High temperature impairs rice (Oryza sativa) grain filling by inhibiting the deposition of storage materials such as starch, resulting in mature grains with a chalky appearance, currently a major problem for rice farming in Asian countries. Such deterioration of grain quality is accompanied by the altered expression of starch metabolism-related genes. Here we report the involvement of a starch-hydrolyzing enzyme, α-amylase, in high temperature-triggered grain chalkiness. In developing seeds, high temperature induced the expression of α-amylase genes, namely Amy1A, Amy1C, Amy3A, Amy3D and Amy3E, as well as α-amylase activity, while it decreased an α-amylase-repressing plant hormone, ABA, suggesting starch to be degraded by α-amylase in developing grains under elevated temperature. Furthermore, RNAi-mediated suppression of α-amylase genes in ripening seeds resulted in fewer chalky grains under high-temperature conditions. As the extent of the decrease in chalky grains was highly correlated to decreases in the expression of Amy1A, Amy1C, Amy3A and Amy3B, these genes would be involved in the chalkiness through degradation of starch accumulating in the developing grains. The results show that activation of α-amylase by high temperature is a crucial trigger for grain chalkiness and that its suppression is a potential strategy for ameliorating grain damage from global warming.
Grain filling of cereals is vulnerable to environmental stress including high temperatures. For japonica cultivars of rice (Oryza sativa L.), a temperature above 26 °C during the first half of the ripening period adversely affects yield through a decrease in grain size (Peng et al., 2004; Tashiro and Wardlaw, 1991) and quality due to impaired deposition of storage materials such as starch (Morita et al., 2004; Yamakawa et al., 2007). Notably, the loss of quality, often recognizable as chalky grain, reduces the commercial value because of increased cracking during polishing (Fitzgerald et al., 2009). Microscopic observation of chalky areas of grain ripened under high temperature revealed that air spaces remain among loosely packed starch granules and reflect light, thus appearing chalky (Yamakawa et al., 2007; Zakaria et al., 2002). Importantly, small pits were observed on the surface of the starch granules, suggesting the involvement of starch-degrading enzymes in grain chalkiness (Iwasawa et al., 2009; Zakaria et al., 2002).
α-Amylase (EC18.104.22.168) is a starch-hydrolyzing enzyme whose gene family in rice has at least eight members classified into three subfamilies, Amy1, Amy2 and Amy3 (Huang et al., 1990; Mitsui et al., 1993; Ranjhan et al., 1991). The peptide sequences encoded by the respective α-amylase genes have four highly homologous domains containing conserved catalytic residues (Nakajima et al., 1986). In germinating cereal seeds, α-amylase is synthesized de novo in the embryo and aleurone layer (Mitsui et al., 1996; Morita et al., 1998) and converts endosperm starch into metabolizable sugars to nourish the young seedling (Beck and Ziegler, 1989; Mitsui et al., 1996). α-Amylase is one of the enzymes involved in the first committed step for degradation of reserve starch. During the process, synthesis of α-amylase is up-regulated by a plant hormone, gibberellin (GA), but down-regulated by another hormone, abscisic acid (ABA), and by sugars (Thomas and Rodriguez, 1994; Zentella et al., 2002). However, the physiological function and mode of induction of α-amylase expression in developing seeds is largely unknown, although the enzymatic activity and gene expression of α-amylase increase during the grain filling period (Baun et al., 1970; Yamakawa et al., 2007).
Previously, we found that the expression of starch-hydrolyzing α-amylase genes, Amy1A, Amy3D and Amy3E, was increased more than twofold by an elevated temperature during the ripening period, while the expression of many genes for starch biosynthesis was down-regulated (Yamakawa and Hakata, 2010; Yamakawa et al., 2007), preventing the accumulation of starch. Additionally, the possible involvement of α-amylase in grain chalkiness through impaired starch accumulation in the developing endosperm has been suggested by the production of chalky grains by Amy1A and Amy3D-overexpressing rice plants (Asatsuma et al., 2006), prompting us to generate α-amylase-suppressed plants to overcome such deteriorations in grain quality at high temperature. In this study, we characterized the altered expression of α-amylase genes as well as the increased enzymatic activity caused by high temperatures in developing seeds, and revealed that suppression of α-amylase ameliorates high temperature-triggered occurrence of chalky grains through analyses with α-amylase knock-down transgenic plants. We propose a strategy to produce a ‘premium grain quality’ rice plant which is tolerant of high temperatures during the ripening period.
Activity of α-amylase and expression of Amy1A, Amy1C, Amy3A, Amy3D and Amy3E were induced by high temperature in ripening grain
The morphology of chalky grains caused by high temperature stress during the ripening stage is associated with impaired starch accumulation. As small pits observed on the surface of starch granules in the chalky areas (Iwasawa et al., 2009; Zakaria et al., 2002) clearly indicate that de novo degradation of starch occurs under high temperature in the course of grain filling, the activity of α-amylase, a typical starch-hydrolyzing enzyme, was investigated in the developing seeds. An in vitro enzyme assay was conducted at 37 °C, the temperature ordinarily employed, and at 30 °C/22 °C for the high/control temperature plots respectively, taking account of the difference in ripening temperature between the two plots. In both temperature regimes, the α-amylase activity in grains ripening under high temperature was approximately twofold higher than that in the control (Figure 1a).
Next, the dynamics of the expression of α-amylase genes was determined in the developing caryopsis. In a public rice genome database, RAP-DB (http://rapdb.dna.affrc.go.jp/), eight α-amylase genes designated Amy1A, Amy1C, Amy2A, Amy3A, Amy3B, Amy3C, Amy3D and Amy3E, whose accession numbers are shown in Table S1, were found to encode products with four catalytic regions conserved and an overall peptide sequence homologous to α-amylases from other plant species, and subjected to expression profiling. A series of quantitative real-time RT-PCR analyses using gene-specific primer sets for respective α-amylase genes revealed that the expression of Amy1C and Amy3A as well as Amy1A, Amy3D and Amy3E whose transcripts increased at high temperature in a previous study (Yamakawa et al., 2007) was up-regulated under high temperature (33 °C/28 °C) in the middle of the ripening stage at around 15 days after flowering (DAF; Figure 1b). Amy1A and Amy1C showed a similar pattern of expression in the high temperature plot, gradually increasing from 8 DAF and peaking at 15–20 DAF. Amy3A and Amy3D expression peaked around 15 DAF, while Amy3E showed high levels at 8–30 DAF in the high temperature plot. In contrast, the expression of Amy2A, Amy3B and Amy3C was not induced by high temperature (Figure 1b). The intensity of the expression of all α-amylase genes in the course of grain filling clearly indicates that the newly characterized Amy3A gene was expressed at a considerable level under high temperature from 15 DAF (Figure 1c), although the data obtained by RT-PCR do not always give absolute intensities of expression because of differences in amplification efficiency with the primers used.
Levels of the plant hormone ABA were decreased by ripening under high temperature
As the activity of α-amylase in germinating seeds is known to be regulated by plant hormones such as GA and ABA, which induces and represses the activity, respectively, levels of the respective hormones as well as the expression of genes related to the biosynthesis, deactivation and signalling of plant hormones in response to high temperature were investigated in ripening caryopses. Simultaneous quantification of plant hormones by liquid chromatography-tandem mass spectrometry revealed that exposure to high temperature (33 °C/28 °C) decreased ABA levels 0.42-fold compared to the levels in caryopses ripened under the control temperature (25 °C/20 °C), while it did not increase bioactive GAs, GA1, GA4 and GA7 (Figure 2 and Table S2). Such a change in ABA levels was well consistent with the altered expression of plant hormone metabolism-related genes revealed by a microarray analysis, where gene expression in high (33 °C/28 °C) and the control (25 °C/20 °C) temperature-ripening caryopses was compared (Yamakawa and Hakata, 2010). In Figure 2, the extent of expression of genes encoding each enzyme in the control plot and changes at high temperature are shown by the square size and colour of the heat map, respectively. To clarify the impact of temperature at each reaction step, the ratio of cumulative expression level, which is the ratio of the sum of expression levels for all genes encoding a corresponding enzyme/factor in the high temperature plot to that in the control plot, was calculated as an index for the change in expression of all of the corresponding genes at high temperature, taking into account the difference in the extent of expression for each gene. Thus, major genes would make more of a contribution to the ratio value than minor expressers among the isoforms in a multigene family. The expression of ABA biosynthesis-regulating genes, ABA4, NCED (9-cis-epoxycarotenoid dioxygenase) and ABA2, was repressed 0.54, 0.38 and 0.67-fold, respectively, while that of ABA-deactivating ABA8ox (ABA 8′-hydroxylase) was induced 1.62-fold by the elevated temperature (Figure 2 and Table S3). For other plant hormones, several cytokinins such as isopentenyl adenine (iP) and an amino acid-conjugate of auxin (IA-Asp) were increased by high temperature, whereas stress-related hormones, jasmonic acid (JA) and salicylic acid (SA), were decreased (Figure S1 and Table S2).
Rice plants with suppressed expression of α-amylase genes produced less chalky grain
As α-amylase genes, whose overexpression resulted in severely chalky grain even under ambient temperature (Asatsuma et al., 2006), were induced to express by high temperature, the effect of the down-regulation of α-amylase genes on the extent of grain chalkiness caused by high temperature was examined. To suppress most of the α-amylase genes expressed in developing seeds simultaneously, a 129-bp nucleotide region of Amy1A and Amy3E corresponding to the fourth catalytic domains of their peptides, which is remarkably conserved among the eight α-amylases (Figure S2a), was employed to construct RNAi binary vectors in combination with the developing endosperm-specific prolamin promoter and terminator by arranging two identical fragments derived from Amy1A or Amy3E in a ‘tail to tail’ or ‘head to head’ manner, yielding four different vectors generating artificial ‘hair-pin’-structured transcripts, designated 1Att, 1Ahh, 3Ett and 3Ehh (Figure S2b).
In every independent series of transformants with each of four vectors, multiple lines producing a high ratio of translucent grains (defined as high grain quality) with less immature, chalky grain than the control lines, transformants with a vector without RNAi-triggering chimeric genes, were obtained when ripened under a moderately high temperature, 31 °C/26 °C. Although the high temperature treatment with plant incubators produced various types of chalky grains; milky white, white berry, white back, etc., visual inspection revealed that milky white type of chalky grains was most decreased by the RNAi-mediated suppression of α-amylase. Among stably transformed lines generated with respective vectors, the lines showing various degrees of grain quality; a high to relatively low ratio of translucent grains, 1Att-13, 1Att-14, 1Ahh-1, 1Ahh-15, 3Ett-5, 3Ehh-2, 3Ehh-4 and 3Ehh-14, were selected in the T2 generation (Figure S3a), and the siblings of their descendants were evaluated for the correlation of expression levels of α-amylase genes and the extent of grain chalkiness. In the T3 generation, a similar tendency was observed for improvement of grain quality. The descendants of 1Att-13, 1Att-14, 3Ehh-2 and 3Ehh-4, which showed high grain quality in the T2 generation (the appearance of 1Att-13, 1Att-14 and control grains shown in Figure 3a), 1Att-13-2, 1Att-13-4, 1Att-14-1, 1Att-14-3, 3Ehh-2-2, 3Ehh-2-4 and 3Ehh-4-3, yielded high ratios of translucent grains again compared to the control and non-transformant, wild-type plants in the T3 generation (Figure S3d), clearly showing inheritance of the improved grain quality. The fertility and grain weight of the lines with improved grain quality were similar to those of the control lines (Figures S3b,c and S3e,f).
Expression levels of eight α-amylase genes in the ripening grain at 12 DAF were examined in five to six T3 plants each for the RNAi lines 1Att-13-2, 1Att-13-4, 1Att-14-3, 3Ehh-2-4, 3Ehh-4-3 and 3Ehh-14-3, covering a wide range of grain quality, and the control lines C-4-2 and C-8-2 (Figure 3b). The RNAi lines showed some reduction in the expression of Amy1A, Amy1C, Amy3A, Amy3B and Amy3D (Figure 3c). However, the extent of repression of Amy3C and Amy3E was relatively small, and expression of Amy2A was hardly affected at all. 1Att-13-2, which produced a high ratio of non-chalky, translucent grains, had the most effectively reduced expression of α-amylase genes such as Amy1A, Amy1C, Amy3A, Amy3B and Amy3D, whereas 3Ehh-14-3, the line with a similar level of grain chalkiness to the control, expressed these genes at similar levels to the control lines.
Then, the correlation between expression levels of respective α-amylase genes and grain quality was investigated by plotting the gene expression and grain quality data for six RNAi and two control lines. The plots clearly showed negative correlations between grain quality and expression levels with particularly high correlations detected for Amy1A, Amy1C, Amy3A and Amy3B (coefficient of determination, R2, of 0.69, 0.75, 0.67 and 0.61, respectively; Figure 4).
To investigate the impact of suppressed expression of α-amylase genes on starch biosynthesis, the amount of a starch component, amylose, whose synthesis is governed by a granule-bound type of starch synthase, was determined in grains of the α-amylase knock-down and control lines. As the ripening temperature increased, the amylose content of non-transformant, wild-type (WT), Nipponbare grains decreased due to impaired starch biosynthesis (Figure S4). However, grains of both α-amylase knock-down and control lines had similar amylose contents to those of the non-transformant, when ripened under the same temperature condition, 31 °C/26 °C, suggesting that suppression of the starch-hydrolyzing enzyme itself does not influence starch biosynthesis, at least for amylose.
The accumulation of starch in rice endosperm is impaired at high temperatures, causing grains to appear chalky because of the imperfect filling of endosperm cells with starch granules (Yamakawa et al., 2007). The present study showed that a starch-hydrolyzing enzyme, α-amylase, was induced in ripening grains by an elevated temperature at both the gene expression and activity levels (Figure 1). Furthermore, the suppression of α-amylase genes by a transgenic approach improved grain filling under high temperature and reduced the amount of chalky grain produced by heat stress (Figure 3), revealing the unexplored physiological significance of α-amylase in developing seeds and a potential new strategy for controlling grain quality by modulating the expression of a starch-metabolizing enzyme.
The high temperature conditions employed in the present study, 33 °C/28 °C and 31 °C/26 °C, closely mimic conditions often encountered by rice plants in paddy fields because of recent global warming. Such temperatures, which produce chalky grains in both the field and cultures in the laboratory, induced the expression of five of eight α-amylase genes, namely Amy1A, Amy1C, Amy3A, Amy3D and Amy3E, as well as starch-hydrolyzing activity in developing seeds (Figure 1), suggesting that starch once accumulated in the endosperm was hydrolysed and utilized as an energy source probably for adaptation to heat stress. Taking account of the relative expression intensity estimated by our quantitative RT-PCR analysis (Figure 1) and microarray data retrieved from a public database for rice global expression profiles (Figure S5, described in detail in the later section), Amy3E and Amy3A might be major high temperature-induced α-amylase genes with relatively high levels of expression. As overexpression of α-amylases such as Amy1A and Amy3D produced chalky grains even under ambient temperature (Asatsuma et al., 2006), several, if not all, α-amylase isoforms induced by the elevated temperature might be involved in the degradation of endosperm starch granules, in which traces of hydrolysis have been observed as small pits (Iwasawa et al., 2009; Zakaria et al., 2002). It has been also microscopically observed that multiple, extraordinarily small amyloplasts were proliferated in a milky stage of endosperm of grains ripening under high temperature (Iwasawa et al., 2001), suggesting that starch granules included inside of the developing amyloplasts would be immature and small. Therefore, in the early filling stage, α-amylase might attack the periphery of developing starch particles before tightly crystallized without making pits, although our previous metabolite analysis of filling grains failed to detect the immediate products released, oligosaccharides (Yamakawa and Hakata, 2010).
Quantification of plant hormones and determination of expression levels of genes related to hormone biosynthesis and deactivation revealed that ABA levels decreased in high temperature-exposed developing seeds (Figure 2), suggesting the activation of α-amylase to be, at least in part, attributed to the removal of the α-amylase-repressing hormone under high temperature in a manner similar to that in germinating seeds, where ABA represses de novo synthesis of α-amylase (Zentella et al., 2002). If α-amylase in developing seeds is under the regulation of ABA, grain quality could be controlled with chemicals such as the bioactive hormone itself or inhibitors of hormone deactivation in rice plants ripening under high temperature. Actually, an improvement in grain quality on the application of ABA to ripening caryopses was reported (Sasaki et al., 2007). More recently, repression of α-amylase by another hormone, JA was reported in germinating seeds (Yang et al., 2012). In the present study, high temperature decreased JA in the developing seeds. The decrease in α-amylase-repressing JA level might partially contribute to the activation of α-amylase under the high temperature condition.
More evidence for the involvement of α-amylase in grain chalkiness under high temperature comes from analyses of α-amylase-suppressed transgenic plants which produced less chalky grain through generations when ripened at high temperature (Figure 3) without affecting the biosynthesis of a starch component, amylose (Figure S4). Although most α-amylase genes were down-regulated simultaneously in the transgenic plants (Figure 3), the extent of reduction in the expression of Amy1A, Amy1C, Amy3A and Amy3B was highly correlated to the degree of reduction in occurrence of chalky grains (Figure 4), suggesting a contribution of these isoforms to grain chalkiness. Among the eight α-amylase genes in rice, Amy3A has a peculiar expression profile. According to the RiceXPro database (http://ricexpro.dna.affrc.go.jp/) which compiles expression data for entire rice genes covering all organs of rice plants in various developmental stages determined by extensive microarray analyses (Sato et al., 2011), Amy3A is expressed exclusively in the developing endosperm, not in the embryo where other α-amylase genes such as Amy3E are expressed intensively (Figure S5). As the tissue where Amy3A is expressed coincides with the site of starch's degradation at high temperature, the α-amylase produced de novo by Amy3A might directly attack starch granules, although the possibility that α-amylase expressed in the embryo or neighbouring tissues such as aleurone layers would be translocated to starchy endosperm and then hydrolyze the reserve starch could not be excluded. It is of particular interest that respective members of the starch-hydrolyzing α-amylase gene family are expressed in different tissues like the starch biosynthesis-related genes which are divided into endosperm-expressing and pericarp-expressing starch synthases (Hirose and Terao, 2004). In the present study, the suppression of α-amylase decreased milky white type of chalky grains mainly, suggesting that α-amylase is involved in chalkiness in the various regions of endosperm tissue. However, it remains to be determined whether the site where the chalkiness occurs under high temperature coincides with the site where each α-amylase gene expresses. Such issue would be clarified by promoter-reporter gene assays.
In the present study, induction of α-amylase expression by high temperature in developing seeds was observed and its suppression ameliorated damage to the grain such as chalkiness and improved grain quality under high temperature ripening. However, the α-amylases attacking starch granules and causing the chalkiness remain to be identified. Evaluation of grain quality in transgenic plants with ‘one by one’ suppression of each α-amylase isoform or mutants lacking the activity of the corresponding gene is required to elucidate the contribution of the respective α-amylases to grain chalkiness. Once mutants deficient in a certain α-amylase gene are found to improve grain quality, they can be used as genetical resources for breeding high temperature-tolerant rice plants with less chalky grain. In such cases, germination efficiency should be investigated along with grain quality, as most α-amylase genes are expressed intensively in germinating seeds and their mutation might impair seed germinability. Notably, an α-amylase gene, Amy3A, is expressed in ripening endosperm, but not in germinating seeds, according to the MSU database of gene expression (http://rice.plantbiology.msu.edu/). Currently, we are collecting α-amylase mutants by the TILLING method, which allows high-throughput screening for chemically induced point mutations (Colbert et al., 2001; Suzuki et al., 2008), in combination with a rapid backcrossing procedure using biotrons (Ohnishi et al., 2011). Analyses with such mutants and transgenic plants should provide a comprehensive understanding of the physiological significance of the starch-hydrolyzing enzyme as well as materials to develop high temperature-tolerant cereal crops.
Plant materials and growth conditions
Rice (Oryza sativa L., cultivar ‘Nipponbare’) plants were grown in plant incubators (model FLI-2000H or FLI-301NH; Eyela, Tokyo, Japan) as described previously (Yamakawa et al., 2007). Each plant was restricted to the main culm by the removal of the tillers. Five days after heading, half of the plants were exposed to 33 °C, 12-h light/28 °C, 12-h dark for the high temperature plot, while the rest were ripened at 25 °C, light/20 °C, dark for the control. Developing seeds were collected 10 and 12 DAF for the high temperature and control plots, respectively, whose increase in fresh weight has been shown to be similar (Yamakawa et al., 2007), and used for determination of transcript levels, enzymatic activity and plant hormone quantity. For the time-course gene expression study, developing seeds were collected at various DAF indicated.
For experiments with transgenic plants, seeds were sterilized by an antiformin solution and germinated for 2 weeks on Murashige and Skoog agar medium supplemented with 50 mg/L hygromycin, and then the resulting plants were transferred to soil and grown as described above. After flowering, plants were exposed to a moderately high temperature, 31 °C, 12-h light/26 °C, 12-h dark and maintained until the harvesting stage. At 12 DAF, an aliquot of developing caryopses was detached from the ear, dehulled, immediately frozen in liquid nitrogen and stored at −80 °C until used for the assay of transcripts. Approximately 45 DAF, the rest of the caryopses were harvested, dehulled, counted, weighed and photographed. Grain quality, defined as the ratio of non-chalky, translucent grains, was determined with a rice grain grader (model ES-1000; Shizuoka Seiki Co. Ltd, Shizuoka, Japan).
Quantitative RT-PCR and DNA microarray analyses
Total RNA was extracted with an RNeasy Plant Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. First-strand cDNA was synthesized from each RNA preparation (250 ng/reaction) with an oligo (dT) primer using a PrimeScript RT reagent Kit (Takara-Bio, Ohtsu, Japan) in a total volume of 10 μL.
Specific primers for the amplification of transcripts for rice α-amylase genes (Table S3) were used in the quantitative RT-PCR analysis. The UBQ5 primer pair (Jain et al., 2006) was used as an internal control for normalization of the data. The primer pairs yielded a strong single band as fractionated on 2.0% agarose gel, and the amplified fragments were those of the intended gene as confirmed by DNA sequencing.
Quantitative RT-PCR was performed with 1 μL of cDNA template per 25-μL reaction with a Thermal Cycler Dice Real Time system TP850 (Takara-Bio) using SYBR Premix Ex Taq (Takara-Bio). The reaction schedule was 10 s at 95 °C, followed by 45–50 cycles of 5 s at 95 °C and 30 s at 60 °C. The threshold cycle (Ct) was auto-calculated by the accompanying analysis software. The expression level, normalized to that of the endogenous control gene (UBQ5), was calculated by the ΔΔCt method.
The DNA microarray assay was performed as described previously (Yamakawa and Hakata, 2010). The complete set of microarray data in this paper has been deposited in the Gene Expression Omnibus (GEO) repository under accession number GSE20345. A list of selected plant hormone-related genes with corresponding function, RAP-DB accessions, and expression data is available as Supporting Information in the online version of this article (Table S3). As the rice 44-K oligo DNA microarray often contains several features for one given gene, the geometric mean of expression intensity data from all the features corresponding to the same gene was taken in such cases.
Determination of α-amylase activity
The activity of α-amylase was measured by a slightly modified method of Okamoto and Akazawa (1978). Three developing grains were macerated by a Multi-beads Shocker (model: MB301; Yasui Kikai Co., Osaka, Japan) after cooling in liquid nitrogen, suspended in an equal amount of extraction buffer (50 mm sodium-acetate, pH 5.3, 0.5 mm CaCl2 and 5 mm EDTA), and slowly agitated at 4 °C for 3 h. A crude enzyme solution clarified by two rounds of centrifugation at 20 000 g for 15 min and then for 5 min was used for the subsequent enzyme assay.
The reaction was started by addition of 10 μL of the enzyme solution to the same volume of substrate solution, 50 mm sodium-acetate, pH 5.3, 0.5 mm CaCl2, 5 mm EDTA and 0.34% solublized starch (product number 32126-64; Nacalai Tesque, Kyoto, Japan). After incubation at the indicated temperature (22, 30 or 37 °C) for 5 min, the reaction was stopped by adding 25 μL of I2-KI solution containing 0.012% I2, 0.12% KI and 0.05 m HCl. Upon addition of 100 μL of distilled water, the decrease in absorbance at 620 nm was measured immediately. The absorbance at 620 nm of the reaction mixture without the enzyme extract was usually approximately 1.5, which was regarded as 100% (control). One enzyme unit was arbitrarily defined as the activity causing a 10% decrease in absorbance at 620 nm during 5 min incubation. The data were corrected by the protein content of the respective extracts determined by a dye-binding procedure of Bradford (1976) with bovine serum albumin as a standard.
Quantification of plant hormones
Approximately 100 mg of developing caryopses at the milky stage (corresponding to five caryopses) was subjected to quantification of plant hormones by liquid chromatography-tandem mass spectrometry as described previously (Kojima et al., 2009). The genes related to the biosynthesis, deactivation and signalling of respective plant hormones were chosen according to previous reports (Hirano et al., 2008; Kojima et al., 2009).
Generation of α-amylase knockdown transgenic plants
To produce endosperm-specific RNAi-triggering vectors for simultaneous suppression of most of the α-amylase genes, a conserved 129-bp region corresponding to nucleotide 949–1077, counting from the translational initiation site of Amy1A (AK101744), or nucleotide 925–1053 of Amy3E (AK064300) (Figure S2a) was amplified by PCR with BamHI and XbaI-SacI restriction sites at the respective ends. The fragments were cloned in a tail-to-tail or head-to-head orientation into the corresponding restriction sites (between XbaI and BglII sites and between BamHI and SacI sites) of the pZH2B10ik binary vector harbouring a developing endosperm-specific expression cassette consisting of the rice 10-kDa prolamin promoter, third intron of a rice aspartic protease gene and rice 10-kDa prolamin terminator (Kuroda et al., 2010), as indicated in Figure S2b.
The RNAi and empty vectors were introduced into Agrobacterium tumefaciens strain EHA101 by electroporation. Rice transformation was performed as described previously (Toki, 1997).
Determination of amylose content
Apparent amylose content was measured by an iodine colorimetric method as described previously (Yamakawa et al., 2007).
We thank Ms H. Ikeda for technical assistance. We also thank Dr Y. Nagamura and R. Motoyama for technical help with the microarray analysis, and the Rice Genome Resource Center at the National Institute of Agrobiological Sciences for supplying full-length cDNA clones. This work was funded by the Ministry of Agriculture, Forestry and Fisheries of Japan (Genomics for Agricultural Innovation, IPG-0020 to H.Y. and IPG-0021 to T. Mitsui) and National Agriculture and Food Research Organization (Development of innovative crops through the molecular analysis of useful genes, No. 1211 to H.Y.). The plant hormone analysis was supported by the Japan Advanced Plant Science Research Network.