Tuning starch granule size distributions in durum wheat using genetic variation at a single locus

Abstract

Key message

Different missense mutations in TtMRC-A1 can be used to fine-tune granule size distributions in durum wheat grains, creating useful alterations in starch properties.

Abstract

The size distribution of starch granules in wheat grains influences bread- and pasta-making quality, as well as nutritional properties. Here, we demonstrate that in durum wheat, wide variation in starch granule size distributions can be induced through missense mutations at a single genetic locus encoding the MYOSIN RESEMBLING CHLOROPLAST PROTEIN on chromosome 6A (TtMRC-A1). We isolated 29 independent TILLING mutants in durum cultivar Kronos, each harbouring a different missense mutation that causes an amino acid substitution in the MRC protein. Compared to the B-type granule content of wild-type Kronos (24%), six of the missense lines had significant increases in B-type granule content (33–42%), although not to the extent observed in the mrc-1 mutant (58%) which carries a premature stop codon mutation. Notably, one missense line had significantly decreased B-type granule content (15%), demonstrating that mutations in TtMRC-A1 can achieve both increases and decreases in B-type granule content. In these lines, A-type granule size decreased as B-type granule content increased, and Rapid Visco Analysis on selected lines demonstrated that both B-type granule content and A-type granule size strongly correlated with pasting parameters (e.g. peak viscosity and pasting temperature). However, strong correlations between pasting properties and A-type granule size were still observed after removing most of the B-type granules via sieving, indicating that A-type granule size is the primary contributor to the observed variation in pasting properties. Overall, we demonstrate that mutations at TtMRC-A1 can greatly extend the range of granule size distributions in durum wheat, creating useful alterations in starch properties.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00122-025-05013-8.

Introduction

Starch granule size distributions influence both the functional and nutritional properties of starch (Li et al. 2021; Lindeboom et al. 2004; Qi and Tester 2016). In grains of wheat and other Triticeae crops, there is a bimodal distribution of starch granule size, consisting of large A-type and small B-type granules. A-type granules are lenticular in shape and are initiated during the early phases of grain development, whereas B-type granules are spherical and initiated about 10–15 days after the A-type granules (Buttrose 1960; Evers 1971).

The effect of starch granule size distributions on end-use quality has been studied using the variation in this trait among different wheat cultivars, or by creating more extreme variation through flour reconstitution after granule size fractionation. For breadmaking, there was an optimum amount of B-type granules (often referred to as B-type granule content) for crumb grain and texture, which depended on the protein content of the grain (Lelievre et al. 1987; Park et al. 2005, 2009). This suggests that starch granule size distributions affect the protein–starch interactions that are important for rheological properties and bread quality. In pasta-making experiments with reconstituted flours, increased B-type granule content enhanced pasta firmness (Soh et al. 2006). In addition to textural qualities, starch granule size affects nutritional properties. In general, larger starch granules are digested more slowly in vitro than smaller granules due to their reduced surface area per volume ratio (Dhital et al. 2010). Indeed, native A-type granules are digested more slowly in vitro than B-type granules (Colussi et al. 2021; Moon and Kweon 2024). The extent to which these trends translate to cooked starch is not known, but it is important to note that granules do not completely gelatinise when cooked in a food matrix. For example, cooked pasta contains both intact and incompletely gelatinised starch granules, particularly towards the internal regions of the noodle (Zou et al. 2015). Thus, increasing the amount of A-type granules in durum wheat, as well as their size, could boost the amount of resistant starch in pasta.

Genetics offers a means to modify granule size distributions in wheat. The range of B-type granule content in durum wheat was successfully increased through crossing with Triticum dicoccoides lines that have substantial variation in this trait (Sissons and Egan 2023). However, for the range of B-type granule content achieved (22–44%), there were no relationships between B-type granule content and pasta quality parameters, suggesting that a wider range of variation could be necessary for achieving the positive effects seen in flour reconstitution experiments. We reasoned that the recent identification of genes required for B-type granule synthesis may help achieve this wider range. For example, reduced activity of B GRANULE CONTENT1 (BGC1), or knockout of ⍺-glucan PHOSPHORYLASE (PHS1), can lead to wheat with reduced B-type granule content (Chia et al. 2020; Kamble et al. 2023; Saccomanno et al. 2022).

Another promising gene for altering B-type granule content encodes the MYOSIN RESEMBLING CHLOROPLAST PROTEIN (MRC). MRC (also referred to as PROTEIN INVOLVED IN STARCH INITIATION, PII1) is a long coiled-coil protein which promotes starch granule initiation in Arabidopsis leaves (Seung et al. 2018; Vandromme et al. 2019, 2023). In wheat, MRC represses B-type granule initiation during early grain development (Chen et al. 2024). Premature stop codon mutations in MRC in the durum wheat Kronos dramatically increased the B-type granule content, up to 68%. This increase resulted from the early initiation of B-type granules during grain development, which also led to a concomitant decrease in the average size of A-type granules. MRC is a particularly useful target in durum wheat because it has only one functional homeolog (TtMRC-A1 on chromosome 6A) due to the chromosome 6B homeolog becoming a pseudogene, allowing manipulation of B-granule content through mutations at a single locus. Here, we screened B-type granule content in 29 durum wheat lines, each carrying different missense mutations in the TtMRC-A1 locus. We identified lines with both increased and decreased B-type granule content, generating a remarkable range of B-granule content (15–58%) within the single isogenic background of cultivar Kronos. We demonstrate using the Rapid Visco Analyser (RVA) that the lines at the extremes of this variation have substantial changes in starch pasting properties. Overall, we demonstrate that mutations at a single genetic locus can be used to tune granule size distributions in durum wheat, offering ideal material for testing how granule size affects functional and nutritional quality.

Materials and methods

Plant materials and growth

Seeds of wheat TILLING lines were obtained from the John Innes Centre (JIC) Germplasm Resources Unit. Mutants from Chen et al. (2024) include mrc-1 (referred to as the BC2 aabb line in the original publication) and mrc-2. KASP markers used to genotype the mutations are provided in Supplemental Table 1. For experiments 1 and 2, plants were grown in a Controlled Environment Room fitted with fluorescent lamps supplemented with LED panels. The chamber was set to provide 16 h light (400 µmol photons m⁻² s⁻¹) at 20 °C and 8 h dark at 16 °C, with constant relative humidity (65%). For bulking grains for the RVA experiment, 96 plants per genotype were grown in 2 L pots in the glasshouse (4 plants per pot) until maturity and seed harvest. The glasshouse was set to provide a minimum 16 h of light at 20 °C and 16 °C during the dark. All plants were grown in John Innes cereal mix—65% peat, 25% loam, 10% grit, 3 kg/m³ dolomitic limestone, 1.3 kg/m³ pg mix, and 3 kg/m³ osmocote exact.

Table 1.

Kronos mrc mutants used in this study

Genotype	Mutation	SIFT score	PROVEAN score	Comments:
Missense mutants included in the study (29)
K3745	A19T	0.18	− 0.856	Exp. 1
K3543	G63S	0.7	0.926	Exp. 1
K1435	S92N	0.06	0.004	Exp. 1
K2447	L149F	0*	− 3.944*	Exp. 1
K2937	A178T	0.33	− 2.114	Exp. 1; Exp. 2
K2124	R196K	0.75	0.12	Exp. 1
K2205	E213K	0.15	− 2.437	Exp. 1
K3499	R230C	0.16	− 2.755*	Exp. 1
K1431	E225K	0*	− 3.911*	Exp. 1
K2397	R230C	0.04*	− 2.755*	Exp. 1
K4060	R255K	0.38	0.05	Exp. 1; Exp. 2
K0598	L289F	0*	− 3.961*	Exp. 1; Exp. 2
K3533	A346V	0.13	− 3.278*	Exp. 1
K2981	A356T	0.11	− 0.575	Exp. 1
K0775	L394F	0*	− 3.5*	Exp. 1
K0265	M412I	0.03*	− 2.153	Exp. 1; Exp. 2
K3424	E491K	0.01*	− 3.267*	Exp. 1
K2408	S503N	1	3.094	Exp. 1; Exp. 2
K1232	S539N	1	1.528	Exp. 1
K3889	D581N	0.13	− 2.011	Exp. 1; Exp. 2
K3505	E606K	0.22	− 1.96	Exp. 1; Exp. 2
K2485	A625T	0.01*	− 2.79*	Exp. 1; Exp. 2
K4297	P656L	0.13	− 1.238	Exp. 1
K3416	S667N	0.24	0.188	Exp. 1
K2348	E679K	0.02*	− 0.233	Exp. 1
K2096	P681S	0.01*	0.835	Exp. 1
K2821	A687V	0.13	− 3.182*	Exp. 1
K3666	D708N	0.02*	− 1.631	Exp. 1
K4193	S726N	0.46	− 0.47	Exp. 1
Missense mutants excluded in the study (16)
K4533	A79V	0.01*	− 3.263*	Insufficient seeds
K2450	A155T	0.12	− 0.42	Poor growth
K4618	R196K	0.75	0.12	Insufficient seeds
K0565	S329F	0.01*	− 3.214*	Unable to genotype
K0665	A356T	0.11	− 0.575	Insufficient seeds
K3286	S375F	0.01*	− 5.338*	Poor growth
K2976	L409F	0.07	− 0.445	Insufficient seeds
K1112	E413K	0.15	− 3.465*	Poor growth
K3145	E425K	0.04*	− 3.8*	Poor growth
K3657	T500I	0.07	− 1.68	Unable to genotype
K1161	T525I	0.1	− 3.878*	Poor growth
K3661	D575N	0.86	0.287	Poor growth
K4205	L591F	0.24	− 0.127	Poor growth
K3137	A603T	1	2.048	Unable to genotype
K0338	R700K	0.48	− 0.325	Insufficient seeds
K4082	D743N	0.52	− 1.675	Unable to genotype
Premature stop codon mutants included in the study (2)
mrc-1	W258STOP	–		Exp. 1; Exp. 2
mrc-2	W550STOP	–		Exp. 1

SIFT and PROVEAN scores that are considered damaging (using default cut-off values of 0.05 and − 2.5 for each programme, respectively) are marked with an asterisk (*). For lines included in the study, the experiment(s) in which phenotypic analyses were conducted is provided in the comments. For lines excluded, the reason for exclusion is provided in the comments

Small-scale starch extraction

Small scale starch extraction was used for the analyses in Experiments 1 and 2. Briefly, seeds (3 seeds per replicate) were cracked with a plier and soaked overnight in 0.5 M NaCl (0.75 mL) at 4 °C and then homogenised for 6 min in a MM-300 ball mill (Retsch) (set to 25 Hz) with a 5-mm steel ball. The homogenate was filtered through Miracloth, then through a pluriStrainer Mini with 70 µm nylon mesh (Pluriselect). Starch in the filtrate was pelleted at 18,000g, then resuspended in 300 µL ddH₂O and placed over a 1-mL cushion of 90% (v/v) buffered Percoll solution (prepared in 50 mM Tris–HCl, pH 8). The starch was spun through the Percoll at 2500g for 15 min. The starch pellet was then washed 3 times in 1 mL of 2% SDS, then once in 1 mL acetone, spinning at 18,000g for 5 min after each resuspension. The starch was air dried.

Large-scale starch extraction and fractionation

Large-scale starch extraction from bulked grains was used to produce starch for the RVA analyses. Grains (20 g per extraction) were soaked overnight in ddH₂O at 4 °C, before homogenising in a mortar and pestle. The homogenate was then washed with excess water through muslin, followed by Miracloth. The starch in the filtrate was collected by centrifugation at 3000g for 5 min and then resuspended in 8 mL ddH₂O before placing over a 36 mL cushion of 90% (v/v) buffered Percoll solution (prepared in 50 mM Tris–HCl, pH 8). The starch was spun through the Percoll at 2500g for 15 min. The starch pellet was washed 3 times in 20 mL of 2% SDS, then once in 20 mL acetone, spinning at 3000g for 5 min after each resuspension. The starch was air dried.

To remove B-type granules from the starch preparation, purified starch was suspended in 70% ethanol and then filtered through a pluriStrainer fitted with a 10 µm nylon mesh (Pluriselect). Filtration was assisted by placing the strainers over a magnetic stirrer, and adding a stir bar within each strainer above the nylon mesh to prevent starch from settling and blocking the filters. Once filtration had completed, more 70% ethanol was added to the strainers to wash remaining B-type granules through the nylon mesh. After filtration and washing were completed, starch enriched in A-type granules in the retentate was collected by placing the strainers into a bottle with 70% ethanol and shaking to resuspend the starch. The starch was then collected by pelleting at 3000 g for 5 min, washed once in acetone, and then air dried.

Total starch and amylose quantification and grain morphometrics

For total starch quantification, grains (5 grains per replicate) were crushed with pliers into a 2-mL centrifuge tube and then milled into flour using a MM-300 ball mill (Retsch) (set to 30 Hz) with a 5-mm steel ball. Quantification was carried out using the Total Starch HK Kit (Megazyme) according to the manufacturer’s instructions, except the starting material was adjusted to 5–10 mg of flour, and all assay volumes were reduced by 6X. Amylose was quantified as described in Fahy et al. (2022). Grain size was quantified using the MARViN seed analyser (Marvitech GmbH, Wittenburg).

Scanning electron microscopy (SEM)

Purified starch suspended in water was air-dried onto a glass coverslip attached onto an SEM stub. Stubs were sputter coated with gold and observed using a Gemini 300 FEG SEM microscope (Zeiss).

Coulter counter analysis

Purified starch granules were resuspended in Isoton II electrolyte (Beckman Coulter) and analysed on a Multisizer 4e Coulter counter (Beckman Coulter), fitted with a 70 µm aperture and 100-mL beaker. We measured a minimum of 50,000 particles per run. Granule size parameters were derived using curve fitting scripts, described in McNelly et al. (2024)

Rapid Visco Analysis

Rapid Visco Analysis (RVA) was carried out on an RVA Tecmaster instrument (Perten, Waltham, MA, USA) running the preinstalled general pasting method, according to AACC Method 76–21. Analyses were performed with 2 g purified starch in 25 mL of water.

Results

Variation in granule size distributions in durum wheat through missense mutations in MRC

Using the wheat TILLING mutant database on Ensembl plants (Krasileva et al. 2017), we identified a total of 45 different mutant lines in the tetraploid durum wheat Kronos that carried different missense mutations in TtMRC-A1 (Table 1). The amino acid substitutions in these lines were in different positions along the length of the MRC protein and were predicted to range from tolerated to deleterious using both the SIFT and PROVEAN programmes (Choi and Chan 2015; Ng and Henikoff 2003). Seeds from all 45 mutant lines (either homozygous or heterozygous for the mutations) were obtained from the Germplasm Resource Unit of the John Innes Centre, and KASP genotyping markers were designed for each of the mutations. For our experiments, we successfully genotyped and isolated 29 homozygous lines (Table 1). The remaining 16 lines were excluded, either because the mutation could not be successfully genotyped using KASP markers, due to poor growth of the mutant line (indicating the presence of background mutations), or because not enough homozygous seeds were obtained in time for the first experiment.

In our first experiment (Experiment 1), we screened the 29 homozygous mutant lines for alterations in starch granule size distribution in the endosperm. Starch was extracted from mature grains harvested from n = 3–4 individual plants per line, which were all grown together in a single growth experiment in the glasshouse. For comparison, the mrc-1 and mrc-2 mutants which carry premature stop codons in TtMRC-A1 were also included (Chen et al. 2024) (Table 1). Granule size distributions were analysed with the Coulter counter, and subsequent curve fitting was used to derive B-type granule content (McNelly et al. 2024).

The mrc-1 mutant had the highest B-type granule content, whereas mrc-2 had a relatively moderate increase in B-type granule content compared to mrc-1 (Fig. 1). These results were consistent with Chen et al. (2024) and likely due to the premature stop codon occurring later in the coding sequencing in mrc-2 compared to mrc-1 (Table 1). Interestingly, three of the lines with missense mutations had B-type granule content that was in between that of mrc-1 and mrc-2. This included K0598 (referred to as mrc-3 in Chen et al. 2024), K4060 and K3889. Strikingly, some of the lines had a lower B-type granule content than the wild type, with the lowest (K2485) having a value almost half that of the wild type.

Fig. 1.

B-granule content from initial screening of Kronos TILLING mutants (Experiment 1). Starch was extracted from mature grains and B-type granule volume was calculated using Coulter counter and curve-fitting analyses. Values are the mean ± standard error of the mean (SEM) from n = 3–4 biological replicates, where each replicate (shown as an individual data point) was prepared from grains harvested from a separate plant. Values marked with an asterisk are significantly different to the Kronos wild type (WT) under a pairwise two-tailed t test (p < 0.05)

Although we observed substantial variation in B-granule content in this experiment, a one-way ANOVA with Tukey’s HSD test showed that only the B-type granule content values of mrc-1 and K0598 were significantly different from the WT. This is due to low statistical power within the experiment, caused by the large number of lines being compared within the experiment, and the relatively low numbers of replicates per line. However, under pairwise two-tailed t tests against the wild type, six of the missense lines (K3505, K2937, K2408, K4060, K3889, K0598), alongside mrc-1 and mrc-2, had significantly higher B-type granule content than the wild type, while two lines (K0265, K2485) had significantly lower B-type granule content. Experiment 1 was therefore successful as a screen that provided candidate lines with altered B-type granule content, but the lack of statistical power meant that more replicates were required for further characterisation.

Therefore, in Experiment 2, we selected the six candidate lines that had higher B-type granule content than the wild type under pairwise t tests, along with both lines with lower B-type granule content, and grew them for a second generation in the glasshouse with a larger number of replicates (n = 7–8 plants per line). Mature grains were harvested for starch granule analyses. The smaller number of genotypes and more replication compared to Experiment 1 increased statistical power, as one-way ANOVA and Tukey’s HSD post hoc test revealed more lines with significantly altered B-type granule content than the previous experiment (Fig. 2A). Like Experiment 1, mrc-1 and K0598 had the largest B-type granule content, at 58% and 42%, respectively. The value obtained for mrc-1 was slightly lower than the 67% increase we reported in Chen et al. (2024). Four of the missense mutants (K4060, K3889, K2408 and K2937) had an intermediate increase in B-type granule content (33–36%), that was significantly different from the wild type (24%). Also similar to Experiment 1, K2486 and K0265 were the only lines to have lower B-type granule content, but the effect was significant only for K2485 (15%). Overall, in this experiment, the different mrc mutations achieved a wide range of B-type granule content between 15 and 58%.

Fig. 2.

Granule size parameters from Kronos TILLING mutants in Experiment 2. Coulter counter analyses were carried out on starches extracted from mature grains, and granule size parameters were calculated using curve-fitting analyses. A B-type granule content (by relative volume). B Mean B-type granule diameter. C Mean A-type granule diameter. D Relative number of small granules, defined as granules < 10 µm. For panels A–D, values are the mean ± standard error of the mean (SEM) from n = 7–8 biological replicates, where each replicate (shown as an individual data point) was prepared from grains harvested from a separate plant. Values marked with an asterisk are significantly different to the Kronos wild type (WT) under a one-way ANOVA and Tukey’s HSD test (p < 0.05). E and F Volumetric granule size distribution from the Coulter counter for the K2485 line with low B-type granule content (panel E), and for lines with high B-type granule content (panel F). These are from the same analyses as those presented in A–D. The solid line represents the mean distribution from n = 7–8 biological replicates, and the shading represents the SEM. Data were acquired with logarithmic bin spacing and are shown on a logarithmic x-axis. G Schematic representation of the TtMRC-A1 polypeptide, with locations of amino acid substitutions in the different mutants indicated. Locations of mutations that significantly increased B-type granule content are marked with a green arrowhead, while the location of the K2485 mutation that significantly decreased B-type granule content is marked with a blue arrowhead

We then assessed the mean size of the A- and B-type granules, another granule size parameter from the Coulter counter and curve-fitting analysis. Consistent with Chen et al. (2024), the mrc-1 knockout mutant and K0598 had significantly larger B-type granules than the wild type (Fig. 2B) and significantly smaller A-type granules (Fig. 2C). We previously proposed that the smaller A-type granules in these mutants are due to the increased amount of B-type granules competing for substrates with A-type granules (Chen et al. 2024). However, the four lines with intermediate increases in B-type granule content (K4060, K3889, K2408 and K2937) did not show consistent decreases in A-type granule size, suggesting that such moderate increases are not sufficient to cause detectable decreases in A-type granule size.

Interestingly, the mean size of B-type granules was significantly larger in line K2485 compared to the wild type, despite its significant reduction in total B-type granule content (Fig. 2B), which is only possible if this line has fewer B-type granules by number. We therefore calculated the relative number of small granules from the Coulter counter data, defining small granules as < 10 µm, which would mostly contain B-type granules. K2485 was the only line with a significant change in this parameter, indeed having significantly fewer small granules (Fig. 2D). K2485 therefore makes fewer but larger B-type granules compared to the wild type, and an overall decrease in B-type granule content by volume. This was reflected in the drastically altered granule size distribution curve compared to the wild type, where the B-type granule peak was much less prominent in the mutant (Fig. 2E). This starkly contrasted all other mrc mutants, which had more prominent B-type granule peaks (Fig. 2F).

To complement the analysis of granule morphology using the Coulter counter, we visually examined the starch granules in these lines using scanning electron microscopy. Although quantitative differences in granule size distribution are difficult to discern using microscopy, we observed that all lines had typical lenticular A-type and round B-type granule morphology (Supplemental Fig. 1). In addition, we measured total starch content in these selected lines. With the exception of one line (K2408) that had a small reduction in starch content compared to the wild type, all other lines had no significant changes in starch content (Supplemental Fig. 2A), indicating that the altered granule size distributions were largely independent of starch content. To check for effects on grain size, we measured the Thousand Grain Weight (TGW) using the Marvin seed analyser (Supplemental Fig. 2B). For unknown reasons, four of the examined lines had elevated TGW (K3505, K4060, K3889 and K2408), but this did not correlate with the variation in starch granule size, as the lines with the most extreme changes in starch granule size (e.g. K2485, K0598 and mrc-1) did not have altered TGW. We also analysed starch composition by quantifying amylose content, but there were no significant differences between the lines (Supplemental Fig. 2C).

Overall, we discovered that mutations in TtMRC-A1 can be used to both increase and decrease B-type granule content without affecting total starch content or grain size, making it an ideal target gene for achieving a wide range of granule size distributions in durum wheat. The amino acid substitutions that led to these changes followed no obvious pattern. Those that increased B-type granule content were scattered along the length of the TtMRC-A1 polypeptide and were not necessarily in a coiled coil region, while the A625T mutation in K2485 that decreased B-type granule content was in a coiled coil towards the C-terminus (Fig. 2G).

The altered granule size distribution affects starch physicochemical properties in the mutants

Our durum lines presented a unique opportunity to study how variation in granule size parameters affects starch functional properties within a single cultivar. We therefore selected lines for bulking grains and for conducting starch functional analyses. We chose the line with the highest B-type granule content (mrc-1), the K0598 line with an intermediate increase in B-type granule content, and the K2485 line with reduced B-type granule content, together with the wild-type Kronos background. Grains were bulked in the glasshouse and we used a large-scale starch extraction method to obtain purified starch. Coulter counter analyses showed that the large-scale starch preparation showed similar trends to what we observed with the small-scale experiments, with K2485 having a lower B-type granule peak than the wild type, whereas both K0598 and mrc-1 had higher B-type granule peaks than the wild type (Fig. 3A). However, the measured B-type granule content for all genotypes was higher than what we observed in the small-scale experiments, ranging from 31% in K2485 to 64% in mrc-1 (Supplemental Fig. 3), although the value for mrc-1 was still in a range that was expected for this genotype from our previous work (Chen et al. 2024).

Fig. 3.

Mutants in TtMRC-A1 with altered granule size distributions have altered pasting properties. A Granule size distributions of starch samples from bulked grains, measured using a Coulter counter. Data were acquired with logarithmic bin spacing and are shown on a logarithmic x-axis. B Rapid Visco Analysis (RVA) showing pasting profiles of the starches shown in panel A. For both panels A and B, the solid line represents the mean from n = 3 replicate measurements, while the shading indicates the standard error of the mean (SEM). C Correlation plots between granule size and pasting parameters. The R² value is shown on all panels

We then analysed starch pasting parameters using the Rapid Visco Analyser (Fig. 3B; Table 2). The viscograms showed substantial variation between the tested lines: both lines with higher B-type granule content (mrc-1 and K0598 lines) had lower peak viscosity than the wild type, while the K2485 line with lower B-type granule content had higher viscosity. To identify which pasting (RVA) parameters were most dependent on granule size distributions, we performed pairwise linear regressions between each pasting parameter and each granule size parameter, using R² values to quantify the strength of these relationships (Supplemental Fig. 4). The strongest relationships with B-type granule content were a negative correlation with peak viscosity (R² = 0.93) and positive correlation with pasting temperature (R² = 0.97) (Fig. 3C). However, as expected from our results in Experiment 2, B-type granule content also strongly negatively correlated with A-type granule size (R² = 0.93) (Fig. 3C); and thus all RVA parameters that negatively correlated with B-type granule content were also positively correlated with A-type granule diameter, and vice versa (Supplemental Fig. 4). For example, A-type granule size was strongly positively correlated with peak viscosity (R² = 1) and negatively with pasting temperature (R² = 0.92) (Fig. 3C). A-type granule size also correlated with breakdown viscosity and peak time (Supplemental Fig. 4). B-type granule size correlated with only few pasting parameters, most notably with setback viscosity. Overall, the altered granule size distributions in the selected mutants affected the pasting properties of the starch, but it was not clear whether this was primarily due to variation in B-type granule content, or A-type granule size.

Table 2.

Starch pasting parameters from RVA experiments

Sample	Peak visc (cP)	Trough visc. (cP)	Breakdown visc. (cP)	Final visc. (cP)	Setback visc. (cP)	Peak time (min)	Pasting temp. (°C)
Whole starch sample
Kronos WT	1176 ± 6	461 ± 18	715 ± 23	1325 ± 11	864 ± 28	5.24 ± 0.09	86.40 ± 0.49
K2485	1239 ± 8	529 ± 40	710 ± 33	1464 ± 19	936 ± 59	5.27 ± 0.10	86.45 ± 0.48
K0598	1043 ± 27*	517 ± 22	526 ± 49	1314 ± 55	796 ± 77	5.38 ± 0.12	88.87 ± 0.83
mrc-1	962 ± 31*	444 ± 25	518 ± 56*	1366 ± 35	922 ± 56	5.44 ± 0.16	91.78 ± 1.50*
B-type granule-depleted sample
Kronos WT	1202 ± 4	627 ± 2	575 ± 3	1296 ± 10	669 ± 8	5.49 ± 0.02	85.82 ± 0.29
K2485	1289 ± 12*	648 ± 6	641 ± 7*	1453 ± 20*	805 ± 14*	5.4 ± 0*	85.03 ± 0.28
K0598	1114 ± 10*	680 ± 9*	434 ± 3*	1267 ± 9	587 ± 3*	5.6 ± 0*	86.73 ± 0.28
mrc-1	1069 ± 7*	663 ± 3*	406 ± 5*	1313 ± 5	650 ± 2	5.8 ± 0*	90.43 ± 0.03*

RVA parameters were calculated from the viscograms shown in Figs. 3B and 4B. Values are the mean ± standard error of the mean (SEM) from n = 3 replicate measurements. Values marked with an asterisk are significantly different to the Kronos wild type (WT) under a one-way ANOVA and Tukey’s HSD test (p < 0.05)

To resolve whether the altered pasting properties were due to the variation in B-type granule content or A-type granule size, we measured pasting parameters on starch samples from the selected mutants after removing most of the B-type granules by sieving the starch through a 10 µm mesh. The retentate after sieving contained starch that was highly enriched in A-type granules, as determined using Coulter counter analysis (Fig. 4A). Compared with the original RVA analyses done on the whole starch preparation (Fig. 3B), the removal of B-type granules slightly increased peak viscosity in all genotypes, while final viscosity remained similar (Fig. 4B; Table 2). Importantly, the relative differences between the genotypes remained similar, with K2485 having higher peak viscosity than the wild type, K0598 having lower peak viscosity than the wild type, and mrc-1 having the greatest reduction in peak viscosity. The strong correlation between A-type granule size and peak viscosity was retained after sieving, and the correlation between A-type granule size and pasting temperature was also retained, although the R² value was slightly weaker than before sieving (Fig. 4C; Supplemental Fig. 4). These data suggest that A-type granule size is the primary contributor to the variation in pasting properties in these mutants.

Fig. 4.

The mutants in TtMRC-A1 retain altered pasting properties after removal of most B-type granules. A Granule size distributions of starch samples after depletion of B-type granules using a 10 µm filter. Size distributions were measured using a Coulter counter. Data were acquired with logarithmic bin spacing and are shown on a logarithmic x-axis. B Rapid Visco Analysis showing pasting profiles of the starches with depleted B-type granules shown in A. For both panels, the solid line represents the mean from n = 3 replicate measurements, while the shading indicates the standard error of the mean (SEM). C Correlation plots between A-type granule size and pasting parameters. The R² value is shown on both panels

Discussion

Tuning granule size distributions using mutations at a single genetic locus

Here, we demonstrate that starch granule size distributions in durum wheat grains can be tuned at a single genetic locus, TtMRC-A1. Mutations at this locus can be used to achieve a wide range of different B-type granule content and A-type granule size. We had previously discovered that TtMRC acts as a repressor of B-type granule initiation during early grain development, such that mutations in TtMRC-A1 substantially increase B-type granule content (Chen et al. 2024). By screening durum wheat lines carrying TtMRC-A1 missense mutant alleles, we characterised the effect of random amino acid substitutions along the length of the protein (Table 1). From the 29 missense mutants examined, we identified 5 lines with varying degrees of increased B-type granule content, and unexpectedly one line (K2485) with significantly decreased B-type granule content (Fig. 2). The previously characterised mrc-1 mutant carrying an early premature stop codon (Chen et al. 2024) had the strongest increase in B-type granule content. Using mutations in TtMRC-A1, we achieved a range of B-type granule content between 15 and 58%—where K2485 had the lowest B-type granule content, and mrc-1 had the highest (Fig. 2). This range is greater than what was achieved by crossing in natural variation from Triticum dicoccoides (22–44%) (Sissons and Egan 2023).

Our results also demonstrate that altering B-type granule content through TtMRC-A1 indirectly modulates A-type granule size. We previously proposed that in the mrc-1 mutant, the B-type granules initiate too early during grain development and hinder A-type granule growth by competing for substrates, thus resulting in smaller A-type granules. Consistent with this, when comparing the mutants with the strongest alterations in B-type granule content (K2495, K0598 and mrc-1), we observed a strong negative correlation between B-type granule content and A-type granule size (Fig. 3). However, in the lines with intermediate increases in B-type granule content (K4060, K3889, K2408 and K2937), A-type granule content was either not significantly changed, or significantly increased in the case of K4060 (Fig. 2). It is likely that such increases in B-type granule content are not sufficient to cause detectable changes in A-type granule size. The reason for the increase in A-type granule size in K4060 is not known but is unlikely to be related to the B-type granules.

Interestingly, significant increases in B-type granule size were observed in the two lines with the strongest increases in B-type granule content (K0598 and mrc-1), as well as in the K2485 line that had lower B-type granule content. Although this might seem counter-intuitive, the increases in B-type granule size likely occur for different reasons. In the lines with increased B-type granule content, the early initiation of the B-type granules could allow more time for B-type granules to grow during grain development, resulting in larger B-type granules. However, in K2485, we observed that there was a significant reduction in the number of B-type granules. Thus, there are fewer granules competing for substrates for granule growth, allowing each individual B-type granule to grow larger.

How the amino acid substitutions in MRC affect its function in granule initiation remains to be investigated. In the lines with increased B-type granule content (K2937, K4060, K0598, K2408 and K3889), the mutations (A178T, R255K, L289F, S503N and D581N, respectively) must have partially inhibited MRC function, since their granule size distributions shifted in the same direction as mrc-1 with the premature stop codon mutation. Conversely, since the K2485 mutation had the opposite phenotype to mrc-1, the mutation (A625T) likely enhanced MRC function. The exact mechanism by which MRC acts is currently not known. However, we previously showed that MRC interacts directly with BGC1 (Chen et al. 2024). It is possible that the mutations in MRC specifically affect its ability to interact with BGC1, or with other granule initiation proteins. It is also possible that the mutations affect MRC protein stability. Protein destabilisation can inhibit function, whereas stabilising mutations can decrease the rate at which the protein is turned over, resulting in a longer protein lifetime. The latter is relevant since we proposed that MRC is expressed only during early grain development when it represses B-type granule initiation, and that the protein is not present during the later stages of grain development when B-type granules are initiating (Chen et al. 2024, 2023). A stabilising mutation may cause MRC protein to be retained during these later stages of grain development, thus inhibiting B-type granule initiation.

Although some significant changes in starch content and grain weight were observed among the lines (Supplemental Fig. 2), the lines with the strongest alterations in B-type granule content (K2495, K0598 and mrc-1) had no change in these parameters. This suggests that the changes in granule size distributions can be achieved in the absence of changes in starch content.

It is important to note that with wheat TILLING mutants, due to the number of background mutations in these lines (Uauy et al. 2017), we cannot rule out that other mutations influenced the observed phenotypes. However, it is unlikely that other mutations are directly responsible for the observed increases in B-type granule content, since this phenotype is consistent with our previously published mrc mutants (Chen et al. 2024). In the case of K2485 where B-type granule content was decreased, the phenotype segregated with the K2485 mutation after a round of backcrossing, decreasing the possibility that background mutations cause this phenotype (Supplemental Fig. 5). Further work with these mutants must also determine how robust these granule size phenotypes are to the environment, particularly by testing them under field conditions.

Granule size distributions affect starch physicochemical properties

Our RVA analyses demonstrated that the changes in granule size distribution induced by TaMRC-6A mutations affect starch pasting properties, particularly peak viscosity and pasting temperature. Although the mutations affected B-type granule content and the size of A- and B-type starch granules, we determined that the effect on A-type granule size is likely to be the major contributor to the altered pasting properties. There was a strong positive correlation between peak viscosity and A-type granule size (Fig. 3), which was still observed after removing the majority of B-type granules (Fig. 4). Peak viscosity is one of the most important RVA parameters as it reflects the texture and thickness during starch cooking or gelatinisation. Our findings are consistent with the general trend that larger granules have higher viscosity (Li et al. 2021; Lindeboom et al. 2004). Also, given that A-type granules constitute the majority of the starch by volume in most of the lines examined, it is reasonable to expect that the pasting parameters are more influenced by the characteristics of the A-type granules. In addition to peak viscosity, A-type granule size (and thus B-type granule content) also strongly correlated with breakdown viscosity, peak time and pasting temperature (Supplemental Fig. 4). All these parameters correlated poorly with B-type granule size. Rather B-type granule size correlated strongly with the setback viscosity, and to some degree with final viscosity.

Our study explored how pasting properties are affected by extensive, induced variation in granule size distributions within a single wheat cultivar. This contrasts the many previous studies that have investigated pasting properties in a narrower range of granule size variation existing between wheat cultivars, achieved through flour reconstitution, or following environmental treatments. Among these previous works, there are conflicting data on the effect of B-type granule content on pasting properties. Consistent with our study, higher B-type granule content in durum wheat was associated with lower peak viscosity, when variation in B-type granule content was achieved through crossing with Triticum dicoccoides or through reconstitution (Sissons and Egan 2023; Soh et al. 2006). This is also consistent with results in bread wheat: Wootton et al. (1998) found that B-type granule content negatively correlated with peak viscosity in a panel of Australian wheats, while Singh et al. (2008) also observed a similar negative correlation among different cultivars following water stress treatments. However, the opposite was found by Peterson and Fulcher (2001) and Dai et al. (2009)—both genotypic variation in B-type granule content among wheat cultivars and variation induced by different watering treatments were positively correlated with peak viscosity. As discussed by Dai et al. (2009), it is likely that these contrasting results are due to pasting parameters being influenced by many other parameters, such as amylose and lipid content, which could also vary among genotypes and after environmental treatments. Indeed, Wootton et al. (1998) attributed the observed correlation between B-type granule content and peak viscosity to variation in amylose content, which also correlated with B-type granule content in the cultivars. An advantage of our study is that the variation in granule size distributions was induced by targeting a gene specifically involved in B-type granule initiation and is unlikely to alter other aspects of grain composition (Chen et al. 2024).

It should also be noted that the extent of variation in A-type granule size in these previous studies, and the degree to which that correlated with pasting properties is not known. The decreasing A-type granule size with increasing B-type granule content is a characteristic of TtMRC-A1 mutations, and it cannot be assumed that A-type granule size correlates negatively with B-type granule content in a broader range of cultivars. Also, although our results suggest that A-type granule size is a stronger determinant of pasting properties than B-type granule content, we emphasise that this is only under the pasting conditions tested. The relative viscosities of fractionated A- and B-type granules are dependent on the concentration of starch used in the RVA, where A-type granules have higher viscosity than B-type granules at high solids content, while B-type granules have higher viscosity than A-type granules at lower solids content (Watanabe et al. 2024). It is therefore possible that we will see stronger contributions of B-type granule size and content to pasting parameters under different starch concentrations and pasting conditions.

The RVA parameters provide proof-of-concept that the altered granule size distributions achieved through TtMRC-A1 affect starch physicochemical properties. Given the interesting correlations between granule size and pasting parameters observed using a subset of the lines, analysing pasting parameters in a larger set of lines in a future study could further strengthen these correlations. Of particular interest could be lines K0265 and K3505, as these had variation in A-type granule size without changes in B-type granule content. Analysing these lines could strengthen the association between A-type granule size and pasting parameters. Also, it should be noted that the limitation of laboratory tests (e.g. RVA) is that they cannot always predict how the starches behave during industrial processing of food and other products. There can also be other benefits of having variation in granule size that are not directly linked to pasting parameters. For example, having lower B-type granule content in malt is beneficial for brewing (Bathgate and Palmer 1973). The digestibility of the starch may also be altered in the mrc mutants since A-type granules are digested more slowly in vitro than B-type granules (Colussi et al. 2021; Moon and Kweon 2024). Our set of mrc mutants provide an ideal panel that can be tested for improved quality in various applications of durum wheat, including pasta making.

Applicability to other Triticeae crops

Given that MRC has orthologs in other Triticeae species (Chen et al. 2024), it is likely that it can also be targeted to alter starch granule size distributions in these species. However, it is an ideal gene target in durum wheat because the B-genome homeolog became a pseudogene shortly after the hybridisation that gave rise to tetraploid wheat (Chen et al. 2024), allowing changes in granule size to be induced by altering the single TtMRC-A1 locus. We expect that a similar benefit will be seen in barley, a diploid Triticeae that also has only one gene encoding MRC. However, in hexaploid bread wheat, there are two active homeologs of MRC on chromosomes 6 A and 6D, and it is likely both homeologs need to be targeted to induce substantial changes to the granule size distribution. Missense mutations may be induced using chemical mutagenesis and identified using TILLING approaches, or induced using CRISPR-based base editing. The mutations that significantly increased (A178T, R255K, L289F, S503N and D581N) and decreased B-type granule content (A625T) were all in amino acids that are conserved in all wheat and barley MRC sequences examined (Supplemental Fig. 6), suggesting that the same mutations can be targeted in bread wheat and barley. Interestingly, the missense mutations that led to the strongest increase (L298F) and decrease (A625T) in B-type granule content were in amino acids that were also conserved in the Arabidopsis and rice proteins, suggesting that these amino acids may play an important role in MRC function in a broader range of species.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions statement

DS conceived the study. BF and DS designed the research. BF and JC performed the research. All authors analysed data. DS wrote the article with input from all authors.

Funding

This work was funded through a John Innes Foundation (JIF) Chris J. Leaver Fellowship (to D.S), JIF Rotation Ph.D. studentships (to J.C.), Biotechnology and Biological Sciences Research Council (BBSRC, UK) research grants BB/W015935/1, BB/W01632X/2 and BB/Z517501/1 (to D.S.), and BBSRC Institute Strategic Programme grants BB/X01097X/1 and BB/X011003/1 (to the John Innes Centre).

Data availability

The data underlying this article are available in the article and in its supplementary material.

Declarations

Competing interests

JC and DS are co-inventors on a patent for modifying starch granule size in crops using MRC.