The redistribution of protein sulfur in transgenic rice expressing a gene for a foreign, sulfur-rich protein


For correspondence (e-mail


Sulfur amino acid composition is an important determinant of seed protein quality. A chimeric gene encoding sunflower seed albumin (SSA), one of the most sulfur-rich seed storage proteins identified so far, was introduced into rice (Oryza sativa) in order to modify cysteine and methionine content of the seed. Analysis of a transgenic line expressing SSA at approximately 7% of total seed protein revealed that the mature grain showed little change in the total sulfur amino acid content compared to the parental genotype. This result indicated that the transgenic rice grain was unable to respond to the added demand for cysteine and methionine imposed by the production of SSA. Analysis of the protein composition of the transgenic grain showed changes in the relative levels of the major seed storage proteins, as well as some non-storage proteins, compared to non-transgenic controls. Changes observed at the protein level were concomitant with differences in mRNA accumulation but not always with the level of transcription. The limited sulfur reserves appeared to be re-allocated from endogenous proteins to the new sulfur sink in the transgenic grain. We hypothesize that this response is mediated by a signal transduction pathway that normally modulates seed storage protein composition in response to environmental fluctuations in sulfur availability, via both transcriptional and post-transcriptional control of gene expression.


Methionine content can limit the nutritive value of seed protein, particularly in the case of grain legumes (Tabe and Higgins, 1998). Transformation with a gene specifically expressed in seeds and encoding a protein ‘sink’ rich in sulfur amino acids is a strategy that has been used successfully to enhance the seed methionine content and protein quality of canola (Brassica napus, Altenbach et al., 1992), narbon bean (Vicia narbonensis, Pickardt et al., 1995), narrow leaf lupin (Lupinus angustifolius, Molvig et al., 1997), and maize (Anthony et al., 1997; Lai and Messing, 2002). In the case of canola, the seed protein is not particularly deficient in sulfur amino acids. However, the use of canola meal in animal feed makes it an attractive target for over-accumulation of methionine to complement deficiency of this essential amino acid in legume seed meals with which canola meal is blended (Tabe and Higgins, 1998). Similarly, in the case of maize, Lai and Messing (2002) concluded that increasing the methionine level in this cereal grain resulted in an enhanced formulation for animal feed. Although not the first limiting amino acid, methionine can be limiting in animal feeds based on rice (Sure, 1955). We assessed rice as a potential production system for protein methionine by transforming it with a gene encoding sunflower seed albumin (SSA), which contains 16% methionine residues and 8% cysteine residues (Kortt et al., 1991).

We introduced a chimeric gene encoding SSA modified to contain an endoplasmic reticulum (ER) retention signal at the carboxy terminus of the protein into Japonica rice (cultivar Taipei). The gene was controlled by a strong, endosperm-specific promoter from a wheat high molecular weight glutenin gene. In the highest expressing transgenic line, SSA accumulated to approximately 7% of total protein in the mature grain. As had been observed in narrow leaf lupin, mature rice grain that accumulated SSA contained no more total sulfur than non-transgenic control grain (Molvig et al., 1997). However, in marked contrast to the transgenic lupins (Molvig et al., 1997), transgenic SSA rice contained essentially the same concentration of reduced (organic) sulfur as non-transgenic controls. We found that the protein composition of the transgenic rice grain was altered in parallel with the SSA accumulation. Changes in the relative abundance of a number of specific rice proteins were documented. We hypothesize that rather than leading to an increase in sulfur amino acid biosynthesis in the developing rice grain, SSA accumulation results in the diversion of limited sulfur reserves away from the synthesis of endogenous proteins. The most likely mediator of such a response is the mechanism that modulates seed storage protein composition in response to sulfur and nitrogen availability (Tabe et al., 2002).


Sulfur content does not change in the SSA transgenic grain

Twenty independent transgenic lines of rice were screened for SSA expression by Western blot, and a variety of expression levels were observed. Using a transgenic lupin expressing SSA at approximately 5% of total seed protein (Molvig et al., 1997) as a quantitative reference, the highest expressing line, 64-90, was estimated to express SSA at a level of approximately 7% of total seed protein (Figure 1b). Pooled de-hulled grain samples from individual mature rice panicles were analyzed for nitrogen and sulfur content (Table 1). Control and SSA transgenic grain showed no significant difference in nitrogen content. Despite an abundant supply of sulfur to the plants, analysis of mature grain showed that the level of reduced (organic) sulfur was not significantly different between control and transgenic lines, while the level of oxidized (inorganic) sulfur was below detectable levels in both samples.

Figure 1.

Sunflower seed albumin (SSA) levels in rice lines transformed with pLT10 and pLT11.

(a) pLT10 and pLT11 constructs containing the 1Dx5 and Bx17 promoters, respectively, were used to transform Japonica rice by microparticle bombardment. The position of the sequence encoding the KDEL endoplasmic reticulum retention signal is shown.

(b) The #63 rice lines (63-8, 63-28, and 63-24) were produced by transformation with the pLT10 construct, while the #64 lines were produced using the pLT11 construct. Samples of approximately 20 µg total extractable protein from pooled grains of 10 of the 20 transgenic lines produced were fractionated by SDS–PAGE. The samples were electroblotted to a nitrocellulose membrane and incubated with antiserum to SSA. Approximately 3 µg SSA transgenic lupin protein (Molvig et al., 1997) was included as a reference.

Table 1.  Nitrogen and sulfur content of mature control and sunflower seed albumin (SSA) transgenic rice grain
 Nitrogen content (µmol g−1 DW)Reduced sulfur content (µmol g−1 DW)Oxidized sulfur content (µmol g−1 DW)
  1. Control and transgenic flour samples were produced from pooled grain of individual panicles. Seed nitrogen (n = 5) was determined by a micro-Kjeldahl method. Reduced (organic) sulfur and oxidized (inorganic) sulfur (n = 6) was determined by X-ray fluorescence spectrometry. Each value represents mean ± SD.

Control778.6 ± 72.829.9 ± 4.3Not detected
SSA transgenic792.9 ± 52.930.4 ± 3.3Not detected

Analysis of total amino acid concentration was performed on flour produced from SSA transgenic and non-transgenic control mature, whole rice grains. By comparing measurements of reduced sulfur with measurements of total cysteine and methionine, we estimated that sulfur amino acids made up approximately 70% of total grain sulfur, the remainder being in other reduced sulfur compounds. There was little difference between the amino acid compositions of the control and transgenic grain. However, the level of cysteine decreased by 15%, and the level of methionine increased by 27% in the transgenic grain (Table 2). Combining this data, total sulfur amino acids increased by 5%. On the basis of the measured expression level of SSA and its known cysteine and methionine content, we predicted that total grain sulfur amino acid content would increase by 40% in the transgenic grain if the sulfur amino acids in SSA were simply additive to those in the parental rice grain. Instead, the results indicated that sulfur amino acids were re-allocated to SSA from other pools in the grain, most probably storage proteins. A 25% increase in aspartic acid + asparagine was also observed in the transgenic grain (Table 2). One possible explanation for this result is that non-protein aspartic acid + asparagine increased in the transgenic grain. Studies in barley and wheat showed an increase in total and non-protein aspartic acid + asparagine in the grain of plants grown in sulfur-deficient soil (Byers and Bolton, 1979; Shewry et al., 1983; Wrigley et al., 1980). These findings would seem to imply that the entire aspartate amino acid biosynthetic pathway may be subject to sulfur availability.

Table 2.  Total amino acid content of mature control and sunflower seed albumin (SSA) transgenic rice grain
Amino acidControl (µmol g−1 DW)SSA transgenic (µmol g−1 DW)
  • Flour samples were produced from pooled grain of plants grown under the same conditions. Amino acids (n = 3) were quantified by oxidation, acid hydrolysis, and separation on an amino acid analyzer followed by derivatization with ninhydrin. Each value represents mean ± SD.

  • a

    Aspartic acid + asparagine.

  • b

    Glutamic acid + glutamine.

Asxa48.1 ± 1.560.1 ± 7.5
Thr20.1 ± 0.821.0 ± 0.8
Ser34.3 ± 1.033.3 ± 1.0
Glxb96.5 ± 17.095.8 ± 12.9
Pro26.9 ± 2.626.9 ± 2.6
Gly38.6 ± 1.337.3 ± 1.3
Ala47.1 ± 5.647.1 ± 6.7
Val35.0 ± 0.933.3 ± 2.6
Ile22.1 ± 1.524.4 ± 3.0
Leu42.7 ± 3.045.7 ± 4.6
Tyr18.2 ± 1.118.2 ± 1.7
Phe22.4 ± 1.821.2 ± 2.4
Lys17.8 ± 0.719.8 ± 1.4
His13.5 ± 0.612.9 ± 0.6
Arg35.6 ± 2.333.9 ± 2.9
Cys10.9 ± 0.89.3 ± 0.8
Met10.1 ± 1.312.8 ± 1.9

We measured the concentrations of three major, reduced sulfur metabolites in the developing endosperm tissue of control and SSA transgenic rice. The transgenic plants contained approximately the same levels of free cysteine and methionine as the control plants. However, transgenic endosperm expressing SSA contained a substantially smaller glutathione pool compared to the control (Table 3).

Table 3.  Glutathione (GSH), free cysteine (Cys), and free methionine (Met) content of developing control and sunflower seed albumin (SSA) transgenic rice endosperm
 GSH content (nmol g−1 DW)Free Cys content (nmol g−1 DW)Free Met content (nmol g−1 DW)
  1. Endosperm tissue extracts from pooled developing grain were analyzed for glutathione and cysteine (n = 4) after treatment with monobromobimane (mBBr), and methionine (n = 3) after treatment with OPA. Each value represents mean ± SD.

Control310 ± 2292 ± 841 ± 2
SSA transgenic111 ± 1696 ± 637 ± 5

Expression of SSA in transgenic rice alters the protein profiles of developing endosperm and mature grain

The protein profiles of both mature grain and developing endosperms were analyzed to determine the relationships between nitrogen, sulfur and amino acids from mature grain, and gene expression in developing endosperm. We observed changes in the levels of six major endogenous proteins (designated 1–6, Figure 2). Proteins of approximately 22 kDa (band 6) and 33 kDa (band 5) were at lower levels in the mature transgenic grain and the developing transgenic endosperms, while proteins of approximately 13 kDa (band 4), 56 kDa (band 3), and 68 kDa (band 1) were at higher levels (Figure 2). A protein of approximately 60 kDa (band 2) was at higher level in the developing transgenic endosperm but was not at discernible level in the mature transgenic grain.

Figure 2.

Protein profiles of developing endosperm and mature grains of control and sunflower seed albumin (SSA) transgenic rice.

Protein samples were extracted from pooled seeds of four to six plants. Endosperm or whole grain protein extracts were separated by SDS–PAGE. Six protein bands of differing intensity (designated 1–6) were cut out of the gel for identification (Table 4).

(a) Protein from 12 DAF control (C) and SSA transgenic (Tg) rice endosperms.

(b) Protein from mature control (C) and SSA transgenic (Tg) rice grain.

Selected polypeptides were excised from polyacrylamide gels, and based on amino acid sequence and peptide mass fingerprint data (Table 4), the 22 kDa (band 6) and 33 kDa (band 5) proteins observed at lower levels in the transgenic endosperm were identified as the basic and acidic subunits of glutelin A. Glutelin A consists of at least three members (types I, II, and III) sharing over 80% identity in their amino acid sequences (Okita et al., 1989; Takaiwa et al., 1987). From the data we obtained, it was not possible to identify which subfamily member we isolated or if more than one subfamily member was represented in the excised band. Based on available sequence information (Okita et al., 1989; Takaiwa et al., 1987), we predicted that all family members would co-migrate using one-dimensional SDS–PAGE. The strong band of approximately 13 kDa (band 4), over-represented in the SSA transgenic endosperm, was identified as either prolamin 7 or prolamin 14 by peptide mass fingerprinting. These two storage proteins share 95% identity in their amino acid sequences (Kim and Okita, 1988). The 56 kDa band (band 3) was identified by MS/MS (Tandem Mass Spectrometry) as the precursor of glutelin B type I or II. Sequences from both the acidic and the basic portions of glutelin B precursor were represented in the excised band, and the migration of the protein corresponded with the predicted size for the glutelin precursor. The 68 kDa (band 1) and 60 kDa (band 2) proteins at higher levels in the transgenic grain did not correspond to major storage proteins and were identified (Table 4) as binding protein (BiP) and protein disulfide isomerase (PDI), respectively. Both BiP and PDI are ER-resident chaperones that aid in the folding and assembly of proteins routed through the endomembrane system.

Table 4.  Identification of polypeptides fractionated by SDS–PAGE
BandPeptides matching mass fingerprintδ mass (p.p.m.)Protein (% coverage)
  1. Bands 2, 4, and 5 were identified by peptide mass fingerprinting. The difference in mass between the significant peak in MS spectra and the theoretical mass of the matching peptide is shown (δ mass) for each peptide. Bands 1 and 6 were identified by N-terminal sequencing, and band 3 was identified by MS/MS sequencing.

2TADEIVDFIK118.29Protein disulfide isomerase (32.6%)
4YGIYPR122.34Prol 7/Prol 14 (35.6/35.8%)
5LQAFEPIRSVR26.68Glutelin A – acidic subunit (25.5%)
Amino acid sequence(s) Match
1KEETKKLGTVIGIDL Immunoglobulin binding protein (BiP)
3PSTNPWHSPR Glutelin B (type I or II) precursor
6GLDETFCTLRVRQNI Glutelin A (type I) basic subunit

Storage protein genes show differences in transcriptional and post-transcriptional regulation in the SSA transgenic grain

Messenger RNA levels for selected proteins were determined in order to investigate the mechanisms of the observed changes in endogenous endosperm proteins of the SSA transgenic grain. Messenger RNA levels were also determined for the ssa transgene and the 10 kDa sulfur-rich prolamin gene (Figure 3). The 10 kDa sulfur-rich prolamin protein contains 20% methionine and 10% cysteine residues (Masumura et al., 1989), and represents a significant pool of sulfur in the rice grain.

Figure 3.

Northern analysis of 12 DAF control (C) and sunflower seed albumin (SSA) transgenic (Tg) rice grains.

Total RNA was extracted from the endosperm tissue of pooled grain from four to six plants and separated on a 1.4% agarose gel. Ribosomal RNA bands are shown for each sample. The RNA was then blotted onto a nylon membrane and hybridized with a different 32P-labeled DNA probe: ssa, sulfur-poor 13 kDa prolamin (Prol7), sulfur-rich 10 kDa prolamin (λRP10), glutelin A (REE61), and glutelin B (REEK1).

Northern analysis of the 13 kDa sulfur-poor prolamin showed that mRNA levels were approximately threefold higher in the transgenic sample compared to the control, while glutelin A mRNA levels were twofold lower (Figure 3). These changes mirror the protein changes and suggest that the differences in protein levels are the result of regulation prior to translation. Message levels of the 10 kDa sulfur-rich prolamin were approximately 2.5-fold lower in the SSA transgenic sample compared to the control (Figure 3). Although this protein was not identified by SDS–PAGE analysis, we predict that it would co-migrate with SSA, making any changes difficult to detect using one-dimensional SDS–PAGE. There was no detectable change in the glutelin B mRNA level in the transgenic rice compared to the control (Figure 3).

To discriminate between transcriptional and post-transcriptional effects on gene expression, nuclear run-on assays were performed to assess the abundance of nascent mRNAs encoding a number of proteins with altered accumulation in the SSA transgenic grain (Figure 4). This method was used to estimate the rate of transcription of these genes. In the case of the 13 kDa sulfur-poor prolamin, there was approximately 2.5-fold more labeled mRNA produced over the 1 h incubation period in transgenic grain than in control, indicating a higher rate of transcription of this gene in the transgenic grain. Conversely, for the 10 kDa sulfur-rich prolamin, approximately threefold less labeled mRNA was produced in extracts from the SSA transgenic grain than from the control (Figure 4). These results correlate with the changes in steady-state mRNA levels for these storage proteins, as shown by Northern blot analysis (Figure 3), indicating that the changes in prolamin abundance in SSA transgenic endosperm were because of changes in the rate of transcription of the corresponding genes. In contrast, the level of labeled glutelin A mRNA was similar in the transgenic and control endosperms (Figure 4). This did not correlate with the steady-state levels of glutelin A mRNA, which were lower in SSA transgenic endosperms compared to the control (Figure 3). We conclude that the reduced accumulation of glutelin A was because of a post-transcriptional mechanism that resulted in a reduction of mRNA stability in the transgenic endosperm. A similar level of labeled glutelin B mRNA was observed in the transcription assay for control and transgenic endosperms (Figure 4), consistent with the observation that total glutelin B mRNA levels did not alter. The fact that unprocessed glutelin B protein levels were increased without a concomitant increase in transcription or mRNA level indicates that there was a translational or post-translational regulation of glutelin B precursor accumulation. Similar levels of labeled mRNA were detected for the ribosomal RNA gene and the starch branching enzyme gene in control and transgenic endosperms (Figure 4). These genes served as controls in that their transcription did not differ between the control and SSA transgenic endosperms.

Figure 4.

In vitro nuclear run-on transcription using isolated nuclei from 12 DAF control (C) and sunflower seed albumin transgenic (Tg) rice grains.

Nuclei were isolated from pooled developing whole grains of four plants and incubated with unlabeled nucleotide triphosphates and 32P-labeled GTP. After incubating for 1 h at 28°C, RNA was isolated and hybridized on a nylon membrane with cloned DNA representing genes for ribosomal RNA (TA71), 13 kDa prolamin (Prol7), 10 kDa prolamin (λRP10), glutelin A (REE61), glutelin B (REEK1), and rice starch branching enzyme (BE1). The experiment was performed twice with consistent results.


Transgenic rice grain that accumulated large amounts of a foreign protein rich in methionine and cysteine was found to contain no more total sulfur or organic sulfur than non-transgenic control grain. The sulfur amino acids required for the foreign protein appeared to be derived largely by redistribution from endogenous sulfur amino acid-containing proteins. These endogenous proteins were replaced by proteins with little or no sulfur amino acids so that, overall, there was no change in grain nitrogen level. These changes in expression were achieved by a combination of transcriptional, post-transcriptional, and post-translational mechanisms. These results imply that the developing rice grain was unable to increase its supply of sulfur for protein synthesis in response to an increased demand from an added, transgenic sulfur sink. This is in contrast to reports of increases in organic sulfur in transgenic oilseeds, grain legumes, and cereals expressing sulfur-rich proteins (Altenbach et al., 1992; Lai and Messing, 2002; Molvig et al., 1997). A comparison of the sulfur fractions in rice and lupin expressing SSA is shown in Figure 5. In the SSA-lupin, reduced sulfur in the seed was increased at the expense of oxidized sulfur. The accumulation of significantly more sulfur amino acids was demonstrated in the transgenic seeds (Molvig et al., 1997). The results were consistent with increased biosynthesis of methionine in the transgenic seeds compared to controls grown under the same conditions (Tabe and Droux, 2002). It is not known whether the developing endosperm of rice is active in reductive sulfur assimilation or if it imports sulfur in a reduced form. The finding that sulfur in rice phloem is predominantly in the form of glutathione supports the latter hypothesis (Kuzuhara et al., 2000). In either case, rice expressing SSA was apparently unable to respond to the increased demand by supplying the required sulfur amino acids.

Figure 5.

The sulfate and organic sulfur composition of control and transgenic seeds expressing sunflower seed albumin (SSA).

The average level of oxidized (sulfate) and reduced (organic) sulfur were determined for control and transgenic lupin (n = 5) and rice (n = 6) seeds by X-ray fluorescence spectrometry analysis. The proportion of organic sulfur represented by SSA in rice (this paper) and lupin (Molvig et al., 1997) was calculated based on estimations of SSA expression levels. The height of each column represents total seed sulfur.

The changes in storage protein composition in the SSA transgenic rice show some similarities to the changes reported in seeds produced under sulfur-deficient conditions. In both cases, a high degree of phenotype plasticity is exhibited with respect to the composition of the storage protein fraction. In general, a constant level of protein nitrogen is maintained, with significant changes in the relative levels of sulfur-rich and sulfur-poor protein components of this fraction. The elevated level of the sulfur-poor prolamin in the SSA transgenic rice is analogous to the upregulation of vicilin in sulfur-deficient peas (Chandler et al., 1984; Randall et al., 1979), conglutin β in sulfur-deficient lupins (Blagrove et al., 1976), and the β-subunit of β-conglycinin in sulfur-deficient soybeans (Gayler and Sykes, 1985; Holowach et al., 1984). The 13 kDa sulfur-poor prolamin was also elevated in the seeds of non-transgenic rice plants grown under sulfur-deficient conditions (Hagan and Randall, unpublished results). Like the pea vicilin, this rice prolamin (Prol7/Prol14) is predicted to be devoid of cysteine and methionine residues (Kim and Okita, 1988; Masumura et al., 1990). Although not clear by SDS–PAGE, protein levels of the 10 kDa sulfur-rich prolamin (λRP10) are likely to be considerably lower in the SSA transgenic grain than in the control grain. Northern blot analysis showed lower levels of 10 kDa prolamin mRNA in the transgenic endosperm than in the control. On one-dimensional PAGE, this protein would co-migrate with SSA, making differences in protein abundance difficult to detect. Decreased accumulation of the two subunits of a glutelin A protein was observed in the transgenic SSA rice grain. Members of the glutelin A subfamily have a moderate concentration of sulfur amino acids: the mature subunits of type I glutelin A together contain 1.9% cysteine and methionine, while both types II and III contain 2.3% cysteine and methionine (Takaiwa et al., 1991). Members of the glutelin A subfamily are present at high levels in rice grains; therefore, despite their relatively modest content of sulfur amino acids, they represent a significant endogenous pool of organic sulfur.

The changes in seed storage protein composition that result from responses to sulfur deficiency have been attributed to both transcriptional and post-transcriptional mechanisms of gene regulation (Beach et al., 1985; Hirai et al., 1995; Morton et al., 1998). Our results indicate that the decreased accumulation of glutelin A in SSA transgenic rice grain is consistent with the mediation by a post-transcriptional mechanism. Similar post-transcriptional regulation of storage protein gene expression has been demonstrated in peas grown under sulfur-deficient conditions. Nuclear run-on experiments in developing pea seeds showed that, under conditions of sulfur-deficiency, the levels of sulfur-rich legumin and pea albumin 1 (PA1) were decreased without any change in the transcription of these genes (Beach et al., 1985; Chandler et al., 1983; Higgins et al., 1986). Our results also indicate that the increased abundance of 13 kDa sulfur-poor prolamin may be because of an increase in transcription of this gene in SSA transgenic rice. Transcriptional control of a sulfur-poor storage protein under conditions of sulfur deficiency has been demonstrated for the β-subunit of soybean β-conglycinin (Holowach et al., 1986). Hirai et al. (1995) demonstrated that the β-subunit promoter retained its sulfur-responsiveness control when fused to the GUS reporter gene and transformed into Arabidopsis thaliana.

The most likely signals sensed by these mechanisms are levels of metabolites, such as O-acetylserine, free methionine, or glutathione. Research into the β-subunit promoter suggested that O-acetylserine, a key intermediate in cysteine biosynthesis, was the signal that stimulated β-subunit gene transcription in response to sulfur deficiency (Kim et al., 1999). However, the application of methionine suppressed the levels of β-subunit mRNA in cultured developing seeds of transgenic petunia (Fujiwara et al., 1992), suggesting that, directly or indirectly, methionine can also act as a signaling molecule. The size of free methionine and free cysteine pools in developing SSA transgenic rice endosperm were unchanged compared to the control endosperm, but glutathione (GSH) levels were found to be threefold lower (Table 3). The decrease in glutathione is consistent with a shortage in sulfur amino acid supply in the developing endosperm of the SSA transgenic line. It is possible that glutathione has a signaling role in the regulation of seed storage protein genes in response to sulfur availability. However, we cannot rule out the possibility that transient changes in free methionine or cysteine may be involved in signaling.

Our findings in SSA transgenic rice are consistent with results reported for transgenic maize expressing a sulfur-rich zein (Anthony et al., 1997). This group observed that methionine levels in some kernels could be increased by up to 30%, although this was not always correlated with the over-expressed zein. As in the case of rice here, they noted that, in at least one line, when the transgenic high-sulfur zein was expressed, there was a concomitant decrease in two endogenous high-sulfur zeins. When the same gene for a sulfur-rich zein was introduced back into maize with cis DNA regions selected for endosperm-specific expression, but with other cis-acting regulatory sites removed, Lai and Messing (2002) described a 15% increase in seed methionine content when expressed as a proportion of total protein. The protein profiles were not shown for the maize kernels, so it was not possible to decide whether this increase was to some extent achieved at the expense of other endogenous sulfur-rich proteins.

The increased levels of the non-storage proteins, BiP and PDI, in association with SSA accumulation in the transgenic rice endosperm may not be directly related to sulfur signal transduction but may reflect ER stress. BiP aids in general polypeptide folding, whilst PDI specifically assists in the formation of disulfide bonds between cysteine residues of newly formed or misfolded proteins. These ER-localized chaperones have been associated with the unfolded protein response, a phenomenon first described in mammalian cells whereby ER-resident proteins are upregulated in response to ER stresses (Kozutsumi et al., 1988). In mammalian, plant, and yeast cells, BiP accumulation is induced by inhibitors of glycosylation and conditions that promote protein denaturation and aggregation. BiP was shown to be upregulated in the seeds of maize floury2 mutants that produce a mutant zein storage protein with a defective signal peptide processing site (Coleman et al., 1995). The aberrant zein processing in floury2 causes anchoring of the protein to the ER membrane and results in severe ER stress accompanied by a dramatic increase in the expression of BiP and other ER chaperones (Boston et al., 1991, 1995). Recent evidence indicates that BiP may also play a significant role in prolamin protein body formation in rice. The majority (90%) of BiP in developing rice endosperm was found in the protein body fraction, with less than 10% in the rough ER microsomal fraction (Li et al., 1993). A further study showed that BiP is localized to the periphery of the developing rice prolamin protein body (Muench et al., 1997). In the SSA transgenic rice endosperm, the elevated level of BiP could be associated with changes in prolamin composition resulting from accumulation of the foreign protein. In order to optimize SSA protein stability, the SSA construct used in this study contained a C-terminal KDEL motif, which has been shown to retain proteins in the ER (Munro and Pelham, 1987; Wandelt et al., 1992). The upregulation of both BiP and PDI in transgenic developing endosperm raises the possibility that at least some of the SSA is misfolded.

Another response to SSA accumulation that may not be directly related to competition for sulfur was the increase in the level of unprocessed glutelin B. From Northern and nuclear run-on analyses, it appeared that glutelin B genes were expressed at similar levels in control and SSA transgenic endosperm. The simplest explanation to these observations is a decrease in the processing of glutelin B polypeptides. Normal processing of glutelin B would produce subunits of a size similar to those of glutelin A (Takaiwa et al., 1991). SDS–PAGE of rice endosperm protein showed that several bands of these sizes were decreased in the transgenic endosperm. The cause of the glutelin B-specific processing failure is unclear. It has recently been reported that rice mutants lacking PDI show a lesion in processing of glutelins, although types A and B were not distinguished (Takemoto et al., 2002). This lesion seemed to arise from defects in prolamin protein body formation and transport, resulting ultimately from illicit intermolecular disulfide bonds between glutelins and prolamins in the mutant grain. In the case of SSA transgenic rice, PDI is actually over-abundant, but it may be hampered in its function by its location, the presence of SSA, or by changes in endogenous storage proteins.

Transgenic rice over-expressing a gene for SSA was not suitable as a potential production system for protein sulfur, as the sulfur amino acid content remained largely unchanged. Instead, the transgenic rice grain showed a redistribution of sulfur from endogenous proteins to SSA. The mechanisms involved were complex in that they involved transcriptional, post-transcriptional, and possibly post-translational controls. We propose that a homeostatic mechanism is active in the transgenic rice grain similar to that involved in modulating seed storage protein composition in response to sulfur and nitrogen availability. Nitrogen and sulfur sensing in seeds is subject to sophisticated regulation, and the signal transduction pathway is still largely unknown.

Experimental procedures

Plant material

Transgenic rice was produced by microparticle co-bombardment of Japonica rice (Oryza sativa) cv. Taipei as described in Li et al. (1997) with a plasmid containing the gene of interest (pLT10 or pLT11) and a plasmid containing the hygromycin-selectable marker gene. The transgene constructs (Figure 1a) consisted of the ssa cDNA (SF8, Kortt et al., 1991) with the KDEL endoplasmic reticulum retention sequence (Munro and Pelham, 1987) at the 3′ terminus of the coding region, using the nos 3′ region and either 1.3 kb of the 1Dx5 wheat high molecular weight glutenin promoter (Lamacchia et al., 2001) (pLT10) or 1.3 kb of the Bx17 wheat high molecular weight glutenin promoter (Reddy and Appels, 1993) (pLT11). The pLT11-derived, homozygous transgenic line 64-90 was selected for detailed study, using non-transgenic plants of the parental line as the control. Plants were grown in the glasshouse in a mixture of 75% potting mix/25% perlite with 1 g l−1 solid calcium sulfate (gypsum) to ensure an adequate sulfur supply to the rice plants. Whole endosperms were harvested at 12 days after flowering (DAF) and whole grains at maturity. Control and transgenic lupins (Molvig et al., 1997) were grown in soil in the glasshouse supplemented with complete fertilizer as described in Tabe and Droux (2002).

Sunflower seed albumin analysis

Pooled mature rice grains were ground to a fine powder in liquid nitrogen. Approximately 100 mg of rice powder was mixed with 6 volumes of a buffer consisting of 62.5 mm Tris–HCl (pH 6.8), 2% (w/v) SDS, 10% (v/v) glycerol. The sample was heated to 80°C for 5 min and then centrifuged at 10 000 g for 10 min. The resulting sample consisted of a gelatinous pellet containing the insoluble starch and the supernatant, which was used for further analysis. Protein was fractionated by SDS–PAGE and electroblotted onto nitrocellulose membrane. SSA was detected with SSA antiserum from goat in combination with rabbit antigoat IgG conjugated to alkaline phosphatase. Polyacrylamide gels were loaded with approximately 20 µg total grain protein.

Chemical analyses

Mature rice or lupin seed samples were milled to fine flour using a puck mill. Total nitrogen was determined using an autoanalyzer after Kjeldahl digestion (Heffernan, 1985). Total sulfur, oxidized sulfur (corresponding to sulfate), and reduced sulfur (corresponding to carbon-bonded sulfur) were determined by X-ray fluorescence spectrometry. Powdered samples were pressed into aluminum planchettes, and then total sulfur, reduced sulfur, and oxidized sulfur were determined using a PW 1404 spectrometer (Philips, Eindhoven, the Netherlands) as previously described (Pinkerton et al., 1989; Tabe and Droux, 2001). The total amino acid composition of mature rice seeds was determined after complete hydrolysis of finely ground flour as previously described (Tabe and Droux, 2001). All analyses listed above were performed on pooled flour samples from mature grain. For quantification of sulfur metabolites, approximately 20 mg of dried, powdered endosperm tissue from pooled developing rice grains was extracted into 1 ml of 25 mm HCl at room temperature for 30 min. Insoluble material was removed by centrifugation in a bench microfuge at 10 000 g for 5 min. For quantification of cysteine and glutathione, a sample of the acid supernatant was reacted with monobromobimane (mBBr) and then analyzed by reverse phase HPLC as previously described for homo-cysteine (Droux et al., 1995). For quantification of free methionine, a sample of the supernatant was reacted with O-phthaldialdehyde (OPA) immediately before separation by reverse phase HPLC as previously described (Tabe and Droux, 2002).

Protein profiles and protein identification

Protein from mature endosperms was extracted as described in the ‘Sunflower seed albumin analysis’ section. Endosperm tissue of developing seeds pooled from four to six plants was ground in liquid nitrogen and homogenized in 4 volumes of a buffer consisting of 62.5 mm Tris–HCl (pH 6.8), 2% (w/v) SDS, 10% (v/v) glycerol, followed by heating and centrifugation as described above. Aliquots of supernatant containing approximately 50 µg protein were mixed with 2 volumes of 7 m acrylamide and the mixture incubated at 25°C in the dark for 1 h to derivatize cysteine residues. One volume of 0.9 m Tris–HCl (pH 8.45), 24% (v/v) glycerol, 8% (w/v) SDS, 0.01% (w/v) Coomassie G, 0.01% (w/v) phenol red, 5 mm DTT was added to this mixture, which was then separated on a 15–30% gradient SDS polyacrylamide gel (acrylamide:bis-acrylamide, 75 : 1) for 24 h at 10 V cm−1. Six major protein bands of differing intensity were cut out of the gel and extracted with a 50% (v/v) acetonitrile, 0.5% (v/v) TFA solution. Edman degradation was performed on all protein samples using a 494 Procise Protein Sequencing System (Applied Biosystems, California, USA). N-terminal protein sequence was obtained for proteins 1 and 6 using this method; the other proteins were either N-terminally blocked or produced ambiguous sequences. A 16 h tryptic digest was performed on these remaining proteins prior to analysis by matrix-assisted laser desorption ionization mass spectrometry using a TofSpec 2E Time of Flight Mass Spectrometer (Micromass, Manchester, UK). Peptide masses were searched against swiss-prot and trembl using peptident software, and good matches were found for proteins 2, 4, and 5. In Table 4, the δ mass for each matching peptide is shown. This measurement represents difference between the mass determined from the MS spectra and the theoretical mass of the matching peptide. The proportion of the protein covered by the matching peptides is typically well below 100% as some peptides may be post-translationally modified or may fall outside the detectable mass range. Protein 3 was analyzed by ESI-TOF tandem mass spectrometry (MS/MS) using a Micromass Q-TOF MS. Identification of samples analyzed by Edman degradation or MS/MS was achieved using standard blastp searches. In each case, an exact match was found in the GenBank database.

Northern hybridization analysis

Total RNA was extracted from endosperm tissue of 4 g fresh, developing (12 DAF) rice grains pooled from four to six plants. Tissue samples were frozen in liquid nitrogen and ground in 2 volumes of 1 m Tris–HCl (pH 9.0), 1% (v/v) β-mercaptoethanol and 3 volumes of phenol:chloroform:isoamyl alcohol (25 : 24 : 1). Samples were vortexed vigorously for 2 min and centrifuged for 10 min at 10 000 g. The supernatant was precipitated with ethanol and re-suspended in diethyl pyrocarbonate-treated water. A solution of LiCl was added to a final concentration of 2 m and the RNA precipitated for 12 h at 0°C. The solution was centrifuged for 10 min at 10 000 g and the pellet re-suspended in diethyl pyrocarbonate-treated water. Approximately 10 µg of each RNA sample was denatured at 65°C for 10 min in 50% (v/v) formamide, 17.5% (v/v) formaldehyde, 1× MOPS, 0.5 mg ml−1 ethidium bromide and separated on a 1.4% agarose gel containing 5% (v/v) formamide. RNA was transferred to a nylon membrane by blotting in 20× SSC. Individual membranes containing one control lane and one SSA transgenic lane were probed with 32P-labeled DNA from the gel-purified cDNAs representing genes for SSA (SF8), 13 kDa prolamin (Prol7), 10 kDa prolamin (λRP10), glutelin A (REE61), and glutelin B (REEK1). The membrane was washed with 2× SSC at room temperature, followed by 2× SSC, 0.1% SDS, 0.1% sodium pyrophosphate at room temperature and 0.1× SSC, 0.1% SDS, 0.1% sodium pyrophosphate at 42°C. These high-stringency conditions were used to ensure virtually no cross-reaction with non-subfamily members. imagequant software (Version 3.3; Molecular Dynamics) was used to quantify bands on the developed blots.

Nuclear run-on transcription

Approximately 2 g of developing whole rice grains (12 DAF) pooled from four plants were chopped with a razor blade to a fine slurry in 4 volumes of 2.5% (w/v) Ficoll 400, 5% (w/v) Dextran T40, 250 mm sucrose, 50 mm Tris–HCl (pH 8), 10 mm MgCl2, and 10 mmβ-mercaptoethanol. The slurry was filtered through Miracloth and the filtrate centrifuged at 2000 g for 5 min. The pellet was re-suspended gently in 20 ml of 50 mm Tris–HCl (pH 8.0), 10 mm MgCl2, 10 mmβ-mercaptoethanol (Tris–Mg-βMe) with 1% (w/v) Triton X-100 and centrifuged again at 2000 g for 5 min. This process was repeated using 20 ml of Tris–Mg-βMe without Triton X-100. The final pellet containing nuclei and starch was re-suspended in 1.9 ml Tris–Mg-βMe and used in the in vitro transcription run-on assay. Hundred microliters of 1 m (NH4)2SO4 was added to the nuclei suspension, in addition to ATP, CTP, and UTP, at a final concentration of 340 µm and GTP at a final concentration of 5 µm. Finally, 400 µCi of 32P-labeled GTP was added and the suspension was incubated for 1 h at 28°C. RNA was isolated and hybridized on a nylon membrane which had been pre-blotted with gel-purified cDNAs representing genes for ribosomal RNA (TA71), 13 kDa prolamin (Prol7), 10 kDa prolamin (λRP10), glutelin A (REE61), glutelin B (REEK1), and starch branching enzyme (BE1). The membrane was washed with 2× SSC at room temperature, followed by 2× SSC, 0.1% SDS, 0.1% sodium pyrophosphate at room temperature and 0.1× SSC, 0.1% SDS, 0.1% sodium pyrophosphate at 42°C. These high-stringency conditions were used to minimize cross-reaction between cDNAs. imagequant software was used to quantify bands on the developed blots.


We thank Dr T. Okita for supplying rice clones Prol7, REE61, and BE1; Dr K. Tanaka for supplying rice clone λRP10; and Dr F. Takaiwa for supplying rice clone REEK1. We also thank Dr P. Randall for supplying sulfur-deficient rice seeds; Judy Gaudron and Belinda Schouten for excellent technical assistance; and Dr D. Spencer for manuscript revisions. This research was facilitated by access to the Australian Proteome Analysis Facility established under the Australian Government's Major National Research Facilities Program.