In silico expression analysis of SSIIIa and SSIIIb genes
Genomic DNA sequences for SSIIIa and SSIIIb homoeologues in the elite bread wheat cultivar Cadenza were identified using the Grassroot Genomics BLAST Search (https://grassroots.tools/service/blast-blastn) to find homologous sequence to the SSIII genomic DNA sequence previously isolated by Li, et al. 8 (GenBank accession no. AF258609). Using the Wheat Expression Browser (http://www.wheat-expression.com/) in silico expression profiling of SSIIIa and SSIIIb paralogs and homoeologues was performed to investigate expression patterns for distinct plant tissues and for different stages during grain development. A heat-map comparing expression of SSIIIa and SSIIIb indicated that expression of SSIIIa is mainly limited to the grain, whereas SSIIIb is expressed in the roots, leaves/shoots, and spike as well as the grain (Figure S1). SSIIIa transcript is highest around 10 and 15 days post anthesis, which is consistent with the mid-development peak of SSIII transcript observed in RNA isolated from developing wheat endosperms by Li, et al. 8. Comparisons of SSIIIa and SSIIIb transcript profiles suggest this peak was almost entirely due to SSIIIa; SSIIIb has high expression very early in development (~2 days post anthesis) but lower expression throughout the rest of development. Given these results we supposed that SSIIIa is the predominant form expressed in the grain and corresponds to the cDNA and genomic DNA sequences previously isolated by Li, et al. 8. In light of these data we focused on developing mutants for SSIIIa homoeologues which we presumed would have the greatest impact on starch properties in the grain.
TILLING for ssIIIa genes
Mutations causing premature stop codons in the A, B and D genome copies of SSIIIa (ssIIIa-A, ssIIIa-B, ssIIIa-D) (Table 1 and Figure S2) were identified using the in-silico Wheat TILLING resource for the cultivar Cadenza (http://www.wheat-tilling.com/). To predict the effect of the mutations on protein function we used the reference SSIII amino acid sequence (1628 residues) from a hexaploid wheat SSIII cDNA clone (GenBank accession AF258608), and previously defined protein domain regions described by Li, et al. 8. Each of our selected mutations was predicted to result in a protein truncated before the catalytic domain or within it, rendering the mutant protein inactive.
Table 1. Selected mutations for SSIIIa homoeologues.
|
Gene
|
Line
|
Codon
|
Protein coordinate and predicted effect
|
Chrom
|
Exon
|
Protein domain
|
SSIIIa-A
|
C2074
|
Cag → Tag
|
Q 531 (stop codon)
|
1A
|
3
|
Variable repeat region
|
SSIIIa-B
|
C0905
|
Caa → Taa
|
Q 1306 (stop codon)
|
1B
|
10
|
Catalytic domain
|
SSIIIa-D
|
C0291
|
Cga → Tga
|
R 435 (stop codon)
|
1D
|
3
|
Variable repeat region
|
Generation of ssIIIa mutants
The TILLING mutants for each homoeologue of ssIIIa (ssIIIa-A, ssIIIa-B, ssIIIa-D) were crossed to create a line lacking active SSIIIa protein. Plants heterozygous for mutations in all three ssIIIa homoeologues were self-fertilized to produce segregating F2 populations. Using KASP markers (Table S1), 724 segregating F2 plants were screened for homozygous single, double and triple mutants (single: ssIIIa-A, ssIIIa-B, ssIIIa-D; double: ssIIIa-AB, ssIIIa-AD, ssIIIa-BD; triple: ssIIIa-ABD). The number of plants selected for each genotype class is included in Table S2. At least eight plants were selected for each genotype class, except for the double mutant ssIIIa-AD for which only five plants were identified. Despite fewer ssIIIa-AD plants, a Chi-square test showed no evidence of segregation distortion (P = 0.8034). In addition to mutant genotypes, siblings without mutations in SSIIIa were selected to provide an ideal control; these lines are expected to have the same level of segregating background mutations left over from the TILLING process as the ssIIIa mutants. Genome-specific sequencing primers were designed to validate the selected mutations (Table S3) and all triple mutant and sibling control plants were sequenced to confirm presence or absence of each mutant allele (ssIIIa-A, ssIIIa-B, and ssIIIa-D).
Immunoblot analysis of SSIIIa
To determine the effect of ssIIIa mutations on SSIIIa protein, immunoblot analysis was performed using maize polyclonal antiserum for SSIIIa (anti-DU1N)23 (Figure 1). The anti-DU1N antiserum recognized multiple proteins in extracts of sibling control grains. However, all the bands except two bands at ~250 kDa and one at ~70 kDa were also present on blots of the triple mutants, thus the common bands were presumed to result from non-specific binding of the antiserum and/or recognition of other wheat starch synthase isoforms. The predicted molecular weights of the mature SSIIIa proteins corresponding to the A, B and D homoeologues were calculated as 183 kDa, 181 kDa and 181 kDa respectively, which is consistent with the 183 kDa molecular weight predicted by Li, et al. 8. Wheat, barley, maize and rice SSIII proteins were previously reported to have anomalously low electrophoretic mobilities; it has been suggested previously that this could be due to intrinsic properties or post-translational modifications8,19,21,23. For example, the SSIIIa protein of maize is predicted to encode a 188-kDa protein; however, in SDS-PAGE of grain extracts the protein migrated with an apparent mass in excess of 200 kDa23. We conclude that the bands at 250 kDa are highly likely to be SSIIIa. Their absence in the triple mutant is consistent with loss of all three homoeologous proteins.
Grain characteristics
Grain physical characteristics
At maturity, most of the triple mutant grains were shrunken and dull in colour compared to plump and yellow sibling control grains (Figure 2). Further examination by dissection showed vitreous endosperms in triple mutant grains and white mealy endosperms in sibling control grains (Figure 2). The grains of single and double mutants were not noticeably different from sibling control grains.
The average thousand grain weight (TGW) of triple mutants was 11% lower than that of sibling controls (37.7 g vs. 42.3 g for sibling controls, P = 0.004); the average width of triple mutant grains was also 5% lower than that of the sibling control (P < 0.001) and the area was 5% smaller (P = 0.04) (Tables 2 and S4). The grains of single and double mutants were not significantly different from the sibling control with respect to these physical characteristics.
Hardness and germination potential
Given the vitreous nature of triple mutant grains, we compared the hardness index and germination potential of these grains with those of the sibling controls. Triple mutant grains were 32% harder than sibling control grains, 79.0 (SD = 4.2) vs. 59.6 (SD = 9.7) for controls, (P = 0.003, Table S4). The hardness of a sample is reported in arbitrary units and values above 45 are termed hard; typical bread wheat ranges from 50 – 8028. The vitreous nature of the triple mutant grains had little effect on their germination potential: the germination index was only 6% lower than that of sibling controls (P = 0.041, Table S4).
Grain protein content
There was no evidence for an effect of genotype on mature grain protein content (estimated by measuring nitrogen released by combustion) (e.g., triple mutant = 11.27% (SD = 2.04%), compared to 9.91% (SD = 0.75%) in the sibling control; ANOVA across all genotypes P = 0.400) (Table S4).
Starch properties
Starch content
Wholemeal flour prepared from triple mutant grains had significantly less starch than the sibling controls (56.1 g/100 g flour compared to 68.1 g/100 g flour in sibling controls; P < 0.001) (Table 2, Figure S4a). The flour starch contents of the double mutants were also significantly lower than the sibling control (Table 2, Figure S4a). Although for the single mutants, there were no significant differences compared to the sibling control, the values were lower suggesting a possible dose-response effect of number of mutations on starch content (Table 2, Figure S4a). The flour starch content was used to calculate the starch content on a per grain basis using thousand grain weight (TGW) values. Starch contents per grain for the triple mutants were significantly lower than for the sibling controls (20.3 mg/grain compared to 28.7 mg/grain in sibling controls; P < 0.001), as were the ssIIIa-BD double mutants (23.3 mg/grain; P = 0.003 compared to sibling controls) (Table 2). Similarly, the starch contents per grain for each of the single and the other double mutants were lower albeit not individually statistically significantly different from sibling controls (Table 2).
Table 2. TGW, protein and starch properties of mature grains of ssIIIa mutants and sibling control.
|
|
TGW
|
Protein
|
Starch
|
Amylose
|
Resistant starch
|
|
g
|
%
|
g / 100 g wholemeal flour
|
mg /grain
|
%
(Iodine method)
|
%
(SEC method)
|
g / 100 g wholemeal flour
|
g / 100 g starch
|
sibling control
|
42.3 (2.8)
|
9.1 (0.5)
|
68.1 (5.6)
|
28.7 (1.1)
|
26.0 (3.2)
|
25.5 (1.7)
|
2.86 (1.15)
|
0.69 (0.11)
|
ssIIIa-A
|
41.3 (3.0)
|
9.4 (0.9)
|
65.8 (4.7)
|
27.0 (3.2)
|
26.7 (1.8)
|
25.3 (1.4)
|
2.40 (0.72)
|
0.72 (0.16)
|
ssIIIa-B
|
43.5 (3.4)
|
8.0 (1.6)
|
64.4 (2.5)
|
27.9 (3.0)
|
27.3 (1.3)
|
27.0 (2.4)
|
2.59 (0.89)
|
0.80 (0.13)
|
ssIIIa-D
|
40.8 (4.2)
|
8.1 (2.1)
|
64.4 (2.1)
|
27.2 (2.6)
|
25.5 (2.8)
|
27.5 (3.0)
|
2.59 (1.28)
|
0.47 (0.09)
|
ssIIIa-AB
|
41.8 (2.0)
|
8.6 (0.5)
|
63.7 (2.5)*
|
26.6 (2.4)
|
28.9 (4.7)
|
29.2 (1.2)*
|
3.76 (0.50)
|
0.87 (0.20)
|
ssIIIa-AD
|
43.7 (4.5)
|
9.2 (2.0)
|
61.4 (3.0)*
|
26.9 (3.9)
|
26.4 (3.1)
|
28.9 (1.0)*
|
3.75 (0.50)
|
0.78 (0.27)
|
ssIIIa-BD
|
39.4 (3.5)
|
8.9 (1.7)
|
59 (2.9)*
|
23.3 (2.3)*
|
26.3 (2.4)
|
29.9 (3.0)*
|
3.98 (1.65)
|
0.69 (0.06)
|
ssIIIa-ABD
|
37.7 (3.0)*
|
10.3 (2.1)
|
56.1 (2.7)*
|
20.3 (2.2)*
|
35.1 (0.56)*
|
33.8 (4.0)***
|
6.01 (0.61) ***
|
1.70 (0.59)***
|
Amylose and resistant starch
The amylose content of purified starch from triple mutant grains measured by iodine binding was higher than that of the sibling controls (35.1% vs. 26.0% in sibling controls; P < 0.001), but the amylose contents of starch from single and double mutants were not different from that of the sibling controls (Table 2, Figure S4b).
The resistant starch content of the triple mutant wholemeal flour (6.01 g/100 g flour) was more than twice that of the sibling control (2.86 g/100 g flour); P < 0.001) (Table 2, Figure S4c). Resistant starch contents were also measured in purified starch samples and similarly the triple mutants (1.70 g/100 g starch) had nearly 2.5-fold more than the sibling controls (0.69 g/100 g starch; P < 0.001) (Table 2).
Starch granule morphology and size distribution
Light microscopy was used to examine isolated starch and grain sections of the ssIIIa mutants and sibling controls. In the triple mutants many of the A-type granules were smaller and more elongated than those of sibling controls, and many granules had protrusions (Figure 3). These characteristics were also observed using scanning electron microscopy of isolated starch (Figure 3). The birefringence pattern of many of the triple mutant A-type granules in polarized light showed distorted cross patterns distinct from the typical ‘Maltese cross’ pattern observed in the sibling control; this was more apparent in granules that had an irregular shape and protrusions. The birefringence pattern of the B-type granules was not notably different (Figure 3).
To quantify starch granule size and size distribution a Coulter counter (Beckman Multisizer 4e) was used to measure particle volumes in purified starch. All of the starch samples showed a bimodal size distribution as expected for wheat29, and these were estimated as mixtures of two Gaussian distributions individually for each replicate. There was a clear shift towards smaller A-granule diameters (between 10 and 20 μm) in the triple mutant compared to the sibling control (Figure 4).
The average size of the A-type granules was significantly smaller in the triple mutant (13.9 µm, SD = 0.75; P < 0.001) and BD double mutant (15.6 µm, SD = 0.71; P = 0.019) compared to sibling controls (16.7 µm, SD = 0.28) (Table S4). There was no significant difference in the proportion of the larger A-type granules across genotypes, or in the diameter of the B-type granules (Table S4).
Starch chain-length distribution
Both double and triple ssIIIa mutants had different chain-length distributions (CLD) compared to the sibling control (measured on debranched starch). Analysis of CLD data indicated fewer long amylopectin chains and more amylose chains in the double and triple mutants compared to the sibling controls. The CLDs showed three distinct peaks, the first around a degree of polymerization (dp) of 10-12 corresponding to short amylopectin chains within a single crystalline lamella of starch, the second between 37 and 100 dp corresponding to longer chains that span multiple lamellae as defined by Hanashiro, et al. 30, and the third around 2500 dp corresponding to long amylose chains 30 (Figure 5).
The relative heights of the first and third peak, and the proportion of areas under the curve in regions corresponding to < 37 dp, 37 – 100 dp and > 100 dp were calculated for starch from individual plants (biological replicates, Table S5) for each genotype (wild-type sibling control, single: ssIIIa-A, ssIIIa-B, ssIIIa-D; double: ssIIIa-AB, ssIIIa-AD, ssIIIa-BD; triple: ssIIIa-ABD). The relative heights of the peaks were significantly different between genotypes (1-way ANOVA, P < 0.001). The proportion of area > 100 dp (corresponding to proportion amylose by mass) was higher in the double (mean = 29.4%) and triple mutants (mean = 33.8%) than the sibling control (mean = 25.5%; P-values for differences = 0.003 and <0.001, respectively). Within chains <100 dp; the proportion of the area corresponding to longer chains (37 to 100 dp) was lower in the double (30.0%) and triple mutants (28.3%) than the sibling control (35.5%; P < 0.001 for both comparisons), indicating a reduction in longer amylopectin chains compared to shorter amylopectin chains in the double and triple mutants (Table S4).
Physicochemical properties
The elevated levels of resistant starch in the triple mutant suggest it may be useful for wheat breeding applications, thus, we examined whether the physicochemical properties of starch from the triple mutant were altered relative to those of the controls. The gelatinization behaviour of purified starch was measured using differential scanning calorimetry (DSC). The peak and conclusion temperatures of gelatinization for the triple mutants were significantly greater than those of the sibling controls (for peak: 61.8°C vs. 58.4°C in sibling controls (P < 0.001); for conclusion: 75.4°C vs. 70.5°C in sibling controls (P = 0.001)) (Table S4). There were no significant differences in enthalpy and onset temperature between the triple mutants and sibling controls (Table S4).
Non-starch Polysaccharides
To determine pleiotropic effects of ssIIIa mutations on non-starch polysaccharides in the grain, enzyme fingerprinting was used to measure relative levels of arabinoxylan oligosaccharides (AXOS) and mixed-linked β-glucan oligosaccharides in wholemeal flour samples following enzymatic digestion. Total AXOS (sum of peak areas of all known AXOS peaks) was significantly higher in the triple mutants (35% more, P < 0.001) and in BD double mutants (17% more P = 0.001) relative to the wild-type sibling controls (Table 3), with some evidence for a dose-response effect among single mutants and the AB and AD double mutants. Both the triple and BD-double mutants also had greater levels of substituted and unsubstituted AXOS compared to the wild-type siblings. Interestingly, the other double mutants (AB and AD) and D-genome single mutants showed greater levels of substituted AXOS , reflected in the small change in total AXOS (Table 3). Only the AD-double mutants had a greater ratio of substituted: unsubstituted AXOS compared to the wild-type sibling controls (0.67 for AD, and 0.56 for sibling controls, P = 0.008) (Table S4).
We also observed higher levels of mixed-linkage β-glucans (MLGs) in the triple mutant (40% more, P < 0.001), AB and BD double mutants (both 17% more, P = 0.016 and P = 0.017 respectively), and D-genome single mutant (19% more, P = 0.009), with some evidence for an increase in the B and AD mutants (Table 3). There were no differences in the G3 to G4 ratio for each genotype compared to the wild-type sib suggesting no changes in the structure of MLGs (Table S4).
Table 3. Arabinoxylan oligosaccharides (AXOS) and mixed-linkage β-glucans (MLGs) in wholemeal flour of ssIIIa mutants and sibling control.
|
|
Relative peak values
|
|
Total AXOS
|
Substituted AXOS
|
Unsubstituted AXOS
|
MLG (G3 + G4)
|
sibling control
|
32.7 (3.6)
|
11.6 (0.4)
|
21.0 (3.3)
|
9.1 (1.2)
|
ssIIIa-A
|
32.8 (2.4)
|
11.7 (0.3)
|
21.0 (2.6)
|
9.3 (1.1)
|
ssIIIa-B
|
34.1 (2.3)
|
12.0 (0.9)
|
22.1 (1.7)
|
9.7 (0.6)
|
ssIIIa-D
|
34.0 (1.9)
|
12.9 (0.4)*
|
21.1 (1.7)
|
10.8 (0.4)*
|
ssIIIa-AB
|
35.3 (2.7)
|
13.2 (1.3)**
|
22.1 (1.5)
|
10.6 (0.6)*
|
ssIIIa-AD
|
35.1 (2.9)
|
14.0 (1.1)***
|
21.2 (2.3)
|
10.0 (1.4)
|
ssIIIa-BD
|
38.2 (3.2)**
|
14.3 (0.7)***
|
24.0 (2.6)*
|
10.6 (0.6)*
|
ssIIIa-ABD
|
44.1 (1.9)***
|
16.7 (0.8)***
|
27.4 (1.6)***
|
12.7 (1.5) ***
|