• Open Access

Recombinant protein yield in rice seed is enhanced by specific suppression of endogenous seed proteins at the same deposit site


*(fax +81-29-838-8397; email takaiwa@nias.affrc.go.jp)


Human IL-10 (hIL-10) is a therapeutic treatment candidate for inflammatory allergy and autoimmune diseases. Rice seed-produced IL-10 can be effectively delivered directly to gut-associated lymphoreticular tissue (GALT) via bio-encapsulation. Previously, the codon-optimized hIL-10 gene was expressed in transgenic rice with the signal peptide and endoplasmic reticulum (ER) retention signal (KDEL) at its 5′ and 3′ ends, respectively, under the control of the endosperm-specific glutelin GluB-1 promoter. The resulting purified hIL-10 was biologically active. In this study, the yield of hIL-10 in transgenic rice seed was improved. This protein accumulated at the intended deposition sites, which had been made vacant through the selective reduction, via RNA interference, of the endogenous seed storage proteins prolamins or glutelins. Upon suppression of prolamins that were sequestered into ER-derived protein bodies (PB-I), hIL-10 accumulation increased approximately 3-fold as compared to rice seed with no such suppression and reached 219 μg/grain. In contrast, reducing the majority of the glutelins stored in protein-storage vacuoles (PB-II) did not significantly affect the accumulation of hIL-10. Considering that hIL-10 is synthesized in the ER lumen and subsequently buds off in ER-derived granules called IL-10 granules in a manner similar to PB-Is, these results indicate that increases in the available deposition space for the desired recombinant proteins may be crucial for improvements in yield. Furthermore, efficient dimeric intermolecular formation of hIL-10 by inhibiting interaction with Cys-rich prolamins also contributed to the enhanced formation of IL-10 bodies. Higher yield of hIL-10 produced in rice seeds is expected to have broad application in the future.


Plant production of recombinant proteins has potential advantages for generating biopharmaceuticals (Daniell et al., 2009). Seeds, especially cereal endosperm, provide ideal production platforms for pharmaceutical and neutriceutical products in terms of ample accumulation, high stability at ambient temperature, non-contamination with human pathogens and direct oral administration. In addition, plant systems have low production costs and the most common such can be easily scaled (Lau and Sun, 2009; Stoger et al., 2005; Takaiwa et al., 2007).

Several cytokines have been produced in the leaves, tubers and seeds of transgenic plants using nuclear transformation. However, the accumulation levels are often less than 1.0% of the total soluble proteins (Sirko et al., 2011). Therefore, progressive improvements in recombinant protein accumulation are necessary for practical application. A great deal of effort has been devoted to optimizing transcription, translation and intracellular targeting (Streatfield, 2007; Takaiwa, 2007). Techniques to improve recombinant protein yields have included the use of plant signal peptides that direct to the secretory pathway, codon optimization and the addition of ER retention signals (H/KDEL) that function in retrieval from cis-Golgi to ER. Depending on the harvested tissues or organs, strong tissue-specific promoters are required to ensure sufficient transcript quantities under the control of temporal- and spatial-specific promoters. Thus, the selection of a suitable promoter is essential to increasing the transgene product yield (Kawakatsu and Takaiwa, 2010; Streatfield, 2007). The intracellular localization of the recombinant product also plays a large role in the protein’s stability, post-translational modification (glycosylation) and accumulation levels (Benchabane et al., 2008). Considering expression in seeds, the use of mutants with reduced amount of storage protein or suppression of seed storage protein expression are known to allow higher accumulation levels of recombinant proteins as compared to the wild-type host (Goossens et al., 1999; Schmidt and Herman, 2008; Tada et al., 2003).

IL-10, a contra-inflammatory regulatory cytokine, has multiple roles in the regulation of immune-responses and has been proposed as a potent anti-inflammatory biological therapy for the treatment of chronic inflammatory bowel diseases and ulcerative colitis, autoimmune diseases and allergic diseases (Asadullah et al., 2003; Mosser and Zhang, 2008). It is primarily produced by activated macrophages and regulatory T cells such as CD4+ CD25+, Tr1 cell and Fox p3 iTreg cells, it acts as an inhibitor of activated macrophages and dendritic cells, and it is involved in the control of innate immune reactions and cell-mediated immunity. In allergen-specific immunotherapy, IL-10 secretion induced by regulatory T cells, B cells and monocytes is involved in counter-regulation of antigen-specific IgE and IgG4 antibody production and in the down-regulation of the mast cell, eosinophil and T cell response (Ozdemir et al., 2009). This results in the suppression of the inflammatory allergenic reaction.

IL-10 has been produced in stable transgenic tobacco plants and transgenic rice seeds (Bortesi et al., 2009; Fujiwara et al., 2010; Menassa et al., 2004). The plant-produced hIL-10 was biologically active in a non-covalent dimer form. Upon efficient refolding and purification, the plant-produced hIL-10 exhibited higher activity than the commercial forms produced by bacteria (Fujiwara et al., 2010). Furthermore, when hIL-10 was expressed in leaves of low alkaloid tobacco and orally administered as a crude leaf extract, the severity of inflammatory bowel disease was significantly reduced in a mouse model (Menassa et al., 2007). The efficacy of plant cells for oral delivery to mucosal tissues in GALT was demonstrated using transgenic plants accumulating IL-10 or IL-4 (Ma et al., 2004; Thanavala et al., 2005). In these cases, the unprocessed plant cell offered increased protection from digestion because of the bio-encapsulation of plant-made pharmaceuticals.

In this study, transgenic rice grains were generated to contain markedly high levels of human IL-10 (11.0 mg/g seed weight at the maximum level). This high accumulation was achieved through the simultaneous suppression of the majority of the prolamins (16, 13 and 10 kDa) stored in the ER-derived PBs. In contrast, the reduction in glutelins did not result in a yield improvement. These results indicate that the reduction in specific endogenous seed proteins that are localized at the same intracellular deposit site as the desired recombinant protein can provide an effective means for enhancing product yield. Furthermore, this approach can be more effective than the general suppression of endogenous seed proteins.


Production of transgenic rice plants

Transgenic rice plants were generated using Agrobacterium-mediated transformation with binary vectors containing pH-IL-10, pProless, pGluless, pProless/H-IL-10 and pGluless/H-IL-10 (Fig. 1). All of the vectors were designed for endosperm-specific expression in transgenic rice seed using 2.3 kb or 1.2 kb of the GluB-1 or 10-kDa prolamin promoter, respectively. pH-IL-10 was designed to produce hIL-10, while pProless and pGluless were intended to reduce the endogenous seed storage proteins prolamins or glutelins, respectively. pProless/H-IL-10 and pGluless/H-IL-10 were designed to produce hIL-10 while simultaneously suppressing either prolamins or glutelins, respectively. A minimum of 25 independent transgenic rice lines were generated for each construct. The champion lines from each construct were advanced to the next generations, and the obtained homozygous lines were tested further. The protein profiles of the transgenic lines were determined by the intensity of Coomassie Brilliant Blue (CBB) staining or by Western blot analyses of proteins extracted from four seeds of individual lines for each construct.

Figure 1.

 Binary vector constructs for rice transformation. pH-IL-10 was designed for the accumulation of human IL-10 in transgenic rice seed endosperm, while pProless and pGluless were designed to reduce the levels of internal prolamins and glutelins, respectively. pProless/H-IL-10 and pGluless/H-IL-10 were designed for the accumulation of h-IL-10 in prolamin-less or glutelins-less transgenic endosperm, respectively. Ag7, Agrobacterium gene 7 terminator; hpt, hygromycin phosphotransferase coding region; CaMV35S P, Cauliflower mosaic virus 35S promoter; GluB1 P-SP, 2.3 kb glutelin B1 promoter and its signal sequence; H-IL-10-His6-KDEL, hIL-10 with His6 tag and ER retention signal KDEL added to its C-terminus. GluB1 T, glutelin B1 terminator; 10-kDa P, 10-kDa prolamin promoter; 10-kDa 5′UTR, 16-kDa 5′UTR, GluC and GluD 5′UTRs; and 13-kDa 3′ UTR and GluA2 3′ UTR are 5′ or 3′ non-translated regions for 10-, 16- and 13-kDa prolamins, and Glutelins C, D and A; RNA Silencing Inducible Sequence (RSIS), a modified glucagon-like peptide-1 sequence used as an RNA silence-inducer (RSIS); LB, left border; RB, right border.

pProless and pGluless reduce the prolamins and glutelins in transgenic rice seed

The major storage proteins of rice seed are glutelins, globulins and prolamins. When rice seed proteins were separated by SDS-PAGE, glutelins, globulins, and 13-kDa Cys-rich as well as Cys-poor prolamins were detected as visible bands after CBB staining. In contrast, 16- and 10-kDa prolamins were not visible on the stained gels (Fig. 2, lane C). Kawakatsu et al. (2010) reported that the modified glucagon-like peptide-1 sequence acting as a RNA Silencing Inducible Sequence (RSIS) suppressed the expression of rice endogenous storage proteins when the 5′ or 3′ untranslated regions (UTRs) derived from the targeted glutelin or prolamin genes were linked to the RSIS and subsequently expressed under the control of the endosperm-specific promoter. These suppressed storage proteins included glutelins of GluA, GluB, GluC and GluD, as well as 10-, 13- and 16-kDa prolamins. In the present study, pProless and pGluless containing the RSIS sequence were designed to produce simultaneous suppression (knock-down) lines of endogenous 10-, 13-kDa (RM2) and 16-kDa prolamins, or GluA, GluB, GluC and GluD in transgenic rice seeds, respectively.

Figure 2.

 Analysis of pProless and pGluless transgenic rice seeds. (a, d) Coomassie Brilliant Blue (CBB)-stained 12% SDS-PAGE of rice seed proteins of pProless and pGluless. Bands for 13-kDa cysteine-rich prolamins are indicated by a white arrow, while 13-kDa cysteine-poor prolamins are indicated by solid bold arrow. Bands for GluBs are indicated by open arrowheads. (b, e) Western blot analysis of rice storage proteins with antibodies specific to 13-kDa Cys-rich prolamins (RM1 and RM9), 13-kDa Cys-poor prolamins (RM2 and RM4), and 10- and 16-kDa prolamins. (c, f) Western blot analysis of rice storage proteins with antibodies specific to glutelins of GluA, GluB, GluC and GluD, as well as 26-kDa globulin. C, Wild-type Kita-ake seed used as a control. Arrow head and arrow indicate GluA precursor and acidic subunit, respectively.

Of 28 independent pProless transgenic lines assayed, 12 lines showed a distinct reduction in 13-kDa prolamin compared to control wild-type rice (Fig. 2a, indicated by a solid black arrow). Judging from the molecular mass, the decreased storage proteins were estimated to be 13-kDa Cys-poor prolamins. This suppression was further confirmed by Western blot analysis using antibodies that were specific to the 13-kDa Cys-poor prolamins RM2 and RM4. As shown in Fig. 2b, RM2 and RM4 were not significantly detected in pProless. However, the 13-kDa Cys-rich prolamins (RM1 and RM9) levels were similar to those of the control. The 10- and 16-kDa prolamins were also decreased in pProless. Notably, the 10-kDa prolamin was scarcely detected via Western blot analysis using anti-10-kDa prolamin antibody. These results indicate that the RSIS sequence knocked down the expression of endogenous prolamins by linking to the 5′ or 3′ UTRs of the targeted prolamin genes as shown in Fig. 1. No significant effects on the expression of the endogenous glutelin GluA, GluB, GluC and GluD genes as well as the 26-kDa globulin gene were detected (Fig. 2c).

When the pGluless transgenic rice seeds were examined, 9 of 25 independent lines exhibited a significant reduction in glutelins when compared to wild-type rice seeds (Fig. 2d, indicated by the open arrowhead). Based on the migration rate of the affected bands on SDS-PAGE, the reduced bands were estimated to be members of the glutelin B family (GluB1, GluB2 and GluB4) (Kawakatsu et al., 2010). Further confirmation with Western blot analyses and antibodies specific to individual GluA, GluB, GluC and GluD revealed that GluB, GluC and GluD decreased dramatically, while GluA was reduced somewhat as compared to the wild-type control. The 26-kDa globulin and the majority of the prolamins containing 13-kDa (RM1, RM2, RM4 and RM9), 16- and 10-kDa were unaffected as compared to wild-type rice (Fig. 2e,f).

Taken together, these results indicate that the RSIS sequence not only suppressed the expression of an independent gene but also efficiently knocked down the expression of multiple target genes when RSIS was expressed fused to the 5′ and 3′ UTRs of the target genes.

Human IL-10 highly accumulates in pProless seeds but not in pGluless seeds

Initially, the hIL-10 accumulation level in pH-IL-10 and pProless/h-IL-10 was compared. Although a relatively high hIL-10 yield was detected in pH-IL-10 seed by Fujiwara et al. (2010), the hIL-10 yield was remarkably enhanced in pProless/H-IL-10 seed (Fig. 3a). To further detail the hIL-10 accumulation levels in pH-IL-10 and pProless/H-IL-10 lines, the hIL-10 levels were analysed by Western blot analysis in a minimum of four positive seeds per line from >20 independent transgenic lines for individual construct. The pProless/H-IL-10 lines accumulated an average of approximately 97 μg hIL-10/grain (Fig. 3b). Approximately 45% of the lines accumulated higher levels, with a maximum of 219 μg hIL-10/grain. In contrast, the pH-IL-10 lines accumulated an average of 28 μg hIL-10/grain.

Figure 3.

 Analysis of pProless/H-IL-10 transgenic rice seeds. (a) SDS-PAGE (upper panel) and Western blot analysis (lower panel) of pH-IL-10 and Proless/H-IL-10 transgenic rice seeds. 13-kDa Cys-rich prolamins are indicated by a white arrow, while 13-kDa Cys-poor prolamins are indicated by a solid arrow. Arrowhead indicates the hIL-10 band. (b) More than 20 lines from each construct were used to determine the accumulation level. A minimum of four positive seeds from each line were analysed by Western blotting using a commercially purchased human IL-10 as a standard control. (c) Western blot analysis with antibodies to 13-kDa prolamins of RM1, RM2, RM4 and RM9, and 10- and 16-kDa prolamins. (d) Western blot analysis with antibodies to GluA, GluB, GluC, GluD and globulin.

The endogenous storage protein profiles of the pH-IL-10 and pProless/H-IL-10 seeds were also compared by Western blot analysis (Fig. 3c,d). As expected, 13-kDa (RM1, RM2 RM4 and RM9), 16- and 10-kDa prolamins were all decreased in the pProless/H-IL-10 seeds, whereas the 16- and 10-kDa prolamins decreased to undetectable levels. In the pH-IL-10 seeds, Cys-rich 13-kDa prolamin (RM9) decreased, while 10-kDa prolamin was slightly increased as compared with the wild type. Meanwhile, GluA, GluB, GluC and GluD remained unchanged. Globulin decreased with both constructs when compared with the wild type (Fig. 3d).

When the hIL-10 accumulation in pH-IL-10 and pGluless/H-IL-10 was compared, the hIL-10 levels were similar. However, most of the glutelins, GluA, GluB, GluC and GluD, were markedly reduced, and the other storage proteins, such as 26-kDa globulin and prolamins, remained unchanged (Fig. 4). This result suggests that the reduction in endogenous glutelins does not provide a significant advantage for the enhancement of hIL-10 accumulation.

Figure 4.

 Analysis of pGluless/H-IL-10 transgenic rice seeds. (a) Coomassie Brilliant Blue (CBB)-stained 12% SDS-PAGE of rice seed proteins (upper panel) and Western blot analysis (lower panel) of pGluless/H-IL-10 and pH-IL-10 transgenic seed. (b) Western blot analysis with antibodies to GluA, GluB, GluC, GluD and globulin. (c) Western blot analysis with antibodies to 13-kDa prolamins of RM1, RM2, RM4 and RM9, and 10- and 16-kDa prolamins. Arrowhead and arrow indicate GluA precursor and acidic subunit, respectively.

Chaperone protein accumulation

Binding Protein (BiP), a major molecular chaperone, interacts with nascent immature secretory proteins synthesized from membrane-bound ER polysomes and assists protein folding along with the help of other chaperones such as disulfide isomerase (PDI) and calnexin (CNX) in the ER lumen (Oono et al., 2010; Wakasa et al., 2011). The effect of hIL-10 accumulation on chaperone expression in transgenic seeds was examined by Western blotting (Figure 5). Of the chaperones examined, the level of PDI 2-3 was substantially up-regulated in pGluless/H-IL-10 and pProless/H-IL-10 seeds, while the levels of BiP1 and CNX in pGluless/H-IL-10 and pProless/H-IL-10 seeds were slightly enhanced. Notably, the level of BiP4&5 was drastically elevated in transgenic seeds containing pGluless/H-IL-10 and pProless/H-IL-10. The BiP4&5 level was also up-regulated in pH-IL-10 seed as compared to the wild-type control seed. This result indicates that hIL-10 accumulation in pGluless/H-IL-10 and pProless/H-IL-10 seeds induces higher levels of ER stress than in pH-IL-10 seeds. Furthermore, in pGluless/H-IL-10 seed, an extra band of approximately 40 kDa was noted. This band is likely a degradation product derived from BiP4&5 as a result of auto-regulation (Fig. 5c). Thus, more severe ER stress may be induced in pGluless/H-IL-10 seeds compared to pProless/H-IL-10 seeds as a response to unfolded proteins.

Figure 5.

 Western blot analysis of chaperone proteins in transgenic rice seeds. (a) and (b) Expression level comparison of the molecular chaparones of Binding Protein 1 (BiP1), BiP4&5, PDI1-1, PDI1-4, PDI2-3 and calnexin (Cal) between pH-IL-10 and pGluless/H-IL-10, and between pH-IL-10 and pProless/H-IL-10, respectively. (c) Expression level comparison of BiP4&5 among pProless, pH-IL-10 and pProless/H-IL-10; and pGluless, pH-IL-10 and pGluless/H-IL-10, respectively.

Intracellular localization of hIL-10 in rice endosperm

Two types of storage protein bodies (PBs) are present in the endosperm of rice seeds (Krishnan and White, 1995; Tanaka et al., 1980). Prolamins are localized in PB-I, an ER-derived protein body that exhibits lower electron density and a roughly spherical structure. Glutelins and globulins are deposited into protein-storage vacuoles called PB-II, which exhibit higher electron density and an irregularly shaped structure.

The intracellular localization of hIL-10 was examined by immune-electron microscopy in the developing endosperm cells of transgenic seeds 18 days after flowering (DAF) using anti-hIL-10 antibody. While a portion of hIL-10 was localized in typical PB-Is, the majority was observed in tiny granules (from <0.1 to 0.2 μm in diameter) that clearly differed from the PB-Is and PB-IIs (Fig. 6Aa–c). These tiny granules exhibited high electron density and were tentatively designated IL-10 granules. The size and morphology of the IL-10 granules were highly variable. As shown in Fig. 6, the IL-10 granules gathered with other spherical tiny granules (0.1–0.2 μm in diameter) of lower electron density. When the origin of the small low-density granules was examined by immune-electron microscopy, the granules reacted specifically to gold particles labelled with anti-RM1 of 13-, 10- and 16-kDa prolamin antibodies. These granules were therefore known as mini-PB-I (Fig 6Ad–f). Numerous mini-PBI and IL-10 granules either loosely assembled at the same location or were engulfed and unevenly fused together to form a unique huge body-like structure (Fig. 6A). The huge body-like structure typically exhibited irregular or deformed shapes with a mean diameter size of 3–5.5 μm (Fig. 6Ab, Bd–f). Gold particles labelled with anti-RM1 of 13-kDa prolamin were located in a lower electron density region of the huge bodies (Fig. 6Cb), whereas those with hIL-10 were located in a higher electron density region of the huge bodies (Fig. 6Ab, Bf). No glutelins were localized in the bodies because the gold particles labelled with anti-GluA antibody were observed exclusively in PB-II (Fig. 6Cd). ER membrane and halo-like structures surrounding the huge body structures were occasionally observed.

Figure 6.

 Immunoelectron microscopy observation of intracellular localization of h-IL-10 in developing rice endosperm cells. (A) Transgenic seed endosperm cells of pH-IL-10. Anti-hIL-10 antibody conjugated with gold particles was specifically located at PBI (a), and higher electron density granules and regions within huge bodies (d) and assemblies (b). Gold particles were labelled to 13-kDa (RM1), 10- and 16-kDa prolamins (e, c and f, respectively). (B) Transgenic seed endosperm cells of pProless/H-IL-10. Anti-hIL-10 antibody conjugated with gold particles was specifically located at higher electron density granules and regions within assemblies (b, c) or huge bodies (e, f). (C) Transgenic seed endosperm cells of pProless/H-IL-10. Anti-RM1 antibody labelled with gold particles was localized at lower electron density granules (mini-PBI) and a region inside of the huge bodies as well as PBIs (a, b and c), while gold particles labelled with GluA antibody were located specifically in protein bodies II (PB-II) (d). IL-10: IL-10 granules; Assemblies: loosely gathered IL-10 granules and mini-PBIs; huge body: tightly fused IL-10 granules and mini-PBIs; PBI, protein body I and mini-PBI; PB-II, protein body II. Bar = 1 μm.

Next, the intracellular structure of the endosperm cells of pProless/H-IL-10 and pH-IL-10 were compared. Although the subcellular structures were similar, the abundance of huge body structures was clearly increased relative to the abundance of loosely gathered multiple mini-PBI and IL-10 granule assemblies. The results of subcellular localization analyses clearly detected hIL-10 in the ER-derived PBs of the transgenic endosperm cells. The ER retention signal KDEL in IL-10 was suggested to be involved in retaining IL-10 in the ER and ER-derived PB.

In vitro digestion of transgenic rice seeds

The digestibility of IL-10 in transgenic seeds was compared with that of 13-kDa prolamin (RM1) stored in PB-I and with glutelin A deposited in PB-II (Fig. 7). IL-10 from both pH-IL-10 and pProless/H-IL-10 seeds was rapidly digested by pepsin within 15 min, and its digestibility was very similar to glutelin A. In contrast, the digestion of RM1 derived from PB-Is required several hours and produced a different digestion pattern.

Figure 7.

In vitro digestibility of transgenic rice seed powder of pH-IL-10 and pProless/H-IL-10 containing h-IL-10. Seed powder (5 mg) was added to a reaction mixture containing 0.1% pepsin and incubated at 37 °C for up to 3 h. Total proteins in each reaction mixture were extracted with urea/SDS buffer and examined by immunoblotting with hIL-10, GluA and RM1 antibodies.

A portion of hIL-10 aggregates with cysteine-rich prolamins

To investigate the structure of hIL-10 deposited in endosperm cells, the conditions for extracting hIL-10 from pH-IL-10 and pProless/H-IL-10 seeds were examined. hIL-10 partially extracted in the glutelin fraction with or without pre-removal of the globulin fraction from seed powders, although the majority remained in unextractable residues (Figure 8, lanes 1 and 2, indicated by an arrowhead). Removal of cysteine-poor prolamins with 60% 2-propanol did not improve the hIL-10 yield (Figure 8, lane 3). Notably, hIL-10 was completely extracted by 1% lactic acid containing 4% SDS only after removal of cysteine-rich prolamins with 60% propanol plus 5% mercaptoethanol (2-ME), and no hIL-10 remained in the residue’s fractions (Figure 8, lane 4). The requirement for pre-removal of cysteine-rich prolamins for the complete extraction of hIL-10 suggests the possibility that at least a portion of the hIL-10 molecules are aggregated with cysteine-rich prolamins via disulphide bonds in the endosperm cells. It is interesting to note that in pH-IL-10 seeds, 10- and 16-kDa Cys-rich prolamins are the main prolamins removed by 60% propanol plus 5% 2-ME. In contrast, in pProless/H-IL-10 seeds, Cys-rich 13 kDa (RM1) was extracted using the same solution because of a lack of 10- and 16-kDa prolamins. In these seeds, the 13-kDa prolamin amount in the residue fraction was reduced (Figure 8, lane 4). These results indicate that hIL-10 interacts with 10- and 16-kDa prolamins in pH-IL-10 seed but interacts with 13-kDa prolamin (RM1) in pProless/H-IL-10 seeds (Figure 8).

Figure 8.

 SDS-PAGE and Western blot analyses of hIL-10 extracted with 1% lactic acid plus 4% SDS after pre-extraction with different solvents. T, total seed proteins; −, without extraction; +, with extraction; 16-, 10-kDa and RM1 are 16-, 10- and 13-kDa cysteine-rich prolamins. Arrows indicate the hIL-10 protein.

Presence of hIL-10 monomers and dimers in rice seed

The active form of hIL-10 is a non-covalently associated homodimeric protein (Zdanov et al., 1995). To determine the molecular structure of the hIL-10 produced in transgenic rice seed, total protein was extracted from pH-IL-10 and pProless/H-IL-10 seeds by the method of Fujiwara et al. (2010). These total proteins were analysed by Western blot analysis using anti-hIL-10 antibody. The results of both reducing and non-reducing SDS-PAGE showed that, in addition to a band corresponding to monomeric hIL-10 with a molecular mass of approximately 19 kDa, there was another major protein band that corresponds to the dimeric form of hIL-10 with a molecular mass of approximately 40 kDa (Figure 9). It is interesting to note that even under strong reducing conditions (containing 2.5% 2-ME), the dimers did not dissociate into monomers. The detection of both dimeric and monomeric forms of hIL-10 in non-reducing SDS-PAGE suggests that a portion of hIL-10 exists in the dimeric form via non-covalent associations in the native state in the ER lumen. The hIL-10 monomers may aggregate with cysteine-rich prolamins via disulphide bonds that are used for the stabilization of accumulated hIL-10 in endosperm cells. These results support data from the stepwise extraction experiment, as complete hIL-10 extraction can be achieved only after the removal of cysteine-rich prolamins (Figure 8; Figure 9, upper panel).

Figure 9.

 SDS-PAGE and Western blot Analyses of h-IL-10 from pH-IL-10 and pProless/H-IL-10 under non-reducing and reducing conditions. Upper panel, non-reducing (without 2-ME) conditions. Lower panel, reducing (with 2-ME) conditions. The positions of dimer and monomer are indicated.


IL-10 inhibits the expression of many pro-inflammatory cytokines and chemokines as well as pro-inflammatory enzymes. IL-10 functions as the main inhibitory cytokine produced by regulatory T cells during allergen immunotherapy. Administration of IL-10 to normal volunteers decreased the number of circulating CD4+ and CD8+ T cells and suppressed mitogen-induced T cell proliferation and endotoxin-derived TNFα and IL-1b production (Chernoff et al., 1995). Recombinant human IL-10 has been developed for the prevention or cure of several autoimmune diseases and allergic diseases, and IL-10 is currently being tested against rheumatoid arthritis, inflammatory bowel disease psoriasis, organ transplantation and chronic hepatitis C.

Human IL-10 has been produced in stable transgenic tobacco plants under the control of the CaMV35S promoter. The overall accumulation level was 0.0055% of the total soluble protein when the recombinant proteins were targeted to the ER (Menassa et al., 2004). The biological activity of plant-derived IL-10 was confirmed in vitro and its ability to induce the anti-inflammatory response has been demonstrated in vivo using a mouse model of colitis (Menassa et al., 2007). Previously, hIL-10 was produced in transgenic rice seeds under the control of the endosperm-specific GluB-1 promoter (Fujiwara et al., 2010). When this recombinant hIL-10 was purified by chromatography and assayed for its biological activity using mouse bone marrow dendritic cells, the non-covalent dimer form of IL-10 exhibited higher activity than the commercial forms produced in bacteria. These results indicated that rice seed is an ideal host for hIL-10 production.

To increase hIL-10 accumulation levels, a fusion of hIL-10 with an elastin-like polypeptide (ELP) was generated. This approach yielded up to 0.27% of the total soluble protein (TSP) in stable transgenic tobacco plants (Patel et al., 2007). In the present study, the codon-optimized hIL-10 was deposited into ER-derived PBs where prolamins were simultaneously suppressed by RNA interference. This approach produced remarkably higher hIL-10 yields up to approximately 15% of the rice TSP (approximately 220 μg/grain). This high hIL-10 concentration in rice grain is expected to induce oral immune tolerance by the oral (mucosal) route, because daily feeding of plants accumulating 2.5–9 μg of IL-10 improved histological level symptoms in the mouse model (Menassa et al., 2007). In fact, oral feeding of this transgenic rice grain to cedar pollen allergy model mouse inhibited the production of specific IgE after vaccination with cedar pollen allergens, indicating the prophylactic effect of rice-based IL-10 on cedar pollen allergy (data not shown).

As compared with systemic injection, increases in accumulation levels are necessary for oral therapeutic products to deliver to mucosal tissues. The bio-encapsulation of therapeutic products by plant cells protects the product from digestion during the delivery process to the mucosal and gut systems (Walmsley and Arntzen, 2000). When produced in seeds, therapeutic agents are deposited in protein bodies. This provides an ideal delivery system to mucosal tissues because of PB and cell wall encapsulation (Takagi et al., 2010; Takaiwa, 2011). Compared to products produced in vegetative tissues such as leaves and roots, seed-based products have many advantages. These include high accumulation levels and high stability at ambient temperature for 2 or 3 years because of low protease activity and dehydration. These advantages can permit extensive transport and storage.

The combination of strong endosperm-specific expression and ER-targeting represents a leading strategy for enhancing the yield of recombinant proteins in transgenic rice seed. The ER lumen contains an abundance of folding enzymes, and chaperones required for proper folding and assembly, thus providing a suitable environment for various recombinant proteins. Choosing a proper intracellular compartment as a deposition site for recombinant proteins can determine their stability, modification, production yield and activity. When recombinant proteins are expressed as secretory proteins and trafficked via the endomembrane system in cereal endosperm tissues, the subcellular destination can be affected by interactions with seed storage proteins in the ER lumen or by the aggregation properties of storage proteins. This can result in localization at unexpected intracellular sites. When several recombinant proteins (Asperigillus niger phytase, human serum albumin) containing the KDEL signal were expressed in transgenic wheat under the control of the ubiquitin promoter, the intracellular deposition site in seeds was altered from the inherent destination site (apoplast or ER) that had been observed in the vegetative tissues. As a result, the proteins were retained in ER-derived PB and PSV by interaction with seed storage proteins in the endosperm (Arcalis et al., 2004; Drakakaki et al., 2006). A similar result was observed in transgenic rice when recombinant proteins were expressed in endosperm tissue. Previously, sequential extraction experiments in transgenic rice seed showed that prolamins such as 10K, 16K and Cys-rich 13K prolamins interacted with several recombinant proteins that contained cysteine residues (Cry j 1, 7Crp, Der f 2) via disulfide bonds. This was demonstrated by the requirement of a reducing agent such as 2-ME for the recombinant protein extraction (Takaiwa et al., 2009; Yang et al., 2007, 2012). The induction of chaperone proteins in transgenic rice seeds by the production of these recombinant proteins may result from the unfolded protein response. This response may be caused by the interaction with seed storage proteins in the ER lumen. The interaction strength with the endogenous seed proteins may depend on the individual inherent physicochemical properties of the introduced recombinant protein. As shown in Figure 5, the expression of BiP4&5 and PBIL 2-3 was significantly induced by hIL-10 production.

Intracellular targeting plays an important role in determining the recombinant protein yield. When hIL-10 was expressed as a secretory protein in rice endosperm, it accumulated as both dimers and monomers (Figure 9). Zdanov et al. (1995) reported that the active form of hIL-10 is a non-covalently associated homodimeric protein. Thus, it is plausible that the accumulated hIL-10 may self-assemble into a more stable and native form of dimeric structure that possesses biological activity. These assemblies, which may occur via non-covalent bonds based on inherent properties, subsequently bud off as IL-10 bodies (Figure 6). As a portion of hIL-10 was extracted only after the removal of the cysteine-rich proteins, this suggests that hIL-10 monomers most likely accumulate in PB-Is by interaction with Cys-rich prolamins, as anti-hIL-10 immuno-gold was detected in typical PB-Is (Figures 6 and 8). Interestingly, the percentage of dimeric hIL-10 was increased in the endosperm cells of pProless/H-IL-10 using RNA interference to knock-down prolamins (Figure 9). These results further demonstrate that the protein’s structural features as well as protein–protein interactions may play crucial roles in intracellular transport and in increasing accumulation levels by hIL-10 stabilization.

hIL-10 was demonstrated to be predominantly deposited in specific ER-derived IL-10 granules (Figure 6). The IL-10 granules, very small particles of <0.1–0.2 μm in diameter, have higher electron densities than PB-Is that contain prolamins. Independently, numerous tiny IL-10 granules and small mini-PB-Is loosely gathered together like ER-derived PB structures. These gathered granules occasionally fused together to constitute a huge unique body-like structure (Figure 6). Furthermore, a portion of hIL-10 accumulated in typical PB-Is via interaction with Cys-rich prolamins. Therefore, hIL-10 that is deposited in mature rice seed is expected to exhibit resistance to digestive enzymes, as hIL-10 is deposited in ER-derived IL-10 granules in a manner similar to the PB-Is. However, hIL-10 that was deposited primarily in IL-10 granules as well as in huge bodies exhibited a lower digestibility than prolamins stored in PB-Is (Figure 7). This digestibility difference may be explained by the fact that several types of prolamins (16-, 10- and 13-kDa Cys-rich prolamins and 13-kDa Cys-poor prolamins) are tightly and regularly packaged in PB-Is as part of the seed maturation process (Nagamine et al., 2011), while IL-10 granules are mainly composed of dimeric IL-10 linked by non-covalent bonds. Furthermore, the hIL-10 monomers that interact with cysteine-rich prolamins to form huge bodies are not programmed precisely like ‘native PB-Is’. Therefore, the dimeric and monomeric hIL-10 forms in rice seeds may be related to their susceptibility to digestive enzymes, although some portions of hIL-10 interact with Cys-rich prolamins.

The suppression of endogenous seed storage proteins has enhanced the accumulation levels of other seed proteins at the transcriptional and translational levels as a compensatory effect that maintains the homoeostasis of seed protein levels (Kawakatsu et al., 2010). In a low glutelin mutant such as LGC-1 or three glutelin structural mutants, the prolamin levels significantly increase in a compensatory manner (Iida et al., 1993, 1997). Furthermore, it was recently demonstrated in transgenic rice that the suppression of individual seed storage proteins or a combination of several seed storage proteins such as glutelins, prolamins and globulin significantly affected the expression of other seed proteins and PB formation by rebalancing the proteome to maintain the protein content of rice seeds (Kawakatsu et al., 2010). These studies suggest that the production yield of foreign proteins can be enhanced by redirecting intrinsic seed protein production to foreign protein production.

Transgene product accumulation was highly enhanced by the suppression of an endogenous gene using co-introduction of an antisense gene in transgenic Arabidopsis seed or by introducing the transgene in soybean seeds with suppressed or knocked down seed storage proteins (Goossens et al., 1999; Schmidt and Herman, 2008). When the soybean glycinin (A1aB1b) gene directed by the endosperm-specific glutelin GluB-1 promoter was introduced into these low glutelin mutants, the accumulation levels of soybean glycinin resulted in an approximate 2-fold increase as compared to wild-type rice transformed with the same construct (Tada et al., 2003). Notably, the soybean glycinin was deposited in the same protein-storage vacuoles (PB-II) as the rice glutelins. In an effort to improve the hIL-10 production yield in transgenic rice seed, the present study specifically expressed hIL-10 under the control of the endosperm-specific promoter along with the simultaneous specific suppression of numerous prolamin or glutelin genes by RNA interference. Upon significant reduction in Cys-rich 16-, 10-kDa and Cys-poor RM2 and RM4 13-kDa prolamins, the hIL-10 yield was enhanced up to 3-fold in transgenic rice seed as compared to the yield in transgenic rice seed transformed with pH-IL-10 construct. In contrast, a marked reduction in most of the glutelins (GluA, GluB, GluC and GluD) did not significantly affect the hIL-10 accumulation level. As shown in Fig. 6, hIL-10 was mainly sequestered into ER-derived PBs called IL-10 granules or PB-I, whereas glutelins were located in different protein-storage vacuoles (PB-II). Therefore, the vacancy of the final destination site (deposition site) of the desired recombinant proteins is very important for the enhancement of the production yield. This is because the competition for deposition is relieved by the reduction in endogenous seed proteins. The source of amino acids for protein synthesis may not be a limiting factor. That is, space for the accumulation of recombinant proteins is more critical than the amino acid source. Therefore, the reduction in seed storage proteins was expected to provide a new strategy for enhancing the accumulation levels of desirable recombinant proteins. The resulting induced compensation effect was expected to enhance the expression of the recombinant protein, and the increase in deposition space was also expected to increase the recombinant protein yield. It should be noted that the supplied vacant space should correspond to the localization of the desired recombinant proteins to alleviate competition for accumulation. The reduced endogenous seed protein may be efficiently exchanged by foreign proteins. As a next step, it is important to examine whether other cytokines can be produced at higher levels via the suppression of prolamins.

Experimental procedures

Plasmid construction and rice transformation

The DNA sequence encoding human IL-10 (accession number: NM-000572) was optimized for translation based on the codon usage in rice seed storage protein genes. Subsequently, IL-10 was synthesized by GenScript Corporation (NJ, USA). The gene was ligated downstream of the 2.3 kb GluB-1 promoter (Qu and Takaiwa, 2004; accession number: AY427569) containing DNA encoding a signal peptide (Fig. 1). The coding sequences of the histidine hexamer tag and a KDEL ER retention signal were attached to the C-terminus of the gene, followed by 0.65 kb of the GluB-1 terminator (accession number: X54314). The gene cassette was then cloned into the HindIII and EcoR1 sites of binary vector pGPTV-35S-HPT (Goto et al., 1999). The resulting plasmid was designated pH-IL-10 (Fig. 1).

The plasmids pProless and pGluless were designed to reduce the levels of internal 10-, 13- and 16-kDa prolamins, and GluA, GluB, GluC and GluD in transgenic rice seeds, respectively (Fig. 1). pProless was constructed by ligating the modified glucagon-like peptide-1 (mGLP-1) sequence, a RNA silence-inducer (RSIS) (Kawakatsu et al., 2010; Yasuda et al., 2005), downstream of the 10-kDa prolamin promoter containing the 10-kDa prolamin 5′UTR plus the 16-kDa prolamin 5′UTR, followed by the 13-kDa prolamin 3′UTR. The prolamin silence-inducing cassette was cloned into binary vector p35SHPTAg7-GW using the MultiSite Gateway system (Fig. 1) (Wakasa et al., 2006). pGluless was also constructed via addition of the RSIS sequence downstream of the 2.3 kb GluB-1 promoter containing its signal sequence as well as GluC and GluD 5′UTRs, followed by GluA2 3′UTR. When prolamin or glutelin silencing-inducing cassette, together with hIL-10 expression cassette, were cloned into binary vector p35SHPTAg7-GW, the resulting plasmids were designated pProless/H-IL-10 and pGluless/H-IL-10, respectively (Fig. 1).

The expression plasmids were introduced into the rice genome (Oryza sativa cv. Kita-ake) via Agrobacterium tumefaciens-mediated transformation. 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). The highest hIL-10 accumulation lines of pH-IL-10, pProless/H-IL-10 and pGluless/H-IL-10, as well as the most suppressed lines of pProless and pGluless, were selected among T1 regenerates, and homozygous lines were selected among the progenies for further analyses.


Human IL-10 antibody was purchased from American Research Products, Inc.™ (Belmont, MA, USA). Antibodies to glutelins (GluA, GluB, GluC and GluD) and 26-kDa globulin peptides, 10-, 13-kDa (RM1, RM2, RM4 and RM9), 16-kDa prolamins, and the rice chaperones of BiP1, BiP4&5, PDI 1-1, PDI 1-4, PDI 2-3 and CNX were prepared previously in the laboratory (Yasuda et al., 2009; Oono et al., 2010; Wakasa et al., 2011). Horseradish peroxidase-conjugated anti-rabbit IgG secondary antibody was purchased from Cell Signaling Technology (Danvers, MA).

SDS-PAGE and Western blotting analysis

A minimum of four mature seed grains from independent transgenic rice plants were harvested and ground separately into a fine powder using a multibeads shocker (Yasui Kikai, Tokyo, Japan). Total protein was extracted from the powder using 600 μL Urea-SDS buffer [50 mm Tris–HCl, pH 6.8, 8 m Urea, 4% SDS, 5% mercaptoethanol (2-ME), 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) for immuno-detection.

To determine the hIL-10 yield in transgenic rice seeds, a minimum of four positive seeds from each line were mixed and subjected to Western blot analysis with anti-hIL-10 antibody using a standard human IL-10 as a control (Wako Pure Chemical Industries, Ltd. Osaka, Japan). Band density was quantified with NIH image J software (National Institutes of Health, Washington, DC).

Immunocytochemical electron microscopy

Immature transgenic rice seeds (18–20 days old) were collected and used for immunocytochemical electron microscopic analysis. The procedure for preparing the samples was exactly as described previously (Kawakatsu et al., 2010; Takaiwa et al., 2007), with the exception that the sections were reacted with primary anti-human IL-10 (dilution 1:500), GluA (1:50000), 13-kDa cysteine-rich prolamin (RM1) (1:100000), 10-kDa (1:50000) and 16-kDa (1:50000) prolamins antibodies, followed by secondary 20-nm gold particle–labelled goat anti-rabbit IgG Fc-specific (EY Laboratories, San Mateo, CA) with 1:200 dilution.

Sequential extraction of proteins

Sequential extraction of proteins was conducted according to the method of Tada et al. (2003), with minor modifications. In brief, glutelins were extracted from 20 mg seed powder with 1% lactic acid containing 4% SDS after the stepwise removal of albumins and globulins with 500 μL saline buffer (0.5 m NaCl, 0.1% Triton X-100, 50 mm Tris–HCl, pH 7.5) followed by the removal of cysteine-poor prolamins with 500 μL 60% 2-propanol solution and cysteine-rich prolamins with the same solution but supplemented with 5% 2-ME. Each extraction step was performed by re-suspending the seed powder in the solution and sonicating on ice for 2 min. After centrifugation at 9100 g for 10 min, the residues were extracted twice more with the same solution.

In vitro digestion of transgenic rice seeds by pepsin

Transgenic rice seeds were subjected to pepsin digestion according to the method of Astwood et al. (1996), with minor modifications. In brief, 150 μL of reaction buffer containing 0.1% (w/v) pepsin (Sigma) and 30 mm NaCl (pH1.2) was added to 5 mg of seed powder and incubated at 37 °C. The reaction was terminated by neutralization with NaOH for 0, 2, 5, 15, 30, 60, 120 and 180 min. After adding 150 mL of urea/SDS buffer, the samples were analysed by 12% SDS-PAGE, followed by Western blot analysis.

Preparation the dimeric hIL-10 protein samples

The protein samples used for non-reducing SDS-PAGE were prepared according to Fujiwara et al. (2010). After extraction with the pre-extraction buffer (50 mm Tris–HCl, pH 7.4, 0.5 m NaCl), 20 mg of seed powder was extracted with 500 μL of cetyltrimethyl ammonium bromide (CTAB) extraction buffer [50 mM Tris–HCl, pH 7.4, 0.5 m NaCl, 1% CTAB, 10 mm 2-ME]. Each extraction step was performed by suspending the seed powder in the solution and sonicating on ice for 2 min. The hIL-10 containing supernatant was obtained after centrifugation at 9100 g for 10 min. After adding the same volume of urea/SDS buffer with or without 5% 2-ME, the samples were analysed by 12% SDS-PAGE and Western blot analysis.


The authors thank Ms M. Utsuno, Ms Y. Ikemoto, Ms H. Yajima and Ms Y. Suzuki for technical assistance. This work was supported by ‘Genomics and Agricultural Innovation, GMC0004’ to F. T. from the Ministry of Agriculture, Forestry and Fisheries of Japan.