for immunocytochemical electron microscopic analysis. The procedure for preparing sample was exactly as described previously (Takaiwa et al., 2007), except that the sections were reacted with primary anti-Cry j 1 (1 : 100 dilution) or anti-13 kDa prolamin antibodies (1 : 3000 dilution), and followed by secondary 20 nm gold-labelled goat anti-rabbit IgG Fc SP (EY Laboratories, San Mateo, CA, USA) with a 1 : 50 or 1 : 200 dilution, respectively.

Acknowledgements

This work was supported by research grants (Research for the utilization and industrialization of agricultural biotechnology; Functional analysis of genes relevant to agriculturally important traits in rice genome (IP2001)) from the Ministry of Agriculture, Forestry and Fisheries of Japan to F. Takaiwa.

Ancillary