Initiation of B-type starch granules in wheat endosperm requires the plastidial α-glucan phosphorylase PHS1

Abstract The plastidial α-glucan phosphorylase (PHS1) can elongate and degrade maltooligosaccharides (MOSs), but its exact physiological role in plants is poorly understood. Here, we discover a specialized role of PHS1 in establishing the unique bimodal characteristic of starch granules in wheat (Triticum spp.) endosperm. Wheat endosperm contains large A-type granules that initiate at early grain development and small B-type granules that initiate in later grain development. We demonstrate that PHS1 interacts with B-GRANULE CONTENT1 (BGC1), a carbohydrate-binding protein essential for normal B-type granule initiation. Mutants of tetraploid durum wheat (Triticum turgidum) deficient in all homoeologs of PHS1 had normal A-type granules but fewer and larger B-type granules. Grain size and starch content were not affected by the mutations. Further, by assessing granule numbers during grain development in the phs1 mutant and using a double mutant defective in both PHS1 and BGC1, we demonstrate that PHS1 is exclusively involved in B-type granule initiation. The total starch content and number of starch granules per chloroplast in leaves were not affected by loss of PHS1, suggesting that its role in granule initiation in wheat is limited to the endosperm. We therefore propose that the initiation of A- and B-type granules occurs via distinct biochemical mechanisms, where PHS1 plays an exclusive role in B-type granule initiation.


Open Access Introduction
Starch, the major storage carbohydrate in plants, is synthesized in plastids as semicrystalline granules composed of 2 glucose polymers (glucans)-amylopectin and amylose.Recent work has revealed new insights into molecular mechanisms that initiate (or prime) the formation of granules and control the number, size, and shape of granules in plastids (Seung and Smith 2019;Mérida and Fettke 2021).However, there is huge variation in the spatiotemporal patterns of granule initiation between different species and different organs (Seung and Smith 2019;Chen et al. 2021).The endosperm of wheat (Triticum spp.), and other Triticeae, has 2 distinct types of granules: large discoid A-type granules and small spherical B-type granules (Langeveld et al. 2000;Chia et al. 2020).A single A-type granule is initiated in each amyloplast during early grain development, and numerous B-type granules are initiated 10 to 15 d after the B-type granules, at least partially in amyloplast stromules (Parker 1985;Langeveld et al. 2000;Esch et al. 2023).The mechanisms underpinning this unique spatiotemporal pattern of A-and B-type granule formation are poorly understood.Understanding such mechanisms is also of industrial significance, since granule size distributions affect grain quality for bread and pasta making (Soh et al. 2006;Park et al. 2009) and brewing (Bathgate and Palmer 1973).
It is not known whether A-and B-type granule initiations occur through similar or distinct biochemical mechanisms.Both STARCH SYNTHASE4 (SS4) and B-GRANULE CONTENT1 (BGC1) are required for proper A-type granule initiation in wheat.Loss of either protein results in supernumerary granule initiations in most amyloplasts during early grain development, which later fuse to form "compound" starch granules (Chia et al. 2020;Hawkins et al. 2021).However, BGC1 also plays an important role during B-type granule initiation.While SS4 expression levels peak during early grain development, BGC1 expression is highest during later grain development, when B-type granules are initiating (Chen et al. 2023).B-type granule formation can be almost completely eliminated by reducing BGC1 gene dosage: by introducing loss-of-function mutations in 2 of 3 homoeologs of BGC1 in hexaploid wheat or by combining a loss-of-function mutation in 1 homoeolog with a hypomorphic missense mutation in the other homoeolog in tetraploid durum wheat (Chia et al. 2020).
The mechanism by which BGC1 acts in B-type granule initiation is not known.BGC1, along with FLOURY ENDOSPERM6 (FLO6) in barley (Hordeum vulgare) and rice (Oryza sativa), is an ortholog of Arabidopsis (Arabidopsis thaliana) PROTEIN TARGETING TO STARCH2 (PTST2) and thus belongs to the PTST family.These proteins have a carbohydrate-binding module (CBM48) that can bind soluble maltooligosaccharides (MOSs) (Seung et al. 2017) but have no known enzymatic domains.However, they interact and act together with enzymes.For example, in Arabidopsis leaves, PTST2 interacts with SS4, and both proteins promote granule initiation in chloroplasts (Roldán et al. 2007;Seung et al. 2017).The elongation of soluble MOS is an important step towards initiating new starch granules (Nakamura 2015;Seung and Smith 2019;Mérida and Fettke 2021).The plastid contains multiple enzymes that can elongate MOS-including SS4 and other starch synthase isoforms, as well as the plastidial α-glucan phosphorylase (PHS1 or PHO1).PHS1 catalyzes a reversible reaction, either degrading α-1,4-linked glucan chains via a phosphorolysis reaction that releases glucose-1-phosphate (G1P) or extending the glucan chain using G1P as a substrate.
The role of PHS1 in granule initiation, or more broadly in starch metabolism, is not clear.In Arabidopsis, the complete loss of PHS1 does not affect growth, starch turnover (Zeeman et al. 2004), or granule number per chloroplast (Malinova et al. 2014).However, double mutants defective in PHS1 and other components of maltose metabolism have severe defects in growth and strong reductions in granule number per chloroplast (Malinova et al. 2014;Malinova and Fettke 2017).A role for PHS1 in granule initiation was also inferred from rice PHS1 knockout mutants, which produced grains ranging from normal to shrunken, where shrunken grains with dramatic decreases in starch content were more prevalent when plants were grown at low temperatures (Satoh et al. 2008).Overall, these findings suggest PHS1 could play a role in granule initiation, but in the species examined, its importance is conditional on the presence of other mutations or temperature.More recent evidence suggests that PHS1 deficiency in potato (Solanum tuberosum) and rice results in strong alterations in MOS levels (Dong et al. 2023;Flores-Castellanos and Fettke 2023).
In this study, we aimed to discover the mechanism of B-type granule initiation.We looked for wheat endosperm proteins that interact with BGC1, and we discovered that BGC1 interacts with PHS1.We therefore investigated the role of PHS1 in the wheat endosperm, and we provide strong genetic evidence that PHS1 is required for normal initiation of B-type granules during grain development but not for normal A-type granules.We therefore propose a model where Aand B-type granules initiate via distinct biochemical mechanisms, the latter requiring PHS1.

IN A NUTSHELL
Background: Starch is the main storage carbohydrate in plants and a vital source of calories in human diets.In the Triticeae tribe, which includes wheat, barley, and rye, starch forms 2 distinct types of granules in grain: large discoid A-type granules and small spherical B-type granules.The A-type granules form early during grain development, and the B-type granules form later.
Question: We asked whether A-and B-type granules are initiated via similar or distinct biochemical mechanisms.To better understand the mechanism of B-type granule initiation, we looked for interaction partners of B-GRANULE CONTENT1 (BGC1), a carbohydrate-binding protein that is important for B-type granule initiation.

Findings:
We discovered that BGC1 interacts with the α-glucan phosphorylase, PHS1, in developing wheat endosperm.PHS1 can efficiently elongate short maltooligosaccharides in vitro, but decades of research have failed to find a clear role for this enzyme in plants.We produced phs1 knockout mutants in wheat and discovered they had fewer B-type granules compared to the wild type.By examining granule formation during grain development in the mutant, as well as double-mutant combinations with bgc1, we discovered that loss of PHS1 only affects the formation of B-type granules, not A-type granules.Our findings reveal an indispensable role for PHS1 in wheat and demonstrate that A-and B-type granule initiations occur through distinct biochemical mechanisms.

Next steps:
We are investigating whether the other BGC1 interaction partners identified in our study are involved in B-type granule initiation and studying how the actions of these proteins are coordinated.Since B-type granules affect the nutritional and functional properties of wheat, we are testing our phs1 mutants in various industrial applications.

BGC1 interacts with proteins involved in starch synthesis, including PHS1
To discover proteins involved in B-type granule initiation, we identified proteins that associate with BGC1 using affinity pulldown and MS.We produced extracts from developing durum wheat (Triticum turgidum) 'Kronos' endosperm harvested at 18 d post anthesis (dpa), a timepoint when B-type granule formation is actively occurring.Following coincubation with recombinant His-tagged BGC1, we pulled down BGC1 and associated proteins using anti-His beads.MS was used to identify interactors.We set a threshold for at least 2-fold significant enrichment (P < 0.05) in the pulldown vs. controls (extracts incubated without TaBGC1).The 118 proteins that met this criterion are presented in Supplemental Data Set 1.
Given the large number of proteins identified in the pulldown, we used the data set to specifically assess which known proteins of starch metabolism pulled down with BGC1.There were 13 starch-related proteins in the filtered list (Table 1).We observed that MAR-binding filament-like protein (MFP1), a known interaction partner of PTST2 (the BGC1 ortholog) in Arabidopsis (Seung et al. 2018), was enriched with the highest abundance in the pulldown-providing strong indication that the pulldown was effective.
MFP1 has been duplicated in cereals (Seung et al. 2018), and interestingly, both paralogs of MFP1 in wheat (MFP1.1 and MFP1.2) were identified in the pulldown, suggesting both are present in the endosperm and can associate with BGC1.In addition, we found multiple starch-branching enzyme 1 (SBE1) isoforms.Only 2 proteins with glucan-elongating activity were detected in the pulldown: the plastidial α-glucan phosphorylase (PHS1) and starch synthase 2a (SS2a).In this study, we focused on the interaction between PHS1 and BGC1, given the prior knowledge that PHS1 elongates MOS and can conditionally act in granule initiation in other species (Satoh et al. 2008;Malinova et al. 2014), and because SS2a already has established roles in other aspects of starch synthesis, particularly in determining amylopectin structure (Morell et al. 2003).Notably, peptides matching SS4 were not detected in this experiment.It is possible that SS4 abundance is relatively low in our extracts at 18 dpa, since SS4 expression is highest during early grain development when A-type granules are initiated and lower at the later stages when B-type granules are initiated (Hawkins et al. 2021;Chen et al. 2023).
To confirm the interaction between BGC1 and PHS1 observed in wheat extracts, we carried out a pairwise immunoprecipitation assay using transiently expressed proteins in Nicotiana benthamiana leaves.We first confirmed that both PHS1 and BGC1 are located in the chloroplast in this heterologous system (Fig. 1A).PHS1 was located in the stroma, whereas BGC1 was located around starch granules and in punctate structures similarly observed for the Arabidopsis ortholog (Seung et al. 2017(Seung et al. , 2018)).We then used YFP/ RFP-tagged proteins for the immunoprecipitation.BGC1: RFP copurified with PHS1:YFP in the immunoprecipitation (IP) with anti-YFP beads when the 2 fusion proteins were coexpressed (Fig. 1B).BGC1:RFP did not purify in control reactions when coexpressed with a chloroplast-targeted YFP, nor did a chloroplast-targeted RFP copurify with PHS1:YFP.This suggests that BGC1 specifically interacts with PHS1.
We then compared the expression patterns of PHS1 and BGC1 during grain development.We used our previously generated RNA-Seq data set for endosperm development in durum wheat 'Kronos' (Chen et al. 2023).PHS1 expression was detectable at all stages of grain development tested between 6 and 30 dpa but was particularly strong between 8 and 13 dpa (Fig. 2B).No differences in expression level or pattern were observed between the A and B homoeologs.PHS1 activity levels in endosperm extracts were visualized on native PAGE gels.Consistent with the transcript levels, robust PHS1 activity was detected at all timepoints tested between 8 and 28 dpa but was stronger at the earlier than later timepoints, especially at 12 dpa (Fig. 2C).To assess PHS1 expression in different tissues, we used public data for bread wheat 'Chinese Spring' in the wheat expression browser (Borrill et al. 2016).PHS1 transcripts were detectable in leaves/shoots, roots, spike, and grains, and all homoeologs showed similar levels of expression across all examined tissues (Supplemental Fig. S1).

The phs1 mutant of wheat has no obvious effects on vegetative growth
To study the function of PHS1, we used the wheat in silico TILLING resource (http://www.wheat-tilling.com)(Krasileva et al. 2017) to identify mutants in tetraploid durum wheat 'Kronos' with mutations in either homoeolog A or B (Fig. 2A).The Kronos4533 (K4533) line had a premature stop codon, and the Kronos4367 (K4367) line had a splice acceptor mutation in the PHS1-A1 homoeolog.The Kronos2864 (K2864) line had a premature stop codon, and the Kronos0238 (K0238) line had a splice donor mutation in the PHS1-B1 homoeolog.To combine A and B homoeolog mutant alleles, K4367 was crossed with K2864 to generate phs1-1 mutant lines, and K4533 was crossed with K0238 to generate phs1-6 mutant lines.We isolated in the F2 generation the following genotypes: AA BB (the wild-type [WT] sibling control), aa BB (single mutant-mutation only in the PHS1-A1 homoeolog), AA bb (single mutant-mutation only in the PHS1-B1 homoeolog), and aa bb (the full double mutant defective in both homoeologs).
We verified the effect of the mutations on PHS1 activity using native PAGE gels, using proteins extracted from leaf tissue.Unlike in the endosperm (Fig. 2C), we observed 4 different phosphorylase activities in leaf protein extracts (Fig. 2, D and E).Further, due to the lower activity in leaves compared to endosperm, we could distinguish 2 closely migrating PHS1 bands for leaf extracts.Both bands were absent in the phs1-1 and phs1-6 aa bb double mutants, and the 2 aa BB and AA bb single mutants of phs1-1 were missing either 1 of the 2 bands.We concluded that the 2 bands represent PHS1-A1 and PHS1-B1 activities, and the selected mutations for phs1-1 and phs1-6 mutant lines eliminate all detectable PHS1 activity.The strong upper activity bands on the gels likely correspond to the cytosolic phosphorylase, PHS2 (Schupp and Ziegler 2004).Overall, we observed no detectable effect of the phs1 mutants on plant growth in glasshouses (Fig. 3A).However, in some experiments, plants from all phs1-1 lines (including the mutants and the WT siblings) were slightly shorter than WT plants.This is likely due to background mutations rather than the loss of PHS1 activity, since it was also observed in the WT sibling from the phs1-1 cross but not in any of the phs1-6 lines.Similarly, there was no effect of phs1 mutations on the grain yield per plant, but a decrease was seen in all phs1-1 lines (Fig. 3B).Again, there was no significant difference in grain yield per plant or seed number between the phs1-1 double mutant and its WT sibling control or between the phs1-6 double mutant and its WT sibling control (Fig. 3B).All lines produced seeds that appeared normal (Fig. 3C), and we did not observe any significant effects of phs1 mutations on thousand grain weight or grain size (Fig. 3, D and E).Since the rice mutant is reported to have a significant proportion of shrunken seeds (Satoh et al. 2008;Dong et al. 2023), we also assessed the distribution of grain sizes harvested from the mutants, but the proportion of small seeds was not greater in the mutants compared to the WT (Supplemental Fig. S2).We therefore conclude that loss of PHS1 activity does not affect vegetative growth, seed yield, or seed morphology under our conditions.

PHS1 is required for normal B-type granule size and number
We investigated the effect of phs1 mutations on starch synthesis in the endosperm.First, we measured the total starch content of grains.The phs1 double mutants had identical starch content to the WT segregant control and the WT (Table 2; Supplemental Fig. S3, A and B).Interestingly, starch in both the phs1-1 and phs1-6 double mutants had a small   reduction in amylose content.The effect was significant in phs1-1 when compared with all WT controls.The effect in phs1-6 was significant only when compared to the WT but not when compared to the phs1-6 WT segregant, likely because this WT segregant had slightly lower amylose content compared to the WT.The reduction in amylose content in the phs1-6 double mutant is still likely to be directly caused by the loss of PHS1, since amylose contents of the phs1-1 and phs1-6 double mutants were nearly identical.Further, reducing the number of statistical comparisons to include only phs1 mutants and the WT controls (not including bgc1-1) made the phs1-6 double mutant significantly different to all WT controls (Supplemental Fig. S3, C and D).Notably, these changes in amylose content occurred without major changes in the abundance of starch-associated granulebound starch synthase (GBSS), the enzyme that synthesizes amylose (Supplemental Fig. S3E).This suggests that PHS1 is likely required for normal amylose content in wheat.The reduced amylose content was not observed in the phs1-1 single homoeolog (aa BB, AA bb) mutants.
We then purified starch granules from mature grains and examined their morphology using scanning electron microscopy (SEM).Interestingly, starch from the phs1 double mutants, phs1-1 and phs1-6, had visibly fewer B-type granules than the WT (Fig. 4A; Supplemental Fig. S4A).Quantification of granule size distribution on the Coulter counter confirmed that the granule size distribution (plotting relative volume over diameter) was drastically altered in both phs1-1 and phs1-6 double mutants, with the B-type granule peak being smaller compared to the WT controls, and shifted towards the larger sizes (Fig. 4B; Supplemental Fig. S4B).Using curve fitting on these plots, we calculated that both phs1-1 and phs1-6 double mutants had less than half of the B-type granule content (by volume) compared to the WT and single mutants, despite having significantly larger B-type granules (Fig. 4,C and D;Supplemental Fig. S4,C and D).As the large size of the B-type granules in the double mutant affects the area of the B-type granule peak when plotted on a volume basis, we also plotted the size distributions on a relative number basis for the phs1-1 double mutant (Fig. 4E), which showed a large shift in the proportion of A-vs.B-type granules in the double mutant.The granule size distributions in the phs1-1 single homoeolog mutants were identical to the WT controls, suggesting that mutations in both PHS1 homoeologs are necessary to alter this phenotype (Fig. 4, A to E).
As the large size of the B-type granules affects the calculation of B-granule content on a volumetric basis, we also investigated whether the number of B-type granules was affected in the double mutant.Curves cannot be reliably fitted to number-diameter plots from the Coulter counter due to the very small size of the A-type granule peak.So instead, as done previously (Chia et al. 2020), we calculated the percentage of starch granules that were smaller than 10 µm, which includes B-type granules and some smaller A-type granules.There was a significant decrease in the relative number of small granules in the double mutant compared to the controls (Fig. 4F; Supplemental Fig. S4E).In contrast to B-type granules, the size of A-type granules was unaffected in both phs1-1 and phs1-6 double mutants (Fig. 4G; Supplemental Fig. S4F).
Taken together, these data suggest that loss of PHS1 results in the synthesis of fewer, but larger, B-type granules.This, together with its interaction with BGC1, points to a role for PHS1 in the initiation of B-type granules, which we investigate further below.Additionally, PHS1 is required for normal amylose content in wheat grains.

PHS1 acts during B-type granule initiation in grain development
To examine if the loss of PHS1 specifically affected B-type granule formation during grain development, we measured starch content, granule number, and granule size distributions in developing endosperm.We dissected endosperms from developing grains of the phs1-1 and phs1-6 double mutants, as well as their corresponding WT controls.Similar to mature grains, we did not observe significant differences in total starch content between the mutants and WT controls at any timepoint during grain development (Fig. 5A).We then used the Coulter counter to quantify the number of starch granules per milligram of endosperm, as well as to assess granule size distributions.The number of starch granules did not differ between the mutants and the WT at 8 or 14 dpa.However, at 18 and 22 dpa, both the phs1 mutants had significantly fewer starch granules than the WT controls (Fig. 5B).
Since B-type granules initiate typically between 15 and 20 dpa, and only A-type granules are present before this timepoint, it is likely that the phs1 mutants contain normal numbers of A-type granules (Fig. 6, A and B) but have fewer B-type granules than the WT (Fig. 6, C and D).This hypothesis was also strongly supported by the granule size distributions from the Coulter counter and in SEM (Fig. 6, A to H).The distributions were unimodal in the mutants and the WT controls at the 8 and 14 dpa timepoints (Fig. 6, A, B, E,  and F).However, at the 18 and 22 dpa timepoints, the distributions turned bimodal-suggesting all genotypes could initiate B-type granules (Fig. 6, C, D, G, and H).However, the B-type granule peaks were smaller and shifted towards the larger sizes in the phs1 mutants (Fig. 6, C and D).Larger and fewer B-type granules were observed in the SEM images of starch from these timepoints (Fig. 6, G and H).There were no defects in A-type granule morphology at any timepoint.These data suggest that PHS1 is required for normal B-type granule initiation but not for the correct number and morphology of A-type granules (Fig. 6).
Our finding that PHS1 is required solely for B-type granule initiation was further supported in a genetic approach.Reduced gene dosage of BGC1 affects the number of B-type granules, such as in the bgc1-1 mutant of durum wheat 'Kronos', which almost has no B-type granules due to a loss-of-function mutation in the BGC1-A1 homoeolog and a missense mutation in the BGC1-B1 homoeolog (Chia et al. 2020).We crossed this bgc1-1 line to the phs1-1 line and isolated a bgc1-1 phs1-1 quadruple mutant in the F2 generation.We also isolated from the cross a WT sibling control and bgc1-1 and phs1-1 mutant siblings as controls.We purified starch granules from mature grains of these lines and examined their morphology using SEM.Starch from the bgc1-1 and bgc1-1 phs1-1 mutants had visibly fewer B-type granules than the WT, WT sibling control, and phs1-1 mutant siblings (Fig. 7A).Coulter counter analysis of granule size distributions showed that the reduction of B-type granule number was stronger in bgc1-1 than in phs1-1 (Fig. 7B), since bgc1-1 had almost no detectable B-type granule peak (when plotted by volume), whereas phs1-1 had a B-type granule peak that was between that of the WT controls and bgc1-1.Importantly, the granule size distribution of bgc1-1 phs1-1 quadruple mutants was almost identical to bgc1-1.This suggests that the loss of PHS1 in the bgc1-1 background (which only has A-type granules) has no further effect on starch granule size distributions.Using curve fitting on these plots, we calculated that the size of A-type granules was unaffected by the phs1 mutations in the bgc1-1 background when compared to the WT sibling control (Fig. 7C).Next, we calculated the percentage of starch granules that were smaller than 10 µm.There was a significant decrease in the relative number of small granules in the bgc1-1 phs1-1 quadruple mutant and bgc1-1 mutant siblings compared to the WT controls (Fig. 7D).Even though the phs1-1 mutant sibling had lower percentage of smaller granules (as observed in Fig. 4), it did not come out as statistically significant from the WT controls in this experiment.However, since its granule size distribution was clearly different to the WT control in Fig. 7B, and comparable to those observed in previous experiments, the lack of statistical significance is most likely due to the smaller sample size analyzed when compared with our other experiments.Interestingly, like the phs1 double mutants, we observed that bgc1-1 also has a small reduction in amylose content (Table 2).Taken together, our results suggest PHS1 has an exclusive role in B-type granule initiation and does not appear to be required for A-type granule initiation.

Loss of PHS1 in wheat does not affect total MOS levels in the endosperm
Given recent evidence that PHS1 deficiency in potato and rice results in strong MOS accumulation in tubers and endosperm (Dong et al. 2023; Flores-Castellanos and Fettke 2023), we investigated whether there was MOS accumulation in the developing endosperm of wheat phs1 mutants and quantified the total soluble glucans in perchloric acid extracts.However, levels of MOS, soluble glucans, and other sugars were unaltered in the phs1 mutants.For all genotypes, soluble glucans were highest at 8 dpa and decreased in abundance during grain development but were always relatively low compared to starch content (∼10% of starch content at 8 dpa and <0.001% at 22 dpa) (Supplemental Fig. S5).There were no consistent differences between the phs1 mutants and their controls.We then quantified the methanol precipitable fraction of soluble glucans (phytoglycogen and long MOS) and calculated the nonprecipitable fraction (short MOS) as the difference between the total soluble glucan and precipitable glucans.Like the total glucans, these decreased in abundance as the grain developed, and there were no consistent differences between the phs1 mutants and controls (Supplemental Fig. S5).Additional high-performance anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD) analyses largely confirmed this result, since maltose and other detectable MOS did not change in abundance or chain length distribution pattern in the phs1 mutants (Supplemental Fig. S6).In addition, glucose, fructose, and sucrose levels remained unchanged in the mutants at all timepoints (Supplemental Fig. S6).

Loss of PHS1 does not affect starch granule number or starch turnover in leaves
To test whether the loss of PHS1 affected leaf starch metabolism, we first quantified the total starch content in leaves of seedlings.There was no significant difference in starch content between the WT and mutants, at either the end of day or end of night (Fig. 8A; Supplemental Fig. S7A).This indicates that PHS1 is not required for normal levels of starch accumulation and nocturnal starch turnover.To examine the number and morphology of starch granules in chloroplasts, we examined semithin sections of leaves using light microscopy (Fig. 8B; Supplemental Fig. S7B).There were no discernible differences in starch granule number per chloroplast in any of the genotypes.Like in Arabidopsis (Zeeman et al. 2004;Malinova et al. 2014), PHS1 appears to be dispensable for normal starch turnover in leaves.

Distinct mechanisms for A-and B-type granule initiation in wheat
After discovering PHS1 when screening for protein-protein interactions for BGC1 in wheat, we demonstrate that PHS1 is required for normal starch granule number and size distribution in the wheat endosperm, due to a specific role of the enzyme in B-type granule formation.We presented 3 lines of evidence to support this role: Firstly, mutants of durum wheat defective in phs1 had fewer but larger B-type granules, with no differences in the number or size of A-type granules (Figs. 4 and 5).Secondly, granule size distributions in the mutants were identical to those of the WT during the early stages of grain development but deviated from the WT at 18 dpa-the timepoint after B-type granules started forming in the WT (Figs. 5 and 6).Finally, the bgc1-1 phs1-1 double mutant had similar granule size distributions to the bgc1-1 mutant, suggesting that the further loss of PHS1 in a background that only has A-type granules has no effect on granule size distribution (Fig. 7).Thus, PHS1 appears to be only required for normal B-type granule formation in the wheat endosperm and not for A-type granule formation.
Based on our findings, we present a model where A-and B-type granule initiation occurs via distinct mechanisms involving different enzymes (Fig. 9).B-type granule initiation is fundamentally different to A-type granule initiation in that B-type granules initiate within amyloplasts that already contain an A-type granule.They can therefore be considered "secondary" granule initiations.The A-type granule can act as a source of substrates that can prime B-type granule formation, particularly as MOSs are released through the process of amylopectin trimming by isoamylase (ISA) (Mouille et al. 1996;Tickle et al. 2009).In contrast, A-type granules are the first granules to form within each amyloplast during early grain development and are more likely to require de novo primer formation.We propose that in wheat, PHS1 may only be important in the secondary B-type granule initiations, perhaps as it processes existing MOS substrates in the plastid in a manner that allows B-type granule initiation.
In this process, PHS1 may be assisted by BGC1.The number of B-type granules in wheat can be greatly reduced either through a loss of PHS1 function or a reduction in BGC1 function (Fig. 7) (Chia et al. 2020).BGC1 contains a glucanbinding CBM48 domain but no known enzymatic domain and so is likely to act by influencing the substrate binding of its interacting enzymes, as previously proposed for PTST2 in Arabidopsis (Seung et al. 2017).Since BGC1 interacts with PHS1 in wheat, it could influence the substrates available for further elongation by PHS1 (Fig. 1 and Table 1).However, it appears that BGC1 is more strictly required for B-type granule formation than PHS1.B-type granules are almost absent in the bgc1-1 mutant, whereas the phs1 mutant can still initiate some B-type granules (Fig. 7).It is very likely that this reflects the degree of redundancy in the function of both proteins.BGC1 is the only PTST2/3 ortholog in wheat and may be indispensable for B-type granule formation.PHS1 is the only plastidial phosphorylase, but it is not the only plastidial enzyme that can elongate MOS.Indeed, all SS isoforms can elongate MOS in vitro (Brust et al. 2013;Cuesta-Seijo et al. 2016) and may to some degree compensate for the loss of PHS1 function.Indeed, previous work in barley noted that mutants of SS1 have altered B-type granule numbers (Sparla et al. 2014).SS2a is also a candidate enzyme that requires more thorough investigation, as it was pulled down with BGC1 (Table 1).Although SS3 can in some circumstances initiate granules in Arabidopsis leaves (Szydlowski et al. 2009;Seung et al. 2016), it is unlikely to play a major role in B-type granule initiation, since SS3a mutants of wheat produce apparently normal B-type granules (Fahy et al. 2022).MOS can also be shortened/elongated by the disproportionating enzyme DPE1 (Critchley et al. 2001;Lütken et al. 2010;Hwang et al. 2016a) and shortened by amylolytic activities.On the one hand, such redundancy in MOS processing activity suggests that multiple enzymes can likely participate in B-type granule initiation in the absence of PHS1.On the other hand, it is remarkable that loss of PHS1 alone results in such drastic reductions in B-type granule number, despite the redundancy.Clearly, PHS1 is a critical player in B-type granule initiation.
Since the initiation of A-and B-type granules is specific to the Triticeae, the role of PHS1 in B-type initiation must be limited to species within the tribe.Interestingly, RNAi suppression of PHS1 in barley led to no noticeable effects on starch synthesis (Higgins et al. 2013).It is unclear whether these findings reflect differences in the requirement of PHS1 between barley and wheat or the incomplete silencing of PHS1 in barley.However, our results are consistent with evidence in other species that PHS1 can affect starch synthesis, although in wheat, the phs1 phenotype appears to be restricted to granule number and less conditional on environment or other mutations.For example, rice mutants deficient in PHS1 produced grains ranging from normal to shrunken with dramatic decreases in starch content, and the shrunken grains were more prevalent when plants were grown at low temperatures (Satoh et al. 2008).In Arabidopsis, an effect on granule number is conditional upon other mutations.Complete loss of PHS1 does not affect growth, starch turnover (Zeeman et al. 2004), or granule number per chloroplast (Malinova et al. 2014)-much like in leaves of our wheat mutants (Fig. 8).However, double mutants defective in PHS1 and other components of maltose metabolism have severe defects in growth and strong reductions in granule number per chloroplast (Malinova et al. 2014;Malinova and Fettke 2017).

Loss of PHS1 does not affect MOS levels but affects products of MOS extension
There is substantial biochemical evidence that PHS1 can act in both biosynthetic and degradative directions in vitro (Kruger and ap Rees 1983;Hwang et al. 2010Hwang et al. , 2016b) ) and has high affinity for soluble MOS in comparison to branched or crystalline substrates (Preiss et al. 1980;Steup and Schachtele 1981;Steup et al. 1983).Several roles of PHS1 have been proposed, including (i) the phosphorolytic degradation of soluble glucans released during starch degradation (Steup et al. 1983), (ii) the phosphorolytic degradation of linear MOS generated through amylopectin trimming during starch synthesis (Tickle et al. 2009), and (iii) the synthesis of MOS used for starch granule initiation (Malinova et al. 2014;Mérida and Fettke 2021).There is currently no genetic evidence that PHS1 is required for starch degradation in any species, whereas there is genetic evidence for a conditional role in starch synthesis in Chlamydomonas, rice, and Arabidopsis (Dauvillée et al. 2006;Satoh et al. 2008;Malinova et al. 2014).Our discovery that PHS1 is required for normal B-type granule initiation in wheat provides further support for a role of the enzyme in starch synthesis.
However, whether PHS1 fulfills its role in B-type granule initiation by acting in a biosynthetic or degradative direction is less clear.Given that elongation of MOS is thought to be a critical step in granule initiation (Nakamura 2015;Seung and Smith 2019;Mérida and Fettke 2021), the simplest scenario is that PHS1 acts in the biosynthetic direction to elongate MOS released from A-type granules.Once the MOS reaches a certain length, it can act as a substrate for branching enzyme, develop more structural complexity, and initiate the formation of a B-type starch granule.However, we cannot completely rule out that PHS1 acts in the degradative direction to support granule initiation, such as trimming longer MOS to a size that is most suitable for other enzymes to act on and prime B-type starch granule initiation.Distinguishing between these 2 possibilities will require further research.
Our results in wheat are in contrast to recent reports that lines deficient in PHS1 have substantial MOS accumulation in rice endosperm and potato tubers (Dong et al. 2023;Flores-Castellanos and Fettke 2023).In wheat endosperm, loss of PHS1 did not alter MOS levels (Supplemental Figs.S5 and S6).Although the total MOS pool was not affected, it appears that 2 processes that utilize MOS as a substrate were affected: granule initiation (as discussed above) and amylose synthesis.The precise reason why amylose content decreases in phs1-1, bgc1-1, and most likely in phs1-6 needs further investigation.Amylose is synthesized by GBSS and is thought to be primed by MOS such as maltotriose (Denyer et al. 1996(Denyer et al. , 2001;;Zeeman et al. 2002;Seung 2020).The phs1 double mutants consistently had lower amylose content than the WT but normal GBSS levels on starch (Table 2; Supplemental Fig. S3).However, the reduced amylose content was also observed in bgc1-1.It is possible that both PHS1 and BGC1 have a role in MOS metabolism that affects amylose priming.The fact that bgc1 (flo6) mutants of rice also have reduced amylose content supports this hypothesis (Peng et al. 2014;Zhang et al. 2021).The alternative possibility is that the lower amylose content of wheat phs1 and bgc1 mutants is a direct consequence of having fewer B-type granules.Further research is required to distinguish these 2 possibilities.

BGC1 can interact with multiple proteins
Our discovery that BGC1 interacts with PHS1 in wheat highlights yet another interaction partner for this protein, where the genetic evidence supports both partners acting in the same process (Fig. 1 and Table 1).In Arabidopsis, PTST2 interacts with SS4, but in the rice endosperm, it is reported to interact with ISA1, SS4, and GBSS (Peng et al. 2014;Seung et al. 2017;Zhang et al. 2021).None of these proteins were pulled down from wheat extracts, which suggests these interactions can be diverse among different species and organs.Diverse interactions could also explain how BGC1 plays multiple roles during grain development in wheat.For example, BGC1 also plays an important role in A-type granule formation during early grain development.Partial reduction in BGC1 function reduces the number of B-type granules but permits A-type granule formation (as in the bgc1-1 mutant), but complete elimination of BGC1 function results in the formation of compound or semicompound granules in place of normal A-type granules (Suh et al. 2004;Chia et al. 2020).These supernumerary initiations suggest that BGC1 is involved in establishing a single initiation in amyloplasts during granule initiation.It is likely that it fulfills this role with SS4, since the ss4 mutant also produces compound granules (Hawkins et al. 2021).
PHS1 itself has several interaction partners reported and participates in protein complexes.In wheat and maize (Zea mays) endosperm, PHS1 has been observed in complex with SBEI and SBEII (Tetlow et al. 2004;Subasinghe et al. 2014).Further, the rice PHS1 interacts with DPE1 (Hwang et al. 2016a).The interactions with BE and DPE1 greatly enhance the ability of PHS1 to make insoluble substrates (Nakamura et al. 2012(Nakamura et al. , 2017;;Hwang et al. 2016a).It is therefore of particular interest that both SBEI and DPE1 were pulled down with BGC1 (Table 1).We are currently dissecting the dynamics with which PHS1 interacts with BGC1 and these other interaction partners.

PHS1 as a target for starch modification in the Triticeae
Since PHS1 mutations did not affect plant growth or grain yield (Fig. 3), PHS1 represents another gene target to reduce the content of B-type granule content in wheat.Varieties with low B-type granules are desirable for bread making (Park et al. 2009).Recent work with BGC1 mutants showed that starch without B-type granules had higher water absorption, reduced grain hardness, and higher protein content (Saccomanno et al. 2022).B-type granules also cause processing problems due to their small size (Stoddard and Sarker 2000;Park et al. 2009).Further, reduced B-type granules are highly desired in barley for brewing, since B-type granules cause filtration problems in brewing and cause a starch haze (Stark and Lynn 1992).Since reduced B-type granules can only be achieved through BGC1 by reducing gene dosage rather than through knockouts, the phenotype is easier to achieve in a polyploid species like wheat than in a diploid species like barley.Thus, PHS1 could be a more suitable gene target for reducing B-type granules in diploid Triticeae as the phenotype can be achieved through a homozygous knockout mutation.

Plant materials and growth
Wheat (T.turgidum) plants were grown in soil (John Innes cereal mix-65% peat, 25% loam, 10% grit, 3 kg/m 3 dolomitic limestone, 1.3 kg/m 3 pg mix, and 3 kg/m 3 osmocote exact) in climate-controlled glasshouses for all grain analyses and in controlled environment rooms (CERs) for the analysis of starch in leaves.The glasshouses were set to provide a minimum 16 h of light at 20 °C and 16 °C during the dark.The CERs were set to provide 16 h light at 20 °C (light intensity was 300 μmol photons/m 2 /s obtained using fluorescent lamps and supplemented with light-emitting diode panels) and 8 h dark at 16 °C.N. benthamiana were grown in climatecontrolled glasshouses set to provide a minimum of 16 h light (200 μmol photons/m 2 /s) and a constant temperature (22 °C).All glasshouses and CERs were set to provide 60% relative humidity.

Isolation, cloning, and plasmid construction for PHS1 and BGC1
Total RNA was extracted from leaves of 2-wk-old wheat seedlings using RNeasy kit (Qiagen) with on-column DNase I digestion (Qiagen).cDNA was synthesized (using 2 µg RNA) using the Go Script Reverse Transcriptase Kit (Promega) following the manufacturer's instruction.The full-length wheat PHS1 cDNA sequence was amplified using gene-specific primers listed in Supplemental Table S2 and inserted into Gateway entry vector, pENTR, using the pENTR/D-TOPO kit (Invitrogen).To produce constructs with a C-terminal YFP or RFP tag, we recombined the coding sequences from PHS1:pENTR and BGC1:pDONR221 (Hawkins et al. 2021) into the plant expression vectors pB7YWG2 (35S promoter; C-terminal YFP) or pB7RWG2 (35S promoter; C-terminal RFP) using Gateway LR clonase (Invitrogen).To generate the chloroplast-targeted YFP and RFP proteins, the Rubisco small subunit (RbcS) transit peptide (Kim et al. 2010) was cloned into pDONR221 and recombined into pB7YWG2 and pB7RWG2.

Transient expression of PHS1 and BGC1 in N. benthamiana and protein localization
Constructs for YFP-tagged PHS1 and BGC1 expression were transformed into Agrobacterium tumefaciens (strain GV3101).Agrobacterium cells were grown at 28 °C in LB medium with the appropriate antibiotics.Cells were pelleted, washed, and resuspended in the infiltration medium (10 mM MES, 10 mM MgCl2, and 0.15 mM acetosyringone, pH 5.6).The cell suspension was then infiltrated into the intercellular spaces between abaxial epidermal cells of intact N. benthamiana leaves using a 1-mL plastic syringe as previously described (Kamble and Majee 2022).Infiltrated plants were incubated overnight in the dark followed by 2 d in a 16 h/8 h light-dark cycle.YFP fluorescence in leaf samples was observed using a laser scanning confocal microscope (SP8; Leica) with a 63× water immersion lens.Images were processed using LAS X software.

Pull-down assay, immunoprecipitations, and MS
For the pulldown assay to identify proteins associating with BGC1, the bait protein (recombinant His-tagged BGC1) was expressed in Escherichia coli BL21 ΔglgAP cells (Szydlowski et al. 2009) as described in Hawkins et al. (2021) and purified in its native state using Ni-NTA agarose beads (Qiagen) as previously described (Seung et al. 2013).To produce endosperm extract, endosperms were dissected from developing grains (collected at 18 dpa) of WT 'Kronos' plants and homogenized in ice-cold extraction medium (50 mM Tris-HCl, pH 8, 1 mM DTT, 1% [v/v] Triton X-100, 150 mM NaCl, and Roche Complete Protease Inhibitor cocktail) at a rate of 1 mL buffer per 100 mg tissue.Insoluble material was removed by centrifugation at full speed for 5 min at 4 °C, and proteins were collected in the supernatant.Recombinant BGC1-His protein (2.5 µg) was added to the supernatant (1 mL) and was incubated for 1 h at 4 °C.µMACS magnetic beads conjugated to anti-His (Miltenyi Biotec) were added and incubated for 1 h at 4 °C to retrieve the bait protein together with interacting proteins.The beads were captured with a µColumn on a magnetic stand (Miltenyi Biotec), washed 3 times with wash medium (50 mM Tris-HCl, pH 8, 1 mM DTT, 1% [v/v] Triton X-100, 300 mM NaCl, and Roche Complete Protease Inhibitor cocktail), and then 3 times with wash medium without Triton X-100, before eluting the bound proteins with elution medium (50 mM Tris-HCl, pH 6.8, and 2% [w/v] SDS).
The eluted proteins were precipitated with chloroform/ methanol (Pankow et al. 2016).Protein pellets were resuspended in 50 µL of 2.5% sodium deoxycholate (SDC; Merck) in 0.2 M EPPS-buffer (Merck), pH 8.5, and reduced, alkylated, and digested with trypsin in the SDC buffer according to standard procedures.After the digest, the SDC was precipitated by adjusting to 0.2% trifluoroacetic acid (TFA), and the clear supernatant subjected to C18 SPE.Samples were dried in a SpeedVac concentrator (Thermo Fisher Scientific, #SPD120), and the peptides dissolved in 0.1% TFA/3% acetonitrile.
Peptides were analyzed by nano-LC-MS/MS on an Orbitrap Eclipse Tribrid mass spectrometer with a FAIMS Pro Duo source, coupled to an UltiMate 3000 RSLCnano LC system (Thermo Fisher Scientific, Hemel Hempstead, UK).The samples were loaded and trapped using a trap cartridge (Pepmap Neo C18, 5 µm, 300 µm × 5 mm, Thermo) with 0.1% TFA at 15 µL/min for 3 min.The trap was then switched in-line with the analytical column (nanoEase M/Z column, HSS C18 T3, 100 Å, 1.8 µm; Waters, Wilmslow, UK) for separation using the following gradient of solvents A (water, 0.1% formic acid) and B (80% acetonitrile, 0.1% formic acid) at a flow rate of 0.2 µL/min: 0 to 3 min 3% B (during trapping); 3 to 10 min linear increase B to 7%; 10 to 100 min increase B to 32%; 100 to 148 min increase B to 50%; followed by a ramp to 99% B and reequilibration to 3% B, for a total running time of 180 min.MS data were acquired with the FAIMS device set to 3 compensation voltages (−35, −50, and −65 V) at standard resolution for 1 s each with the following MS settings in positive ion mode: MS1/OT: resolution 120 K, profile mode, mass range m/z 300 to 1,800, spray voltage 2,800 V, AGC 4e 5 , and maximum injection time 50 ms; MS2/IT: data-dependent analysis was performed using higher energy collisional dissociation fragmentation with the following parameters: cycle time of 1 s in IT turbo for each FAIMS CV, centroid mode, isolation window 1.0 Da, charge states 2 to 5, threshold 1.0e 4 , collision energy = 30, normalized automatic gain control target 100%, maximum inject time set to auto, dynamic exclusion 1 count, 15 s exclusion, and exclusion mass window ±10 ppm.
The acquired raw data were processed and quantified in Proteome Discoverer 3.0 (Thermo) using the incorporated search engine CHIMERYS (MSAID Munich, Germany).The processing workflow included recalibration of MS1 spectra (RC) and the Minora Feature Detector for quantification with minimum trace length = 7 and signal/noise threshold = 3.The Top N Peak Filter (10 per 100 Da) was applied, and the CHIMERYS search was performed with the prediction model inferys_2.1_fragmentation,enzyme trypsin with tolerance 0.5 Da, variable modification oxidation (M), and fixed modification carbamidomethyl (C).Percolator was used for validation using q-value and FDR 0.01 (strict) and 0.05 (relaxed).
In the consensus, workflow quantification was performed with a maximum retention time shift of 3 min and a mass tolerance of 4 ppm between runs.Protein quantification was based on the top 3 most abundant unique peptides per protein group.Missing values were replaced by low abundance resampling.Protein abundance ratios were calculated from the 3 replicates per sample.The hypothesis test was performed by a background-based t test, and the P-values adjusted according to Benjamini-Hochberg.

Native gel analysis of PHS activity
The visualization of α-glucan phosphorylase activity by native PAGE was carried out using the method of Zeeman et al. (2004) with minor modifications.Leaf samples were homogenized in extraction medium (100 mM MOPS, pH 7.2, 1 mM DTT, 1 mM EDTA, 10% [v/v] ethanediol, and Complete protease inhibitor cocktail [Roche]) and then spun at 20,000 × g for 10 min.Soluble proteins were collected in the supernatant, and the protein concentration of the extracts was determined using the Bradford protein assay.Equal amounts of protein (∼100 to 180 µg) were loaded per lane on native PAGE gels (7.5% acrylamide and 0.3% oyster glycogen in resolving gel; 3.75% acrylamide in stacking gel) and were electrophoresed at 100 V for 4 h at 4 °C.Gels were washed twice in 100 mM Tris-HCl, pH 7.0, and 1 mM DTT and then incubated in 100 mM Tris-HCl, pH 7.0, 1 mM DTT, and 50 mM Glc-1-P overnight at 20 °C.Activity bands were stained with Lugol's iodine solution (L6146, Sigma, St. Louis).

Grain morphometrics
Grain yield per plant was quantified as the total weight of grains harvested from each plant.The thousand grain weight and grain size were quantified using the MARViN seed analyzer (Marvitech GmbH, Wittenburg).

Starch content of leaves and mature grains
Starch content of leaves and mature grains (in glucose equivalents) were quantified as described in Hawkins et al. (2021).Briefly, for leaves, 2-wk-old seedlings were harvested at the base of the lowest leaf and flash frozen in liquid N 2 and then homogenized in 0.7 M perchloric acid.Insoluble material was collected by centrifugation, washed 3 times in 80% ethanol, and then resuspended in water.Starch was digested using α-amylase/amyloglucosidase (Roche, Basel), and the released glucose was assayed using the hexokinase/glucose-6phosphate dehydrogenase assay (Roche).For mature grains, flour (5 to 10 mg) was dispersed in 100 mM sodium acetate buffer, pH 5, and the starch was digested with thermostable α-amylase at 99 °C for 7 min.Amyloglucosidase was added to the digestion and was further incubated at 50 °C for 35 min.Both the thermostable α-amylase and amyloglucosidase were from the Total Starch Assay kit (K-TSTA, Megazyme, Bray).The digested sample was centrifuged to remove insoluble material, and glucose was measured in the supernatant as for the leaf starch quantification.

Starch purification and determination of granule morphology, size distribution, and amylose content
For starch purification from mature grains, grains (3 to 5 grains per extraction) were soaked overnight in ddH 2 O at 4 °C and then homogenized in a mortar and pestle with excess ddH 2 O.The homogenates were filtered through a 70-µm nylon mesh and then centrifuged at 3,000 × g, 5 min, before resuspending the starch pellet in water.The starch suspension was centrifuged at 2,500 × g, 5 min on a cushion of 90% (v/v) Percoll, 50 mM Tris-HCl, pH 8.The pellet was washed twice in 50 mM Tris-HCl, pH 6.8, 10 mM EDTA, 4% SDS (v/v), and 10 mM DTT and then twice in ddH 2 O, before finally resuspending in ddH 2 O.
The morphology of purified granules was examined using a Nova NanoSEM 450 (FEI, Hillsboro) scanning electron microscope.To quantify starch granule size distributions, the purified starch was suspended in Isoton II electrolyte solution (Beckman Coulter, Indianapolis), and particle sizes were measured using a Multisizer 4e Coulter counter (Beckman Coulter) fitted with a 70-µm aperture tube.A minimum of 100,000 particles were measured per sample.These data were used to produce relative volume vs. diameter and relative number vs. diameter plots.A mixed bimodal distribution (2 log-normal distributions) was fitted to the relative volume vs. diameter plots (Python script available at https://github.com/DavidSeungLab/Coulter-Counter-Data-Analysis) to calculate mean diameters of A-and B-type granules and the B-type granule volume percentage (defined as volume occupied by B-type granules as a percentage of the total volume of starch).
Amylose content of the purified starch granules was estimated using an iodine colorimetry method (Washington et al. 2000).Briefly, starch (1 mg) was dispersed overnight at room temperature in 1 M NaOH.The solution was neutralized to pH 7 using 1 M HCl, and 5 µL of this solution was diluted in 220 µL of water and 25 µL Lugol's iodine solution (L6146, Sigma, St. Louis).The reaction was incubated at room temperature for 10 min prior to absorbance measurements at 535 and 620 nm.Apparent amylose content was estimated using the formula: Apparent amylose content = 1.4935 * exp [2.7029 * (Absorbance 620/Absorbance 535)] .

Quantification of starch, sugars, MOS, and granule number and size in developing grains
Endosperms were dissected from developing grains collected at 8, 14, 18, and 22 dpa.In each extraction, 3 to 5 individual endosperms of known fresh weight were used.After homogenization in 0.7 M perchloric acid, homogenates were spun at 10,000 × g for 5 min, and the supernatant was immediately neutralized using neutralization buffer (2 M KOH and 400 mM MES).This neutralized soluble fraction was used for sugar and MOS quantification (see below).
The pellet was resuspended in ddH 2 O and equally divided into 2 fractions.One fraction was used to quantify starch content following the method described above for leaves.The other was used for starch purifications (as described above for mature grains), and all granules purified within this fraction were resuspended in known volumes of Isoton II electrolyte solution (Beckman Coulter, Indianapolis).The suspension was analyzed in a Multisizer 4e Coulter counter (Beckman Coulter) fitted with a 70-µm aperture tube and running in volumetric mode (analyzing 2 mL of the suspension).This gave the number of granules in the suspension, which could be used to calculate the number of granules per starting fresh weight of endosperm, as well as granule size distribution plots.
For the quantification of soluble glucans, total soluble glucans (neutralized soluble fraction without precipitation) and methanol-precipitable soluble glucans (after precipitation) were quantified.For methanol precipitation, 1 volume of neutralized soluble fraction was mixed with 4 volumes of pure methanol, mixed, and incubated overnight at −20 °C.Precipitated glucans were collected by centrifugation at 10,000 × g, 5 min, then washed with 75% methanol, and dried.Dry pellets were suspended in 1 volume of ddH 2 O.To quantify glucans, the neutralized soluble fraction and resuspended precipitated glucans were digested using α-amylase/amyloglucosidase (Roche, Basel), and released glucose was assayed using the hexokinase/glucose-6-phosphate dehydrogenase assay (Roche).
HPAEC-PAD was used to quantify glucose, fructose, sucrose, and maltose and to visualize MOS accumulation patterns.The neutralized soluble fraction was purified on sequential columns of Dowex 50W and Dowex 1 (Sigma) as described previously (Seung et al. 2013).Purified samples were separated on an ICS-5000 HPLC fitted with a CarboPac PA20 column (3 × 250 mm, CV = 1.06 mL; Dionex).The mobile phase consisted of eluate A (100 mM NaOH) and eluate B (150 mM NaOH and 500 mM sodium acetate), following a gradient program of the following: 0 to 7 min, 0% B; 7.0 to 26.5 min, a concave gradient to 80% B; 26.5 to 32.0 min, 80% B; 32.0 to 32.1 min, linear gradient to 0% B; and 32.1 to 40.0 min, 0% B with flow rate of 0.25 mL/min.spectrometry, JIC Bioimaging for providing access to microscopes, JIC Genotyping for providing support for KASP genotyping, and JIC Chemistry for providing access to HPAEC-PAD instruments.

Figure 1 .Figure 2 .
Figure 1.Interaction between wheat PHS1 and BGC1 in N. benthamiana.A) Plastidial localization of YFP-tagged PHS1 and BGC1 transiently expressed in N. benthamiana leaves.Bars = 10 µm.B) Pairwise IP of PHS1:YFP and BGC1:RFP coexpressed in N. benthamiana leaves, using anti-YFP beads.Input and IP samples were blotted with YFP (top panels) and RFP antibodies (bottom panels).Chloroplast-targeted YFP (cTP:YFP) and RFP (cTP:RFP) were used as controls to exclude unspecific binding to the fluorescent protein tags.

Figure 3 .
Figure 3. Plant and grain phenotypes of phs1 mutants.The WT sibling control (AA BB), single mutants (aa BB and AA bb), and the double mutant (aa bb) from phs1-1 and phs1-6 lines were compared with the 'Kronos' WT.A) Photograph of plants at maturity.Bar = 5 cm.B) Grain yield per plant.C) Photographs of grains showing both dorsal and ventral sides, as well as a cut section through the middle of the grain.Bar = 1 cm.D) Thousand grain weight.E) Grain size measured as 2D area.For B), D), and E), data represent the mean ± SEM from n = 15 to 16 plants.Values with different letters are significantly different under a 1-way ANOVA and Tukey's post hoc test at P < 0.05.

Figure 4 .
Figure 4. Endosperm starch from the phs1-1 mutant has fewer and larger B-type granules.A) Scanning electron micrographs of purified endosperm starch.Bar = 10 µm.B) Granule size distributions were determined using a Coulter counter, and the data were expressed as relative % volume (of total starch) vs. granule diameter plots.C, D) B-type granule volume (% of total starch) and the average diameter of B-type granules were extracted from the relative volume vs. diameter plots by fitting a bimodal mixed normal distribution.E) Same as B) but expressed as relative starch granule number (% of total granule number) vs. diameter plots.F) The percentage of small granules by number (smaller than 10 µm) was calculated from the Coulter counter data.G) The average diameter of A-type granules was calculated from the relative volume vs. diameter plots, as for C) and D).For B to G), plots show the mean from the analysis of n = 4 to 6 replicate starch extractions, each from grains from a separate plant.The shading (on B and E) and error bars (on C, D, F, and G) represent the SEM.Values with different letters are significantly different under a 1-way ANOVA with Tukey's post hoc test (P < 0.05).

Figure 5 .
Figure 5. Loss of PHS1 affects granule number in midgrain development but not total starch content.The endosperm was dissected from developing grains of WT, phs1-1 and phs1-6 double mutants, and corresponding WT controls, harvested at 8, 14, 18, and 22 dpa, with n = 3 individual plants for each genotype per timepoint.A) Starch content of the endosperm.Values are expressed relative to the fresh weight of the dissected endosperm.B) Starch granule number in the endosperm.Starch was purified from dissected endosperm, and the number of granules was determined using a Coulter counter running in volumetric mode (analyzing 2 mL of the suspension).Values are expressed relative to the fresh weight of the dissected endosperm.Error bars represent the ±SEM.Values with different letters are significantly different under a 1-way ANOVA with Tukey's post hoc test (P < 0.05).

Figure 6 .
Figure 6.PHS1 affects granule size distributions during midgrain development.The endosperm was dissected from developing grains of WT, phs1-1 and phs1-6 double mutants, and corresponding WT controls, harvested at A and E) 8 dpa, B and F) 14 dpa, C and G) 18 dpa, and D and H) 22 dpa, with n = 3 individual plants for each genotype per timepoint.A to D) Granule size distributions were analyzed on the Coulter counter, and the data were expressed as relative % volume (of total starch) vs. granule diameter plots.The shading represents the ±SEM.E to H) Starch granule morphology observed using SEM.Bars = 15 µm.

Figure 7 .
Figure 7. Loss of PHS1 does not affect granule size distribution in the bgc1-1 mutant.A) Scanning electron micrographs of purified endosperm starch.Bar = 15 µm.B) Granule size distributions were determined using a Coulter counter, and the data were expressed as relative % volume (of total starch) vs. granule diameter plots.C) A-type granule diameter extracted from the relative volume vs. diameter plots by fitting a bimodal mixed normal distribution, except for genotypes with bgc1-1, where a unimodal distribution was fitted.D) The percentage of small granules by number (smaller than 10 µm) was calculated from the Coulter counter data.For B to D), plots show the mean from the analysis of n = 3 replicate starch extractions, each from grains from a separate plant.The shading (in B) and error bars (in C and D) represent the ±SEM.Values with different letters are significantly different under a 1-way ANOVA with Tukey's post hoc test (P < 0.05).

Figure 8 .
Figure 8. Leaf starch content of phs1-1 mutants.The WT segregant control (AA BB), single mutants (aa BB and AA bb), and double mutants (aa bb) were compared with the 'Kronos' WT.A) Total starch content.Seedlings were grown for 2 wk under 16 h day/8 h night and harvested at the end of day (ED) and end of night (EN).Values are the mean ± SEM from n = 5 plants (each represented by a data point), and those with different letters are significantly different under a 1-way ANOVA and Tukey's post hoc test at P < 0.05.B) Light micrographs of leaf sections stained with periodic acid-Schiff (PAS) staining to visualize starch granules.Leaf segments were harvested halfway along the length of the older blade in 2-wk-old seedlings at the ED, prior to fixation and embedding.Bars = 10 µm.

Figure 9 .
Figure 9. Model of PHS1 action in B-type starch granule initiation in wheat endosperm.During wheat grain development, a single A-type granule initiates in each amyloplast during early grain development by around 6 to 8 dpa and grows.B-type granules initiate later in grain development, at least partially in stromules and vesicle-like amyloplasts at around 15 to 20 dpa.We propose PHS1 acts on MOSs, likely released from A-type granules by ISA, to initiate B-type granules.This process is disrupted in mutants lacking a functional PHS1 protein, resulting in fewer initiations of B-type starch granules.Fewer B-type granules means each one has a greater share of substrates for granule growth, leading to larger B-type granules in the mutant at grain maturity.

Table 1 .
Starch metabolic proteins significantly enriched in the BGC1 pulldown

Table 2 .
Starch and amylose content of phs1 mutantsStarch content was quantified relative to the dry weight of whole flour.Values are the mean ± SEM on grains from n = 3 to 4 different plants.Amylose content was measured on purified starch.Values are the mean ± SEM on starch extracted from n = 3 to 5 different plants.Values with different letters are significantly different under a 1-way ANOVA and Tukey's post hoc test at P < 0.05.Note that the plants for the 2 experiments were grown independently in different glasshouses, and therefore, the statistical analyses compare genotypes within each experiment.