Introduction

Starch is a major storage component of rice grains. The endosperm starch is composed of 20–30 % amylose and 70–80 % amylopectin (Juliano 1985). Amylose has a mainly linear structure with very few α-(1, 6) linkages, whereas amylopectin is highly branched (for review, see Preiss and Sivak 1996). In breeding, the amylose content (AC) needs to be taken into account, as it is the most important factor affecting the eating quality of cooked rice (Webb 1980; Juliano 1985). Rice cultivars with specific AC are required to match the preferences in different countries.

AC is a complex trait in rice (Ikeno 1914) and is controlled by many genes, such as Waxy (Wx) (Sano 1984), Du1 (Satoh and Omura 1981), Du2 and Du3 (Satoh and Omura 1986), and Du4 and Du5 (Yano et al. 1988). The Waxy (Wx) gene encodes granule-bound starch synthase I (GBSSI), one of the enzymes involved in amylose synthesis, and is located on rice chromosome 6 (Sano 1984). In the rice, two functional alleles, Wx a and Wx b, are reported, Wx b is mainly found in japonica cultivars, and Wx a is found in indica cultivars and various wild rice species (Sano 1984; Sano et al. 1991). Wx a and Wx b were initially defined on the basis of the amount of their gene product (Sano 1984). The Wx a allele produces about tenfold higher levels of mRNA and protein than those of Wx b. As the result, AC of japonica cultivars is almost lower than 20 % while that of indica cultivars is higher than 20 %. Genetic regulation of the amylose synthesis pathway was also studied in the dull (du) mutants (Isshiki et al. 2000, 2008). The Du1 gene encodes an mRNA splicing factor (Prp1; Zeng et al. 2007) and is located on chromosome 7; it is not allelic to du2 (Yano et al. 1988). The du1 and du2 mutations reduce the Wx b transcript levels in endosperm (Isshiki et al. 2000). The Du3 gene encodes the rice homolog of cap-binding protein 20 kD subunit (CBP20) playing a role in pre-mRNA splicing, RNA nuclear export and nonsense-mediated decay (Isshiki et al. 2008), and is located in the middle of the long arm of chromosome 2. However, our knowledge of the functions of, and the relationships between, the genes controlling AC is still limited.

Recently, quantitative trait locus (QTL) mapping has contributed to the elucidation of the genetic basis of AC in rice. In addition to the mutant genes at the wx locus such as Wx-mq and Wx1-1, several other QTLs for AC have been detected (Sato et al. 2002; Ando et al. 2010). One QTL for AC each has been detected on chromosome 1 (Takeuchi et al. 2007), chromosome 3 (Li et al. 2003), chromosome 4 (Li et al. 2003), chromosome 5 (He et al. 1999; Li et al. 2003), chromosome 6 (Septiningsih et al. 2003) and chromosome 12 (Wan et al. 2004). Two QTLs each have been detected on chromosome 8 (Wan et al. 2003, 2004; Li et al. 2011) and chromosome 9 (Wan et al. 2004; Ando et al. 2010). These QTLs have not been confirmed by fine mapping, and no epistatic interactions between these genes have been reported.

Here we report a novel QTL, qAC2, in the near-centromere region of the long arm of chromosome 2, detected by QTL analysis using 191 BC1F4 lines derived from crosses between japonica cultivars Itadaki and Kuiku162 (a low AC line). We mapped this QTL within a 74.9-kb region by high-resolution mapping using backcross-derived progeny. The effect of qAC2 was confirmed with the developed near-isogenic line (NIL) 110. The genetic interaction of qAC2 was investigated with BC3F2 plants derived from crosses between NIL110 and four lines related to AC (Wx a, du1, du2 and du3).

Materials and methods

Plant materials

For QTL detection, 191 BC1F4 lines were developed from crosses of Itadaki/Kuiku162//Itadaki (Supplementary Figure S1). Kuiku162 is a low AC line, which developed from japonica cultivar Kokuhorose. All plants and lines including the parental and the BC1F4 lines were grown at NARO National Institute of Crop Science (Tsukubamirai, Ibaraki, Japan). The lines were seeded on April 25 and transplanted on May 27, 2005. Seeds were harvested at maturity and used for AC measurement.

For high-resolution mapping of qAC2 (Supplementary Figure S1), a BC3F1 plant from the advanced backcross progeny was selected on the basis of the genotype of the simple sequence repeat (SSR) markers. The selected BC3F1 plant was heterozygous for only one segment that included qAC2 in the Itadaki background. The self-pollinated progeny of the BC3F1 plant (125 BC3F2 plants) and Itadaki were seeded on July 5 and transplanted on July 31, 2006. Seeds were harvested at maturity and used for AC measurement. Among 125 BC3F2 plants, 13 BC3F2 plants were selected in which recombination had occurred between the SSR markers flanking qAC2, RM1313 and RM3688. The BC3F3 lines were produced by self-pollination of the selected 13 BC3F2 plants. In 2009, three BC3F3 lines seeded on April 28 were transplanted on May 24 (early planting), and four BC3F3 lines seeded on May 25 were transplanted on June 19 (late planting). In 2010, six BC3F3 lines seeded on April 30 were transplanted on May 21. Five non-recombinant (control; C) and five recombinant homozygous plants (tester; T) were selected from each BC3F3 line by marker-assisted selection of 11 SSR and seven insertion and deletion (InDel) markers in the flanking region of the QTL. Seeds were harvested at maturity and used for AC measurements. The qAC2 genotype of each BC3F2 plant was determined on the basis of the mean AC of the C and T selected from each BC3F3 line.

To further refine the position of qAC2, eight BC3F2 plants in which recombination occurred between SSR markers RM13268 and RM13276 were selected from 1,000 BC3F2 plants. The BC3F3 lines, produced by self-pollination of the eight BC3F2 plants, were seeded on April 28 and transplanted on May 25, 2011. C and T were selected from each BC3F3 line as above. Seeds were harvested at maturity and used for AC measurements. The qAC2 genotype of each BC3F2 plant was determined on the basis of the mean AC of the C and T of each BC3F3 line.

To verify the genetic effects of qAC2, we selected one near-isogenic plant from the BC3F3 lines on the basis of the genotype of 109 SSR markers covering all 12 chromosomes (McCouch et al. 2002) and seven InDel markers located in the qAC2 flanking region. The NIL (designated NIL110) was produced by self-pollination of the selected one near-isogenic plant. The NIL110 was homozygous for the Kuiku162 allele of qAC2 (qAC2 Kuiku) in the Itadaki background. The seeds of NIL110 and Itadaki were sown on May 1 and seedlings were transplanted on May 25, 2013. Seeds of each of the five plants of NIL110 and Itadaki were harvested at maturity and used for AC measurements, physical property measurements and amylopectin structure analyses.

To investigate the genetic interactions between qAC2 Kuiku and four loci related to AC, we crossed NIL110 (qAC2 Kuiku) with a japonica cultivar Nipponbare NIL7 (Wx a) and three japonica cultivar Kinmaze du mutant lines. All the du mutant lines (EM12, du1; EM2, du2; and EM23, du3) had the Wx b allele and were provided by the Laboratory of Plant Genetic Resources, Kyushu University, Japan. Using the SSR markers flanking qAC2 (RM13263 and RM1211) and the du phenotype (for the du mutants), we selected BC3F2 plants with four genotypes: qAC2 Kuiku/Wx a; qAC2 Kuiku/du1/Wx b; qAC2 Kuiku/du2/Wx b; and qAC2 Kuiku/du3/Wx b. The BC3F2 seeds were sown on May 1 and seedlings were transplanted on May 25, 2013. For each genotype, seeds of five plants were harvested at maturity and used for AC measurements. AC of five plants of each genotype was used for the calculation of the mean value and standard error.

Determination of apparent amylose content

Rice seeds were dehulled, and brown rice was polished to 90 % to remove the embryo, pericarp and seed coat. Polished rice was crushed using a motor mill. The powder (100 mg) was suspended in distilled water (5 mL) for 30 min, diluted with 10 mL of 0.5 M NaOH, and left for 24 h at room temperature. Apparent AC was determined by a colorimetric method (680 nm) (Juliano 1971; Nishi et al. 2001) using a Technicon Autoanalyzer II (Bran & Luebbe, Norderstedt, Germany). The calibration line was obtained using varying ratios of purified amylose from potato (Sigma, St. Louis, MO, USA) and rice amylopectin extracted from japonica glutinous cultivar Kogane-mochi in the iodine solution. In this paper, apparent AC was presented as AC.

Determination of physical property

Polished rice (10 g) was added with distilled water (16 g) in an aluminum cup (55 mm × 40 mm × 55 mm). After soaking for 1 h at room temperature, the rice was cooked in an electric rice cooker (RC183, Toshiba Corporation, Tokyo). To prevent moisture loss after cooking, cup was covered with an outer sheet of aluminum foil and an inner sheet of paper for the absorption of excess vapor, and sealed in a plastic and airtight vessel. The cooked rice was kept in the vessel for 2 h at room temperature. The physical property of stickiness in a surface layer of a single cooked rice grain was measured using a Tensipresser (Myboy System, Takemoto Electric Corporation, Osaka).

Determination of the length distribution of α-1,4-glucan chains in α-polysaccharides by high-performance anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD)

The embryo, pericarp and seed coat were removed from three dehulled grains of average size and the endosperms were ground with a mortar and pestle. The resulting powder (5 mg) was suspended in methanol (5 mL) and boiled for 10 min. The homogenate was centrifuged at 2,500g for 5 min. The pelleted polyglucan was washed twice with 5 mL of 90 % (v/v) methanol, resuspended in 5 mL of distilled water, and boiled for 60 min. An aliquot (1.0 mL) of gelatinized polyglucan was added to 50 µL of 0.6 M sodium acetate buffer (pH 4.4) and 10 µL of 2 % (w/v) NaN3, hydrolyzed with 10 µL of Pseudomonas amyloderamosa isoamylase (1,400 units; Seikagaku Corporation, Tokyo) at 37 °C for 24 h. The hydroxyl groups of the debranched glucans were reduced with 0.5 % (w/v) sodium borohydride under alkaline conditions for 20 h by the method of Nagamine and Komae (1996). The preparation was dried under vacuum, dissolved in 20 µL of 1.0 M NaOH for 60 min, and diluted with 180 µL of distilled water. An aliquot (25 µL) was injected into a BioLC System (model DX-500; Dionex, Sunnyvale, CA, USA) equipped with a PAD and a Carbopac PA-1 column (4-mm i.d. × 25 cm). Size fractionation of α-1,4-glucans was performed with a linear gradient of sodium acetate (50–500 mM) in 0.1 M NaOH at a flow rate of 1 mL/min.

Mapping quantitative trait loci

Linkage map

The genotypes of 191 BC1F4 lines and 125 BC3F2 plants were determined using 109 SSR markers covering all 12 chromosomes (Supplementary Table S1) (McCouch et al. 2002), and 11 SSR markers of the International Rice Genome Sequencing Project (IRGSP) genome sequence data (http://www.rgp.dna.affrc.go.jp/cgi-bin/statusdb/status.pl) (Fig. 3a), respectively. The linkage and map distances were determined using MAPMAKER/EXP 3.0 software (Lander et al. 1987).

Development of new markers

Seven InDel markers on the long arm of chromosome 2 were developed (Table 1) on the basis of the IRGSP genome sequence data. Primers were designed with the online primer design tool Primer 3 (http://frodo.wi.mit.edu/) to generate 100–300 bp amplicons at approximately 20-kb intervals across the region on the long arm of chromosome 2. The ability of primers to detect polymorphisms between Itadaki and Kuiku162 was verified.

Table 1 The list of InDel markers used in the fine mapping of qAC2
Full size table

QTL analysis

Quantitative trait locus analysis in the 191 BC1F4 lines and 125 BC3F2 plants was performed using genotype data for 109 and 11 SSR markers, respectively. Putative QTLs were detected by one-way ANOVA with SAS GLM PROC in a single-point analysis (SAS Institute, Cary, NC, USA). A threshold LOD score was determined at a significance level of 0.05 %. MAPMAKER/QTL software (the “f2 backcross” mode; Lander and Botstein 1989) was used to confirm the presence of a putative QTL and to estimate the additive effects and percentage of variance explained.

Analysis of SSR and InDel markers

Total DNA was extracted from small leaf pieces of BC1F4, BC3F2, and BC3F3 plants, and NIL110, and BC2F2 plants with four genotypes (qAC2 Kuiku/Wx a; qAC2 Kuiku/du1/Wx b; qAC2 Kuiku/du2/Wx b; and qAC2 Kuiku/du3/Wx b). Each leaf piece was crushed in a 2.0-mL tube, incubated with 1.0 mL of a solution containing 1.5 % CTAB, 75 mM Tris–HCl (pH 8.0), 1.05 M NaCl and 20 mM EDTA at 65 °C for 20 min, and centrifuged. The supernatant was mixed with 1 mL chloroform:isoamylalcohol (24:1) for 20 min and then centrifuged. The aqueous phase was transferred to a new 2.0-mL tube. DNA was precipitated with 1.5 mL of a buffer containing 1.0 % CTAB, 50 mM Tris–HCl (pH 8.0) and 10 mM EDTA. The pellet was washed with ethanol and redissolved in 50 μL of a buffer of 0.1 × TE containing 10 mM Tris–HCl (pH 8.0) and 1 mM EDTA. For PCR, 1 μL DNA (50 ng/μL) was combined with 0.7 μL 10 × PCR buffer (Promega, Madison, WI, USA), 0.7 μL of a solution containing 2 mM each dNTP (Boehringer Mannheim, Mannheim, Germany), 0.1 μL of 5 U Taq DNA polymerase (Promega), 0.3 μL of a 20 pM solution of each primer, and 3.2 μL H2O. Amplification was performed for 30 cycles at 94 °C for 30 s, 55 °C for 1 min, and 72 °C for 1 min; followed by a final cycle at 72 °C for 7 min. Amplified fragments were separated by electrophoresis in 3.5 % agarose gel, and stained with ethidium bromide.

Results

QTL mapping for AC in BC1F4 lines

In 2005, AC of the parental lines was 20.2 % for Itadaki and 15.3 % for Kuiku162. The AC frequency distribution for the 191 BC1F4 lines is shown in Fig. 1. The BC1F4 lines showed a continuous AC distribution from 14.4 to 21.4 %. We detected two QTLs for AC in the 191 BC1F4 lines. One QTL was detected near RM1211 in the near-centromeric region of the long arm of chromosome 2 (Table 2; Fig. 2). This QTL explained 13.9 % of the total phenotypic variance (Table 2). The Kuiku162 allele at this QTL decreased AC. Previously, the Du3 locus was mapped between the SSR marker RM3515 and the CAPS marker 5300 in the middle of the long arm of chromosome 2 (Isshiki et al. 2008). Because the QTL on chromosome 2 detected in this study was located at a different position from the Du3 locus, we designated it as qAC2.

Fig. 1
figure 1

Frequency distribution of amylose content in 191 BC1F4 (Itadaki/Kuiku162//Itadaki) lines. Black and white arrows indicate the mean values for Kuiku162 and Itadaki, respectively. Horizontal lines under the arrows indicate SD

Full size image
Table 2 QTL for AC in 191 BC1F4lines
Full size table
Fig. 2
figure 2

Chromosomal positions of putative QTLs for amylose content detected in 191 BC1F4 lines. SSR markers are shown on the left of the map. Black circles indicate the nearest SSR marker loci revealed by one-way ANOVA. Bold lines indicate the most likely chromosomal regions for the putative QTLs within a certain confidence interval (defined by a decrease of 1.0 from the peak LOD values). 1) Position of Du3 as reported by Isshiki et al. (2008), 2) Positions of qAC8-1 and qAC8-2 as reported by Li et al. (2011)

Full size image

The other QTL was detected near RM1235 on the short arm of chromosome 8 (Table 2; Fig. 2). The Kuiku162 allele of this QTL increased AC. This QTL explained 9.5 % of the total phenotypic variance (Table 2). Previously, qAC8 was also mapped on chromosome 8 between the restriction fragment length polymorphism markers G1149 and R727 (Wan et al. 2004). qAC8 was further resolved into qAC8-1, mapped between the SSR markers RM7356 and RM7756, and qAC8-2, mapped between the SSR markers RM23510 and RM23479 (Li et al. 2011). The QTL on chromosome 8 detected in this study differed from both qAC8-1 and qAC8-2, and was designated as qAC8-3.

Fine mapping of qAC2

In 2006, the 125 BC3F2 plants showed a continuous AC distribution from 19.4 to 22.7 % (Supplementary Fig. S2). One QTL qAC2 was detected a near RM13329 in the long arm of chromosome 2. Thirteen BC3F2 plants were selected in which recombination had occurred between the SSR markers RM1313 and RM3688 flanking qAC2. The position of qAC2 was refined using BC3F3 lines derived from self-pollination of 13 BC3F2 plants. Seven BC3F3 lines derived from self-pollination of seven BC3F2 plants recombinant in the region between RM1313 and RM3688 were grown in 2009, and six such BC3F3 lines were grown in 2010. In 2009, four of the BC3F3 lines were planted early and three were planted late. Among the seven BC3F2 plants with progeny grown in 2009, 09-IK5 and 09-IK6 were homozygous for the Kuiku162 allele of qAC2, 09-IK9 was homozygous for the Itadaki allele and four (09-IK10, 09-IK22, 09-IK7 and 09-IK20) were heterozygous (Table 3, Supplementary Table S2). Among the six BC3F2 plants with progeny grown in 2010, 10-IK22 was homozygous for the Kuiku162 allele of qAC2, 10-IK25 and 10-IK29 were homozygous for the Itadaki allele, and three (10-IK33, 10-IK35 and 10-IK36) were heterozygous. These results showed that qAC2 is located in the region between SSR markers RM13268 and RM13276 on chromosome 2 (Table 3; Fig. 3a).

Table 3 Genotypes of DNA markers and qAC2 of BCaF2plants
Full size table
Fig. 3
figure 3

Physical maps of the region surrounding qAC2 on the long arm of chromosome 2. a Physical map of the region surrounding qAC2. Physical positions of the markers, based on the build 5.0 rice genome sequences (International Rice Genome Sequencing Project) are shown on the left side of the map. Numerals in parentheses indicate the number of recombinant BC3F2 plants in 2009 and 2010. b High-resolution physical map of the region surrounding qAC2. qAC2 was localized within the 74.9-kb region between two InDel markers, KID3001 and KID5101. Numerals in parentheses indicate the number of recombinant BC3F2 plants in 2011

Full size image

In 2011, the position of qAC2 was further refined using BC3F3 lines derived from self-pollination of eight BC3F2 plants recombinant between RM13268 and RM13276. Among eight BC3F2 plants, 11-IK7, 11-IK8 and 11-IK13 were homozygous for the Kuiku162 allele of qAC2, and 11-IK4, 11-IK5, 11-IK6, 11-IK9 and 11-IK14 were heterozygous (Table 3, Supplementary Table S2). These results showed that qAC2 is located in a 74.9-kb region between two InDel markers, KID3001 and KID5101 (Fig. 3b). This region included at least seven genes, detected using Knowledge-based Oryza Molecular biological Encyclopedia (KOME: http://cdna01.dna.affrc.go.jp/cDNA/). In the qAC2 region, three genes annotated to encode glucose/ribitol dehydrogenase family, glycosyl transferase, and auxin responsive proteins. The other genes were hypothetical proteins. None of these genes encoded proteins related to starch-synthesizing enzymes.

Effect of qAC2

To verify the genetic effects of qAC2, we selected a near-isogenic plant from the BC3F3 lines by genotype analysis using 109 SSR and seven InDel markers (Fig. 4a). The NIL110 produced by self-pollination of the near-isogenic plant had the Kuiku162 allele of qAC2 (qAC2 Kuiku) in the Itadaki genetic background. The culm length, flowering date and grain yield of NIL110 were similar to those of Itadaki (Supplementary Table S3). AC of NIL110 (18.2 %) was lower than that of Itadaki (19.3 %) (p < 0.01) (Fig. 4b), confirming the effect of qAC2 Kuiku. The stickiness in a surface layer of a single cooked rice grain of NIL110 (27.3 × 102N/m2) was higher than that of Itadaki (20.3 × 102N/m2) (p < 0.01) (Fig. 4c). On the other hand, chain length distribution of amylopectin in NIL110 and Itadaki was almost identical (Fig. 5).

Fig. 4
figure 4

Genotype and amylose content of NIL110, carrying qAC2 Kuiku. a Graphical genotype of NIL110. Black block denotes a region derived from Kuiku162; white blocks denote those from Itadaki. b Amylose content of NIL110 in comparison with Itadaki. Amylose content of NIL110 and Itadaki represent the mean score of each five plants. Error bars indicate SD. ** indicate significance at p < 0.01. c The surface stickiness of NIL110 in comparison with Itadaki. The surface stickiness of NIL110 and Itadaki represent the mean score of each of the 15 cooked rice grains. Error bars indicate SD. ** indicate significance at p < 0.01

Full size image
Fig. 5
figure 5

Comparison of amylopectin structure between NIL110 and Itadaki. The chain length distributions of amylopectin of NIL110 and Itadaki were measured using high-performance anion exchange chromatography with pulsed amperometric detection. The chain length distributions of amylopectin of NIL110 and Itadaki represent the mean score of each of two replication

Full size image

Genetic interaction between qAC2 and other loci related to AC

To study the genetic interaction between qAC2 Kuiku and five alleles of four other loci related to AC, we developed BC3F2 plants derived from crosses between NIL110 and each of four lines: Nipponbare NIL7 (Wx a), and Wx b lines EM12 (du1), EM2 (du2) and EM23 (du3). AC was similar in qAC2 Kuiku/Wx a (30.7 %) and Nipponbare NIL7 (30.7 %), qAC2 Kuiku/du1/Wx b (6.1 %) and EM12 (6.2 %), and qAC2 Kuiku/du2/Wx b (7.5 %) and EM2 (7.3 %), respectively (Fig. 6). AC in qAC2 Kuiku/du3/Wx b (11.4 %) was remarkably lower than in EM23 (12.1 %) (p < 0.01). These results indicate that the qAC2 Kuiku has epistatic interaction with Wx a. The qAC2 Kuiku has epistatic interactions with two loci, du1 and du2, on Wx b, whereas the genetic effect of qAC2 Kuiku has an additive to that of du3 on Wx b. These results indicate that the expression of Wx a would be not affected by qAC2 Kuiku, whereas the expression of Wx b affected.

Fig. 6
figure 6

Amylose content of BC3F2 plants derived from crosses between NIL110 and each of four lines: Nipponbare NIL7 (Wx a), EM12 (du1), EM2 (du2) and EM23 (du3). Each of the progeny genotypes tested is shown below the graph. For each genotype, five plants were used, which had the same date of flowering as the parental line. Error bars indicate SD. ** indicate significance at p < 0.01

Full size image

Discussion

Identification of the novel QTL controlling low amylose content

Amylose synthesis is catalyzed by GBSSI, encoded by the Wx gene located on chromosome 6. AC is determined by the GBSSI level (Sano 1984; Umemoto and Terashima 2002). Wide variations in AC have been reported (Morishima et al. 1992; Juliano and Villareal 1993). Several other QTLs related to AC have also been reported. To clarify the details of the genetic control of amylose synthesis, we carried out fine mapping of a novel QTL for AC, and analyzed its genetic interaction with five other loci related to amylose synthesis.

Kuiku162, used in this study, has low AC. We identified a QTL for low AC (designated qAC2) in this cultivar, located in a 74.9-kb region between the InDel markers KID3001 and KID5101 on the near-centromeric region of the long arm of chromosome 2. This position differed from that of the Du3 locus, which is located in the middle of the long arm of chromosome 2 (Isshiki et al. 2008), suggesting that the qAC2 is a novel QTL for AC. In the qAC2 region, three annotated genes to encode glucose/ribitol dehydrogenase family, glycosyl transferase, and auxin responsive proteins and four genes annotated to encode conserved hypothetical proteins unrelated to starch biosynthesis were found using KOME.

We also identified a QTL for high AC (designated qAC8-3) in Kuiku162 on the short arm of chromosome 8. This position of the qAC8-3 differed from those of the qAC8-1 and qAC8-2, which are located in the long arm of chromosome 8 (Li et al. 2011). This suggests that the qAC8-3 is a novel QTL for AC.

In this study, 109 SSR markers were mapped onto all 12 chromosomes, while a few markers were mapped on the short arm of chromosomes 4 and 10. To detect all QTLs involved in AC, more high density markers on these chromosomal regions would be required.

Characterization of qAC2

The genetic effect of qAC2 Kuiku was verified by the development of an NIL110 with a significantly decreased AC (Fig. 4a, b). AC in the NIL110 was determined using a colorimetric method (Fig. 4b). In the ae mutant, AC determined by a colorimetric method is affected by altered amylopectin structure (Nishi et al. 2001; Takeda and Hizukuri 1987). As NIL110 and Itadaki in our study had similar chain length distributions of amylopectin (Fig. 5), low AC in NIL110 would be not related to the amylopectin structure. Low AC in NIL110 was probably caused by the decrease in actual AC.

Genetic control of amylose synthesis

To understand the function of qAC2 Kuiku in the amylose synthesis pathway, we investigated the genetic interaction between qAC2 Kuiku and five alleles of four loci related to amylose synthesis: Wx a, Wx b, du1, du2 and du3. Our interaction analysis indicates that Wx a has epistatic interaction with qAC2 Kuiku (Fig. 6). This result indicates that the Wx a expression and GBSSI protein level would be not affected by qAC2 Kuiku. On the other hand, AC in NIL110 with qAC2 Kuiku/Wx b was lower than that of Itadaki with Wx b (Fig. 4b). This result indicates that the Wx b expression would be affected by qAC2 Kuiku. qAC2 Kuiku also has epistatic interactions with two loci, du1 and du2, on Wx b (Fig. 6). Isshiki et al. (2000) reported similar genetic interaction between Wx alleles and both du1 and du2; the level of the GBSSI protein in the du mutants decreased because of inefficient splicing of Wx b pre-mRNA, but neither of these mutants affected the expression of the Wx a allele. The Du1 gene encodes the Prp1 protein, which is a component of the U4/U6 snRNP required for spliceosome assembly (Zeng et al. 2007); du2 may be also related to mRNA splicing (Isshiki et al. 2000). The decrease in AC by qAC2 Kuiku may be caused by the spliceosome factors similar to Du1 and Du2 (Fig. 7). Our results also indicate that qAC2 Kuiku has an additive effect with du3 on Wx b (Fig. 6). Du3 encodes the rice homolog of CBP20, which plays a role in pre-mRNA splicing (Isshiki et al. 2008). We think that the function of qAC2 in the amylose synthesis pathway would be different from that of Du3 (Fig. 7).

Fig. 7
figure 7

Hypothetical function of qAC2. The vertical arrows show the pathway of amylose synthesis. The analysis of genetic interactions indicated the possibility that the function of qAC2 is related to the regulation of Wx b (like du1 and du2)

Full size image

Further studies such as map-based cloning, functional analysis of qAC2 and the analysis of its interaction with other loci related to starch synthesis would clarify the role of qAC2 in amylose synthesis.

qAC2 as a genetic resource for breeding

Wx a, Wx b and du loci have been used as genetic resources for modification of AC. These genes strongly affect AC and starch properties, and considerable modification of AC may result in undesirable eating qualities. The effect of qAC2 Kuiku on AC is moderate (1.1 % points). Cooked NIL110 rice is stickier than Itadaki (Fig. 4c). Thus, qAC2 Kuiku is a useful genetic resource for the improvement of eating quality of cooked rice.

Author contribution statement

Yoko Takemoto-Kuno, Keitaro Suzuki, Hideyuki Hirabayashi, Takuro Ishii, Ikuo Ando, Tokio Imbe, and Yoshinobu Takeuchi designed research; Yoko Takemoto-Kuno, Hiroki Mitsueda, Keitaro Suzuki, Hideyuki Hirabayashi, Osamu Ideta, Noriaki Aoki, Takayuki Umemoto, and Yoshinobu Takeuchi performed research; Yoko Takemoto-Kuno, Hideyuki Hirabayashi, Takayuki Umemoto, and Yoshinobu Takeuchi contributed new reagents/analytic tools; Yoko Takemoto-Kuno, and Yoshinobu Takeuchi analyzed data; Yoko Takemoto-Kuno, Keitaro Suzuki, Takayuki Umemoto, Ikuo Ando, Hiroshi Kato, Hiroshi Nemoto, Tokio Imbe, and Yoshinobu Takeuchi wrote the paper.