|Home | About | Journals | Submit | Contact Us | Français|
In this study of barley starch synthesis, the interaction between mutations at the sex6 locus and the amo1 locus has been characterized. Four barley genotypes, the wild type, sex6, amo1, and the amo1sex6 double mutant, were generated by backcrossing the sex6 mutation present in Himalaya292 into the amo1 ‘high amylose Glacier’. The wild type, amo1, and sex6 genotypes gave starch phenotypes consistent with previous studies. However, the amo1sex6 double mutant yielded an unexpected phenotype, a significant increase in starch content relative to the sex6 phenotype. Amylose content (as a percentage of starch) was not increased above the level observed for the sex6 mutation alone; however, on a per seed basis, grain from lines containing the amo1 mutation (amo1 mutants and amo1sex6 double mutants) synthesize significantly more amylose than the wild-type lines and sex6 mutants. The level of granule-bound starch synthase I (GBSSI) protein in starch granules is increased in lines containing the amo1 mutation (amo1 and amo1sex6). In the amo1 genotype, starch synthase I (SSI), SSIIa, starch branching enzyme IIa (SBEIIa), and SBEIIb also markedly increased in the starch granules. Genetic mapping studies indicate that the ssIIIa gene is tightly linked to the amo1 locus, and the SSIIIa protein from the amo1 mutant has a leucine to arginine residue substitution in a conserved domain. Zymogram analysis indicates that the amo1 phenotype is not a consequence of total loss of enzymatic activity although it remains possible that the amo1 phenotype is underpinned by a more subtle change. It is therefore proposed that amo1 may be a negative regulator of other genes of starch synthesis.
The endosperm of barley (Hordeum vulgare L.) grain typically contains ~50–60% starch composed of ~25% amylose and 75% amylopectin. Amylose is a mostly linear α-(1–4)-linked glucosyl chain with a few α-(1–6)-linked glucan chains and has a mol. wt of 104–105Da. Amylopectin is a highly branched glucan in which α-(1–4)-linked glucosyl chains (with 3– 60 glucosyl units) are connected by 4–5% α-(1,6) linkages, and has a mol. wt of 105–106Da. A suite of enzymes are involved in cereal starch biosynthesis, including ADP-glucose pyrophosphorylases (EC 220.127.116.11), starch synthases (EC 18.104.22.168), starch branching enzymes (SBEs; EC 22.214.171.124), and starch debranching enzymes (EC 126.96.36.199 and 188.8.131.52) (see reviews by Kossmann and Lloyd, 2000; Rahman et al., 2000; Ball and Morell, 2003; James et al., 2003; Tetlow et al., 2004; Morell et al., 2006).
Starch synthases transfer glucose from ADP-glucose to the non-reducing end of pre-existing α-(1–4)-linked glucosyl chains of starch. In higher plants, five classes of starch synthases are consistently present; granule-bound starch synthase (GBSS), starch synthase I (SSI), SSII, SSIII, and SSIV (Li et al., 2003). Ten starch synthase genes have been identified in the rice genome (Hirose and Terao, 2004); two GBSS isoforms (GBSSI and GBSSII), one SSI isoform, three SSII isoforms [SSIIa (also defined as SSII-3), SSIIb (SSII-2), and SSIIc (SSII-1)], two SSIII isoforms [SSIIIa (SSIII-2) and SSIIIb (SSIII-1)], and two SSIV isoforms [SSIVa (SSIV-1) and SSIVb (SSIV-2)] (Hirose and Terao, 2004; Fujita et al., 2007). Proteins corresponding to SSI, SSIIa, and GBSSI have been detected within the starch granules from endosperm, whereas SSIIIa has only been detected in the soluble phase of amyloplastids (Li et al., 2000).
Significant progress has been made in identifying roles of the starch synthases individually and cooperatively in determining the fine structure of starch. GBSS is essential for the biosynthesis of amylose (Nelson and Rines, 1962; Murata et al., 1965; Eriksson, 1969; Delrue et al., 1992; Nakamura et al., 1995), and also contributes to the synthesis of the long chains of amylopectin (Maddelein et al., 1994; Denyer et al., 1996). GBSSI is present in endosperm, whereas GBSSII is present in leaves and stems.
Mutational analysis of the roles of SSI, SSIIa, and SSIIIa suggests that while each of these enzyme classes is primarily involved in amylopectin synthesis, each enzyme class is preferentially involved in the extension of specific subsets of chain lengths within the amylopectin molecule. Recent studies showed that SSI is involved in synthesis of the shorter outer chains of the amylopectin [degree of polymerizaton (DP)8–12] in leaf starch of Arabidopsis (Delvalle et al., 2005) and in the endosperm starch of rice (Fujita et al., 2006). Starch from barley and wheat SSIIa mutants showed an increase in chains of DP3–8, indicating that the SSIIa enzyme plays a role in extending shorter glucan chains of DP3–8 to longer glucan chains of DP12–35 (Yamamori et al., 2000; Morell et al., 2003; Konik-Rose et al., 2007). Loss of SSIIIa in maize and rice confers an increased amylose phenotype, with a reduction in the proportion of very long chains in amylopectin (>DP50 in maize or >DP30 in rice), and slightly reduced gelatinization temperature (Jane et al., 1999; Fujita et al., 2007). Arabidopsis mutants defective for SSIV appear to have fewer, larger starch granules within the plastid, and a role in priming starch granule formation has been postulated for the SSIV protein (Roldan et al., 2007).
The barley amo1 (high amylose Glacier, HAG) mutant was identified on the basis of an increased amylose phenotype, with an amylose content of up to 45% (Banks et al., 1971). The amo1 mutation locus was mapped at 56.5cM on chromosome 1H (by Andris Kleinhofs, Barley-BinMap 2005, GrainGene database). Although a number of groups have attempted to define the mutation locus, the nature of the amo1 locus remains to be defined (Boren et al., 2008). The only gene involved in the starch synthesis pathway located on chromosome 1H is ssIIIa (Li et al., 2000), and unpublished data (Z Li et al.) map this gene close to the amo1 locus, suggesting that ssIIIa is a candidate gene underpinning the amo1 mutation. The barley sex6 mutant is also a high amylose phenotype mutant, coupled with reduced starch content and reduced grain weight due to a reduction in starch biosynthesis (Morell et al., 2003). This phenotype has been shown to result from the loss of the function of SSIIa enzyme in the endosperm (Morell et al., 2003) which is encoded by the ssIIa gene on chromosome 7 of barley. The sex6 mutant barley produces starch with more short chains (DP6–11) and fewer intermediate chains (DP12–30) compared with the wild-type line. The high amylose content phenotype was due to the preferential reduction of the synthesis of amylopectin compared with that of amylose (Clarke et al., 2008). Foods produced from the sex6 mutant barley give a low glycaemic index and contain high levels of resistant starch (Topping et al., 2003).
In this study, the characterization of a barley population generated between the sex6 mutant (Himalaya292) and the amo1 mutant (HAG) in order to evaluate interactions between these recessive mutations in starch biosynthesis is described. The study resulted in the surprising observation that the combination of the recessive mutations results in a partial recovery of starch content and grain weight in the amo1sex6 double mutant, but without a statistically significant further increase in amylose content. The functions of the amo1 locus as a negative regulator of starch synthesis at elevated amylose levels are discussed.
Barley lines used were from a population generated through three backcrosses between Himalaya292 (male, containing the sex6 mutation allele) (Morell et al., 2003) and HAG (female, containing the amo1 mutation allele) (Banks et al., 1971), followed by three generations of single seed descent. Grains from the third backcross were denoted as BC3F1 and from the third single seed descent were named as BC3F4. To increase the quantity of grains for each line, two further generations were grown (designated BC3F6 generations), which were used for this study.
In total, 71 BC3F6 barley lines were produced. These barley lines together with the parent and control barley lines, HAG, Glacier, Himalaya292, and Himalaya, were grown in the glasshouse at CSIRO Plant Industry, Canberra in pots in 2005 (with natural light and temperatures of 18°C during the night and 24°C during the day).
For genomic DNA sequencing, genomic DNA from Himalaya, Himalaya292, Glacier, and HAG were used for the PCR amplification of fragments using three pairs of primers which were from the cDNA and genomic DNA sequence of the wheat ssIIIa gene (Li et al., 2000; GenBank accession nos AF258608 and AF258609). Three pairs of primers were ZLSSIIIa-P1F (5′-ATGGAGATGTCTCTCTGGCCA-3′, located at nucleotide 29 of wheat ssIIIa cDNA), and ZLSSIIIa-P1R (5′-TCTGCATACCACCAATCGCCGT-3′, located at nucleotide 3806 of wheat ssIIIa genomic DNA); ZLSSIIIa-P2F (5′-ATCGTGACCTAACAGCTTTGGCG-3′, located at nucleotide 3189 of wheat ssIIIa genomic DNA) and ZLSSIIIa-P2R (5′-GACAGAAGAACCCAAATCTGCGGTC-3′ located at nucleotide 7189 of wheat ssIIIa genomic DNA); and ZLSSIIIa-P3F (5′-GGAGGTCTCGGGGATGTTGTTAC-3′, located at nucleotide 6038 of wheat ssIIIa genomic DNA) and ZLSSIIIa-P3R (5′-CCACAAATGTAAATATCATTGATGTAT-3′, located at nucleotide 9524 of wheat ssIIIa genomic DNA).
For cDNA sequencing, total RNA was extracted from the developing endosperm [15 days post-anthesis (DPA) was used due to the comparative high level of expression of starch synthase genes at this stage] of barley, Himalaya. The procedures for RNA extraction were as detailed in Clarke and Rahman (2005). First-strand cDNA was synthesized and used for PCR amplification of ssIIIa cDNAs. Primers used for amplification of the full-length of cDNA sequence were ZLSSIIIa-P1F and ZLSSIIIa-P4R (5′-ACGTCACTGCGGTTCTTATCTCG-3′, located at nucleotide 9403, after the stop codon of the genomic DNA sequence of wheat ssIIIa).
For each PCR (20μl), 50ng of cDNA or genomic DNA, 1.5mM MgCl2, 0.125mM of each dNTP, 10pmol of primers, 0.5M glycine betaine, 1μl of dimethylsulphoxide (DMSO), and 1.5U of Advantage 2 Taq polymerase mix (Clontech) were used. The PCRs were conducted using a HYBAID PCR Express (Integrated Sciences) with one cycle of 95°C for 5min, 35 cycles of 94°C for 45s, 59°C for 30s, and 72°C for 3min, one cycle of 72°C for 10min, and one cycle of 25°C for 1min. The PCR fragments (1μl) were cloned into a pCR2.1 TOPO cloning vector (Invitrogen). DNA sequencing was performed using the automated ABI system with dye terminators as described by the manufacturers at JCMRS, Australian National University, Australia.
DNA sequences were analysed using the GCG suite of programs (Devereaux et al., 1984) to detect single nucleotide polymorphisms (SNPs) in genomic DNAs of the ssIIIa gene from Himalaya292, Himalaya, Glacier, and HAG. A CAPS (cleaved amplified polymorphic sequence) marker was designed based on one SNP at nucleotide 6323 of barley ssIIIa genomic DNA between two parental lines which created an EcoRI site in the ssIIIa gene from HAG, but not from Himalaya292. Primer SSIIIa-P5F (5′-GGAGGTCTCGGGGATGT-3′) located at nucleotide 7442bp and primer SSIIIa-P5R (5′-GGTTCCAGGAAGTAAACGGTCAGG-3′) located at nucleotide 7893 of the barley ssIIIa genomic DNA (GenBank accession numbers: Himalaya genomic DNA, JN256944; Himalaya292 genomic DNA, JN256945, Glacier genomic DNA, JN256946; HAG genomic DNA, JN256947) were used for the PCR amplification of the CAPS marker for the ssIIIa gene, generating a 464bp product which was then digested with EcoRI.
Young barley leaves from the BC3F6 population were collected and freeze-dried (freezer FTS systems, Stone Ridge, New York). Genomic DNA was isolated with a fast DNAR kit (Q-BIOgene, CA, USA).
For genotyping for the presence or absence of the ssIIa mutation from the sex6 mutant in Himalaya292, primer SSIIaF (5′-CCTGGAACACTTCAGACTGTACG-3′) starting at nucleotide 1616 and primer SSIIaR (5′-AGCATCACCAGCTGCACGTCCT-3′) starting at nucleotide 2044 of the ssIIa cDNA (GenBank accession no. AY133249) were used for the PCR amplification of a 451bp product spanning the ssIIa mutation site (nucleotide 1829 of Himalaya292 as described; Morell et al., 2003).
The PCR products for the detection of the ssIIa mutation (for the sex6 locus) were digested with the restriction enzyme NlaIV at 37°C overnight. Digested PCR fragments were separated on 2% agarose gels and visualized with gel documentary (UVitec) after GelRed (Biotium) staining. Three types of PCR fragment patterns were evident after electrophoresis on a 2% agarose gel that differentiated the mutated and wild-type ssIIa gene. Two DNA fragments (347bp and 104bp) indicated the occurrence of the mutated ssIIa gene from Himalaya292, three DNA fragments (236, 111, and 104bp) indicated the presence of the wild-type ssIIa gene from HAG, and both 347bp and 236bp fragments (in addition to 111bp and 104bp fragments) indicated heterozygous genotype lines. The 104bp DNA fragment was not used as it is not specific to the mutation. The 111bp fragment could not be separated from the 104bp fragment.
The amo1 mutation locus was mapped at 56.5cM on chromosome 1H (by Andris Kleinhofs, Barley-BinMap 2005, GrainGene database). To genotype for the presence or absence of the region containing the amo1 mutation, 12 SSR (simple sequence repeat) markers (Ramsay et al., 2000) located between 56.00cM and 64.60cM and one EcoRI CAPS marker for the ssIIIa gene (this work), on the short arm of chromosome 1H, were selected for the amplification of PCR products from the two parental lines, Himalaya292 and HAG. These SSR markers were EBmac0405, Bmag0105, Bmac0063, HVM20, Bmac0090, EBmac0560a, EBmac0501, Bmac0044, Bmac0032, Bmag0113, Bmag0211, and Bmag0350. The primers for these SSR markers were synthesized according to the sequences listed in the GrainGenes Database.
The PCRs were assembled as above except for a changes of primers. PCR for detection of the ssIIa mutation and ssIIIa gene was carried out as above using a HYBAID PCR Express (Integrated Sciences) except using GoTaq Hot Start Polymerase (Promega) instead of Advantage 2 Taq polymerase. The PCR conditions for the 12 SSR markers was one cycle of 95°C for 4min, 15 cycles of 94°C for 30s, from 65°C to 50°C with a 1°C decrease in each cycle for 30s, and 72°C for 1min 20s, 30 cycles of 94°C for 15s, 50°C for 15s, and 72°C for 45s, and one cycle of 25°C for 1min.
The PCR products for the ssIIIa gene were digested with the restriction enzyme EcoRI at 37°C overnight. Digested PCR fragments were separated on 2% agarose gels and visualized with gel documentary (UVitec) after GelRed (Biotium) staining. Two types of PCR fragment patterns were evident after 2% agarose gel electrophoresis that differentiated the ssIIIa gene from HAG or Himalaya292. A 464bp DNA fragment only indicated the occurrence of the ssIIIa gene from Himalaya292, while both 303bp and 161bp DNA fragments indicated the presence of the ssIIIa gene from HAG. Three fragments (464, 303, and 161bp) were detected in the heterozygous lines.
For the amo1 mutation, PCR products from the 12 SSR markers were separated on 2% agarose gels and also on an 3130×l Genetic Analyser (following the instructions of Applied Biosystems). Two out of the 12 SSR markers, EBmac0501 and Bmac0090, gave clearly different sized PCR fragments. On 2% agarose gels, the EBmac0501 marker gave three PCR fragmentation patterns for the BC3F6 population. A 167bp fragment was detected from HAG, a 141bp fragment was detected from Himalaya292, and both 167bp and 141bp (EBmac0501) fragments were detected in the heterozygous lines. Using the 3130×l Genetic Analyser, the Bmac0090 microsatellite marker gave three PCR fragmentation patterns for the BC3F6 population. A 234bp fragment was detected from HAG, a 236bp fragment was detected from Himalaya292, and both 236bp and 234bp fragments were detected in the heterozygous lines.
For real-time PCR analysis, total RNA was extracted from the developing endosperm (at 15 DPA) of barley from five lines of each of the four genotypes. The procedures for RNA extraction and the first-strand cDNA synthesis were as described above. Primers used for amplification of ssIIIa cDNA between nucleotides 3649 and 3814 (Himalaya cDNA, JN256948; Himalaya292 cDNA, JN256949, Glacier cDNA, JN256950; HAG cDNA, JN256951) were ZLSSIIIa-RTF (5′-TGGACAAGTCGAAAACCTGACCG-3′) and ZLSSIIIa-RTR (5′-GGCAATGTATTATATGTGGAGAAAGTCC-3′). Primers used for amplification of ssIIIb cDNA between nucleotides 434 and 590 (GenBank accession no. FN179378) were ZLSSIIIb-RTF (5′-GGAAAGGTTGAAGGCATCTCTG-3′) and ZLSSIIIb-RTR (5′-TGATATCTGGAGAAGAGCCACTC-3′). Primers for the amplification of a housekeeping gene, α-tubulin 2, cDNA between nucleotides 658 and 831 (GenBank accession no. Y08490) were ZLbTUB2F (5′-AGTGTCCTGTCCACCCACTC-3′) and ZLbTUB (5′-CAAACCTCAGGGAAGCAGTCA-3′).
For each PCR (20μl), 100ng of cDNA, 3.5mM MgCl2, 0.2mM of each dNTP, 5pmol of primers, 1×SYBR Green I (Invitrogen), 1× PCR buffer (–MgCl2), and 0.25U of Platinum Taq DNA polymerase (Invitrogen) were used. The PCRs were conducted and analysed using a RotorGene 6000 (Corbett Life Science, Australia). The real-time PCR conditions were one cycle of 95°C for 10min, 45 cycles of 95°C for 20s, 58°C for 20s, and 72°C for 30s, and melting from 72°C to 95°C rising by 1°C for each step. Comparative quantitation analysis was used to calculate the comparative expression. The comparative expression for each genotype was calculated by mean values of the comparative concentration for mRNAs for ssIIIa and ssIIIb individually divided by the mean value of the comparative concentration for mRNAs for α-tubulin 2.
The analysis of the starch and grain composition of the BC3F6 population was conducted using all homozygous lines for both ssIIa and three markers for the amo1 locus lacking recombinations at the amo1 locus. These lines included all wild-type lines, lines containing the amo1 locus alone, lines containing the sex6 locus alone, and lines containing both the amo1 and sex6 mutations. For the analysis of water-soluble carbohydrates (WSCs), starch chain length distribution, and grain morphology a subset of four lines per genotype was used.
Barley grains were first ground to wholemeal using a Cyclone mill machine (Cyclote 1093, Tecator, Sweden). Starch content was assayed using the AACC method 76.13 with 20mg of wholemeal for each of three replicate samples (Konik-Rose et al., 2007). Starch was isolated by a protease extraction method (Morrison et al., 1984) followed by washing with water and removal of the tailings. Starch was then freeze-dried. Amylose content was measured using a small-scale (2mg of starch) iodine adsorption method (Morrison and Laignelet, 1983) with some modifications as described by Konik-Rose et al. (2007).
Starch chain length distribution was measured by the method of O'Shea et al. (1998) using a P/ACE 5510 capillary electrophoresis system (Beckman Coulter, NSW Australia) with argon laser-induced fluorescence (LIF) detection.
Grain weights for each of the 71 BC3F6 lines were determined as the average grain weight of 100 grains with three replicates. Grain plumpness was genotyped as three categories: shrivelled, semi-plump, and plump.
Transverse sections (~1mm thick) of the middle part (the largest diameter section) of the barley grains were produced from four genotypes of lines (wild-type lines, amo1 mutants, sex6 mutants, and amo1sex6 double mutants) by cutting with razor blades. The sections and intact grains were then photographed with a Leica MZFLIII (Leica Microsystems, Germany) at ×12.5 amplification.
Protein content was determined by measurement of nitrogen content using a mass spectrometry method employing a Europa 20-20 (electronics rebuilt by Sercon in 2006) isotope ratio mass spectrometer with an automated nitrogen and carbon analyser preparation system. From 3mg to 8mg of wholemeal of barley was used. A nitrogen to protein conversion factor of 6.25 was used for the calculation of the protein content in barley grains (Mosse, 1990). Lipid content was measured using the AOAC 983.23 method and AACC method 08-01 (AOA Chemists, 1990), respectively.
β-Glucan was measured using the AACC method 32.23 using 20mg of wholemeal for each of three replicate samples. Pentosan content was measured using the method from Bell (1985) employing 20mg of wholemeal for each of three replicate samples.
WSCs were extracted from wholemeal following the method of Lunn and Hatch (1995) with some modifications. Barley wholemeal (100mg) was extracted three times with 10ml of 80% ethanol (v/v) in a boiling water bath for 10min. The supernatants from each extraction were pooled, freeze-dried, and resuspended in 2ml of milliQ water. The quantities of sucrose, glucose, fructose, maltose, and fructo-oligosaccharides were analysed by high-performance anion exchange chromatography (HPAEC) (Ruuska et al., 2006).
Starch granule-bound proteins were isolated and separated by SDS–PAGE gel as described (Rahman et al., 1995). The proteins were then stained by silver staining (Li et al., 1999). The protein gels were scanned (Epson Perfection 2450 PHOTO; Epson America Inc., CA, USA).
Developing endosperms at 15 DPA were isolated and ground in a mortar and a pestle with 3 vols of extraction buffer [20mM TRIS-HCl, pH 7.5, 5mM dithiothreitol (DTT), and 1mM pefabloc SC (Roche)] at 4°C. The homogenate was then centrifuged at 10000g for 20min at 4°C and the supernatant (containing 20μg of proteins) was used for analysis of SSIIIa activity by zymogram (Abel et al., 1996). A two-dimensional gel system was applied with an 8% acrylamide native gel as the first dimension gel and a zymogram gel (Abel et al., 1996) as the second dimension gel.
Statistical analyses were performed using Genstat version 9. Analysis of variance was performed for grain weight, starch content, amylose content, amylopectin content, protein content, lipid content, β-glucan content, pentosan content, and WSC content to obtain the least significant difference (LSD, P <0.05), looking at variation between the genotypes.
A population of lines segregating for the presence or absence of mutations at the sex6 and amo1 loci was generated by performing three backcrosses from a sex6 mutant donor line (Himalaya292) into an amo1 mutant line (HAG). Three generations of single seed descent were performed from the BC3F2 lines in order to generate sufficient fixed genotypes to investigate the relative impact of the sex6 and amo1 mutation loci (alone and in combination) on starch synthesis, grain composition, and morphology.
Central to this study is the ability to assign the progeny lines accurately to genotypes. Because the genetic change underpinning the sex6 phenotype has previously been demonstrated to be a lesion in the ssIIa gene, the status of all lines at the sex6 locus could be unambiguously defined (Supplementary Table S1 available at JXB online). However, the amo1 mutation is defined by phenotype alone given that the causal gene has yet to be defined. Two lines of evidence (genotype grouping and phenotype grouping) were used to assign lines to the four possible genotype classes (wild type, sex6, amo1, and amo1sex6 double mutant).
Given that the causal gene at the amo1 locus has not been identified, identification of the closest linked markers available was sought. SSR markers identified from barley mapping populations within ~10cM of the amo1 locus were examined for polymorphism in this population, and two markers (EBmac0501 located at 58.0cM, and Bmac0090 located at 58.0cM) gave clear polymorphisms between the Himalaya292 (sex6) and HAG (amo1) parents. As the ssIIIa gene is located in this region of chromosome 1H (Z Li et al., unpublished data), an ssIIIa marker based on an EcoRI restriction polymorphism at nucleotide 6323 of barley ssIIIa between the Himalaya292 and HAG ssIIIa genes was also developed (Supplementary Table S2 and Fig. S1 at JXB online). Of the 71 lines from the backcross population, 13 lines had a wild-type phenotype at both the sex6 and amo1 loci, 13 carried the sex6 mutation alone, nine had the amo1 mutation alone (containing three DNA markers for the amo1 locus), and there were 13 lines with both the sex6 and amo1 mutations (Supplementary Table S1). In each of these lines, the three markers at the amo1 locus (EBmac0501, Bmac0090, and the ssIIIa marker) showed no recombination between with wild type and amo1 alleles. A further five lines were identified that were homozygous at the sex6 locus, and were homozygous for the three amo1 markers but contained recombination between these three markers (Supplementary Table S1). The remaining 18 lines were heterozygous at one of the four markers used and were excluded from the phenotypic analysis.
Figure 1 shows the relationship between amylose content and starch content for the 53 homozygous BC3F6 lines. For lines containing the wild-type ssIIa gene (triangular symbols) a clear separation into two phenotypic groupings could be made. The grouping with elevated amylose consistently contained the three markers from the amo1 locus (filled triangles) while the group with lower amylose content were wild type for all three amo1 markers (open triangles). The lines with the sex6 genotype (non-functional SSIIa lines) also revealed the presence of two phenotypic groupings, separated in this case not by amylose content but by starch content. The lower starch content group contained the sex6 allele and the wild-type amo1 locus (open diamonds), while the higher starch content group contained both the sex6 and amo1 loci (filled diamonds). Five lines with recombinations between the amo1 markers (three with wild-type alleles at sex6, filled triangles indicated by arrows; two with the mutant sex6 allele, filled diamonds indicated by arrows) were included in the phenotypic analysis. Alignment of the phenotypic and genotypic data for the five recombinant lines provided evidence demonstrating that of the three markers, the ssIIIa SNP marker is more tightly linked to the amo1 locus than either of the EBmac 0501 or Bmac0090 markers (see Fig. 1; Supplementary Table S1 at JXB online). Additional studies with larger populations are being used to examine whether further recombinants can be identified and to quantify the linkage between markers in the amo1 region and the trait (see later section in this study).
Grain weight data for the four genotype groups are given in Fig. 2A. There were no statistically significant differences between grain weights of the amo1 mutant lines and wild-type lines (P <0.05). However, there were statistically significant differences (P <0.05) between each of the sex6 mutants and amo1sex6 double mutant lines and each of the respective three remaining genotypes.
Intact grains from four representative lines for the four genotypes were examined by stereoscopic microscopy on both the dorsal and crease sides (Fig. 3). Consistent with the grain weight data, the wild-type (Fig. 3A) and the amo1 mutant (Fig. 3B) lines produce plump well-filled grains, while sex6 mutant lines produce shrunken grains (Fig, 3C). The amo1sex6 double mutants (Fig,3D) yield grains with an intermediate phenotype, plumper than the shrunken grains of sex6 mutants yet not as well filled as the grains of the amo1 mutant and the wild-type line.
To illustrate the plumpness of the grains from these genotypes further, transverse sections at the widest mid section of the grain are shown in Fig. 3. The wild-type (Fig. 3E) and amo1 genotypes (Fig. 3F) yield fully filled endosperms, while the sex6 mutant line (Fig. 3G) produces incompletely filled grains with a considerable reduction in endosperm packing density. Amo1sex6 double mutant lines (Fig. 3H) showed an intermediate phenotype, with larger grain sections containing an endosperm that is more fully filled than the sex6 mutant and yet less well filled than that of the wild-type or amo1 mutant lines.
Starch content for the four genotypes is shown in Fig. 2B. Compared with the wild-type lines, amo1 mutants, amo1sex6 double mutants, and sex6 mutant lines contained 10.0, 23.7, and 45.1% less starch, respectively. These values were statistically different among the four genotypes (P <0.05).
Consistent with the data in Fig. 1, the sex6 mutant lines and amo1sex6 double mutants contained significantly higher amylose contents than the amo1 mutant and wild-type lines (Fig. 2B). However, the amylose contents of sex6 mutant lines and amo1sex6 double mutants were not significantly different (P <0.05). The amylose content from amo1 mutant lines was also statistically significantly higher than that from the wild type (P <0.05).
To examine the effects of genotype on starch chain length distribution, starch was isolated from four lines from each genotype of the BC3F6 cross population, debranched with isoamylase, and analysed by fluorophore-assisted carbohydrate electrophoresis (FACE). Figure 4 shows a difference plot in which the normalized chain length distribution (expressed as a percentage of molar molecules) for wild-type lines is subtracted from that for normalized amo1 mutants, sex6 mutants, and amo1sex6 double mutants. Lines containing the mutant sex6 allele (sex6 mutants and amo1sex6 double mutants) had more short chains (DP6–14) and fewer intermediate chains (DP15–24) compared with lines containing the wild-type sex6 allele (amo1 mutants and wild-type lines) (Fig. 4). The amo1 mutants had fewer short chains (DP9–14) and more intermediate chains (DP15–24) than the wild-type lines.
Previous analysis of the sex6 mutation in Himalaya292 showed that the mutation was associated with pleiotropic effects on grain composition (Clarke et al., 2008) considered to result from redirection of carbohydrate in the grain as a result of alterations in total starch synthesis (Morell et al., 2003). Therefore, an analysis of the major carbohydrate, protein, and lipid fractions of the grain for each of the four genotype groups was conducted.
The protein content of the four genotypes is shown in Fig. 5. The protein content of the amo1sex6 lines was significantly higher than the protein content of each of the remaining three genotypes, while the sex6 mutant lines contained significantly more protein than both wild-type lines and amo1 mutant lines (P <0.05). There was no statistically significant difference (P <0.05) between amo1 mutants and wild-type lines.
Previous studies showed that the lipid content of sex6 lines was elevated (Clarke et al., 2008). Lipid content data shown in Fig. 5 indicate that each of the four genotypes has a statistically significant difference in lipid content, with the sex6 genotype having the highest lipid content, followed by the amo1sex6, amo1, and wild-type genotypes, respectively.
β-Glucan and pentosan contents of the four barley genotypes are shown in Fig. 5. Interestingly, these studies suggest that the amo1 and sex6 genotypes have differing impacts on cell wall polysaccharide classes. While the amo1 and sex6 genotype had higher β-glucan content than the other two genotypes, sex6 had the highest increase in pentosan levels. The amo1sex6 genotype had intermediate levels of both β-glucan and pentosans.
Compared with the WSC composition of the grains of wild-type lines, amo1 mutant lines did not contain significantly different levels of total WSC, free glucose, free fructose, or maltose, but did contain significantly different levels of sucrose and fructan in their grains. However, both sex6 mutant lines and amo1sex6 double mutant lines contained significantly greater amounts of each of these carbohydrates than wild-type or amo1 lines (but not for sucrose) (P <0.05) (Fig. 6). Compared with amo1sex6 double mutant lines, sex6 mutants contained significantly more glucose, fructose, sucrose, fructan, and total WSCs, but did not have statistically significant differences in levels of maltose compared with the amo1sex6 double mutant lines (Fig. 6).
Expressing grain composition data on a percentage basis can be misleading when the major grain constituent, starch, differs significantly in content between genotypes, leading to apparent increases in grain constituents that do not reflect underlying synthesis rates. In order to examine the absolute levels of synthesis of the various grain components in each of the genotypes, Table 1 presents the composition data on a per caryopsis basis. This analysis confirms that, as expected, modification of the starch synthesis level is the major driver of grain weight differences between these genotypes. However, analysis of the data on this basis indicates that there have been major changes in starch synthesis among the genotypes. Each of the mutant genotypes has a decrease in amylopectin synthesis, with the sex6 genotype having a severe suppression of amylopectin synthesis. In contrast, the amo1 genotype (amo1 mutants and amo1sex6 double mutants) has a significant increase in amylose synthesis while the sex6 genotype (sex6 mutants) has a small decrease. The data, however, demonstrate that in the amo1sex6 genotype, the major driver of a restoration of starch content relative to the sex6 genotype is concomitant increases in both amylose (79% increase) and amylopectin content (61% increase).
Analysis of the levels of other grain constituents on a per caryopsis basis suggests that the protein content may be increased in the amo1sex6 genotype relative to other genotypes, lipid content is specifically increased in both the sex6 and amo1sex6 genotypes, β-glucan and pentosan synthesis levels remain unaltered in all genotypes with the potential exception of β-glucan synthesis in the amo1 genotype, and WSC levels are inversely proportional to starch synthesis.
To investigate the mechanism underpinning the increase in amylose content on a per caryopsis basis, the level of GBSSI protein in the starch granules in barley grains was analysed from the selected lines from each of the four genotypes as GBSSI is the major enzyme involved in amylose synthesis in cereal grains. Figure 7 shows that starches from amo1 genotypes contain markedly higher levels of GBSSI, SSI, and SSIIa/SBEIIa/SBEIIb proteins compared with the wild type. As reported previously (Morell et al., 2003), starches from sex6 mutants contain significantly less GBSSI protein than the other three genotypes as well as no detectable SSI, or SSIIa/SBEIIa/SBEIIb proteins. Starches from amo1sex6 double mutants also contain more GBSSI protein than sex6 mutants, but have a similar amount of GBSSI protein to wild-type lines. However, starches from amo1sex6 double mutants do not contain any detectable traces of SSI, and SSIIa/SBEIIa/SBEIIb proteins.
The observation that the ssIIIa gene encodes the only known starch synthetic enzyme closely linked with the amo1 locus suggests ssIIIa as a potential candidate gene whose mutation is causal for the amo1 phenotype. The presence of SSIIIa activity was examined by zymogram (enzymatic activity) from selected lines from each of the four genotypes. The results showed that there was no consistent change in SSIIIa activity (slow migration activity band as described by Wang et al., 1993; Fujita et al., 2007) among lines from the four genotypes (Fig 8A). Real-time reverse transcription-PCR (RT-PCR) data showed that the comparative expression for both ssIIIa and ssIIIb was similar among the four genotypes. The comparative expression for each genotype was calculated by mean values of the comparative concentration for mRNAs for ssIIIa and ssIIIb individually divided by the mean value of the comparative concentration for mRNAs for α-tubulin 2. The comparative expression levels for ssIIIa were between 0.15 and 0.20, while for ssIIIb they were between 0.018 and 0.039 (Supplementary Fig. S4 at JXB online). While RT-PCR suggests that both ssIIIa and ssIIIb transcripts are expressed in barley endosperm, it has not been possible to date to differentiate between SSIIIa and SSIIIb activity in zymogram analysis. However, two-dimensional affinity electrophoresis has allowed four starch synthase activities with a molecular weight consistent with SSIII to be separated (Fig. 8B). To date, it has also not been possible to assign these activities unambiguously to specific genes, and it is possible that multiple spots derive from a common gene through post-translational modifications. However, all four activities were detected from HAG (amo1 mutant), Glacier (a wild-type parent line for HAG), and Himalaya, providing further evidence strengthening the conclusion that the observed amo1 phenotype is not due to the loss of SSIIIa activity because SSIIIa activity is present in the amo1 mutant (Fig. 8B).
Three out of the five amo1sex6 double mutants tested showed a higher level of SSI activity compared with the amo1 and sex6 mutants. Wild-type lines had very low SSI activity (Fig. 8A). A higher level of SSI activity was also observed in the amo1 mutant compared with wild-type lines in two-dimensional gels (Fig. 8B). Variation in SSI and α-amylase activities was observed in the zymogram gels. The variation of α-amylase activities may be due to the different genetic backgrounds of the lines or from slight variations in the developmental stage of the endosperm material harvested.
In order to investigate whether other polymorphisms (besides the previously described SNP marker at nucleotide 6323) in the ssIIIa gene could underpin the amo1 phenotype, the cDNA from a wild-type barley Himalaya, and genomic DNA sequences from wild-type barley (Himalaya, a barley cultivar used for mutagenesis for Himalaya292), sex6 mutant (Himalaya292), amo1 mutant (HAG), and wild-type barley (Glacier, a barley cultivar used for mutagenesis for HAG) were PCR cloned and sequenced. The ssIIIa genomic DNAs from Himalaya, Himalaya292, HAG, and Glacier lines contained 9550bp sequences with 16 exons and 15 introns (Supplementary Table S2 at JXB online, GenBank accession numbers: Himalaya genomic DNA, JN256944; Himalaya292 genomic DNA, JN256945, Glacier genomic DNA, JN256946; HAG genomic DNA, JN256947). The ssIIIa cDNA sequence from Himalaya was 5088bp long, encoding a polypeptide with 1590 amino acid residues from nucleotide 1 to 4770 (GenBank accession number Himalaya genomic DNA, JN256948; Himalaya292 genomic DNA, JN256949, Glacier genomic DNA, JN256950; HAG genomic DNA, JN256951). Comparison between genomic DNA sequences from Himalaya292 (and Himalaya) and HAG (and Glacier) showed that there were eight SNP variations for HAG (and seven SNPs for Glacier) (Supplementary Fig. S1 and Table S2 at JXB online). Among them, four SNP variations for HAG (and three SNPs for Glacier) were in exons (Supplementary Fig. S2 and Table S2). Three SNPs are in exon 3 (for HAG and Glacier) and one is in exon 14 of HAG only (Supplementary Table S2). Three SNP variations at nucleotide 1084 (for HAG and Glacier), 1676 (for HAG and Glacier), and 4439 (for HAG only) of the cDNA sequences produce a change in the amino acid sequence of the protein. The first change is conservative, changing the hydrophobic amino acid methionine (Himalaya and Himalaya292) to another hydrophobic amino acid valine (HAG and Glacier) at position 362 of the protein, and the second is a conservative change from the non-polar amino acid alanine (Himalaya and Himalaya292) to a hydrophobic amino acid valine (HAG and Glacier) at position 559 of the protein, whereas the third is a non-conservative change from the hydrophobic amino acid leucine (Himalaya, Himalaya292, and Glacier) to the basic amino acid arginine (HAG) at position 1480 of the protein (Supplementaty Fig. S3).
In the five lines that showed that the linkage between the amo1 locus and the EBmac0501 or Bmac0090 marker was broken, all showed a linkage association between the ssIIIa marker and amo1 locus (Supplementary Table S1), indicating that the ssIIIa gene is the closest marker to the amo1 locus. An attempt was therefore made to identify lines containing a recombination between the ssIIIa marker used and the gene underpinning the amo1 phenotype, by genotyping progeny from each of the 18 heterozygous but sex6 mutant lines which were not used for the genotypic grouping analysis above. Grains harvested from those plants were morphologically phenotyped and all 190 lines analysed showed no recombination between the ssIIIa gene and the amo1 locus, confirming that the ssIIIa gene is very tightly linked to the casual gene for the amo1 phonotype. Further studies are required to define whether the ssIIIa gene is the basis of the amo1 mutation through down-regulation of the expression of the ssIIIa gene and complementation of the amo1 mutation by overexpression of a wild-type ssIIIa gene.
The initial objective of this study was to examine the impact on amylose content of combining recessive mutations at the sex6 and amo1 loci. Each of these mutants has increased endosperm starch amylose content, and combining these mutants might be expected to produce a higher amylose content and a concomitant decrease in starch content. However, combining the amo1 mutation with the sex6 mutation significantly restores starch synthesis in the endosperm of sex6 mutants through parallel increases in both amylose and amylopectin content.
The study provides further insights into the effects of the sex6 and amo1 mutations on starch synthesis. The effect of the sex6 mutation was to decrease amylopectin synthesis predominantly (75% reduction on a per caryopsis basis) while amylose synthesis is decreased by 28% (Morell et al., 2003). Given that amylose synthesis requires the granular matrix to be present (Wattebled et al., 2002), the impact of the loss of SSIIa through the sex6 mutation on amylose synthesis may be a secondary consequence of the major decrease in amylopectin synthesis in this mutant. It is evident that the high amylose content of the starch of sex6 mutants is a consequence of the differential inhibition of amylopectin synthesis. In contrast, the data presented in Table 1 suggest that the amo1 mutation promotes a shift from amylopectin synthesis to amylose synthesis, resulting in the elevated amylose phenotype of the mutant. When combined with the sex6 mutant, the impact of amo1 is to increase the synthesis of both amylose and amylopectin, resulting in the significant restoration of starch synthesis and therefore grain weight in the amo1sex6 mutant, but no further increase in amylose content on a percentage basis is observed (Fig. 2). However, the amylose content on a per seed basis is significantly increased in the amo1sex6 mutant compared with the sex6 mutant alone (Table 1). Therefore, the amo1 mutation in both amo1 mutants and amo1sex6 mutants promotes the synthesis of more amylose than that from the wild-type lines. As GBSSI is essential for the biosynthesis of amylose in the endosperm (Nelson and Rines, 1962; Murata et al., 1965; Eriksson, 1969; Delrue et al., 1992; Nakamura et al., 1995), and also contributes to the synthesis of the long chains of amylopectin (Maddelein et al., 1994; Denyer et al., 1996), the amount of GBSSI in the starch granules in the endosperm was examined. This showed that the levels of expressed GBSSI protein in the starch granules from both amo1 mutants and amo1sex6 mutants were significantly increased compared with the wild-type lines and the sex6 mutant, respectively.
Several studies have indicated that ssIII is a negative regulator of starch synthesis and that a mutation in this gene can increase amylose levels. In Arabidopsis, studies investigating the role of the ssIII gene in transient starch synthesis led Zhang et al. (2005) to conclude that ssIII was a negative regulator of starch synthesis in leaves. In Chlamydomonas, mutations in the ssIIIa gene increased the amount of GBSSI protein and the level of the transcripts of GBSSI, resulting in the synthesis of more long chains in amylopectin (Ral et al., 2006). In rice, two ssIII genes are known, ssIIIa and ssIIIb. Of these genes, ssIIIa is expressed in the endosperm during starch synthesis whereas ssIIIb is expressed early in endosperm development but sharply reduces during periods of highly active starch synthesis later in grain filling (Ohdan et al., 2005). Mutations in the ssIIIa gene in rice did not change the endosperm starch content; however, there was an increase in amylose content (from 15% to 20%) suggesting that a decrease in amylopectin synthesis and a concomitant increase in amylose synthesis and the amount of GBSSI had occurred (Fujita et al., 2007). In maize the dull 1 mutant which disrupts the function of the ssIII gene also has an increased amylose content (Wang et al., 1993).
The similarities in phenotype between the ssIIIa and amo1 mutants, and the present observation that the ssIIIa gene maps in the same region of the genome as the amo1 locus in barley, suggested that ssIIIa is a candidate gene underpinning the amo1 mutation. This conclusion is strengthened by further mapping and sequencing data presented here. First, it is shown that the ssIIIa gene is very tightly linked to the amo1 locus as the ssIIIa gene is the closest linked marker to the amo1 locus and no recombinant between this gene and amo1 was obtained in 190 progeny lines. Secondly, sequencing of the entire ssIIIa gene shows that there is a difference in the amino acid sequence of the SSIIIa protein in the amo1 mutant compared with the reference Himalaya (or parent Glacier) gene, and this substitution (from leucine to arginine) is at conserved motif 7 within the catalytic domain of ssIIIa genes (Li et al., 2000). The activity of SSIIIa as assayed by zymogram is, however, comparable in amo1 and wild-type endosperm, illustrating that the total loss of SSIIIa activity does not underpin the amo1 phenotype.
The role of SSIIa in the elongation of amylopectin chains of DP15–24 has been previously demonstrated in barley (Morell et al., 2003) and extensively reviewed elsewhere for other species (Umemoto et al., 2004; Yamamori et al., 2004; Zhang et al., 2004; Konik-Rose et al., 2007). The results presented in this study suggest that the amo1 mutation has statistically significant effects in the chain length range DP9–14 (and DP15–24). The chain length distributions of the sex6 and amo1sex6 amylopectins did not show statistically significant differences across the entire distribution.
Previous studies showed that pleiotropic effects on grain composition were observed in sex6 mutants and that this has ramifications for understanding the impact of manipulation of starch synthesis on grain functionality and end-use, including human nutrition outcomes (Morell et al., 2003; Topping et al., 2003; Clarke et al., 2008). Such changes were also observed for the amo1 mutants and amo1sex6 double mutants. The analyses of other grain components showed that lines containing the amo1 mutant allele in a wild-type background had a significantly higher (P <0.05) β-glucan content on a per grain basis (Table 1) than all wild-type lines. Consistent with data presented in Clarke et al. (2008), the sex6 genotypes analysed in this study had significantly higher (P <0.05) lipid content and WSC content, and lower protein content on a per grain basis than the wild-type lines. With the exception of protein content (which was highest in amo1sex6 lines), the amo1sex6 double mutants had levels of β-glucan, fibre, lipids and WSC lower than those of sex6 alone but inconsistent changes compared with amo1 mutants and the wild type (Figs 5, ,6).6). Data presented in Table 1 suggest that only in the case of lipid and WSC are the changes in the sex6 and amo1sex6 genotypes associated with an increase in net synthesis per caryopsis rather than being a consequence of the dilution effect of major changes in starch content per caryopsis.
The analysis of mutations affecting starch biosynthesis in a wide range of systems (including Chlamydomonas, rice, maize, pea, barley, and Arabidopsis) has been highly informative in defining the key genes involved in the synthetic process and understanding their respective roles. However, it has frequently been noted that when combinations of mutations are generated, the effects of the mutations are not additive, suggesting that interactions between genes and gene products are important. One explanation for non-additive effects is that there is strong evidence now to demonstrate that complexes of starch biosynthetic enzymes exist and are functionally important, involving multiple isoamylases (Zeeman et al., 1998; Kubo et al., 1999) in one class of complex and branching enzymes and starch synthases in a second class of complexes (Hennen-Bierwagen et al., 2008; Hennen-Bierwagen et al., 2009; Tetlow et al., 2008). In barley, loss of the SSIIa protein through mutation at the sex6 locus results in concomitant loss of SSI, SBEIIa, and SBEIIb protein from the starch granule (Morell et al., 2003), possibly because of the disruption of a complex between these proteins (Hennen-Bierwagen et al., 2009). Furthermore, it has been noted that mutations in one starch biosynthetic gene lead to the modulation of the expression levels of other genes in the starch biosynthetic pathway (Rahman et al., 1995; Hylton et al., 1996; Yamamori et al., 2000; Morell et al., 2003; Boren et al., 2004; Fujita et al., 2007; Kosar-Hashemi et al., 2007). The amo1 mutation also increases the SSI protein level in the starch granules of the amo1 mutants. In maize, the dull 1 mutant also shows an increase in SSI activity as a result of a deficiency in SSIII (Cao et al., 1999). Further work needs to be carried out to resolve whether the increased content in starch granules of GBSSI (in amo1 and amo1sex6 genotypes), and SSI, SSIIa/SBEIIa/SBEIIb (in amo1 genotypes) is mediated at the gene expression, protein translation, or protein complex formation level, or are effects resulting from altered partitioning of the expressed proteins between the soluble and granular fractions.
In summary, this work demonstrated that the mutation of the amo1 locus up-regulates the expression of SSI, SSIIa, SBEIIa, and SBEIIb in the starch granules in barley endosperm. Genetic mapping studies indicate that the ssIIIa gene is tightly linked to the amo1 locus, and the SSIIIa protein from the amo1 mutant has a leucine to arginine residue substitution in a conserved domain. Zymogram analysis indicates that the amo1 phenotype is not a consequence of total loss of enzymatic activity, although it remains possible that the amo1 phenotype is underpinned by a more subtle change. It is therefore proposed that amo1 may be a negative regulator of other genes of starch synthesis in barley.
Supplementary data are available at JXB online.
Figure S1. Single nucleotide polymorphisms (SNPs) of barley ssIIIa genomic DNAs from HAG, Glacier, Himalaya292, and Himalaya.
Figure S2Single nucleotide polymorphisms (SNPs) of cDNA sequences of the barley ssIIIa gene from HAG, Glacier, Himalaya292, and Himalaya.
Figure S3. Changes in polypeptide sequences of barley SSIIIa from HAG, Glacier, Himalaya292, and Himalaya.
Figure S4. Real-time reverse transcription-PCR analysis of the expression of mRNAs for barley ssIIIa and ssIIIb of four genotypes of barley lines from the BC3F6 backcrossing population Himalaya292×HAG.
Table S1. Genotypes and phenotypes of BC3F6 lines of barley.
Table S2. Intron and exon structure of the barley ssIIIa gene and SNPs among barley ssIIIa genes from HAG, Glacier, Himalaya292, and Himalaya.
The authors wish to acknowledge Russell Heywood and Mark Cmiel for crossing and growing of barley, Emma Anschaw and Steve McMaugh for protein gels, Lynette Rampling for support of the analysis of the SSR marker, David Lewis for the analysis of water-soluble carbohydrates, Alex Zwart, Colin Cavanagh, Marcus Newberry, and Ahmed Regina for their help with statistical analysis, Tony Bird for lipid analysis, Mark Talbot and Rosemary White for assistance with microscopy, Dick Phillips for the assay of protein content, Hunter Laidlaw for suggestions on pentosan assay, and Sadequr Rahman and Jean-Philippe Ral for critical reading of the manuscript. ACVL Ltd is acknowledged for financial support.