Allelic effects on starch structure and properties of six starch biosynthetic genes in a rice recombinant inbred line population

The genetic diversity of six starch biosynthetic genes (Wx, SSI, SSIIa, SBEI, SBEIIa and SBEIIb) in indica and japonica rices opens an opportunity to produce a new variety with more favourable grain starch quality. However, there is limited information about the effects of these six gene allele combinations on starch structure and properties. A recombinant inbred line population from a cross between indica and japonica varieties offers opportunities to combine specific alleles of the six genes. The allelic (indica vs japonica) effects of six starch biosynthetic genes on starch structure, functional properties, and abundance of granule bound proteins in rice grains were investigated in a common genetic background using a recombinant inbred line population. The indica Wx (Wxi) allele played a major role while indica SSI (SSIi), japonica SSIIa (SSIIaj) and indica SBEI (SBEIi) alleles had minor roles on the increase of amylose content. SSIIaj and japonica SBEIIb (SBEIIbj) alleles had a major and a minor role on high ratio of ∑DP ≤ 10 to ∑DP ≤ 24 fractions (RCL10/24), respectively. Both major alleles (Wxi and SSIIaj) reduced peak viscosity (PV), onset, peak and end gelatinization temperatures (GTs) of amylopectin, and increased amylose-lipid complex dissociation enthalpy compared with their counterpart-alleles, respectively. SBEIIai and SBEIIbj decreased PV, whereas SSIi and SBEIIbj decreased FV. SBEIi reduced setback viscosity and gelatinization enthalpy. RCL10/24 of chain length distribution in amylopectin is negatively correlated with PV and BD of paste property and GTs of thermal properties. We also report RILs with superior starch properties combining Wxi, SSIj, SSIIaj, SBEIi and SBEIIbj alleles. Additionally, a clear relation is drawn to starch biosynthetic gene alleles, starch structure, properties, and abundance of granule bound starch biosynthetic enzymes inside starch granules. Rice Wxi and SSIIaj alleles play major roles, while SSIi, SBEIi, SBEIIai and SBEIIbj alleles have minor roles in the determination of starch properties between indica and japonica rice through starch structural modification. The combination of these alleles is a key factor for starch quality improvement in rice breeding programs. RCL10/24 value is critical for starch structure and property determination.


Background
Rice is the most important cereal crop in developing countries. It feeds over half of the world's population, which is critical to sustain population growth (Toriyama et al. 2005). In rice grains starch is the major component that primarily controls rice quality (Umemoto et al. 2008). There are two major rice subspecies, indica and japonica. The diversity of starch properties between indica and japonica varieties have been widely studied, and the alleles of genes that affect the traits have been found through a number of approaches (Han et al. 2004;Tian et al. 2009Tian et al. , 2010Kharabian-Masouleh et al. 2011Zhao et al. 2011). In terms of rice traits, cooking and sensory properties are the major determinants for consumers' acceptance of a rice variety. In the absence of a clear knowledge of sensory properties, amylose content, gelatinization temperature and gel consistency of rice flour have been widely used as important indicators of quality for rice varietal development (Cuevas and Fitzgerald 2012).
Amylose content (AC) of rice starch, affecting the cooking and eating properties, is determined by the activity of granule-bound starch synthase I (GBSSI) encoded by the Wx gene (Wang et al. 1995;Umemoto and Terashima 2002). A number of alleles were reported to be associated with starch AC variation. However, Wx a and Wx b dominate the functional alleles among indica and japonica subspecies, with high and intermediate GBSSI protein production, respectively (Sano 1984;Wang et al. 1995). They can be distinguished mainly by a G/T single nucleotide polymorphism (SNP) at 5′ splicing site of the first intron (Hirano et al. 1998) that resulted in differential splicing of the first intron. Gel consistency in rice is also related to AC (Tan & Corke 2002;Septiningsih et al. 2003;Zheng et al. 2007) and its quantitative trait locus mapped to the Wx locus has been reported (He et al. 1999;Lanceras et al. 2000).
Gelatinization temperature of rice starch is mainly determined by starch synthase IIa (SSIIa) (Govindaraj et al. 2009;Cuevas et al. 2010;Gao et al. 2011). Allelic effects of SSIIa in rice were reported earlier, and up to 9 SNP alleles were identified (Nakamura et al. 2005b;Umemoto and Aoki 2005;Bao et al. 2006;Yu et al. 2010). Two replacements of amino acid residues at the C-terminal region of the SSIIa protein were found to affect SSIIa activity and amylopectin structure and properties in grains (Nakamura et al. 2005b).
Aside from Wx and SSIIa, other starch synthetic genes are also involved in the modification of amylopectin structure and starch properties in rice varieties. The expression of SSI gene in indica rice is lower than that of japonica rice which reduces the synthesis of short chains in indica rice amylopectin (Takemoto-Kuno et al. 2006). A mutation in the SBEI gene leads to a decrease in long chains (DP ≥ 37) and intermediate chains (DP 12 to 21), increases in short chains (DP ≤ 10) and intermediate chains (DP 24 to 34) and a reduction in the onset, peak and end gelatinization temperature (Xie et al. 1999;Satoh et al. 2003). For the SBEIIb gene, a SNP was determined by Han et al. (2004) in the 3′UTR, between indica and japonica varieties of rice. Taking advantages of nucleotide polymorphisms, genetic markers have been developed for genotyping a series of starch biosynthetic genes to improve grain starch quality in rice breeding programs (Han et al. 2004;Yan et al. 2007;Tian et al. 2010). Nonetheless, nearly all these studies are carried out in isolated individual breeding lines with diverse genetic backgrounds.
The starch properties are highly related to the composition and chemical structure of starch not only in rice but also other cereal grains. Two types of glucan polysaccharides, amylose and amylopectin, comprise starch. The former molecule (approximately 20~30% of starch) is a long linear glucose polymer (~1% branched points) which is synthesised by GBSSI, through adding ADP-glucose to existing α-(1-4) glucan chains (Nelson and Rines 1962;Tsai 1974;Fedoroff et al. 1983;Shure et al. 1983). The latter (approximately 70~80% of starch) is a much larger molecule with frequent α-(1-6) branches formed by multiple enzymes, for example, starch synthases (SSs) and starch branching enzymes (SBEs) (Tetlow et al. 2004;Hannah and James 2008). The isoforms of SSs and SBEs differ in their activities in amylopectin synthesis in a developing endosperm. SSI elongates very short glucan chains to a degree of polymerisation (DP) 10 (Commuri and Keeling 2001;Fujita et al. 2006), while SSIIa is endosperm specific and is involved in the polymerization of intermediate chains (DP 10~24) by elongating short chains (DP < 10) of amylopectin (Nakamura et al. 2005b). The formation of α-(1-6) linkage in amylopectin molecules is mainly performed by SBEI and SBEII (Gao et al. 1996). SBEI transfers glucan chains with a wide range of DP ≤ 35, while SBEIIa is involved in the transferring of chains with DP 6~15 and SBEIIb forms almost only short chains of DP 6 and 7 (Mizuno et al. 2001;Nishi et al. 2001;Nakamura et al. 2010). The activities of these enzymes potentially affect starch structure and properties.
In this study, the polymorphic effects of six starch biosynthetic genes (Wx, SSI, SSIIa, SBEI, SBEIIa and SBEIIb) on starch structure and properties were studied in a common genetic background, using a recombinant inbred line (RIL) population derived from crossing IR64 (indica, i) and Nipponbare (japonica, j). With two alleles (i or j) for six different genes there are 64 potential combinations theoretically. In this work we have compared i or j alleles for each gene in a common background, with 12 combinations representing 64 as a whole. Different approaches such as amylose content assay using size-exclusion chromatography, chain length determination by capillary electrophoresis (CE), rapid viscosity analysis (RVA), and differential scanning calorimetry (DSC) were performed to analyse the starch structure and properties of these samples. Starch granule bound proteins (GBPs) were analysed to study the relation between those proteins and starch structure and properties.

Grouping of recombinant inbred lines
Six allele groups were selected from recombinant inbred lines corresponding to the 6 major starch synthetic enzyme genes, and parental lines (IR64 and Nipponbare) differing in their alleles (i vs j) for each of the 6 genes. Thus we segregated the RILs into six allele groups according to the allele combinations ( Figure 1) to analyse the effects of i vs j alleles of different starch biosynthetic genes on starch structure and properties. The allele types of these six genes of each RIL are described in Table 1, and are designated as follows: line 3-14-12 and 3-14-13 as Wxi and Wxj; line 3-5-15-11 and 3-6-20 as SSIi and SSIj; line 3-6-9 and 3-6-1 as SSIIai and SSIIaj; line 3-5-1-2 and 3-5-1-8 as SBEIi and SBEIj; line 3-6-20 and 3-6-9 as SBEIIai and SBEIIaj; and line 3-5-2-2 and 3-5-2-1 as SBEIIbi and SBEIIbj. Meanwhile, we also separated RILs in Wxi and Wxj allele groups in some of the following analyses, because Wx alleles were well known in determining various starch characteristics.

Amylose content
Although the Wxi allele played a major role in the determination of increased AC, other alleles also showed impacts on AC. The range of AC of selected RILs was 11~35% approximately of rice grain starch ( Figure 2). Six allele groups can be classified into two AC groups depending on the Wx allele. The Wxi allele group contained 17.7~35.3% amylose in starch (including Wxi allele and SSI, SSIIa and SBEIIa groups), whereas Wxj allele group contained 11.1~18.4% (including Wxj allele and SBEI and SBEIIb groups), with parental lines IR64 (21.2%) and Nipponbare (11.8%) fitting in their respective Wx groups. Between the Wxi allele group (containing average 22.0% amylose) and the Wxj allele group (containing average 13.1% amylose), the former contained~9% significantly higher AC than the latter. Within Wxi allele group, starch samples from Wxi allele lines contained significantly higher AC (35.3%) than all other allele groups, and starch from SSIi (22.8%) and SSIIaj (20.9%) alleles contained higher AC than other four alleles (SSIj, SSIIai, SBEIIai and SBEIIaj) ( Figure 2). Within Wxj allele group, SBEIi allele lines (18.4%) contained significantly higher AC than those from other four allele groups, among which AC was not significantly different.
distribution of fewer chains at DP10~13 than SBEIIbj allele ( Figure 3F). Based on Nakamura's study (Nakamura et al. 2002), indica and japonica type amylopectin from the majority of cultivated Asian rice strains can be distinguished by the ratio of ∑DP ≤ 10 to ∑DP ≤ 24 fractions (R CL10/24 ) of CLD. Our present study showed the R CL10/24 value was 0.16 for IR64 and 0.29 for Nipponbare (Table 2), and most of the selected RILs were smaller than 0.14, suggesting that they were indica type. With the only exception of SSIIaj allele, its R CL10/24 value was slightly over 0.20. The highest value of SSIIaj resulted from a high percentage of short chain fraction (∑DP ≤ 10, significantly higher than that of SSIIai allele, and 436 7% higher than the other RILs) and low percentage of short and intermediate chain fraction (∑DP ≤ 24, statistically significantly lower than that of SSIIai allele) in the starch. In terms of the longer chain fraction (∑DP > 24), SSIIaj was the highest but not significantly different to a few other alleles in the group (Table 2). Besides SSIIaj, SBEIIbj allele was also detected with more short chains in starch than the counterpart-allele SBEIIbi.

Starch paste viscosity
By analysing RVA characteristics pasting properties of starch were determined as reported in previous studies (Sasaki et al. 2000;Han and Hamaker 2001;Chen et al. 2003). The RVA characteristics involved in this study were peak viscosity (PV), trough (Tr), breakdown (BD), final viscosity (FV), setback (SB) and peak time (PT). The values of these characteristics were statistically analysed and shown in Table 3. The RVA result revealed that i and j alleles from each of six allele groups  contributed to parameters of starch paste viscosity properties differentially. IR64 and Nipponbare showed significant differences in all parameters except for FV (Table 3). RILs in Wx, SSIIa, SBEIIa and SBEIIb allele groups had significantly distinct PV between i and j alleles, respectively. PV values of Wxj and SSIIai alleles were~60 RVU higher than Wxi and SSIIaj alleles, whereas the increase in SBEIIaj and SBEIIbi alleles was smaller (31~35 RVU) comparing to SBEIIai and SBEIIbj alleles. Tr was the least diverse RVA parameter in this analysis, showing only about 30 and 10 RVU significant increase in SSIj and SBEIIbi alleles comparing to their counterpartalleles, respectively. Significant differences in BD were obtained in Wx, SSI, SSIIa and SBEIIa allele groups, with the greatest (~70 RVU) between SSIIai and SSIIaj alleles. Compared to their counterparts, the values of BD were increased by~40 RVU in Wxj allele, and~30 RVU in SSIi and SBEIIaj alleles. In terms of FV, significant variations were found in Wx, SSI and SBEIIb allele groups, with the highest (~40 RVU) between SSIi and SSIj alleles,~30 and 15 RVU in Wx and SBEIIb allele groups, respectively. The significant variations of SB were 42 RVU in Wx group and 28 RVU in SSIIa group. Less than 10 RVU increase of SB in both SSIj and SBEIj alleles were also significantly different. Significant differences in PT were determined between SSI alleles (0.6 min) and Wx group (0.2 min). The PT values were mainly clustered into the two Wx alleles groups, while IR64 was the lowest. Notably, the SSIIaj allele starch exhibited remarked differences in paste viscosities from all the others including parental controls.

Starch thermal properties
Consistent with previous studies, two primary peaks were observed in DSC curves for rice wholemeal samples: the first one was the initial gelatinization peak at 65~85°C of amylopectin; the second one at around 105°C was the amylose-lipid complex dissociation peak (Biliaderis et al. 1985). Onset, peak and end temperatures of the first peak were named as 1 To, 1 Tp and 1 Te, and those of the second peak as 2 To, 2 Tp, 2 Te. 1 ΔH was named for the amylopectin gelatinization enthalpy, and 2 ΔH for the dissociation enthalpy of amylose-lipid complexes. As shown in Table 4, the DSC result indicated that SSIIaj allele had a major effect, while Wxi allele played a minor role in decreasing 1 Tp. However, SSIj, SSIIaj and Wxi alleles increased 2 ΔH. Regarding amylopectin gelatinization temperatures, IR64 had significantly higher values in 1 To, 1 Tp and 1 Te than Nipponbare, whereas the RILs exhibited divergent values. Wxj allele group observed~3°C higher 1 To than those from Wxi allele group. There was barely any variation in 1 To among different alleles within Wxi allele group except SSIIaj allele having~11°C decrease compared with SSIIai allele. In terms of 1 Tp, Wxj allele group overall was~3°C higher than Wxi allele group. Within Wxi allele group, SSIIai was~11°C higher than SSIIaj, while SSIj and SBEIIai were~1 and 0.5°C higher than SSIi and SBEIIaj alleles, respectively. For 1 Te, Wxj allele group overall was~2°C higher than Wxi allele group. There was~9°C increase in SSIIai allele comparing to SSIIaj allele. In terms of 1 ΔH, Wxj allele group was~0.4 J/g higher than Wxi allele group, except for SBEIi allele 0.3 J/g lower than the counterpart-allele.
Regarding dissociation temperatures of amylose-lipid complexes, IR64 wholemeal had significantly higher values of 2 To, 2 Tp and 2 Te than Nipponbare. In terms of 2 To, the only significant variations were observed in SSIIaj and SBEIIai alleles which both had~2°C increase compared with their counterpart-alleles. In terms of 2 ΔH, SSIIaj and SSIj alleles were up to 0.5 J/g and 0.24 J/g higher than SSIIai and SSIi alleles, respectively. Wxi allele group overall exhibited higher energy than Wxj allele group. Within Wxi group, SSIIaj and Wxi were the highestand lowest, whereas in Wxj allele group SBEIi allele had higher 2 ΔH than the others. Interestingly, similar to the observation in RVA examination, SSIIaj allele starch also showed remarked differences in thermal properties from others.

Protein analyses in starch granules of mature grains
The analysis of GBPs prepared from purified starch of mature rice grains showed that four major protein bands with 60 kDa and above were detected in most of the RILs (Figure 4). In parental lines, the top bands at~88 kDa were identified as SSIIa,~83 kDa as SBEIIb,~75 kDa as SSI and~60 kDa as GBSSI by immunoblotting using specific antibodies ( Figure 5).
Comparing the two parental lines, IR64 contained a higher amount of GBSSI, SSI, SSIIa and SBEIIb in the GBPs than Nipponbare, and SSIIa was barely detectable by immunodetection in Nipponbare ( Figure 5A). Consistent with parental lines, GBSSI abundance of Wxi allele was significantly higher than that of Wxj allele, while the abundance of SSI, SSIIa and SBEIIb in Wxi and Wxj alleles remained at same levels, respectively ( Figures 4A, 5B). In SSIIa alleles, the amount of SSI, SSIIa and SBEIIb protein was significantly decreased in SSIIaj allele ( Figure 5C). Similarly, only faint bands of SSIIa were detected in SSIIaj RILs ( Figure 4C), whereas GBSSI remained at the same level. No significant changes were observed in abundance of the four proteins in SDS-PAGE gels of other groups.

Discussion
In this work, six starch synthetic gene alleles were used for genotyping progenies from one crossing line between indica and japonica rice. We aimed at understanding the roles of those alleles on starch properties of rice grain. The results showed that both amylose content and amylopectin structure were affected by multiple gene alleles, although some of them play major roles, others play minor roles. However, as matter of time, the roles of other genes involved in starch biosynthesis could not be studied, eg. SSIIIa gene affects amylose content (Gao et al. 1998;Fujita et al. 2007;Li et al. 2011) and starch debranching enzymes (Kubo et al. 1999). Therefore, more research work is required to define the functions of other genes on starch property variation between indica and japonica rice.

The relation between amylose content and starch functional properties
Starch AC has been reported to be negatively correlated with the gelatinization ( 1 Tp) and pasting properties (PV) of starches (Sasaki et al. 2000;Chen et al. 2003;Varavinit et al. 2003;Park et al. 2007). In the current study, among the six allele groups, the Wxi allele is determined as a major allele for AC increase. This result for Wxi allele is consistent with early works on Wx a , a high mRNA expressing Wx allele, which has been identified as a major gene controlling amylose content (Sano 1984;Wang et al. 1995). Wx a allele explains most of the difference in increasing amylose content, decreasing the PV and BD of RVA in an indica rice variety, which greatly affects the quality of cooked rice (Umemoto et al. 2008). In addition, the current study also shows the impact of SSIi, SSIIaj and SBEIi alleles in determining difference of amylose content in RIL lines comparing to their corresponding allele pairs. The nature of high AC for Wxi allele group remains to be studied further. Analysis of a complete set of 64 potential combinations may explain the nature of its high AC. In the perspective of starch functional properties, high values were determined for paste and thermal properties in Wxj allele and other three relatively low amylose alleles (SSIj, SSIIai and SBEIj) compared to their counterpart-alleles. This correlation again confirms starch paste and thermal properties are negatively regulated by the ratio of amylose produced in the grains. To, 1 Tp and 1 Te are onset, peak and end gelatinization temperatures, and 1 ΔH is the gelatinization enthalpy at the amylopectin gelatinization peak. 2 To, 2 Tp, and 2 Te are onset, peak and end dissociation temperatures of amylose-lipid complexes, and 2 ΔH is the dissociation enthalpy at the dissociation peak of amylose-lipid complexes. Each value is the mean of three biological replicates. Mean values within the same column with different letters are significantly different at p < 0.01.

The relation between amylopectin chain length distribution and starch functional properties
Previous studies showed that short chains (DP 6~10) and intermediate (DP 11~24) chains of amylopectin are the major components of amylopectin structure in rice endosperms (Hizukuri 1985(Hizukuri , 1986Hizukuri et al. 1989). The variations in short and intermediate chains were related with starch pasting and thermal property changes ). Thus, it is interesting to investigate the impact of indica and japonica alleles on the relation of amylopectin CLD and starch functional properties. Since R CL10/24 was the same between Wxi and Wxj alleles, it suggests Wx protein was not involved in the synthesis of short chains in amylopectin. The roles of SSIj allele in increasing starch R CL12/24 value and starch properties in rice grain were reported by early researchers (Takemoto-Kuno et al. 2006) through comparing starches from progeny lines derived from the crossing population between indica and japaonica rice. However, such increasing starch R CL10/24 value was not observed from SSIj allele in this work although starch from SSIj allele contained more short chains less than DP 24 comparing to that from SSIi allele. The discrepancy between early publication (Takemoto-Kuno et al. 2006) and our current work may be due to the different varieties used to develop RIL population or the effects from SSIj and SBEIIbj combination. Umemoto et al. (2008) studied a near-isogenic line-NIL (SSI k ) containing a SSI allele from an indica rice on a japonica rice background and concluded the variation in SSI alleles between indica and japonica rice hardly affected starch peak viscosity and gelatinization temperatures. The SSIIaj allele starch showing more short chains and less intermediate chains was determined with a remarkably high R CL10/24 . This variation led to distinctive characteristics of starch properties with the lowest PV, BD, and lowest gelatinization temperatures. Despite other alleles within Wxi allele group showed lower PV and BD than IR64, they were comparable in R CL10/24 values. Consequently, these alleles barely contribute to the variations of RVA and DSC characteristics within the Wxi allele group. Within the Wxj allele group, the SSIIai allele was prevalent showing similar CLD patterns, which was significantly different from Nipponbare. The SBEIIbj allele had a significantly higher R CL10/24 value comparing to SBEIIbi, but it was much lower than that of Nipponbare. The SBEIIbj allele had abundant short chains in amylopectin with lower PV, Tr and lower FV and lower 1 Te than SBEIIbi allele. Given all the analyses above, R CL10/24 is negatively correlated with starch viscosity and thermal gelatinization properties. An increase of R CL10/24 can reduce the PV values of RVA and gelatinization temperature of DSC due to the increase of short chains proportion and/or decrease of intermediate chains proportion. SSIIaj is a major allele and SBEIIbj is a minor allele in the determination of R CL10/24 in this study. Interestingly, starch PV and gelatinization temperatures of the SSIIaj allele are even lower than Nipponbare, though the SSIIaj allele R CL10/24 is smaller. This is because SSIIaj allele combines with Wxi allele in the corresponding RILs. Such changes of starch properties were also reported by Umemoto et al. (2008) through study of near-isogenic line-NIL (Wxa) which contained a chromosome segment containing a Wx allele from an indica rice (Kasalath) on a janonica (Nipponbare) rice background. Our analyses here demonstrate that both high amylose content and high R CL10/24 regulate starch properties by lowering starch PV and gelatinization temperatures. SSIIaj and SBEIIbj alleles contribute to high R CL10/24. Therefore, the novel starch functional properties of SSIIaj allele is probably the result of the combination of major alleles, SSIIaj (high R CL10/24 ) and Wxi (high AC), with minor alleles, SBEIi and SBEIIbj.

The relation between starch structure and starch granule bound proteins
The Wx allele group showed significant variation in the abundance of GBSSI in the starch granule of rice grains. RILs with the Wxj allele contained a relatively low abundance of GBSSI and a low level of starch AC compared with those with the Wxi allele. This was consistent with previous studies which demonstrated that Wxj allele produces low levels of mature enzymes (Cai et al. 1998;Hirano et al. 1998), and rice varieties with Wxi and Wxj alleles have different levels of starch AC in rice grains (Sano 1984;Wang et al. 1995). However, the change of GBSSI abundance inside starch granules did not affect the abundance of other starch biosynthetic enzymes in starch granules. SSIIa was the other protein with varied abundance in starch granules between indica and japonica alleles. Moreover, SSIIaj allele starch contained a reduction of SSI and SBEIIb. According to Umemoto et al. (2004), the SSIIa protein from IR64 and Nipponbare differ by two amino acid substitutions, the protein from Nipponbare has reduced SSIIa activity. Furthermore, the reduced SSIIa activity resulted in the decrease of SSIIa abundance in starch granules, and ultimately produced amylopectin with increased R CL10/24 values and slightly increased AC. Although SSIi, SBEIIaj and SBEIIbj alleles showed limited effects on starch structures and properties, the changes in the expression levels of these proteins were hardly detected from current study.
A recent study in rice and maize suggested that the majority of GBPs are involved in starch biosynthesis (Koziol et al. 2011). It indicated that the GBPs are important for the starch structure and properties. In our study, the remarkable changes in GBPs of SSIIa and Wx alleles observed in starch granules is associated with the pronounced changes in starch structure and functional properties as described earlier. This is consistent with the observations by numerous previous studies on single or double starch biosynthetic gene-recessive mutants of SSI, SSIIa and SBEIIb proteins, not only in rice but other cereals (Morell et al. 2003;Nakamura et al. 2005a;Kosar-Hashemi et al. 2007;Grimaud et al. 2008;Li et al. 2011). Therefore, detecting significant changes in the abundance and composition of GBPs is an alternative approach to screen variants with favorable starch structure and functional properties.

Conclusions
This study showed that rice Wxi and SSIIaj alleles are major contributors, whereas SSIi, SBEIi, SBEIIai and SBEIIbj alleles are minor contributors to rice starch properties between indica and japonica rice. The combination of SSIIaj and Wxi with SBEIIbj and SBEIi alleles has accumulative effects lowering PV and gelatinization temperature, which leads to the novel functional properties of SSIIaj allele starch in the study. Starch AC and the amylopectin CLD determine a number of starch functional properties. The R CL10/24 in amylopectin is negatively correlated with PV and BD of paste property and GTs of thermal property. Moreover, we suggest that the alteration of GBPs inside starch granules is a potential indicator for significant changes in starch structure and functional properties. A future study involving 64 possible combinations for the 6 gene groups in a common genetic background in rice will lead to an in-depth understanding of starch structures and functional properties.

Plant materials
Two rice (Oryza sativa) parental lines [cultivars IR64 (indica pollen donor) and Nipponbare (japonica acceptor)] were used for producing a RIL population. Four F2 seeds from one F1 seed were produced (kindly provided by Dr. Narayana Upadhyaya). Only one F2 plant was used for producing all RILs in this study. The detailed steps for selecting RILs for six allele groups are listed in Figure 1. All plants including RILs and their parental lines were grown in a glass house of a plant growth facility at CSIRO Agriculture Flagship (Canberra, ACT, Australia) at 27°C under natural light. Mature seeds were harvested for starch preparation, and starch structure and properties analysed.

Genomic DNA preparation
The genome DNA was extracted with a quick DNA extraction method. Fresh young leaves of 2 cm length were sampled into microfuge tubes on ice and immediately ground in 66 μl of 0.5% NaOH using a micropestle. Samples were then incubated in a boiling water bath for one minute, and diluted with 120 μl of 0.1 M Tris-HCl, pH8.0. DNA was precipitated by adding 180 μl isopropanol and 18 μl 3 M sodium acetate, pH5.2 to sample tubes and gently mixed by reversing the tube, and incubated at room temperature for 5 min. The supernatant was discarded after the centrifugation for 30 min at 3000 g, and the pellet was air dried at room temperature for 1 min and washed with 180 μl of 70% (V/V) ethanol. After centrifugation at 3000 g, the genomic DNA in the pellet was then dissolved in 50 μl MilliQ water.

Genotyping of selected genes
The genotypes of RILs for the selected genes in the study were identified based on a PCR approach using published markers (Yan et al. 2007;Tian et al. 2010). For each allele, five individual plants were selected for generating data for statistical analysis. Due to the limitation of material availability, only three of five plants with more grains of each allele were used for starch property characterization. Alleles of indica derived from IR64 were assigned as Wxi, SSIi, SSIIai, SBEIi, SBEIIai and SBEIIbi, and corresponding japonica alleles derived from Nipponbare were named as Wxj, SSIj, SSIIaj, SBEIj, SBEIIaj and SBEIIbj. The alleles were segregated in different combinations during subculture, and F5 homozygotes were used for further analyses.

Starch preparation
Whole grains of each line (approximately 100~150 mg) were ground in a capsule with ball bearing using an ESPE CapMix™ (model 3 M, AU). Grinding was done for 30 sec three times. The wholemeal flour was then washed with 0.005% NaOH by vigorous vortexing for 2 minutes followed by filtration through 0.5 mm nylon sieves. Each sample was then washed three times by vortex mixing and centrifugation at 5000 g for 5 min. The starch pellet was resuspended in a phosphate buffer (50 mM, pH7.5) containing proteinase K (50 μg/ml) and incubated at 37°C for 2 hrs. After 5 min centrifugation at 5000 g, the pellet was suspended in water and centrifuged. Thus water wash was repeated three times. The starch pellet was then washed with acetone, centrifuged and the residual starch was air dried at 37°C overnight.

Amylose content assay
The amylose content of samples was determined by analysing debranched starches using size-exclusion chromatography (SEC) with Ultrahydrogel (Waters, Milford, MA, USA) as described previously (Butardo et al. 2011). Pullulan standards (Shodex P-82) calibrated with the Mark-Houwink-Sakaruda equation were used for the estimation of the molecular weight from the elution time (Castro et al. 2005;Ward et al. 2006). Samples were prepared and analysed in triplicate.

Chain length distribution of debranched starch
Samples were prepared as previously described with modifications (O'Shea and Morell 1996). For each sample, rice wholemeal (10 mg) was first presoaked in 40 μl ethanol and then incubated with 200 μl NaOH (0.25 M) and 600 μl MilliQ water in a boiling water bath with stirring for 10 min. The samples were then neutralized in the sodium acetate buffer (0.2 M sodium acetate and 3.6% V/V glacial acetic acid). Afterwards, 10 μl of isoamylase (1000 U/ml, Megazyme) was added to each sample to debranch the starch by incubating at 50°C for 2 h. The samples were boiled again for 10 min to denature isoamylase and centrifuged at 15000 g for 10 min. A 50 μl aliquot of supernatant was taken and dried down in a speed vacuum. The dried samples were dissolved in APTS (8-Aminopyrene-1,3,6-trisulfonic acid trisodium salt, BECKMAN COULTER) sodium cyanoborohydride buffer (containing 5 mg APTS labeling dye in 48 μl 15% acetic acid) and incubated overnight at 50°C. After the incubation, the samples were boiled in a urea solution (6 mM) for 1 min and filtered through Wizard mini-Columns (Promega). Determination of the chain length distribution (CLD) of amylopectin was conducted by fluorescence-activated CE as previously described (O'Shea and Morell 1996).

Starch paste viscosity
The polished rice grains (about 5 g) were ground into flour in a stainless steel capsule on a ball bearing machine (model MM300, MEP instruments Pty Ltd, NSW, AU) for starch property analysis. A Rapid Visco Analyzer (model RVA-4SA, Newport Scientific, Sydney, NSW, AU) was used to determine starch paste viscosity parameters of the selected RILs. Three grams of rice grain wholemeal for each of three plants of each RIL were used for analyses. The program setting for the RVA comprised the following stages: hold at 60°C for 2 min, heat to 95°C over 6 min, hold at 95°C for 4 min, cool to 50°C over 4 min, and hold at 50°C for 4 min.
The software Thermocline (Newport Scientific PtyLtd, Warriewood, NSW, AU) was used for data collection and analysis.

Differential scanning calorimetry
A Differential Scanning Calorimeter (model Pyris 1DSC, Perkin Elmer, Norwalk, CT, US) was used to determine the calorimetry profiles of flour samples of 3 different plants with triplicate for each RIL. For each line, 60 mg of flour was weighed and mixed with water at ratio of 1.76 : 1. Afterwards, 3 aliquots with 45 mg of each prepared from the starch-water premixed product were transferred to DSC pans separately, and used as replicates. The pans were then sealed and placed on the bench at room temperature to equilibrate overnight. A heating rate of 10°C min was used to heat the samples at 30~130°C. Data was analysed using the instrument software provided by the manufacturer.

Preparation of granule bound proteins from mature grains
The starch prepared from mature grains by the method described above was used for preparing GBPs of each line. Following the procedures by Rahman et al. (1995) with some modifications, 4 mg of starch was boiled in a protein denaturing extraction buffer (50 mM Tris buffer, pH 6.8, 10% glycerol, 5% SDS, 5% β-mercaptoethanol, and bromophenol blue) at a ratio of 15 μl/mg starch for 5~10 min. After the centrifugation at 13000 g for 20 min, the supernatant was ready for SDS-PAGE analysis.

SDS-PAGE and gel staining
Supernatants (25 μl) containing GBPs of each sample together with 5 μl of marker were loaded onto SDS-PAGE gels, respectively. SDS-PAGE was conducted in Nu-PAGE 4-12% gradient gels (Invitrogen) via MOPS-Tris-SDS buffer system containing (50 mM MOPS, 50 mM Tris, 0.1% SDS and 1 mM EDTA) and operated at 20 mA constant current per gel in an Xcell SureLock™ Mini Cell (Invitrogen). After electrophoresis, gels were stained by Sypro Ruby stain (Bio-Rad) following the manufacturer's instructions, and then visualized under a UV transilluminator (Uvitec, UK). MagicMark™ XP Western Standard protein ladders (Invitrogen) were used to estimate the molecular weight of protein bands.

Statistical analyses
Statistical analyses were performed using Genstat version 9. Analysis of variance was performed for starch AC, amylopectin structure, RVA and DSC characteristics to obtain the least significant differences at p < 0.01, looking at variations among the RILs.