• Open Access

Gelatinization temperature of rice explained by polymorphisms in starch synthase

Authors


* Correspondence (fax +61 2 66222080; e-mail rhenry@scu.edu.au)

Summary

The cooking quality of rice is associated with the starch gelatinization temperature (GT). Rice genotypes with low GT have probably been selected for their cooking quality by humans during domestication. We now report polymorphisms in starch synthase IIa (SSIIa) that explain the variation in rice starch GT. Sequence analysis of the eight exons of SSIIa identified significant polymorphism in only exon 8. These single nucleotide polymorphisms (SNPs) were determined in 70 diverse genotypes of rice. Two SNPs could classify all 70 genotypes into either high GT or low GT types which differed in GT by 8 °C. ‘A’ rather than ‘G’ at base 2412 determined whether a methionine or valine was present at the corresponding amino acid residue in SSIIa, whilst two adjacent SNPs at bases 2543 and 2544 coded for either leucine (GC) or phenylalanine (TT). Rice varieties with high GT starch had a combination of valine and leucine at these residues. In contrast, rice varieties with low GT starch had a combination of either methionine and leucine or valine and phenylalanine at these same residues. At least two distinct polymorphisms have apparently been selected for their desirable cooking qualities in the domestication of rice.

Introduction

Rice (Oryza sativa) starch is the major energy source for many millions of people around the world (Fitzgerald, 2004). Starch is a polymer of glucose which presents as a mixture of two forms, amylose and amylopectin. Amylose is principally composed of a linear polymer of α(1–4)-linked glucose with some α(1–6) linkages, whereas amylopectin is a more complex mixture of both α(1–4)-linked glucose extensively branched by α(1–6) linkages. In its native state, rice starch has a semicrystalline structure which is disrupted by cooking, transforming the starch into a softer, edible, gel-like material. Because it is associated with the cooking time and texture of cooked rice and cool cooked rice, the temperature at which rice starch gelatinizes is an important component of rice eating quality (Maningat and Juliano, 1978).

Starch is synthesized by the activity of several enzymes and has been the subject of extensive recent reviews (Ball and Morell, 2003; Fitzgerald, 2004). Regardless of the subsequent synthetic path, glucose is first activated in preparation for starch synthesis by adenosine-5′-diphosphate glucose pyrophosphorylase (AGPase), and the adenosine-5′-diphosphate glucose so produced becomes the substrate for the starch synthases (SSs). Granule-bound starch synthase (GBSS) is exclusively responsible for the synthesis of amylose, and the SSs extend the α(1–4) links of amylopectin. Starch branching enzymes (SBEs) insert α(1–6) branches in the amylopectin chains, whereas starch debranching enzymes (SDBEs) cleave the growing chains at α(1–6) linkages in what is believed to be a necessary part of the process of remodelling the growing amylopectin, allowing it to crystallize (Ball and Morell, 2004). Each of these enzymes occurs as a number of different isoforms which display tissue-specific expression.

A link between the gelatinization temperature (GT) of rice starch and enzymes of starch biosynthesis has been made, with the finding that a major gene that controls rice GT, as determined by the indirect measure of alkali spreading, genetically maps to a region of chromosome six that is clearly different from the waxy locus (Umemoto et al., 2002). This gene co-segregates with starch synthase IIa (SSIIa) and a gene that affects amylopectin structure (Umemoto et al., 2002), suggesting that the cause and effect or genotype to phenotype hierarchy of GT is as follows: SSIIa, amylopectin structure, GT. Analysis of near-isogenic rice lines (NIL), in which a narrow genomic region surrounding the SSIIa gene of the high-GT variety Kasalath was introgressed into the low-GT variety Nipponbare, provides further support for the hypothesis that SSIIa is the enzyme responsible for the natural variation in GT (Umemoto et al., 2004). Western analysis for SSIIa in two rice varieties that differed in terms of starch disintegration in both urea and alkali found that the amount of SSIIa was reduced in the variety that was more sensitive to both alkali and urea (Jiang et al., 2004).

The data from rice are supported by data derived from peas and barley. Craig et al. (1998) found that the degree of polymerization (DP) of the A chains was reduced in peas that lacked SSII activity, in a similar manner to that reported by Umemoto and co-workers for rice. It has been shown more recently that mutant barley SSIIa, which lacks its catalytic domain as a result of the occurrence of a premature stop codon in the coding sequence, has a higher proportion of chains with a DP of 6–11 relative to a DP of 12–30 in comparison with wild-type barley amylopectin (Morell et al., 2003). In addition to this, the starch from the mutant barley had a lower GT (Morell et al., 2003), suggesting that amylopectin structure is a determinant of starch GT.

Umemoto et al. (2004) reported four haplotypes of the gene that codes for SSIIa, based on four combinations of three functional single nucleotide polymorphisms (SNPs). However, GT and chain length distribution were not unique to all haplotypes in this study. Rices of haplotype 2 were mainly indica types and varied widely in GT. Rices of haplotype 3 were composed of both low and high GT types, with corresponding structural differences of amylopectin. Thus, as suggested by Umemoto et al. (2004), it is likely that the basis of natural variations in the rice SSIIa gene has not been fully elucidated. Working on the sequence encoding the SSIIa gene in a set of rice genotypes different to that utilized by Umemoto and co-workers, we have confirmed two of the SNPs reported by Umemoto et al. (2004), SNP2 and SNP3, and have found an additional functional SNP in exon 8. This additional SNP affects the SSIIa coding sequence and allows differentiation of the rice varieties examined into two discrete GT classes.

Results

Polymorphism identification and assay

The DNA sequence of cDNA clone C73554, which was sequenced and genetically mapped by Umemoto et al. (2002), was used as the query sequence in a blastn search. blastn identified several database entries with a DNA sequence similar to that of C73554. Alignment of these sequences with each other and comparison of the conceptual translations with the translated sequence in Umemoto et al. (2002) determined that the cDNA AF419099 [National Center for Biotechnology Information (NCBI) accession] was a full-length version of C73554. The first base pair of C73554 aligned to base pair 1239 of AF419099. A blastn enquiry of the Gramene database with both AF419099 and C73554 found that the BAC clone AP003509 contained the gene from which AF419099 and C73554 were transcribed. clustalw alignment of AP003509 and AF419099 confirmed that the gene contained eight exons (Figure 1). All eight exons were amplified and sequenced in the Australian rice varieties Opus, Doongara and Langi, and the sequences were aligned and compared with each other and with the rice variety Nipponbare SSIIa coding sequence (BAC clone AP003509) using clustalw. The GTs of Nipponbare, Opus, Doongara and Langi were 68, 69, 74 and 78 °C, respectively (Table 1). A single SNP was identified: a ‘G’ to ‘A’ transition in exon 8 corresponding to base pair 2412 of AF419099. Langi and Doongara carried the ‘G’ allele, whereas Opus and Nipponbare carried the ‘A’ allele. This SNP was also identified by Umemoto et al. (2004) and corresponded to SNP3 in their analysis. In order to maintain consistency, the nomenclature of Umemoto and co-workers is used in this analysis. SNP3 resulted in a conservative amino acid change from methionine (ATG) to valine (GTG). Sequence analysis of a further 66 varieties that differed by GT, using polymerase chain reaction (PCR) primers which amplified a fragment of exon 8 between base pair 1880 and 2643 of AF419099 (the terminal TGA ‘A’ is base pair 2636 and the first base of exon 8 is base pair 1681 of AF419099), allowed the identification of another polymorphism which led to an amino acid change. This polymorphism consisted of two adjacent bases that corresponded to base pairs 2543 and 2544 of AF41099, which were found to be either ‘GC’ or ‘TT’. Only the second base of the two affected the amino acid sequence of the protein, with the alternative amino acids being leucine (CTC) or phenylalanine (TTC). Although this is technically not an SNP, it is called SNP4 for convenience.

Figure 1.

Schematic representation of the gene that codes for rice starch synthase IIa (SSIIa) showing the positions of exons, introns and single nucleotide polymorphisms (SNPs).

Table 1.  Gelatinization temperature (GT) and single nucleotide polymorphism (SNP) genotypes of 70 rice varieties. For named cultivars, the country of origin is listed; breeding lines developed within the Australian Rice Breeding Programme are listed as originating from Australia
VarietyGT (°C)SNP2SNP3SNP4Country of origin
Akitakomachi71AAGCJapan
Ali Combo68AAGCItaly
Ali Combo/Vialone nano71AAGCItaly
Ardito68AAGCItaly
Haenukai69AAGCItaly
Jarrah69AAGCAustralia
Koshihikari73AAGCJapan
Matsuribare71AAGCJapan
Millin68AAGCAustralia
Nipponbare68AAGCJapan
Opus69AAGCAustralia
Opus/Matsuribare71AAGCJapan
Somewake74AAGCJapan
Bluebelle76AGGCUSA
Bluebelle///M9/Pelde//YRL30/4/YRF20779AGGCAustralia
Dawn75AGGCUSA
Della76AGGCUSA
Dellmont74AGGCUSA
Domsiah I75AGGCIran
Doongara74AGGCIran
Goolarah79AGGCAustralia
Gulfmont78AGGCAustralia
Hungarian No. 179AGGCHungary
L20280AGGCUSA
L203//71048.200/Hungarian No.l///Rexmont78AGGCAustralia
Langi78AGGCAustralia
M201/YRM3//Bogan///H989-4-S82AGGCAustralia
Milagrosa79AGGCPhilippines
Moroberekan77AGGCGuinea
Pelde80AGGCAustralia
Rexmont79AGGCUSA
YRF20378AGGCAustralia
YRF207/L20276AGGCAustralia
YRL10279AGGCAustralia
YRL11881AGGCAustralia
YRL118///Inga/M9//213D.2581AGGCAustralia
Amaroo67AGTTAustralia
Bogan70AGTTAustralia
Calmochi 20271AGTTUSA
Calrose63AGTTAustralia
Echuca72AGTTAustralia
Haenukai/Illabong67AGTTAustralia
Illabong67AGTTAustralia
IRBB5973AGTTPhilippines
Kairyo Mochi70AGTTJapan
M10168AGTTUSA
M10371AGTTUSA
M201//YR196/Ardito///YRM5470AGTTAustralia
M201/Bogan//Wakamizu71AGTTAustralia
M20272AGTTUSA
M40168AGTTUSA
M969AGTTUSA
Shimuzi mochi65AGTTJapan
Vialone Nano62AGTTItaly
Wakamizu68AGTTJapan
YRL10173AGTTAustralia
YRM4269AGTTAustralia
YRM5466AGTTAustralia
YRM54/Akitakomachi70AGTTAustralia
YRM54/Rexmont72AGTTAustralia
YRM6267AGTTAustralia
YRW473AGTTAustralia
Amber79GGGCIraq
Basmati 37080GGGCPakistan
Domsiah II78GGGCIran
Dumsorkh73GGGCIran
IRBB6083GGGCPhilippines
Moosa Tarom76GGGCIran
RIL26680GGGCUSA
Teqing79GGGCChina

Umemoto et al. (2004) identified an additional SNP (SNP2) in exon 8 which resulted in an amino acid change. This SNP (A/G), which corresponded to base pair 2013 of AF41099, was assayed in these same 70 genotypes. Because Umemoto and co-workers clearly demonstrated that SNP1 did not affect GT (Umemoto et al., 2004), SNP1 was not assayed in this work. Analysis of SNP2, SNP3 and SNP4 in all 70 different genotypes found that only four of the eight possible combinations or haplotypes were represented: G/G/GC, A/G/GC, A/A/GC, A/G/TT.

GT determination and statistical analysis

When grouped by haplotype class [test of statistical significance; no significant difference in variance, one-way analysis of variance (anova), Tukey HSD post hoc test, P = 5.968 × 10−20], two GT classes were evident (Figure 2). The high-GT class had an average GT of 78 °C and was composed of haplotype 1 (G/G/GC; eight varieties) and haplotype 2 (A/G/GC; 23 varieties). The low-GT class had an average GT of 70 °C and was composed of haplotype 3 (A/G/TT; 26 varieties) and haplotype 4 (A/A/GC; 13 varieties).

Figure 2.

Relationship between haplotype and gelatinization temperature (GT) for 70 rice genotypes. Mean (M) GT (°C) and standard deviation (SD) for each haplotype (from left to right) were as follows: haplotype 1 (G/G/GC), M = 78.50, SD = 2.976; haplotype 2 (A/G/GC), M = 77.96, SD = 2.246; haplotype 3 (A/G/TT), M = 69.00, SD = 2.966; haplotype 4 (A/A/GC), M = 70.00, SD = 2.000. Bars represent the 95% confidence interval around the mean for each haplotype.

Discussion

Targeted sequence analysis of exon 8 of the SSIIa gene in 70 different rice varieties that differed by GT found three polymorphisms that resulted in a changed amino acid sequence and, of these three, two that were correlated with GT differences. Four of the possible eight haplotypes were found in these varieties and, when grouped by haplotype, a low-GT class and a high-GT class were evident. The presentation of two GT classes is consistent with the finding that the fine structure of Asian rice varieties falls into one of two categories: either the L-type, which has more long cluster chains, or the S-type, which has more short cluster chains (Nakamura et al., 2002). In cultivated rice, it is possible that the combination of ‘G’ (valine) at SNP3 and ‘GC’ (leucine) at SNP4 is required for the production of L-type rice starch, and this has a higher GT relative to S-type starch. Changing the valine to methionine or the leucine to phenylalanine may change starch from the L-type to the S-type which, in turn, reduces the GT of the starch. Measurement of the chain length profile will determine whether this is indeed the case. In order to gauge the extent to which phenotypic changes are exerted through direct effects on SSIIa, or pleiotropic effects on other SSs, it would be necessary to undertake SS activity and binding assays of the SSs, including SSIIa, SBEs and SDBEs, in representatives of each of the haplotypes.

Umemoto et al. (2004) found that haplotype 3 of their analysis contained two groups that differed in their tolerance to alkali, amylopectin chain length distribution and amount of granule-bound SSIIa. One group showed high tolerance to alkali, a longer chain length distribution and more granule-bound SSIIa; the other showed low tolerance to alkali, a shorter chain length distribution and less granule-bound SSIIa. Excluding SNP4 from the analysis reported here would mean that rice that most probably corresponds to Umemoto haplotype 3 (haplotypes 2 and 3 in this work) would also form two groups based on GT. However, the addition of SNP4 to the analysis provides a genetic explanation for the occurrence of these two groups, and SNP4 may be the unidentified SNP referred to by Umemoto and co-workers. Thus, SNP3 and SNP4 may be the important genetic polymorphisms which lead to both a biochemical change, an amino acid substitution which appears to affect SSIIa starch granule binding (Umemoto et al., 2004), and a functional change, which is manifested in altered GT.

Comparison of the amino acid sequence of rice SSIIa with orthologous sequences from other cultivated cereals deposited in the NCBI database, and identified using tblastx (Table 2), showed that the amino acids identified here in rice, which were correlated with low-GT starch, have not been captured in other cereals. It could be that, during the domestication of rice, two independent mutations have been captured because they both impart to rice desirable cooking characteristics in the form of a low starch GT. Unlike other cereals, a significant proportion of the rice harvested is consumed directly, and so the cooking qualities are important parameters which impact on the selection process.

Table 2.  A comparison of the amino acids (AA) and their codons found at equivalent positions to rice starch synthase IIa (SSIIa), single nucleotide polymorphism 2 (SNP2), SNP3 and SNP4 in wheat, barley, maize and rice
SpeciesSNP2 (AA/codon)SNP3 (AA/codon)SNP4 (AA/codon)
  1. GT, gelatinization temperature.

WheatGlycine/GGCValine/GTGLeucine/CTC
BarleyGlycine/GGCValine/GTGLeucine/CTC
MaizeGlycine/GGAValine/GTGLeucine/CTC
High-GT riceAGC/serine or GGC/glycineValine/GTGLeucine/CTC
Low-GT riceAGC/serineMethionine/ATGLeucine/CTC
Low-GT riceAGC/serineValine/GTGTTC/phenylalanine

Low-amylose rice starch results from the relative inactivity of GBSS (Sano et al., 1986; Hirano and Sano, 2000; Mikami et al., 2000), whereas rice starch with a low GT results from the presence of a relatively inactive SSIIa (Umemoto et al., 2002, 2004). It seems that both attributes, a low-amylose rice starch and rice starch with a low GT, arise from mutations that diverge from the wild-type and have been captured through artificial selection. It is interesting to note that mutations in the SDBEs produce a similar phenotype (reduced starch GT and a polysaccharide that is less crystalline with amylopectin composed of a higher proportion of short chains relative to the wild-type; Kubo et al., 1999; Fujita et al., 2003; Wong et al., 2003) to that which has been observed with SSIIa. This supports the notion that rice starch with a high GT is the wild-type and that mutations which arise in the starch biosynthetic enzymes, SSIIa and SDBEs, have the potential to disrupt starch structure in such a way as to reduce starch GT.

The mutations identified here, which explain GT class, are removed by only 30 and 74 amino acids from the carboxyl terminal of the SSIIa protein. Maize SSI requires the C-terminus of the protein to be intact for catalytic activity and starch binding (Commuri and Keeling, 2001). Umemoto and co-workers found SSIIa to be absent from the granule-bound fraction of Nipponbare (Umemoto et al., 2004). Although the degree of similarity between SSI and SSIIa is not high, it is possible that the conservative variations in the SSIIa amino acid sequence observed here occur in a domain of SSIIa which is critical for enzyme activity.

Variations in SSIIa may exert their influence on starch structure directly by altering the capacity of SSIIa to synthesize amylopectin, or via pleiotropic effects. Although mutations within SSIIa in maize were not found to affect the branching enzymes (BEs) or debranching enzymes (DBEs) (Zhang et al., 2004), Morell et al. (2003) found that mutations that led to truncation of the SSIIa protein in barley resulted in a loss from the granule-bound fraction of not only SSIIa, but also BEIIa, BEIIb and SSI. This finding led them to suggest that SSIIa, BEIIa, BEIIb and SSI may form a complex that is disrupted by the sex6 SSIIa mutation, a hypothesis that is supported by the observation that SBEIIb and SBEI form a complex in wheat (Tetlow et al., 2004). Mutation in an SSII encoding gene in peas (rug5) has the effect of increasing the activity of both soluble SSs and GBSSs (Craig et al., 1998). Therefore, it is possible that mutations within rice SSIIa may manifest themselves via pleiotropic effects on other starch synthesis enzymes, either through a physicochemical interaction between the enzymes or by affecting substrate availability to other starch synthesis enzymes.

A diet composed of high glycaemic index (GI) foods is associated with an elevated risk of developing type 2 diabetes, cardiovascular disease and, in particular, cancers (Foster-Powell et al., 2002). Although the relationship between rice starch and GI is complex, amylose content is generally accepted as being the principal determinant of GI in rice (Larsen et al., 2000). However, Panlasigui et al. (1991) have demonstrated that, in rice varieties that differ by GI yet have similar amylose contents, starch gelatinization properties, including GT, cooking time, volume expansion and amylograph consistency, are the major predictors of GI. In addition to their impact on cooked rice texture and starch retrogradation properties, the SNPs identified here in SSIIa may go some way in assisting in GI prediction via their association with starch GT.

Perfect molecular markers are now being used in rice breeding programmes for a range of traits, including amylose content (Ayres et al., 1997), blast disease resistance (Jia et al., 2004) and fragrance (Bradbury et al., 2005). A knowledge of the SNPs in the SSIIa encoding gene which affect GT will allow accurate selection for rice varieties that have desirable starch gelatinization properties within breeding programmes.

Experimental procedures

Plant material

Rice (O. sativa) lines that represent a range of GTs were obtained from the Australian Rice Improvement Programme at Yanco, NSW, Australia. These included several varieties and a series of F3 derived F4 (F3:4) breeding lines (Table 1).

GT determination

Rice grains were dehulled (Satake, Hiroshima, Japan), milled (60 s Magill No. 2) and ground to flour (Cyclotec, particle size less than 0.5 mm; Cyclotec 1093, Tecator, Hoganas, Sweden) for the measurement of GT. GT was measured on flour of all varieties using differential scanning calorimetry (DSC) (Mettler-Toledo, Columbus, OH). Flour (4 mg) was mixed with water (10 mg) and the pan was sealed. The heating rate was 10 °C/min, from 25 to 120 °C. GT is reported as the maximum at the peak of the endotherm.

Bioinformatics and statistical analysis

blastn and blastx were used to identify sequences with homology to C73554, which was sequenced and mapped by Umemoto et al. (2002), within the NCBI (http://www.ncbi.nlm.nih.gov/) and Gramene (http://www.gramene.org/) databases. Sequence alignment was undertaken using clustalw within MacVector™ 6.5.1 (Oxford Molecular Group, Cambridge, UK). Primers were designed using MacVector™ 6.5.1. Statistical analysis was undertaken using the software package SPSS (Chicago, IL).

DNA extraction, PCR, sequence analysis and genotyping

Genomic DNA was extracted using a Qiagen Dneasy® 96 Plant Kit (Qiagen GmbH, Hilden, Germany). DNA preparations were diluted in TE buffer (10 mm Tris-HCl, 1 mm EDTA, pH 8) to a final concentration of approximately 10 ng/mL. Oligonucleotide primers were synthesized by Proligo Australia Pty Ltd (Lismore, NSW, Australia). PCR was performed using a Perkin-Elmer Gene Amp PCR system 9700 (Wellesley, MA). The reaction volume was 25 µL containing 10 ng of extracted genomic DNA, 2.5 mm of MgCl2, 200 µm of total deoxynucleoside triphosphates (dNTPs), 1 unit of Platinum® Taq DNA Polymerase (Gibco BRL, Invitrogen, Carlsbad, CA), 1 × Gibco® PCR Buffer (minus MgCl2) and 0.2 µm of each forward and reverse primer. Cycling conditions were 94 °C for 2 min, followed by 30 cycles of 94 °C for 30 s, 55 °C for 30 s and 72 °C for 1 min, followed by a final extension of 72 °C for 7 min.

Prior to sequencing, PCR products were purified using a Montage PCR filter device (Millipore Corporation, Billerica, MA). Sequence reactions were performed on PCR products with both forward and reverse PCR primers using BigDye Terminator version 3.1 (Applied Biosystems, Foster City, CA), and the completed reactions were purified by ethanol precipitation. The reaction products were analysed on an Applied Biosystems 3730 Genetic Analyser.

Acknowledgements

The authors wish to thank Dr Lyndon Brooks and Dr Peter Bundock for advice provided on statistical analysis and Judy Dunn for the starch GT determination. The Australian Rural Industries Research and Development Corporation provided the funds for this research.

Ancillary