This paper describes the use of TILLING to identify mutant alleles of SBEIIa in both durum and bread wheat varieties and conventional breeding to determine their combined effect on wheat starch composition. The union of these mutant alleles led to the reduction of both SBEIIa RNA expression and protein levels resulting in novel non-transgenic wheat lines with high levels of amylose and resistant starch content.
TILLING is a targeted mutation breeding technology that combines efficient mutagenesis with functional genomics to accelerate crop improvement [43
]. TILLING is especially useful in plants with polyploid genomes where direct selection for phenotypes can be hindered by functional redundancy of genes, and also because polyploid species have a higher tolerance for mutation density [34
]. TILLING is most readily applied to traits that can be affected by the down-regulation of a gene (e.g., by knocking out a critical enzyme in a biosynthetic pathway). However, traits can also be stacked to confer more complex characteristics, and some traits can even be up-regulated using TILLING by targeting genes that code for inhibitory factors. In particular, the application of TILLING in wheat provides access to an abundant source of new alleles in each of the A, B and D genomes within the same cultivar for the development of new traits.
Expression of SBEIIa in the bread wheat mutant line was reduced relative to control lines in both developing endosperm and leaves as measured by qPCR. The mutations in this line caused premature stop codons in the coding regions of all three SBEIIa genes. These premature stop codons, or nonsense mutations, often lead to reduced RNA levels due to nonsense-mediated mRNA decay (NMD) of the transcript [45
]. NMD is a surveillance pathway triggering degradation of mRNAs containing premature stop codons that functions in many organisms including plants [46
]. In wheat, reduced RNA levels have been reported for multiple genes containing premature stop codon mutations including a HMW glutenin subunit [48
], a waxy gene [49
], and a polyphenol oxidase gene [50
In addition to stop mutations, splice junction mutations are often considered to be severe mutations and are expected to have a deleterious effect on gene expression due to the production of alternate splice products that often lead to the formation of premature stop codons. We combined an SBEIIa splice junction B genome mutation with a stop mutation in the A genome to develop a high amylose durum wheat line. We found that the SBEIIa message produced from the gene containing the splice junction mutation was alternately spliced in multiple distinct ways. The most common form of alternative splicing identified in the mutant was due to exon skipping, but intron retention and alternative donor splice selection as well as alternative acceptor splice selection were also detected. All but one splice variant led to the formation of premature stop codons in the resulting transcript, but one variant was predicted to encode a protein with a single amino acid change (see Additional file 1
). This result indicates that the splice junction mutation could still potentially produce some protein and could account for the difference in amylose content of the durum wheat line compared to the bread wheat mutation line. For example, two mutant alleles in the amylose extender (ae
) gene in maize (SBEIIb) have recently been reported with very different effects on amylose content even though both alleles are non-functional [51
]. One mutant line, ae1.1
, completely lacks SBEIIb protein and produces starch that is 65.6% amylose. The second mutant line, ae1.2
, produces a non-functional protein missing a 28 amino acid region and contains starch with lower levels of amylose of 49.3%. Even though both alleles encode non-functional SBEIIb genes, the presence of the mutant ae1.2
protein, which was found to be associated with starch granules, influenced the amylose content in a different way than the complete lack of SBEIIb protein. Our results also suggest that some splice junction mutations may prove useful when reduced, but not complete, loss of gene expression is desired.
As part of the mutation library development for TILLING, mutations are induced throughout the genome and these additional background mutations are carried along in crosses that combine mutations in the target gene of interest. Background mutations may cause undesirable phenotypes such as the mutation causing stunted growth described in this report. Such undesirable phenotypes can often be excluded from being caused by the target mutation using segregation analysis. For example, if a phenotype is found arising in both mutant and wild-type segregants of the target gene it is more likely to be caused by a background mutation. Phenotypes caused by background mutations can also be dissociated from a target mutation by evaluating different combinations of mutant alleles in the target genes [52
]. In practice, background mutation loads are reduced by repeated backcrosses of plants with mutations of interest to the parent line or by marker assisted backcross selection to a different variety. In this report, both durum and bread wheat mutation lines and their wild-type siblings had reduced seed weights compared to their parental un-mutagenized lines indicating that the background mutation load is likely having an effect on this phenotype. In addition, the high amylose mutant lines had reduced total starch levels compared to the parental and sibling control lines, but it is unclear how much the background mutation load is contributing to this result. Future evaluation of backcrossed lines with a reduced background mutation load will help clarify the effect of SBEIIa mutations on both starch content and seed weight.
The proportions of amylopectin and amylose in starch have an effect on starch granule morphology. Starch is packaged in granules as compact insoluble structures, and the clustered branch points of amylopectin allow its glucan chains to form ordered arrays of double-helixes that can pack together. Due to this ordered structure, amylopectin forms semicrystalline arrays that have a characteristic birefringence pattern under polarizing light. Reduced birefringence pattern in response to polarized light reflects the disrupted starch crystallinity expected for starch with increased ratio of amylose to amylopectin. For example, the increased amylose content in a SSIIa null mutant combination line caused a substantial decrease in birefringence of wheat starch granules [21
]. Starch granule morphology is altered and birefringence reduced in high amylose lines of wheat and maize [28
]. We found that the high amylose SBEIIa TILLING mutant lines formed starch granules with reduced birefringence and had a high proportion of crescent shaped granules similar to those reported for high amylose hp-SBEIIa RNAi lines in wheat [28
]. The altered shape of starch granules is likely due to the increased proportion of amylose in the granule.
The combination of the TILLING mutation alleles led to the reduction of both SBEIIa RNA expression and protein levels resulting in bread wheat lines with high levels of amylose and resistant starch content. Reduction of SBEIIa RNA by combinations of mutation alleles did not affect the RNA expression levels of other starch biosynthetic enzymes SSI, GBSSI or SBEIIb in the developing endosperm or SSI in the leaves. However, differences between proteins identified in control and mutant samples were apparent. We determined that SBEIIa was reduced by greater than 90% of the level of wild-type protein in starch granules from SBEIIa bread wheat mutation lines based on limit of detection analysis and mass spectrometry. SBEIIb protein was identified in all samples at similar levels indicating that SBEIIb can remain stable even with drastically reduced levels of SBEIIa in wheat. This was a different result than that reported in lines lacking SBEIIa due to an RNAi construct in bread wheat [28
]. In the bread wheat RNAi line with the hp-SBEIIa construct, loss of SBEIIa RNA and protein resulted in additional loss of SBEIIb protein, suggesting that the hp-SBEIIa construct might be affecting expression of both genes. However, SBEIIb RNA was still detectable in the hp-SBEIIa RNAi line. The authors suggested that protein stability of SBEIIb might be affected in the absence of SBEIIa or that there may be an effect of the hp-SBEIIa construct on the translation of SBEIIb RNA to account for the additional loss of SBEIIb protein [28
In the bread wheat TILLING mutant lines, relative quantification of proteins identified by mass spectrometry was performed using spectral counting. In this technique, the number of identified MS/MS spectra from the same protein is compared over multiple datasets. Spectral counts have been shown to strongly correlate with relative protein abundance [54
]. Relative quantification indicated that reduced levels of SBEIIa led to increased representation of SBEI proteins. This result is interesting in light of increasing evidence of functional interactions between starch biosynthetic enzymes and their formation into high molecular weight protein complexes in maize and wheat [55
]. In developing wheat endosperm, SBEIIa and SBEIIb have each been found to be associated in protein complexes with SSI and SSIIa, but not with each other [56
]. Consistent with this result, SBEIIa and SBEIIb have been isolated as homodimers but not as heterodimers despite the high level of sequence conservation between these proteins (74% identity at the protein level). Protein abundance and complex formation in the endosperm can be dramatically affected by the presence or absence of a protein. For example, the elimination of SBEIIb in maize (the major SBEII branching enzyme in maize endosperm) increased the abundance of SBEI, SBEIIa, SSIII, and SP in the starch granule, without affecting SSI or SSIIa [58
]. In the absence of SBEIIb, SSI and SSIIa are complexed with a different group of proteins consisting of SBEI, SP, and SBEIIa in maize [59
]. Our results suggest that a similar mechanism may occur in wheat since reduced levels of SBEIIa led to increased levels of SBEI protein without affecting SSIIa protein levels.
As a result of homozygous SBEIIa mutation combination in all three bread wheat genomes and both durum wheat genomes, the amylose content of the starch was significantly elevated 194-229% relative to controls. As a proportion of total starch, the bread wheat mutant lines had a higher amylose content of 55% compared to durum wheat mutant lines at 47% amylose. Evaluation of wheat lines containing only one wild-type gene and two mutated genes indicated that a single functional SBEIIa gene from any of the A, B or D genomes was sufficient to provide enough branching activity to yield starch with similar composition as wild-type plants having three functional genes. Although there were slight differences in amylose content of double mutant lines with amylose levels increasing from 22.9% in wild-type siblings up to 26.4% in double mutant lines (Figure ), these differences were not statistically significant. In contrast, a recent report in which bread wheat lines with single and double mutant combinations in SBEIIa were evaluated, a 5-6% increase in amylose content was found when two homoeologs were mutated (from 33.2% in wild-type lines up to 38.6-39.9% in double mutant lines) [60
]. Different wheat varieties and different amylose quantification methods were used between these two studies, and could account for these differences.
While we found a major effect on amylose content in bread wheat when TILLING lines with mutations in all three SBEIIa genes were evaluated, this level of amylose was not as high as the 74% amylose reported using RNAi suppression of SBEIIa [28
]. As mentioned previously, the wheat RNAi line with 74% amylose resulted in simultaneous loss of both SBEIIa and SBEIIb proteins. Loss of both these proteins may account for the higher level of amylose in the RNAi line compared to the TILLING mutant lines, which still have detectable SBEIIb protein. A similar phenomenon was observed in barley transgenic plants using the same hp-SBEIIa RNAi construct as in wheat [61
]. In barley, some RNAi lines had decreased SBEIIa protein only, while others had decreased levels of both SBEIIa and SBEIIb protein. In these two types of RNAi lines, those with only SBEIIa reduction had elevated amylose of 38% compared to wild-type levels of 28%, whereas barley hp-SBEIIa RNAi lines with a reduction of both SBEIIa and SBEIIb had very high amylose levels of 65% [61
Along with increased amylose content, resistant starch levels were elevated in the SBEIIa TILLING mutant lines consistent with previous reports that increased amylose content is correlated with higher levels of resistant starch. High amylose starch is more thermally stable than native starch, and amylose molecules have an increased tendency to aggregate and crystallize during retrogradation, which may make them more resistant to digestion [62
]. Resistant starch is classified into different types (RS1-RS5) depending on the basis of their resistance to digestion [1
]. RS1 is starch that is physically inaccessible to digestion such as that found in whole or partially milled grains and intact seeds. RS2 is starch in granular form such as in green bananas. Cooking of starch based foods leads to the formation of RS3 due to retrograded amylose. RS4 refers to chemically-modified starches, and RS5 refers to amylose-lipid complexed starch. We found that the bread wheat SBEIIa mutant line had the highest level of resistant starch at 5.4 or 11.2% of whole grain flour. The durum wheat mutant line containing the splice junction mutation had a lower resistant starch level of 4.7 or 6.2% in the flour consistent with the lower amylose levels in this mutant line. The RS value obtained was dependent on the lot of pancreatic alpha amylase used for the analysis of the whole grain samples. Control standard samples had RS values very similar to their expected values using both lots of PAA (Table ). The difference in RS value may indicate that some component of the whole grain sample is interfering with digestion of starch when using one lot of PAA because the RS values for controls, which were mostly purified starches, were not affected. This is also supported by the similar RS values obtained using either lot of PAA on purified starch granules from high amylose bread wheat (Table ). The analysis showed that SBEIIa mutant lines had significantly increased RS values of at least 5.4% for the high amylose bread wheat and 4.7% for high amylose durum wheat. These whole grain flour RS values likely reflect RS2 levels as they are based on evaluation of starch in granular form. Future experiments will investigate the starch structure and characteristics in these different high amylose lines and the contribution of these high amylose wheat flours to resistant starch formation in bread, pasta and other products made with them.