Unprocessed wheat γ‐gliadin reduces gluten accumulation associated with the endoplasmic reticulum stress and elevated cell death

Summary

• Along with increasing demands for high yield, elite processing quality and improved nutrient value in wheat, concerns have emerged around the effects of gluten in wheat‐based foods on human health. However, knowledge of the mechanisms regulating gluten accumulation remains largely unexplored.

• Here we report the identification and characterization of a wheat low gluten protein 1 (lgp1) mutant that shows extremely low levels of gliadins and glutenins.

• The lgp1 mutation in a single γ‐gliadin gene causes defective signal peptide cleavage, resulting in the accumulation of an excessive amount of unprocessed γ‐gliadin and a reduced level of gluten, which alters the endoplasmic reticulum (ER) structure, forms the autophagosome‐like structures, leads to the delivery of seed storage proteins to the extracellular space and causes a reduction in starch biosynthesis. Physiologically, these effects trigger ER stress and cell death.

• This study unravels a unique mechanism that unprocessed γ‐gliadin reduces gluten accumulation associated with ER stress and elevated cell death in wheat. Moreover, the reduced gluten level in the lgp1 mutant makes it a good candidate for specific diets for patients with diabetes or kidney diease.

Introduction

Bread wheat (Triticum aestivum L.) is one of the most commonly consumed staples worldwide, with widespread uses in foods, such as breads, steamed breads, noodles, cakes and cookies (Veraverbeke & Delcour, ²⁰⁰²). Wheat represents an important source of energy and provides significant amounts of protein, dietary fibre, B vitamins and mineral micronutrients (Shewry & Hey, ²⁰¹⁵). Thus, there is increasing interest in improving wheat‐based foods for human health (Hazard et al., ²⁰²⁰).

Gluten proteins (i.e. gliadins and glutenins) confer the unique viscoelastic properties of wheat dough (Payne et al., ¹⁹⁸⁷). Notwithstanding its essential contribution to wheat end‐use quality, gluten is associated with several digestive disorders and gluten‐related pathologies. For example, gliadins are major allergens associated with coeliac disease, an autoimmune‐related disease that causes inflammation in the small intestine (De Re et al., ²⁰¹³). Glutenins, especially high‐molecular‐weight glutenins, have poor digestibility because they are composed of many long‐chain amino acids and are enriched in proline (15%) and glutamine (35%), which limits their proteolysis and complete degradation by gastrointestinal enzymes (Heredia‐Sandoval et al., ²⁰¹⁶). These properties make foods with high glutenin or gluten levels unsuitable for patients with diabetes or kidney disease (De Angelis et al., ²⁰¹⁰).

For those with gluten‐related disorders, such as coeliac disease, the only effective ‘treatment’ is a lifelong gluten‐free diet, achieved by eliminating grains such as wheat, barley, rye and triticale and substituting them with gluten‐free cereals such as rice, maize, millet and quinoa (El Khoury et al., ²⁰¹⁸). However, this strategy results in products with low micronutrient levels (e.g. B vitamins, iron and dietary fibre) compared to their wheat‐containing counterparts (do Nascimento et al., ²⁰¹³). Thus, the use of gluten‐free or low‐gluten products based on wheat and the development of wheat cultivars with little or no gluten could greatly benefit human health.

Wheat gluten proteins are first synthesized on polyribosomes attached to the rough endoplasmic reticulum (RER) and the N‐terminal signal peptide (SP) is required for their passage into the endoplasmic reticulum (ER) lumen (Tosi et al., ²⁰⁰⁹). After SP cleavage, two gluten protein trafficking pathways occur in wheat. In the first, the gluten proteins are transported via the Golgi apparatus into the protein storage vacuoles (PSVs) (Kim et al., ¹⁹⁸⁸). In the second pathway, the gluten proteins are assembled within the ER to form the protein bodies (PBs), which are subsequently internalized into the PSVs (Levanony et al., ¹⁹⁹²; Tosi et al., ²⁰⁰⁹). At present, the precise molecular and cellular mechanisms regulating gluten protein trafficking and deposition in wheat remain to be fully elucidated.

Here, we identified an ethyl methanesulfonate (EMS)‐induced wheat mutant (low gluten protein 1 (lgp1)) showing strong reductions of both glutenin and gliadin levels, due to a missense mutation in the SP sequence of a single γ‐gliadin gene. The presence of mutated γ‐gliadin bearing the unprocessed peptide leads to accumulation of an excessive amount of unprocessed γ‐gliadin, altered ER structure, delivery of seed storage proteins (SSPs) to the extracellular space, as well as reduced levels of gluten and starch biosynthesis. In addition, the lgp1 mutant exhibits features that make it a good candidate for specific healthy diets, especially for patients with diabetes or kidney disease.

Materials and Methods

More detailed information about the Materials and Methods reported below and additional methods can be found in Supporting Information Methods S1.

Plant materials and growth conditions

M₃ seeds of an EMS‐treated wheat population were used for mutant screening. The wild‐type (WT) was synthetic hexaploid wheat line 3672 (pedigree: Gy/C85) with the high‐molecular‐weight glutenins (HMW‐GSs) (1, 7 + 9, 4 + 12). Zhoumai16 (ZM16) is a common wheat cultivar with the HMW‐GSs (N, 7 + 9, 2 + 12). Wheat plants were grown in Shangzhuang, Beijing, China, for phenotypic analyses (116°E, 40°N).

Phenotypic analysis

Gliadins and glutenins were extracted and separated as previously described (Gil‐Humanes et al., ²⁰¹²). Reversed‐phase high‐performance liquid chromatography (RP‐HPLC) and quality analysis were conducted as previously described (Chen et al., ²⁰¹⁹). Thousand‐grain weight, grain length and grain width were determined using a camera‐assisted phenotyping system (Wanshen Detection Technology, Hangzhou, China). Starch content was determined using a Megazyme Total Starch (AA/AMG) Assay Kit (Megazyme, Bray, Wicklow, Ireland). Starch was extracted as previously described (Liu et al., ²⁰⁰⁷) and images were taken with an Hitachi S‐3400N scanning electron microscope (Tokyo, Japan). Three biological replications were performed.

Map‐based cloning

For bulked segregant RNA‐sequencing (BSR‐seq), RNA of 30 homozygous mutant and normal bulks from ZM16×lgp1 were sequenced on the Illumina NovaSeq 6000 platform. For genome resequencing, genomic DNAs of ZM16 and lgp1 were sequenced with 6× coverage using a NovaSeq 6000 platform. High‐quality reads were aligned to a wheat reference genome (IWGSC RefSeq v.2.0) using Star (Dobin et al., ²⁰¹³). Variant calling was performed using the HaplotypeCaller module. In total, 152 F₂ plants and a larger F₂ population of 8746 plants were used for mapping. Single nucleotide polymorphisms (SNPs) and INDELs were used to develop molecular markers (Dataset S1). Linkage analysis was performed using JoinMap 4.0.

Signal peptide identification

The amino acid sequences of typical SSPs from maize, sorghum and wheat were identified in the NCBI database. The SPs of these sequences were predicted using the SignalP5.0 Server.

Production of transgenic plants

The entire LGP1 coding region and its putative native promoter (2.4 kb upstream of the start codon) from lgp1 (Fig. S4) was inserted into the pWMB110 vector (Liu et al., ²⁰²⁰). The construct vector was transformed in wheat cv Fielder via Agrobacterium‐mediated (EHA105) transformation (Ishida et al., ²⁰¹⁵). All primers are listed in Dataset S1.

Light microscopy

Transverse sections of fresh seeds at 10, 12, and 18 d after pollination (DAP) were fixed in 2.5% (v/v) glutaraldehyde and then fixed in 0.5% (v/v) OsO₄. After being dehydrated and embedded in resin (Yu et al., ²⁰¹⁷), samples were cut with a tissue slicer (RM2265; Leica, Wetzlar, Germany), stained with 0.03 M toluidine blue and observed via a DMLS light microscope (Leica). Three individual replicate seeds were used.

Transmission electron microscopy (TEM)

For conventional chemical fixation (CCF), transverse sections of 12 DAP fresh seeds were fixed in 0.5% (v/v) glutaraldehyde and 3% (w/v) paraformaldehyde and then fixed in 2% OsO₄, dehydrated, embedded in LR white resin (London Resin, Berkshire, UK) and cut with a tissue slicer (RM2265; Leica). For high‐pressure freezing (HPF), 12 DAP fresh seeds were frozen in an HPF apparatus (EM PACT2; Leica); infiltration with Lowicryl HM20 (Electron Microscopy Sciences, Hatfield, PA, USA), embedding and UV polymerization were performed stepwise at −35°C. For immunogold localization, sections were blocked with 1% BSA for 30 min, incubated with primary and secondary antibodies for 2 and 2 h at room temperature, respectively, double stained with 2% (w/v) uranyl acetate and 2.5% (w/v) lead citrate and examined under a Hitachi H‐7600 TEM. Three individual replicate seeds were performed.

Antibodies

Antibodies of anti‐low‐molecular‐weight glutenins‐glutenins (LMW‐GSs), HMW‐GSs and γ‐gliadins were synthesized as previously described (Loussert et al., ²⁰⁰⁸) and used at 1 : 100 for immunogold labelling. Anti‐BiP (AS09 481; Agrisera, Vännäs, Sweden), and gold‐conjugated anti‐rabbit IgG (10 nm, EM.GAR10; BBInternational, Cardiff, UK) were used at 1 : 100 for immunogold labelling. Anti‐ATG8 (AS14 2769; Agrisera), anti‐GFP (HT801‐01; TransGen Biotech, China), anti‐BiP (AS09 481; Agrisera), anti‐eEF1α (AS10 934; Agrisera), anti‐actin (A0480; Sigma), anti‐mouse IgG (BE0102‐100; Easybio, China) and anti‐rabbit IgG (BE0101‐100; Easybio) antibodies were all used at 1 : 5000 for immunoblot analysis.

Immunoblot analysis

Seeds of the WT and lgp1 at different development stages were collected, and protein extraction and immunoblot analysis were performed as previously described (Gao et al., ²⁰²¹). Three biological replicates were performed.

Immunoprecipitation assay

Total proteins were extracted from wheat leaf protoplasts using lysis buffer (50 mM Tris–HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA (pH 8.0), 0.1% NP‐40, 1% Triton X‐100, 0.6 mM PMSF and 20 μM MG132 with protease inhibitor cocktail; Roche). Lysates were incubated at 4°C for 30 min and then centrifuged at 13 000  g for 20 min. The supernatants were incubated with 30 μl anti‐green fluorescent protein (anti‐GFP) magnetic beads (Easybio) at 4°C for 2 h. Immunoprecipitated samples were separated in sodium dodecyl sulphate–polyacrylamide gel electrophoresis (SDS‐PAGE) gels and detected by anti‐GFP (HT801‐01; TransGen Biotech) antibody. Three biological replicates were performed.

Transient expression in protoplasts and confocal imaging

LGP1‐C‐GFP (WT), LGP1‐T‐GFP (lgp1) and LGP1‐noSP‐GFP (without SP), the ER marker RFP‐HDEL and the autophagy marker mCherry‐ATG8e were coexpressed in wheat protoplast cells according to the method previously described (Shan et al., ²⁰¹⁴). Fluorescence signals were observed using a Zeiss LSM 880 laser scanning confocal microscope. The GFP signal was excited by a 488 nm laser, while the mCherry/red fluorescent protein (RFP) signal was excited by a 552 nm laser. Three biological replicates were performed. All primers are listed in Dataset S1.

Yeast two‐hybrid assay

The LGP1 coding sequence (without SP) was fused to pGBKT7 as a bait. Coding sequences (without SPs) of other SSPs were fused to pGADT7 as preys. All primers are listed in Dataset S1.

Luciferase complementation image assay

The coding sequences of LGP1 (without SP) and the other four SSPs (without SPs) were separately cloned into JW771 (NLUC) and JW772 (CLUC) and coexpressed in Nicotiana benthamiana leaves. All primers are listed in Dataset S1. The LUC signals were collected using NightShade LB 985 (Berthold Technologies, Bad Wildbad, Germany).

RNA‐seq

Total RNA was extracted from 15, 18, 23 and 28 DAP WT and lgp1 endosperm. RNA‐seq libraries were prepared with a TruSeq RNA Sample Preparation Kit v.2 (Illumina, San Diego, CA, USA) and sequenced on a NovaSeq 6000 platform. After screening and trimming, clean reads were aligned to the wheat reference genome (IWGSC RefSeq v.1.1) using Star (Dobin et al., ²⁰¹³). DESeq2 v.1.24.0 was used for differential expression gene analysis (Love et al., ²⁰¹⁴) and genes with log₂(fold change) ≥ 1 and false discovery rate (FDR) < 0.05 were identified as differentially expressed genes. Three biological replicates were performed.

TMT proteasome analysis

Wild‐type and lgp1 15 DAP endosperm was collected. Tandem mass tag (TMT) proteasome analysis was conducted by Beijing Protein Innovation Co. Ltd (http://www.proteomics.org.cn/). Proteins displaying a 1.2‐fold change between the WT and lgp1 (P < 0.05) were consider to be differentially expressed proteins. Three biological replicates were performed.

Metabolic profiling

Endosperm from the WT and lgp1 15, 20 and 35 DAP was pulverized using a TissueLyser II (Qiagen) at 29 Hz for 1 min. The metabolites were extracted and analysed using an LC‐electrospray ionization (ESI)‐MS/MS system as previously described (Chen et al., ²⁰¹³). Three biological replicates were performed.

Viability staining

Median longitudinal and cross‐sections of 8, 12, 16 and 20 DAP fresh kernels were incubated in 0.1% (w/v) Evans blue (Coolaber, China) for 2 min, washed with water and photographed. Three biological replicates were performed.

Measurement of ABA and ROS contents

The abscisic acid (ABA) level was determined by Wuhan Greensword Creation Technology Co. (Wuhan, China) via LC‐MS/MS analysis (Huang et al., ²⁰¹²). The reactive oxygen species (ROS) content was measured using a plant ROS enzyme‐linked immunosorbent assay (ELISA) determination kit (Misbio, Wuhan, China). Three biological replicates were performed.

Determination of wheat bran, dietary fibre and gluten contents

Wheat bran content was calculated as the flour weight subtracted from the grain weight, divided by grain weight. Dietary fibre content was analysed using an Integrated Total Dietary Fibre Assay Kit (K‐INTDF; Megazyme). Gliadin immunoreactive peptide content was determined using an R5 Ridascreen^® gliadin competitive ELISA (R‐Biopharm, Darmstadt, Germany). Three biological replicates were performed.

Cooking of cookies and breads

For cooking of cookies, 200 g of flour (or starch), 160 g of butter, 80 g of sugar, 1.67 g of baking powder, 1.67 g of salt and one egg white were stirred evenly and the cookies were squeezed and baked at 180°C for 15 min. For bread cooking, 100% flour (with or without 2% hydroxypropyl methylcellulose (HPMC)), 1.8% yeast, 1.5% salt, 6% sugar, 4% skimmed milk powder, 3% shortening and water were mixed for 15 min, fermented for 90 min, left to proove in a mould for 45 min and baked at 210°C for 20 min.

Results

The lgp1 mutant accumulates low levels of seed storage proteins

To identify gluten‐free or low‐gluten wheat mutants, the hexaploid wheat line 3672 was mutagenized with EMS to achieve an EMS‐mutant population of 1640 lines. Mature seeds from this population were screened via SDS‐PAGE and acid‐PAGE to detect glutenin and gliadin accumulation, respectively. Mature seeds of one mutant showed extremely low levels of gliadins and glutenins, including both HMW‐GSs and LMW‐GSs, compared to WT plants (Fig. 1a). Hence, we named this mutant low gluten protein 1 (lgp1). The reduced levels of gluten proteins in lgp1 were confirmed by RP‐HPLC. Compared to the WT, lgp1 showed a strong decrease of HMW‐GSs, ω‐gliadins, α/β‐gliadins and γ‐gliadins (Fig. S1a). In WT seeds, HMW‐GSs started to accumulate at 10 DAP and continued until seed maturation, whereas, at the same developmental stages, HMW‐GS levels were quite low in lgp1 seeds (Fig. 1b). Consistent with the low glutenin and gliadin levels, dough made from lgp1 flour did not produce any gluten (Fig. 1c). In addition, quality parameters (reflecting gluten strength) including SDS sedimentation volume, dough development time and dough stability time were lower in lgp1 compared to the WT (Dataset S2; Fig. S1b–e). These results indicate that lgp1 is a low‐gluten wheat mutant with low contents of both glutenins and gliadins.

Fig. 1.

lgp1 showed a decreased gluten protein level and defective protein body morphology compared to the wild‐type (WT). (a) Glutenin and gliadin levels in the WT and lgp1 were detected by SDS‐PAGE (left) and acid‐PAGE (right), respectively, followed by Coomassie blue staining (CBS). HMW‐GSs and LMW‐GSs represent high‐ and low‐molecular‐weight glutenin subunits, respectively; B, C, and D represent the three LMW‐GS types; α, β, γ and ω represent the four gliadin types. Fifty milligrams of whole wheat flour was used for glutenin and gliadin extraction for each sample. (b) HMW‐GS level in WT and lgp1 seeds harvested at different developmental stages was analysed by SDS‐PAGE followed by CBS. DAP, days after pollination. Fifty milligrams of whole wheat flour was used for glutenin and gliadin extraction for each sample. (c) Representative photographs of dough and washed‐out gluten from flour derived from WT and lgp1 seeds. Bar, 1 cm. (d) Representative light micrographs of toluidine blue‐stained transverse sections of the middle regions of WT and lgp1 seeds at 10, 12 and 18 DAP. PB, protein body; S, starch granule; V, vacuole; DAP, days after pollination. Bar, 50 μm.

The lgp1 mutant produces irregular protein bodies

Since SSPs primarily form ER‐derived PBs, which are subsequently internalized into PSVs, we investigated the distribution and configuration of PBs in lgp1 vs WT endosperm cells at 10, 12 and 18 DAP (Fig. 1d). At 10 DAP, both WT and lgp1 cells contained large central vacuoles in inner endosperm cells. Small dots, probably representing small PBs, were detected in both the WT and lgp1, while larger assembled PBs were only observed in WT cells. At 12 DAP, a central vacuole and many smaller vacuoles containing numerous aggregated PBs were detected in WT cells, whereas the majority of lgp1 cells lacked central vacuoles and contained fewer PBs. At 18 DAP, the PBs of WT cells merged into large aggregates, which were absent in lgp1 (Fig. 1d). Overall, these results indicate that the PB formation pathway is altered in lgp1.

We compared the PB ultrastructure in WT vs lgp1 endosperm cells at 12 DAP by TEM based on CCF. In WT cells, spherical PBs surrounded by a membrane of ER origin were located near the ER in extensive amounts. In lgp1 cells, many of the ER‐surrounded PBs were misshapen and irregularly clumped (Fig. 2a). In addition, the central vacuoles of WT cells contained large aggregates of PBs with some high‐electron‐density inclusion bodies on the surface, whereas these aggregates were smaller in lgp1 cells (Fig. 2b). These irregular PBs contained SSPs, as determined by immunogold labelling with anti‐γ‐gliadin, anti‐HMW‐glutenin and anti‐LMW‐glutenin antibodies (Fig. 2c). These results indicate that lgp1 accumulates irregular PBs.

Fig. 2.

lgp1 showed altered ultrastructures of protein bodies in developing starch endosperm cells. (a) Representative TEM micrographs of ER‐derived protein bodies in 12 d after pollination (DAP) inner starchy endosperm cells of the wild‐type (WT) and lgp1. PB, protein body; RER, rough endoplasmic reticulum; S, starch granule. Bar, 1 μm. (b) Representative TEM micrographs of protein bodies in the central vacuoles of 12 DAP inner starchy endosperm cells of the WT and lgp1. PB, protein body (black arrow); V, vacuole; S, starch granule. Bar, 5 μm. (c) Representative TEM micrographs showing immunogold labelling of anti‐γ‐gliadin, and HMW‐ and LMW‐glutenin antibodies in 12 DAP inner starchy endosperm cells of the WT and lgp1. Labelling can be seen in protein bodies. Bar, 200 nm.

The lgp1 phenotype is associated with a missense mutation in a single γ‐gliadin gene

To study the genetic inheritance and actionof the lgp1 mutation, we performed reciprocal crosses of lgp1 plants with Zhoumai16 (ZM16), a cultivar with normal gluten levels. Since HMW‐GSs are more easily detected than gliadins, we used HMW‐GS content as a marker. Because the endosperm is triploid with two genomes inherited from the female, the two resulting hybrids contained one (ZM16×lgp1) or two (lgp1×ZM16) lgp1 copies. In addition, self‐pollinating ZM16 and lgp1 plants resulted in endosperm with zero (ZM16) and three (lgp1) copies of lgp1, respectively. We found that lgp1 is a semidominant mutation because ZM16×lgp1 seeds exhibited lower HMW‐GS content than ZM16 and higher HMW‐GS content than lgp1 (Fig. S2a). Moreover, the lgp1 mutation exhibited a dosage effect, as lgp1×ZM16 displayed lower HMW‐GS content compared to ZM16×lgp1 seeds. To assess lgp1 mutation segregation, ZM16×lgp1 plants were self‐pollinated and progeny was examined. The results were in agreement, with a 1 : 2 : 1 segregation ratio (χ² = 0.039 < χ²(0.05, 2) = 5.991) (Fig. S2b,c), indicating that lgp1‐related phenotypes are associated with a single locus mutation.

Bulked segregant RNA‐seq, followed by SNP calling for the homozygous mutant and normal bulks, identified 67.65% SNPs associated with the mutant phenotype and located in the 0.18–8.34 Mb region of Chromosome 1D (Dataset S2; Fig. 3a). Genetic linkage analysis using 152 F₂ plants revealed that the lgp1 mutation was located between markers CS2805420 and CS3183041, at a genetic distance of 0.3 and 0.7 cM, respectively. Using a larger F₂ population of 8746 plants, we narrowed down the region containing the lgp1 mutation to a 53.03 kb region, delimited by markers CS2928184 and CS2981214 (Fig. 3b). The entire genomic region between these two markers was sequenced in the WT and lgp1, and only one annotated high‐confidence gene, TraesCS1D03G0010000, was identified. This gene encodes a γ‐gliadin, and sequencing analysis of the TraesCS1D03G0010000 coding region in lgp1 revealed a C‐to‐T mutation at position 56, downstream of the translational start site, which resulted in an alanine (Ala)‐to‐valine (Val) amino acid substitution at position 19 of the amino acid sequence (Fig. 3c). This Ala residue is located at the end of the SP. It was reported that the Ala residue at the end of the SP is critical for protein processing through cleavage which occurs at the C‐terminal region containing the signal peptidase cleavage site, because an Ala‐to‐Val substitution in an alpha‐zein of the maize floury2 mutant impairs SP‐mediated cleavage (Gillikin et al., ¹⁹⁹⁷). The Ala residue is also conserved in all wheat γ‐gliadins, as well as in rye γ‐secalins and barley γ‐hordeins (position 17) (Dataset S3; Fig. S3a). These findings indicate that the lgp1 phenotype is due to a semidominant mutation within a single gene encoding a γ‐gliadin; thus, we named the TraesCS1D03G0010000 gene LGP1.

Fig. 3.

Map‐based cloning of the gene associated with the lgp1 mutation. (a) Percentage of single nucleotide polymorphisms (SNPs) identified on each chromosome (left) and Chinese Spring v.2.0 chromosome 1D (right) by bulked segregant RNA‐seq (BSR‐seq) analysis. Red rectangle indicates the SNPs with allele frequency difference (AFD) value > 0.6. (b) Schematic diagram of the results achieved from genetic linkage analysis indicating the localization of the locus associated with the lgp1 mutation. The 53.03 kb region (in red) identified through fine linkage analysis only contained one high‐confidence annotated gene, TraesCS1D03G0010000 (in red), encoding a member of the wheat γ‐gliadin family. The numbers between the molecular markers in the population with 152 individuals indicate the genetic distance, and the numbers in the population with 8746 individuals indicate the number of recombinants. (c) Schematic diagram of the γ‐gliadin protein domains encoded by TraesCS1D03G0010000/LGP1. The C‐to‐T mutation leading to an alanine‐to‐valine substitution at position 19 of the amino acid sequence of the mutated gene is indicated in red. (d) Grain morphology of cv Fielder (control) and lgp1 transgenic plants. Three transgenic lines, Prolgp1: lgp1#1, Prolgp1: lgp1#2 and Prolgp1: lgp1#3, were photographed. Bar, 0.5 cm. (e) Comparison HMW‐GSs in Fielder (control) and lgp1 transgenic plants detected by SDS‐PAGE followed by Coomassie blue staining. Two dry seeds are shown for each line.

To verify the functional effect of the lgp1 mutation, the wheat cv Fielder was transformed with the entire LGP1 coding region, containing the C‐to‐T mutation, and its putative native promoter (2.4 kb upstream of the start codon) from the lgp1 mutant genotype (LGP1‐T) (Fig. S4). All three transgenic lines Prolgp1: lgp1#1, #2 and #3 showed a phenotype similar to that observed for lgp1 (see Fig. 7a–c for further details regarding seed phenotype of lgp1). In particular, significant decreases of glutenins and gliadins, reduced grain size and partially empty pericarp were observed (Fig. 3d,e). Therefore, we conclude that the C‐to‐T mutation of LGP1 is sufficient to generate the drastic phenotypic alteration observed in lgp1 seeds.

Fig. 7.

Starch biosynthesis is inhibited in lgp1 endosperm. (a–c) Phenotypes of the wild‐type (WT) and lgp1 grains representing grain length (a), grain width (b) and transverse section (c). Bars: (a, b) 1 cm; (c) 0.5 cm. (d–f) Differences in thousand‐grain weight (d), grain length (e) and grain width (f) between the WT and lgp1. Three replicates were measured for trait analysis, with 200 grains per replicate. Values are given as mean ± SD from three replicates. Statistical significance was determined by Student’s t‐test. **, P < 0.01. (g–i) Differences in starch (g), amylose (h) and amylopectin content (i) between the WT and lgp1. Values are given as mean ± SD from three biological replicates. Statistical significance was determined by Student’s t‐test. **, P < 0.01. (j, k) Representative scanning electron micrographs of isolated starch granules from the flours of WT (j) and lgp (k). White and red arrowheads indicate A‐type (10–35 μm) and B‐type (< 10 μm) starch granules, respectively. Bar, 50 μm. (l) Gene Ontology (GO) enrichment analysis of RNA‐seq data reporting significantly enriched GO terms of downregulated genes in 15 d after pollination (DAP) endosperm of lgp1 compared to the WT. Carbohydrate metabolic process is the most significantly enriched GO term. (m) Expression levels, expressed as fragments per kilobase per million (FPKM), of TaAGPS1, TaAGPS2, TaAGPL1 and TaAGPL3 detected in 15, 18, 23 and 28 DAP endosperm of the WT and lgp1. Each bar corresponds to the average value of three biological replicates. Values are given as mean ± SD from three biological replicates. (n) GO enrichment analysis of tandem mass tag (TMT) data reporting significantly enriched terms of downregulated proteins in 15 DAP endosperm of lgp1 compared to the WT. Cellular carbohydrate metabolic process is the most significantly enriched GO term.

Significant ER stress occurs in lgp1 endosperm

To investigate whether the C‐to‐T mutation of LGP1 leads to the failure of SP cleavage, we transiently transformed wheat leaf protoplasts with GFP‐tagged full‐length LGP1 from the WT (LGP1‐C) and lgp1 (LGP1‐T). LGP1‐T is c. 2 kDa larger than LGP1‐G. Thus, the Ala‐to‐Val substitution causes failed SP cleavage in lgp1 (Fig. S3b).

Gliadin is synthesized and processed in RER, suggesting that the mutation leading to unprocessed γ‐gliadin alters the ER structures (Wang et al., ²⁰¹⁴). To search for possible ER structure alteration, the ultrastructure of 12 DAP endosperm cells of WT, lgp1, and transgenic lines Prolgp1: lgp1#1/#2 and Fielder was analysed using both CCF and HPF TEM (Figs 4, S5). We observed that the central vacuoles in the cells of Fielder (transgenic lines control) contained large PB aggregates (Fig. S5a), compared to Prolgp1: lgp1#1/#2 (Fig. S5b,c), thus resembling the differences previously observed in the WT and lgp1 (Fig. 2b). A decrease in ER formation was observed in lgp1 and Prolgp1: lgp1#1/#2 compared to the WT and Fielder (Figs 4a,b, S5d–f). The continuous and sheet‐like structures of ER were observed in the WT and Fielder (Figs 4a, S5d), whereas in lgp1 and Prolgp1: lgp1#1/#2, the ER tubules were significantly fragmented (Figs 4b, S5e,f). In addition, altered ER structures were detected in lgp1 and Prolgp1: lgp1#1/#2 (Figs 4c–e, S5e,f). For example, in lgp1 the ER was swollen with PBs attached to the membrane (Fig. 4c). Many ring‐like structures with ribosome particles attached to the surface were observed in lgp1 and Prolgp1: lgp1#1/#2, which might represent distorted RER (Figs 4d, S5e,f). Furthermore, in lgp1, the dilated RER distended to a network‐like structure (Fig. 4e). To gain insight into the nature of these altered ER structures, we performed immunoelectron microscopy using an antibody against the ER lumen binding protein (BiP), an ER chaperone (Yao et al., ²⁰¹⁶). Large numbers of gold particles were detected in the ER lumen of the WT (Fig. 4f). In lgp1, in addition to localization in normal ER, gold particles were also observed around the ring‐like, swollen, dilated and fragmented RER (Fig. 4g–i).

Fig. 4.

Significant ER stress occurs in lgp1. (a, b) Representative TEM micrographs of ER in 12 d after pollination (DAP) starchy endosperm of the wild‐type (WT) (a) and lgp1 (b). PB, protein body; RER, rough endoplasmic reticulum. Black arrows represent fragmented RER. Bar, 1 μm. (c–e) Representative TEM micrographs of the altered ER structures in lgp1 cells. (c) Dilated RER; (d) ring‐like RER; (e) network‐like RER. RER, rough endoplasmic reticulum; PB, protein body. Bar, 1 μm. (f–i) Representative TEM micrographs showing immunogold labelling using anti‐BiP antibody detected in 12 DAP WT (f) and lgp1 starchy endosperm cells (g–i). Black arrows represent gold particles. Bar, 200 nm. (j, k) Immunoblot analysis of the ER molecular chaperone BiP in 20 DAP (j) and 25 DAP (k) endosperm of the WT and lgp1. Red arrows represent BiP. Anti‐eEF‐1α was used as a loading control (black arrows). Three independent biological replicates were performed for each sample. (l) Representative images of subcellular colocalization of LGP1‐C‐GFP, LGP1‐T‐GFP and LGP1‐noSP‐GFP fusions with the ER marker RFP‐HDEL expressed in wheat protoplasts. LGP1‐C, LGP1‐T and LGP1‐noSP represent the LGP1 protein from WT, lgp1 mutant and LGP1 without the signal peptide, respectively. Constructs were transiently expressed in wheat protoplasts for 12 h before confocal microscopy observation. Bar, 5 μm.

Upregulated expression of BiP has been indicated to be a common feature of ER stress (Wang et al., ²⁰¹⁴). The level of BiP was examined using a BiP‐specific antibody in protein extracts from lgp1 and WT 20 and 25 DAP endosperm. BiP protein content was markedly increased in lgp1 endosperm, suggesting that ER stress is significantly triggered (Fig. 4j,k). Moreover, transcriptome analysis (see Fig. S10a for details about RNA‐seq) showed that 30 ER‐stress‐associated genes, including the unfolded protein response (UPR) signal activator genes INOSITOL REQUIRING ENZYME 1 (IRE1) and bZIP60 (Humbert et al., ²⁰¹²), and genes encoding proteins taking part in the ER‐associated protein degradation (ERAD) pathway (Wang et al., ²⁰¹⁴), including protein disulphide isomerase (PDI)‐like genes, Derlin family proteins genes and Sec61‐like genes, etc., were expressed at higher levels in lgp1 compared to the WT (Fig. S6). These results indicate that significant ER stress occurs in lgp1.

To further determine whether the lack of LGP1 SP cleavage induced its ER retention, LGP1‐C‐GFP (WT), LGP1‐T‐GFP (lgp1) and LGP1‐noSP‐GFP (without SP) constructs were transiently coexpressed with the ER marker RFP‐HDEL in wheat protoplasts. Confocal microscopy revealed a different GFP signal pattern in LGP1‐C‐GFP and LGP1‐T‐GFP, with the latter exhibiting an irregular aggregation located around the nuclei (Fig. 4l). Both LGP1‐T‐GFP and LGP1‐C‐GFP showed strong correlation with the ER marker; however, the expression of LGP1‐T‐GFP, but not of LGP1‐C‐GFP, led to ER aggregation (Fig. 4l). By contrast, LGP1‐noSP‐GFP showed a random distribution in the cytoplasm and was not associated with the ER, indicating that the SP was required for LGP1 ER localization (Fig. 4l). Together, we speculate that the ER stress is probably associated with the ER retention of the unprocessed γ‐gliadin in lgp1.

Formation of autophagosome‐like structures in lgp1

Significant ER stress is able to induce autophagy (Liu et al., ²⁰¹²). We performed both CCF and HFP TEM to detect whether autophagy occurs in lgp1. Double‐membrane autophagosome‐like structures (ALSs) were observed in lgp1 and Prolgp1: lgp1#1/#2 cells, and these structures contained a variety of cargos and unidentified cytoplasmic contents (Figs 5a–d, S5g,h). In addition, some ALSs contained several small vesicles (Fig. 5e). We also observed that some single‐membrane autolysosomes contained an ER‐like structure inside with membranes decorated with electron‐dense ribosome particles (Fig. 5f) and unidentified contents (Fig. S5i). Some of them contained PBs with highly electron‐dense materials attached to the surface (Fig. 5g,h). These ALSs generally have a diameter of c. 1 μm, in agreement with the previously reported autophagosome size (Zhuang et al., ²⁰¹³). Importantly, these abnormal structures were absent in WT cells. In addition, when treated with autophagy inhibitor (2S,3S)‐trans‐epoxysuccinyl‐l‐leucylamido‐3‐methylbutane ethyl ester (E‐64d), the accumulation of HMW glutenin in lgp1 was significantly increased, while there was no obvious change in the WT (Fig. 5i), suggesting that the decrease of SSPs in lgp1 might be associated with autophagy.

Fig. 5.

Autophagosome‐like structures in lgp1. (a–h) Representative transmission electron microscopy (TEM) images of autophagosome‐like structures (ALSs) in 12 d after pollination (DAP) starchy endosperm cells of lgp1. (a–d) Double‐membrane ALSs observed by TEM based on conventional chemical fixation (a, b) and high‐pressure freezing (c, d). (e) ALSs contained small vesicles. (f) Single‐membrane ALSs contained ER‐like structures decorated with electron‐dense ribosome particles (black arrows). (g, h) ALSs contained protein bodies with highly electron‐dense materials attached to the surface. Bar, 200 nm. (i) Immunoblotting analysis of protein extracts prepared from wild‐type (WT) and lgp1 seeds treated with E‐64d (0.1 mM for 2 d) using an anti‐HMW glutenin antibody. Values are given as mean ± SD from three biological replicates. Statistical significance was determined by Student’s t‐test. **, P < 0.01. The accumulation of anti‐HMW glutenin in lgp1 was significantly increased. Two seeds are shown for each sample. (j) Immunoblotting analysis of ATG8 in 12 DAP seeds of the WT and lgp1. ATG8‐PE indicates the phosphatidylethanolamine (PE)‐conjugated form of ATG8. Anti‐actin was used as a loading control. Two seeds are shown for each sample. (k) Representative images of the subcellular colocalization of LGP1‐C‐GFP, LGP1‐T‐GFP and LGP1‐noSP‐GFP with the autophagy marker mCherry‐ATG8e expressed in wheat protoplasts. LGP1‐C, LGP1‐T and LGP1‐noSP represent the LGP1 protein from WT, lgp1 mutant and LGP1 without the signal peptide, respectively. Constructs were transiently expressed in wheat protoplasts for 12 h before confocal microscopy observation. White arrowheads indicate the colocalization of GFP with mCherry. Bar, 5 μm.

The autophagy‐related 8 (ATG8) protein plays a central role in decorating autophagosomes and binding to specific cargo receptors to recruit cargo to autophagosomes (P. Wang et al., ²⁰²⁰). We observed that lgp1 shows strong increased levels of lipidated ATG8, a phosphatidylethanolamine (PE)‐conjugated form of ATG8 (ATG8‐PE), which labels the forming/completed autophagosomes (Fig. 5j). To further examine the accumulation of ALSs in lgp1 associated with ATG8, we performed immunoelectron microscopy using antibodies against ATG8. However, the signal‐to‐noise ratio was quite high in multiple attempts under various experimental conditions, rendering it difficult to unambiguously identify the association of ATG8 with ALSs, which was possibly caused by nonspecific reactivity of the anti‐ATG8 antibody in wheat (Fig. S7). As an alternative assay, a colocalization analysis was performed using wheat protoplasts transiently coexpressing LGP1‐C‐GFP (WT), LGP1‐T‐GFP (lgp1) or LGP1‐noSP‐GFP (without SP) constructs with an autophagosome marker mCherry‐ATG8e (Zhuang et al., ²⁰¹³). mCherry‐ATG8e punctate was rarely seen in either LGP1‐C‐GFP or LGP1‐noSP‐GFP, whereas two distinct mCherry‐ATG8e dots were visualized and colocalized with LGP1‐T‐GFP (Fig. 5k), suggesting that the LGP1 mutant protein is associated with the autophagosome marker. Overall, these findings suggest that ALS formation is triggered in lgp1.

Elevated cell death occurs in lgp1

Unmitigated or prolonged ER stress can activate cell death (Cai et al., ²⁰¹⁸). To assess cell death in lgp1, Evans Blue staining was performed to detect cell death or membrane damage (Li et al., ²⁰¹⁰) in WT and lgp1 endosperm harvested at 8, 12, 16 and 20 DAP. More intensive signal of Evans Blue staining was detected in 16 and 20 DAP endosperm of lgp1 vs the WT (Fig. 6a). Consistently, the level of the stress‐related hormone ABA was higher in 15 and 20 DAP endosperm of lgp1 compared to that in the WT (Dataset S2; Fig. 6b). Furthermore, the level of ROS, which act as cell death initiators and mediators (Morgan & Liu, ²⁰¹⁰), was higher in lgp1 vs WT endosperm throughout all developmental stages analysed (Dataset S2; Fig. 6c). Tandem mass tag analyses were performed to assess whole‐proteome differences in 15 DAP WT and lgp1 endosperm (Dataset S4; Fig. S8a). Gene Ontology (GO) analysis of differentially expressed proteins in the biological process category indicated that proteins involved in cell disruption, hypersensitive response and programmed cell death (PCD) were highly enriched in lgp1 compared to WT (Fig. S8b). For example, two pathogenesis‐related proteins of the PR‐4 family, wheatwin‐1 and wheatwin‐2, having RNase activity and inhibiting pathogen invasion (Caporale et al., ²⁰⁰⁴; Altenbach et al., ²⁰⁰⁷), were highly upregulated in lgp1 (Fig. S8c). These data suggest that cell death is significantly elevated in lgp1.

Fig. 6.

lgp1 shows elevated cell death and seed storage protein degradation. (a) Representative images of viability staining of developing wild‐type (WT) and lgp1 seeds harvested at 8, 12, 16 and 20 d after pollination (DAP). Cells that underwent programmed cell death are stained in blue. Bar, 1 mm. (b) Analysis of abscisic acid (ABA) content in WT and lgp1 seeds at 15 and 20 DAP. Values are given as mean ± SD from three biological replicates. Statistical significance was determined by Student’s t‐test. **, P < 0.01. (c) Analysis of reactive oxygen species (ROS) content in WT and lgp1 seeds at 10, 15, 20 and 25 DAP. Values are given as mean ± SD from three biological replicates. Statistical significance was determined by Student’s t‐test. **, P < 0.01. (d–f) Representative TEM images of cell wall in 12 DAP endosperm cells of the WT and lgp1. Protein body‐like structures accumulated in the cell wall and the extracellular space in lgp1 (d) and (e), while normal cell walls were observed in the WT (f). CW, cell wall; Black arrows represent protein body‐like structures. Bar, 1 μm. (g–i) Representative TEM micrographs showing immunogold labelling of cell walls of lgp1 mutant 12 DAP endosperm with anti‐HMW glutenin (g), anti‐LMW glutenin (h) and anti‐γ‐gliadin (i) antibodies. Labelling can be seen in protein bodies (black arrows). CW, cell wall. Bar, 200 nm. (j) Metabolic analysis of amino acid derivatives upregulated in lgp1 at 15, 20 and 35 DAP seeds compared to the WT. Data presented were calculated using means of three biological replicates.

lgp1 endosperm cells show SSP aggregation in the extracellular space

High‐pressure freezing TEM images showed that lgp1 and Prolgp1: lgp1#1/#2 transgenic lines contained many PB‐like structures that accumulated peripherally at the cell wall and in the extracellular space (Figs 6d,e, S5k,l). These structures were not observed in WT or cv Fielder cells (Figs 6f, S5j). Detailed examination of lgp1 cells by immunoelectron microscopy with anti‐HMW‐glutenin, LMW‐glutenin and γ‐gliadin antibodies indicated the PB‐like structures were SSPs (Fig. 6g–i). In addition, yeast two‐hybrid and luciferase complementation image assays showed that LGP1 protein lacking SP interacted with other SSPs including HMW‐GS, LMW‐GS, γ‐gliadin and α‐gliadin (Fig. S9a,b). These observations suggest that the failure of SP cleavage does not affect the interaction between LGP1 and other SSPs, but does produce abnormal PBs and alters their cellular trafficking because the abnormal PBs are delivered to the extracellular space.

Gene Ontology analysis of differentially expressed proteins in the cellular component category from 15 DAP WT and lgp1 TMT data indicated that proteins upregulated in lgp1 were enriched for the GO term localization in the extracellular space (Fig. S9c). Among these proteins there were members of serpin family proteins and proteasomes (Fig. S9d). Some immune‐responsive serpins are acute‐phase proteins that are strongly upregulated in response to infection or injury (Meekins et al., ²⁰¹⁷). Proteasomes have been found in the extracellular space and might function in the ubiquitin‐independent degradation of unfolded proteins (Tsimokha et al., ²⁰¹⁷). These findings suggest that lgp1 endosperm is under stress and that SSPs located in the extracellular space may be subjected to protease‐mediated degradation. We then compared the metabolic profiles of WT and lgp1 endosperm at 15, 20 and 35 DAP (Dataset S5). Of the 486 metabolites detected, 20 metabolites representing amino acid derivatives and short peptides were highly upregulated in lgp1 compared to WT (Fig. 6j). These findings indicate that the accumulation of free amino acids and short peptides is associated with the degradation of SSPs in lgp1. Together, these results indicate that lgp1 endosperm cells are under severe stress and some SSPs are delivered to the extracellular space for protease‐mediated degradation.

Starch biosynthesis is inhibited in lgp1 mutant endosperm

Compared to the WT, the lgp1 mutant showed smaller kernels with reduced thousand‐grain weight, grain width and length, and a partially empty pericarp (Dataset S2; Fig. 7a–f); starch and amylose contents were also reduced in lgp1 (Dataset S2; Fig. 7g,h), whereas the amylopectin content was higher (Dataset S2; Fig. 7i). Scanning electron microscopy revealed that WT starch granules (SGs) were composed of larger A‐type and smaller B‐type SGs, whereas lgp1 contained fewer B‐type SGs (Fig. 7j,k). These results indicate that starch synthesis is inhibited in lgp1.

We performed transcriptome analysis using RNA extracted from WT and lgp1 endosperm at 15, 18, 23 and 28 DAP (Fig. S10a). Gene Ontology analysis revealed that the genes downregulated in lgp1 were highly enriched for carbohydrate metabolism, glycogen biosynthesis and starch catabolic processes (Figs 7l, S10b). In particular, 47 genes involved in starch biosynthesis were downregulated in lgp1 at 15 DAP (Dataset S6; Fig. S11). These included TaAGPS1/2 and TaAGPL1/3, encoding two small and two large subunits of ADP‐glucose pyrophosphorylase (AGPase), an enzyme that catalyses a rate‐limiting step in starch biosynthesis (Qu et al., ²⁰¹⁸). All four genes were downregulated in lgp1 15 DAP endosperm and were generally downregulated at all developmental stages analysed (Fig. 7m). These results indicate that AGPase activity is affected in lgp1 endosperm during all stages of development. Tandem mass tag results confirmed the RNA‐seq data, showing that proteins involved in carbohydrate, glucan, starch and glycogen metabolism or biosynthesis were significantly downregulated in lgp1 vs the WT (Fig. 7n). KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis indicated that 24 enzymes involved in starch and sucrose metabolism were downregulated in lgp1 vs the WT (Fig. S12). These data support the conclusion that the accumulation level of genes and proteins related to starch biosynthesis is inhibited in lgp1 seeds, which is consistent with observed phenotypes, indicating that not only SSPs but also starch accumulation is impaired in lgp1.

lgp1 has high dietary fibre content and low gluten

Wheat bran is highly nutritious and an excellent source of dietary fibre. It may benefit digestive and heart health and could even reduce breast and colon cancer risk (Katileviciute et al., ²⁰¹⁹). The wheat bran content in lgp1 contains 82.2% of total grain weight, which is significantly higher than that of WT (29.6%) (Dataset S2; Fig. 8a), and the dietary fibre content of lgp1 was significantly increased (27.5%) compared with that of the WT (16.5%) (Dataset S2; Fig. 8b). Thus, despite the reduction in grain size, the wheat bran of lgp1 could be added to baked goods and casseroles as a natural fibre supplement to increase daily intake.

Fig. 8.

Wheat bran, fibre and gluten contents and cooking performances in lgp1 mutant. (a–c) Differences in wheat bran (a), dietary fibre (b) and gluten (c) contents between the wild‐type (WT) and lgp1. The dietary fibre content was tested using the whole wheat flour derived from the WT and lgp1. The gluten content was estimated by examining immunoreactive gliadin peptides in WT and lgp1‐derived flour using the R5 monoclonal antibody. Values are given as mean ± SD from three biological replicates. Statistical significance was determined by Student’s t‐test. **, P < 0.01. (d) Morphological differences in cookies made with flour derived from the WT, lgp1, potato, rice, sweet potato and maize. Three representative cookies were photographed. Bar, 1 cm. (e) Morphological differences in breads made with WT flour, lgp1 flour and lgp1 flour with the addition of 2% hydroxypropyl methylcellulose (HPMC). Transverse sections, lateral sections, and top views are shown. Bar, 1 cm.

For a gluten‐free diet, the levels of immune reactive peptides in wheat flour must be strongly reduced (García‐Molina et al., ²⁰¹⁹; Scherf et al., ²⁰²⁰). The monoclonal antibody (mAb) R5 method is the major technique used to evaluate immune reactive peptide contents because it detects a conserved domain (QQPFP) found in most immune‐reactive gliadins (Valdés et al., ²⁰⁰³; Sanchez‐Leon et al., ²⁰¹⁸). Using this system, we observed a 31% lower gluten content in lgp1 vs the WT (Dataset S2; Fig. 8c). This suggests that, although the mutant showed a strong reduction in total gliadin contents, small fragments of degraded gliadins were still present and probably accumulated in the extracellular space of lgp1 endosperm.

The performance of lgp1 flour for the preparation of different foods was evaluated. Compared to potato, rice, sweet potato and maize flours, which are used as wheat substitutes in gluten‐free products, lgp1 wheat flour showed better cookie‐making performance with an intact appearance, unbreakable texture and the unique wheat‐product smell (Fig. 8d). Compared to WT flour, bread of lgp1 flour had a decreased loaf volume and nonporous crumb structure (Fig. 8e). However, the addition of HPMC, a bread improver agent (Bárcenas & Rosell, ²⁰⁰⁵), improved the bread‐making performance of lgp1. Therefore, although lgp1 is unsuitable for a gluten‐free diet, it exhibits generally reduced gluten and glutenin levels, while maintaining good cooking performance, making it a good candidate for specific low‐gluten diets.

Discussion

In this study, we showed that an Ala‐to‐Val mutation in the SP of a single wheat γ‐gliadin (LGP1) induced a drastic change in the intracellular trafficking of SSPs. In lgp1, LGP1 was found in the ER; however, the PB formation pathway was completely altered because the SP was not cleaved. The lgp1 mutant exhibited very low levels of SSP deposits, and ER‐surrounded PBs lost their spherical shape and were irregularly clumped, demonstrating that SP cleavage is essential for correct PB formation (Fig. 9). It has been previously reported that suppression of gliadins in wheat results in altered PB morphology (Gil‐Humanes et al., ²⁰¹¹), and irregular PBs were also observed in corn (Zhang & Boston, ¹⁹⁹²; Wu & Messing, ²⁰¹⁰) and rice (Kawakatsu et al., ²⁰¹⁰) when storage protein levels were altered. It remains to be determined whether SP retention of only LGP1 γ‐gliadin and a few other SSPs affects the PB pathway or whether the correct posttranslational processing of any SSP is required to allow correct PB formation.

Fig. 9.

Model describing how the γ‐gliadin with uncleaved signal peptide (SP) leads to the dramatic phenotype of lgp1 seeds. Wheat seed storage proteins (SSPs) are synthesized on polyribosomes attached to the rough endoplasmic reticulum (RER). Following its passage into the ER lumen, the N‐terminal SP of LGP1 in the wild‐type (WT) is cleaved by signal peptidase (SPase) and the proteins form protein bodies (PBs), which are deposited into the protein storage vacuole (PSV). In the lgp1 mutant, this pathway is inhibited when the γ‐gliadin (lgp1) contains an Ala‐to‐Val substitution at position 19, impairing SP cleavage. This mutation leads to the formation of misshapen PBs and retention in the ER, altered ER structures, the development of potential autophagosome‐like structures (ALSs), elevated cell death, SSP secretion into the extracellular space, and a strong reduction in SSPs content and starch biosynthesis (i.e. the severe phenotype observed in lgp1 seeds). Dotted arrows indicate that unprocessed γ‐gliadin may trigger ALSs, altered ER and elevated cell death in lgp1.

The observation that ER stress occurs in storage protein mutants with SP retention in the ER has only been reported in maize seed; however, our understanding of the mechanism and outcome of ER stress remains limited (Wang et al., ²⁰¹⁴). Endoplasmic reticulum stress activates two protein degradation pathways: the ubiquitin‐proteasome via the ERAD pathway and lysosome‐mediated protein degradation via autophagy (Sano & Reed, ²⁰¹³). The degradation of ER components by autophagy acts as a back‐up system for the inefficient proteasomal degradation of ER proteins through the ERAD pathway, due to accumulation of misfolded or unfolded proteins exceeding ER capacity (Grumati et al., ²⁰¹⁸). In lgp1, ER stress is stimulated in lgp1 endosperm, as illustrated by dilated RER and markedly upregulated BiP content and ER‐stress‐associated genes. Three maize mutants with impaired seeds, floury2 (fl2), defective endosperm 30 (DE‐B30) and floury4 (fl4), are associated with mutations at or before the SP cleavage sites of specific zein genes affecting the PB pathway, and ER stress occurs (Coleman et al., ^{1995, 1997}; Kim et al., ²⁰⁰⁴; Wang et al., ²⁰¹⁴). In sorghum (Sorghum bicolor) seeds, an Ala‐to‐Thr amino acid substitution in the kafirin SP also leads to defective PB formation (Oria et al., ²⁰⁰⁰; Wu et al., ²⁰¹³). However, elevated cell death has not been observed in either maize or sorghum. This difference may be related to the existence of different PB trafficking and deposition pathways in wheat compared to maize. In wheat, two PB trafficking pathways have been observed: the preferential accumulation of gliadins and LMW‐glutenins in the vacuole and the preferential accumulation of HMW‐glutenins in the ER lumen (Tosi et al., ²⁰⁰⁹; Tosi, ²⁰¹²; Yu et al., ²⁰¹⁷). Conversely, in maize, SSPs are only translocated into the ER lumen, where they assemble into PBs (Bagga et al., ¹⁹⁹⁷). In lgp1, large amounts of gliadins were stuck in the ER lumen, an incorrect destination for gliadins. More severe ER stress occurs compared to maize and cannot be reversed, and cellular functions then deteriorate, finally leading to elevated cell death.

Autophagy is essential for plant responses to several developmental and environmental cues, functioning in diverse processes such as senescence, male fertility, root meristem maintenance, responses to nutrient starvation, and biotic and abiotic stress (Qi et al., ²⁰²¹). Significant ER stress is able to induce autophagy (Liu et al., ²⁰¹²). We observed many ALSs in lgp1 cells by HPF TEM. We also found strongly increased levels of lipidated ATG8 and the colocalization of ATG8e and mutated LGP1. However, additional experiments in vivo in wheat endosperm cells supporting autophagy are not possible at present, due to peculiar features of endosperm cells (e.g. high accumulation of starch granules), making it a challenge to perform immunolocalization experiments with available antibodies detecting autophagy markers such as autophagy‐related 8 (ATG8) (Fig. S7). Endoplasmic reticulum‐phagy has been reported previously in plant vegetative organs as a mechanism triggered by various cues including abiotic and biotic stress (Liu et al., ²⁰⁰⁹; Zeng et al., ²⁰¹⁹; Bao & Bassham, ²⁰²⁰). In all these cases, autophagosomes are delivered to the vacuole for degradation by hydrolases. PSVs in seeds are functionally different from lytic vacuoles in most vegetative tissues: PSVs function in protein storage, whereas lytic vacuoles are lysosome‐like degradative organelles (Feeney et al., ²⁰¹⁸). Hence, in seeds, a vacuole‐independent ER‐phagy mechanism could be necessary. The secretion of SSPs to the extracellular space was observed in lgp1. The unconventional protein secretion pathway delivers leaderless secretory proteins to the cell exterior, bypassing typical organelles involved in secretion (Robinson et al., ²⁰¹⁶; Wang et al., ²⁰¹⁷). In addition, increasing evidence supports the existence of ER–plasma membrane fusion, especially under stress conditions (Morel, ²⁰²⁰). Both mechanisms might be involved in delivering proteins destined for degradation to the external cellular space in lgp1.

Programmed cell death, the genetically regulated disassembly of cells, occurs in the endosperm during seed maturation (Kobayashi et al., ²⁰¹³). We observed the secretion of SSPs to the apoplast space and increased cell death in lgp1 (Fig. 9). This results in the accumulation of free amino acids and short peptides, as well as reduction of starch biosynthesis. Starch granule composition is also affected, with an increase in A‐ compared to B‐type granules. Since B‐type granules are formed later in seed development than A‐type granules (Parker, ¹⁹⁸⁵), it is likely that the cell death‐associated impairment of general metabolism primarily and more efficiently affects B‐type SG formation. Notably, increased cell death can lead to a dramatic lgp1 phenotype (i.e. reduction of all gluten proteins and starch, as well as partially empty pericarp) by two nonmutually exclusive mechanisms. In both plants and animals, severe or chronic ER stress can stimulate PCD to kill unwanted cells and protect others (Walter & Ron, ²⁰¹¹; Yang et al., ²⁰¹⁴). However, even excessive accumulation of SSPs in the extracellular space can initiate oxidase stress, increase ROS and ABA levels, and stimulate cell death. The last mechanism resembles that observed in plants under pathogen attack or abiotic stress, where PCD is induced to control the extracellular proliferation of pathogens or to counteract negative stress effects (Hara‐Nishimura & Hatsugai, ²⁰¹¹).

The lgp1 mutant exhibited a strong reduction in the levels of all gluten proteins; however, R5 mAb‐based detection of gliadin immunoreactive peptides revealed a reduction of only 31%, probably due to the presence of these short immunoreactive peptides in the extracellular space, following SSP degradation. This observation indicates that lgp1 is not useful in gluten‐free diets for coeliac disease patients because gluten level (estimated by the R5 method) was above the threshold of 20 mg kg⁻¹ of gluten (Codex Standard 118‐1979) (Scherf et al., ²⁰²⁰). Nevertheless, lgp1 exhibits peculiar qualities, making it a good candidate for diets required for specific disorders. Almost all gluten‐free and low‐gluten wheat genotypes identified to date showed reduced gliadins but not glutenin levels (García‐Molina et al., ²⁰¹⁹; Jouanin et al., ²⁰¹⁹; Rustgi et al., ²⁰¹⁹; D. Wang et al., ²⁰²⁰). However, a low glutenin content is also important, especially for patients with diabetes or kidney failure, because glutenins contain high amounts of proline and glutamine residues, making their complete digestion difficult (Shan et al., ²⁰⁰²). Since lgp1 exhibited the simultaneous reduction in glutenin and gliadin levels, it could be suitable for patients with these diseases. In addition, reducing the rate and extent of starch digestion in the gastrointestinal tract leads to healthy blood glucose levels (Hazard et al., ²⁰²⁰). The levels of B‐type SGs were specifically reduced in lgp1. Such a reduction is relevant for improving the health impacts of wheat‐based foods because small SGs are digested more rapidly than large granules (Franco et al., ¹⁹⁹²). Importantly, only limited variation occurs in the proportions of A‐ and B‐granules in various wheat cultivars (Chia et al., ²⁰²⁰). Hence, lgp1 is expected to be beneficial for increasing healthy blood glucose levels. Finally, although a reduction in glutenin content affects the end‐use quality of flour, lgp1 flour maintained good cooking performance. Indeed, for cookie preparation, lgp1 flour performed better than flour derived from potato, rice, sweet potato or maize. The bread‐making performance of lgp1 flour was worse than that of WT flour; however, a similar performance was achieved with the addition of a bread improver agent, HPMC. In general, these results indicate that lgp1 flour could be used to prepare wheat foods for specific diets, thus avoiding the deleterious effects related to the complete elimination of wheat. Accordingly, experiments are in progress for introgressing the lgp1 mutation into wheat elite lines, including those exhibiting a further reduction in the levels of immunoreactive gliadin peptides (Juhász et al., ²⁰¹⁸; Li et al., ²⁰¹⁸), to enable the industrial production of wheat flour with the unique properties associated with the lgp1 mutation. Since approaches based on RNA interference and targeted gene editing have been successful for reduction of specific or multiple members of the gliadin gene family (Sanchez‐Leon et al., ²⁰¹⁸; Jouanin et al., ²⁰¹⁹), LGP1 also represents an additional candidate for reducing immunoreactivity for gluten‐intolerant consumers through biotechnology‐based strategies.

Author contributions

YY and QS conceived the project; QC and CY performed the experiments; ZZ, ZW and YC performed bioinformatics analysis; WC provided technological assistance; VR, MX, ZS, JD, WG, ZH, JL, HP and ZN provided theoretical contributions to the project; QC, CY, YY and QS analysed the data and wrote the article. QC and CY contributed equally to this work.

Supporting information

Dataset S1 Primers used for gene mapping and vector construction.

Dataset S2 Data for all statistical analyses.

Dataset S3 The amino acid sequences of wheat γ‐gliadins, rye γ‐secalins and barley γ‐hordeins.

Dataset S4 List of differentially expressed proteins in lgp1 vs the WT.

Dataset S5 List of different metabolite levels in lgp1 vs the WT.

Dataset S6 List of differentially expressed transcripts in lgp1 vs the WT.

Fig. S1 RP‐HPLC chromatograms and flour quality parameters of the WT and lgp1.

Fig. S2 Phenotypic and genetic analysis of F₁ and F₂ progenies from ZM16×lgp1 seeds.

Fig. S3 Phylogenetic analysis of prolamins in wheat, rye and barley, and peptide cleavage of LGP1 protein from the WT and lgp1.

Fig. S4 Sequence of lgp1 coding region and its putative native promoter.

Fig. S5 Representative TEM micrographs of 12 DAP endosperm cells in Fielder and 1 transgenic lines Prolgp1: lgp1#1/#2.

Fig. S6 Heat map of endoplasmic reticulum stress‐associated genes between the WT and lgp1.

Fig. S7 Representative TEM micrographs showing immunogold labelling using anti‐ATG8 antibody detected in 12 DAP endosperm cells of the WT and lgp1.

Fig. S8 Proteome analysis of WT and lgp1 15 DAP endosperm.

Fig. S9 Interactions between LGP1 and other seed storage proteins and proteome analysis of WT and lgp1 15 DAP endosperm.

Fig. S10 Transcriptome analysis of 15, 18, 23 and 28 DAP endosperm of the WT and lgp1.

Fig. S11 Heat map of starch‐biosynthesis‐associated genes differentially expressed between the WT and lgp1.

Fig. S12 KEGG pathway analysis of proteins downregulated in lgp1.

Methods S1 Supplementary materials and methods.

Please note: Wiley Blackwell are not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

Data availability

RNA and genome sequencing data are available at the Sequence Read Archive (SRA) under accession no. PRJNA744310. The TMT data have been deposited with the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the iProX partner repository (Ma et al., ²⁰¹⁹) with the dataset identifier PXD027683.