Barley cv. Golden Promise (GP) seeds were mutagenized using EMS as previously published methods (Slade et al., 2005). Briefly, about 1,500 dried seeds were divided into two groups, with each group wrapped in cotton yarn and soaked in a 100 ml EMS solution for about 18 h. To ensure that the germination rate of the mutated barley seeds was about 50%, germination experiments were performed in 0.3%–0.9% EMS concentrations, from which the most suitable EMS concentration was selected. After EMS treatment, the seeds were washed with tap water for 3 h. The EMS-treated seeds (M1) germinated in the greenhouse in November 2017, then were planted in the field in November 2017. The M2-generation seeds were harvested from one ear of each M1 plant in June 2018.

Granule-bound starch synthase I protein and total protein extraction was performed as previously described (Yamamori and Endo, 1996; Hurkman and Tanaka, 2007; Hebelstrup et al., 2017). Total proteins and GBSSI protein were separated using sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in a discontinuous Tris–Hcl-SDS buffer system (pH: 6.8/8.8) with a polyacrylamide concentration of 12% (w/v; Ayala et al., 2015). Electrophoresis was performed at a constant current of 100 V/gel for 2.5 h immediately after the tracking dye had migrated off the gel. The total proteins and GBSSI proteins were detected via silver staining of the protein gels (Chevallet et al., 2007).

The total protein extraction method was the same as described above. Protein and starch granules were extracted from the supernatant. Briefly, 60 mg of developmental endosperm was ground and mixed with a 200 μl of buffer (10 mM Tris–HCl, pH 8.0) solution and centrifuged at 8,000g for 5 min. Next, the supernatant was collected and centrifuged three times. The resulting pellet, containing starch granules, was then washed 10 times with the buffer solution. Protein isolation was performed using SDS-PAGE as described above, following which proteins were transferred to a polyvinylidene fluoride membrane. GBSSI proteins were detected using GBSSI antibodies from rabbit serum (ABclonal, Anhui, China) at a dilution of 1:1,000 as the primary antibody. Horseradish peroxidase-labeled goat anti-rabbit IgG(H + L) (11000; Beyotime, Shanghai, China) was used as the secondary antibody. The blots were visualized using BeyoECL Star (Beyotime, Shanghai, China; Hebelstrup et al., 2017).

Genomic DNA was extracted from the leaves of 2-week-old seedlings in M3-generation using the cetyltrimethylammonium bromide (CTAB) method (Murray and Thompson, 1980). To obtain Wx-1 transcripts, seeds collected 15 days after flowering were selected for total RNA extraction using the Plant Genomic RNA Kit (BIOFIT, China). The PrimeScript™ reagent kit with gDNA Eraser (TAKARA, Dalian, China) was used for the reverse transcription of RNA. Based on the reference Wx-1 gene (GenBank accession: AF486514) sequence in the NCBI database, the primers Wx-1-F and Wx-1-R were designed to amplify the DNA and cDNA of Wx-1 (Supplementary Table S1). PCR amplification was performed using Phanta Max super-fidelity DNA Polymerase (Vazyme, Nanjing, China) to avoid sequence introduction errors. The PCR products were isolated on 1% agarose gel. All DNA fragments were recovered, purified, and ligated to pEasY-T1 (TransGen Biotech, Beijing, China). The DNA sequencing was conducted by a commercial company (Sangon Biotech, Shanghai, China). The final nucleotide sequence of each fragment was determined from the sequencing results of at least three independent clones.

The nucleotide sequences were assembled and aligned using the DNAMAN software package (v. 9.0; Lynnon Biosoft). The gene sequences of Wx-1 from wild barley GP and the mutant line M2-1105 were BLASTed and analyzed using ESEfinder3.0 webserver¹ to identify local shear factors (Cartegni et al., 2003; Smith et al., 2006). Phyre2² (Kelley et al., 2015) and the AlphaFold Protein Structure Database (Jumper et al., 2021) were used to predict the protein structure of HvGBSSI in the wild-type GP and mutant line M2-1105.

The expression of Wx-1 transcripts in M2-1105 and GP grains 15 days after flowering was determined by using qRT-PCR. Primers specific to Wx-1 transcripts were designed (Supplementary Table S1), and the amplification was performed using ChamQ™ Universal SYBR^® qPCR Master Mix (Vazyme) on a CFX 96 Real-Time PCR System (Bio-Rad, Hercules, United States). The qRT-PCR data were analyzed using the CFX 96 Real-Time System, and the relative expression was calculated using the 2^−ΔΔC_T method. Relative expression of the candidate genes was normalized using β-Actin and GAPDH as the reference genes.

The primers HvWx-1-PE-F and HvWx-1-PE-R were used to amplify the entire length of GP and Type I cDNA of mutant M2-1105 (Supplementary Table S1). The target gene fragment was amplified through PCR using the Phanta Max super-fidelity DNA Polymerase, and the expression vector pET-32 (a+) was digested with Hind III and Sac I enzymes. The digested products were ligated with the T4 ligase, and the ligated mixtures were transformed into competent Escherichia coli DH5a cells. The recombinant constructs pET32(a+)-GP and pET32(a+)-Type I were transformed into E. coli Rosetta strain (DE3) competent cells. Single-cell colonies were inoculated in Luria-Bertani broth medium containing 100 μg ml⁻¹ ampicillin at 37°C until an OD₆₀₀ of approximately 0.4–0.6 was achieved. Next, 0.3 mM isopropyl-β-d-thiogalactoside (IPTG) was added to the culture, and the incubation was continued for 3–5 h. The culture was centrifuged at 4°C and 12,000g to obtain E. coli cells. The bacterial cells were then lysed with BugBuster (Novagen, San Diego, CA), and fusion proteins were purified on an Ni-NTA column according to the manufacturer′s instructions (Qiagen, Dusseldorf, Germany). The expressed proteins were detected by SDS-PAGE.

Starch-binding assay was performed as previously described (Liu et al., 2015). Purified recombinant GBSSI (final concentration of 5 μM) was added to a 10 μl 10% (w/v) corn starch suspension in 100 mM citrate–phosphate buffer (pH 5.0) up a final volume of 60 μl. Each mixture was incubated with gentle shaking (10 rpm) at 4°C for 1 h and centrifuged at 12,000g for 5 min at 4°C. The upper material was designated as “supernatant.” After treatment of the samples with the SDS buffer, the bound GBSSI was released from the starch granules using the SDS buffer, and these were designated as “bound” ones. The amount of GBSSI in each determined supernatant washed, and the bound complex was analyzed by Western blotting using an anti-GBSSI antibodies.

Starch granules were extracted as previously described (Zhang et al., 2022). A scanning electron microscope (SEM), ZEISS Gemini SEM300 (Carl ZEISS AG, Oberkochen, Germany), was used to analyze and record the morphology of the starch granules. The starch granules were evenly distributed on the tape stuck on the loading platform for observation. As per the instrument instruction, each sample was viewed under three different fields and the morphology of the starch granules was observed at 1,000× and 2,000× magnification.

Starch granule size distribution was measured using the Mastersizer3000 laser diffraction particle size analyzer (Malvern Panalytical, England). Briefly, 140 mg of starch granules were suspended in 1 ml of anhydrous ethanol, and the starch granule suspension was mixed with 180 ml of distilled water in a beaker. After the shading degree reached the desired range of 12–17, the particle size distribution was measured using ultrasonic circulation for 2 min. The measurements were repeated three times for each sample. The volume and quantity of starch granules were determined using the theMastersizer 3000 operating software.

Granule-bound starch synthase I enzymatic activity was determined using the GBSSI enzymatic activity assay kit (Solarbio Science & Technology, Beijing, China) according to the manufacturer′s instructions. Total starch, amylose, and resistant starch contents in GP and M2-1105 were determined using the total starch (AA/AMG) Assay Kit, Amylose/Amylopectin Assay Kit, and Resistant Starch Assay Kit (Megazyme, Wicklow, Ireland), respectively. Protein content was determined using the automatic Kjeldahl apparatus equipment (Foss Analytical Kjeltec™ 8400 series), as previously described (Yang et al., 2019). Each sample was analyzed in triplicate.

Three replicates were used for each test. All data samples were expressed as means ± SD. One-way analysis of variance (ANOVA) was used to compare the differences between the values obtained from the mutant and wild-type lines. ANOVA was performed using IBM SPSS Statistics 26 (SPSS Inc., Chicago, IL, United States). p < 0.05 was considered statistically significant and p < 0.01 was considered extremely statistically significant.

Using SDS-PAGE, a mutant with defect in HvGBSSI expression in starch granules was identified in the M2 mutagenized population. The mutant, termed as M2-1105, lacked the corresponding band of HvGBSSI protein, in contrast with GP (Figure 1A). In terms of appearance, the seeds of M2-1105 were not different in length, width, fullness, or color from those of GP (Supplementary Figure S1).

We investigated the protein profiles in the developing grains of GP and the M2-1105 mutant. The HvGBSSI protein bands were present in the total proteins of M2-1105 as same with GP. In contrast, the HvGBSSI protein bands were not present in the protein profiles extracted from the M2-1105 starch granules but were present in those of GP (Figure 1B). Western blotting was then used to detect HvGBSSI protein in the total, the supernatant fraction, and purified starch granules from developing grains of GP and the M2-1105 mutant (Figure 1C). In total protein samples, HvGBSSI protein was detected in the M2-1105 mutant, indicating that M2-1105 produced the HvGBSSI protein. To further localize HvGBSSI protein, the total proteins were divided into soluble proteins, which was present in the supernatant, and particle binding proteins that were in starch granules. HvGBSSI protein was only detected in the supernatant but not in starch granules, suggesting that the HvGBSSI protein in the M2-1105 mutant failed to bind to the starch granules.

To further characterize the HvGBSSI mutation in the M2-1105 mutant, we amplified the genomic DNA fragment of Wx-1 gene from GP and the mutant M2-1105 by PCR. The final sequences were determined to be 2,793 bp for both GP and the mutant M2-1105 line. However, the M2-1105 had two single-nucleotide mutations: one change was from G to A (G1086A) at the nucleic acid 1,086; the other one (A2424G) was from G to A at the nucleic acid 2,424 (Figure 2A; Supplementary Figure S2). G1086A was located at the end of the 3′ of fourth intron as part of the splicing receptor site. The ESEfinder3.0 program predicated that G1086A disrupted the exonic splicing enhancer (ESE) elements ASF/SF2 with sequence 5′-TGCAGG-3′ and SRP55 with sequence 5′-CAGGGGA-3′ in the mutant line. In contrast, A2424G was located in the middle of intron 9 and had no effect on the ESE elements.

We focused on the Wx-1 transcripts in the M2-1105 mutant since the G1086A mutation occurred at the RNA splicing site. We sequenced 62 cDNA clones of Wx-1 from the M2-1105 mutant and found that these cDNAs were from two types of transcripts. The type I transcript, which was represented by 55 clones, had a size similar to that of GP, while the type II, with 5 clones, was longer than the GP transcript (Figure 1D). Sequencing results showed that transcript lengths of type I and II were 1,806 bp and 1,948 bp, respectively, different from the length of 1,812 bp in GP. The type I transcript had a 6 bp deletion at the 5′ of fifth exon, resulting in a deletion of two amino acid glycine at 218 (G218) and glutamate at 219 (E219). In the type II transcript, the fourth intron was not cleaved, leading to a premature termination codon (TGA) at nucleic acids 994–996 and a truncated protein of 311 amino acids (Figures 2C,D; Supplementary Figures S2, S3).

We analyzed the consequences of the splicing-related mutation G1086A and the resulting mRNA variants. G1086A occurred at the 3′ splicing site of the fourth intron, changing the sequence from AG to AA. This substitution resulted in a failed binding to the 3′3′ splicing site by the U2AF factor. Subsequently, U2AF incorrectly bound to the AG sequence that was the 5th and 6th nucleic acids of the fifth exon, leading to a 6 bp deletion at the beginning of the fifth exon. This is consistent to the sequencing results of the Type I transcript (Figure 2B). On the other hand, as the mutation of G1086A led to the loss of the exonic splicing enhancer (ESE) elements (ASF/SF2 and SRP55), U2AF was not always able to recognize the AG site within the fifth exon, and thus, the type II mRNA with an intact fourth intron was produced. Based on this analysis and our results, we revealed that U2AF must not combine with the mutated 3′ splicing site of the fourth intron, but combine with the wrong AG site of fifth exon randomly, which resulting erroneous transcripts type I and II of Wx-1 in M2-1105.

Further analysis of the expression levels of the two types of transcripts relative to β-Actin and GAPDH showed that the sum of the two transcripts was not significantly different from that of the wild type, but there was a huge difference between the two types of transcripts, with the expression level of the type I transcript was significantly higher than that of the type II transcript (Figure 1E), suggesting that U2AF prefers an AG site.

Two different Wx-1 transcripts in M2-1105 were translated to form two different protein types. The type II protein was terminated prematurely, likely with a complete loss of function. The type I transcript directs the expression of HvGBSSI mutant lacking two residues, G218 and E219. We hypothesized that the absence of two amino acids might affect the higher structure of the HvGBSSI protein in the M2-1105 mutant because only deletion of the two amino acids in the mutant should not change protein size too much. Both G218 and E219 were part of the β-pleated sheet inn the N-terminus and were not close to Adenosine 5′-Diphosphoglucose (ADPG) binding pocket of HvGBSSI. The deletion affected the polypeptide folding and subsequently formed a varied β-pleated sheet, leading to a conformational change leading to conformational change (Figures 3A,B).

HvGBSSI is tightly bound to starch granules, and the binding to starch granules stimulates its enzymatic activity. We hypothesized that the type I protein in the M2-1105 mutant may have lost its ability to bind starch granules. To confirm, we conducted a starch binding assay. The Wx-1 coding sequences of GP and the M2-1105 protein were expressed in the E. coli Rosetta strain (DE3; Supplementary Figure S4A). The recombinant proteins were purified to homogeneity (Supplementary Figure S4B) and incubated with 1% corn starch to determine their ability to bind to starch (Figure 3C). Our results showed that the amounts of the type I HvGBSSI protein were higher in the supernatant fraction but lower in the bound fraction than those of GP, supporting the notion that the binding capacity of the type I HvGBSSI variant was significantly reduced compared to GP.

To evaluate the effects of the HvGBSSI sequence change on starch granule properties, we analyzed the starch granule morphologies of GP and the M2-1105 mutant using a scanning electron microscope (SEM). The SEM results showed that the M2-1105 starch granules had some pores on their surface, and some granules had broken edges compared to those of the parent line (Figures 4A–D). Further analysis of granule size by laser diffraction showed that start granules from GP and the M2-1105 mutant had no significant differences in the quantity or volume distribution of A-type and B-type starch granules (Figures 4E,F).

We further measured the enzymatic activity of wild-type and the type I variant of HvGBSSI using the purified recombinant proteins in vitro. The enzymatic activity of the type I HvGBSSI variant from the M2-1105 mutant was 0.778 mol/g/min, a dramatic decrease from that of GP at 5.78 mol/g/min (Figure 5A). Because HvGBSSI is the sole enzyme for amylose synthesis, we further analyzed the starch composition of grains in the M2-1105 mutant. There was a notable difference in the total starch, amylose, and resistant starch contents between GP and the M2-1105 seeds (Figures 5B–E). The contents of the total starch, amylose, and resistant starch in the M2-1105 mutant were 48.49%, 7.45%, and 0.24%, respectively, which were significantly lower than those of GP at 58.60%, 23.91%, and 0.41%, respectively. The protein levels in the mutant line were higher than those in GP, with M2-1105 at 13.71%, while GP was only 11.31%.