Genetic and Phenotypic Characterization of a Novel dull1 Allele Affecting Starch Accumulation in Maize

Abstract

Background: Starch accumulation contributes substantially to maize grain yield and quality. Starch synthase III (SSIII) is a key component of the starch biosynthetic enzyme complex. However, its regulatory role in starch accumulation in maize endosperm remains incompletely understood. Methods: The du1-2018 mutant arose spontaneously during a conventional maize breeding program. Phenotypic characterization, storage compound contents, and starch structure were compared between the mutant and wild-type lines. BSA-seq, genetic linkage analysis, and transcriptomic analysis were employed to identify the candidate gene responsible for the mutant phenotype. Transcriptome sequencing was performed on developing kernels to evaluate the genome-wide effects of the du1-2018 mutation. Results: The du1-2018 mutant exhibited dull, glassy, and mildly shrunken kernels, with decreased starch levels and elevated soluble sugar and protein contents. The du1-2018 mutation disrupted starch accumulation, resulting in smaller, irregularly shaped starch granules and significant changes in starch composition and fine structure. This mutation was identified as a severe loss-of-function allele of the dull1 (du1) gene, evidenced by almost undetectable Du1 transcripts in developing kernels. Notably, transcriptomic analysis revealed that a substantial proportion of differentially expressed genes (DEGs) were involved in amino acid and protein metabolism. Conclusions: The novel du1 allelic variant, du1-2018, disrupts starch biosynthesis in maize endosperm, leading to reduced starch accumulation, altered starch structure, and transcriptional changes in nitrogen-related metabolic pathways. Our results provide new insights into the regulatory mechanisms underlying SSIII function in starch synthesis and endosperm development, and suggest potential links to carbon/nitrogen balance, with implications for future genetic improvement of maize grain quality.

1. Introduction

Maize, a major cereal crop cultivated worldwide, is of considerable economic and biological importance [1]. Beyond its vital role in supplying food, feed, fuel, and industrial materials, it also serves as an important model organism for plant biological research, with particular relevance to studies of endosperm development [2,3,4]. Starch constitutes the dominant component of the maize kernel, accounting for roughly 75% of its total dry matter and contributing significantly to grain yield and quality [3,5]. Thus, starch is a critical determinant of maize utilization in food-related and non-food industrial sectors.

Starch consists primarily of two distinct glucose polymers: linear amylose and branched amylopectin. Amylopectin constitutes ~75% of starch in normal maize kernels, while amylose accounts for ~25% [3,6,7]. Amylopectin is essential for the proper formation of starch granules, whereas amylose appears to be dispensable for this process [8]. Substantial advances have been achieved in elucidating the molecular, genetic, and biochemical bases of starch biosynthesis in crops. The biosynthesis of crystalline starch is mediated by the coordinated action of starch synthases (SSs), starch branching enzymes (SBEs), and starch debranching enzymes (DBEs) [8]. Among these enzymes, SSs utilize ADP-glucose (ADPG) to extend linear glucan chains, SBEs catalyze the formation of branch linkages, and DBEs hydrolyze these linkages [9,10]. In maize, five distinct classes of starch synthases have been characterized, comprising granule-bound starch synthase (GBSS) and soluble SSs (SSI, SSII, SSIII, and SSIV) [11]. GBSS primarily catalyzes amylose biosynthesis, whereas soluble SSs, in combination with SBEs and DBEs, mainly mediate amylopectin biosynthesis [8].

SSIII is one of the principal soluble SS isoforms in the developing endosperm of maize [12] and rice [13], with activity second only to that of SSI. Mutations in the maize dull1 (du1) gene, encoding SSIII, lead to mature kernels displaying the characteristic dull, glassy, and tarnished appearance, known as the “dull” phenotype [14]. du1 mutant kernels contain slightly less total carbohydrate than the wild-type [15,16] and exhibit reduced starch content, as well as varying degrees of elevated amylose content, depending on the genetic background [17,18,19]. Distinctive, irregularly shaped starch granules have been observed in du1 mutant kernels, with a reduced average granule size compared with the wild-type [19,20]. SSIII deficiency results in a reduced proportion of long B chains in amylopectin [19,21,22]. Furthermore, SSIII deficiency appears to influence the entire range of chains in amylopectin, in contrast to the effects of mutations in other SS isoforms, particularly SSI and SSII [21]. About 15% of starch in the du1 mutant of the maize Oh43 inbred line occurs in an intermediate form, distinct from amylose and amylopectin [19,23]. In addition, previous analyses have demonstrated that starch from the du1 mutant exhibits a higher degree of branching than the wild-type [19,23,24].

Initially, the du1 gene was discovered as a recessive genetic modifier of sugary1 (su1) mutations [14]. Using transposon tagging, the gene was cloned in 1998 and determined to encode a 1,674-aa protein with a predicted mass of 188 kDa [25]. The maize DU1 protein sequence is divided into three discrete regions: an N-terminal region unique to DU1, a central region conserved among SSIII proteins, and a C-terminal region containing a catalytic domain [25]. Furthermore, maize SSIII features the longest N-terminal extension of all SS isoforms, containing conserved carbohydrate-binding modules (CBMs) and coiled-coil motifs [26,27]. A previous study found that the C-terminal 450 residues of DU1 exhibited SS activity when expressed in Escherichia coli [12]. Besides its enzymatic function, SSIII has been suggested by several lines of evidence to function as a regulator of starch biosynthesis. For instance, extracts from du1 mutant kernels were found to contain reduced activities of both SSIII and SBEIIa [28]. Total soluble SS activity in du1 mutant endosperm has been reported to increase, likely due to a specific enhancement of SSI associated with SSIII deficiency [12]. The N-terminal extension has been reported to participate in protein—protein interactions, possibly through conserved coiled-coil motifs [26,29,30].

Analyses of maize endosperm mutants have greatly aided the identification of starch biosynthetic enzyme genes and the elucidation of regulatory mechanisms in carbohydrate metabolism [25,31,32]. Previous studies on du1 mutants, together with in vitro biochemical analyses, have partially elucidated the functional roles of Du1/SSIII in maize. Nevertheless, the precise metabolic defects associated with Du1/SSIII remain to be fully elucidated, particularly with respect to its broader regulatory roles in starch biosynthesis and other biological processes. Moreover, the effects of du1 allelic variants across diverse genetic backgrounds have yet to be thoroughly investigated. In the current study, we characterized a natural kernel mutant, du1-2018, which exhibits a “dull” phenotype. Gene mapping and sequence analysis confirmed du1 (Zm00001eb413290) as the causal gene underlying the mutant phenotype. The du1-2018 allele likely represents a severe loss-of-function mutation caused by an aberrant sequence alteration within the first 685 bp of the 5′ transcribed region. The du1-2018 mutation results in defective starch granule formation and extensive alterations in storage compound contents, starch composition and structure, as well as global gene expression patterns. The systematic investigation of a novel du1 allelic variant in this study broadens our understanding of the regulatory mechanisms of Du1/SSIII in starch biosynthesis and endosperm development. The study also provides valuable insights into its potential broader regulatory roles in carbon/nitrogen balance, thereby offering a theoretical basis for the genetic improvement of maize grain quality.

2. Materials and Methods

2.1. Plant Materials

The du1-2018 mutant was identified during a conventional maize breeding program. It was crossed with the B73 inbred line to generate F₁ hybrids. The F₁ plants were self-pollinated or backcrossed with du1-2018, producing F₂ and BC₁F₁ populations for genetic analysis and gene mapping. The du1-2018 mutant, the corresponding wild-type line, and the segregating populations were grown under field conditions in Chengdu (Sichuan Province, China) and Jinghong (Yunnan Province, China).

2.2. Determination of Carbohydrate and Protein Contents

Mature maize kernels were dried, ground, and sieved through a 100-mesh screen. Total starch content was determined using AOAC Method 996.11 [33]. Amylose content was determined colorimetrically according to GB/T 15683-2008 [34]. Soluble sugars were quantified using the anthrone–sulfuric acid method. Total protein content was calculated based on total nitrogen determined by the Kjeldahl method. All experiments were conducted in triplicate using kernels from three independent ears for each genotype.

2.3. Light and Scanning Electron Microscopy

Kernels at 20 days after pollination (DAP) were collected for paraffin sectioning, following the method of Nie et al. [35]. Kernels were fixed in a formalin–acetic acid–alcohol (FAA) solution overnight. The fixed samples were serially dehydrated, cleared, embedded in paraffin, and then sectioned. The sections were baked at 42 °C and subsequently dewaxed and rehydrated. Endosperm cell starch accumulation and distribution were visualized by staining sections with toluidine blue O and safranin O/fast green [35]. Stained sections were scanned with a Pannoramic Flash 250 slide scanner (3DHistech, Budapest, Hungary). Mature kernels were manually fractured longitudinally along the central axis. The fractured kernels were mounted on the sample stage with the fracture surfaces oriented upward. After sputter-coating the exposed fracture surfaces with gold, the samples were examined and imaged using a scanning electron microscope (SEM, Inspect F50, FEI, Hillsboro, OR, USA).

2.4. Bulk Segregant Analysis Sequencing (BSA-seq) and Gene Mapping

Fresh leaf tissue from 30 mutant and 30 wild-type F₂ individuals was pooled in equal amounts to create two bulks. Genomic DNA was isolated from the two bulks and parental lines using a modified cetyltrimethylammonium bromide (CTAB) method [36]. DNA libraries were prepared and sequenced on the Illumina HiSeq X Ten platform, yielding 150 bp paired-end reads. The average sequencing depth was approximately 30× for the two bulks and 15× for the parental lines. After adapter trimming and quality filtering, clean reads were aligned to the maize B73 reference genome (AGPv4) with the Burrows–Wheeler Aligner (BWA) and further processed with SAMtools [37]. The Genome Analysis Toolkit (GATK, v3.3) was employed to identify single-nucleotide polymorphisms (SNPs) and insertion–deletion (InDel) variants in the parental lines and two bulks [38]. Allele frequencies of each SNP were calculated in both bulks, and ∆(SNP-index) was derived by subtracting the wild-type bulk SNP-index from the mutant bulk [39]. A sliding-window approach with a 5 Mb window and 10 kb step was used to identify genomic regions associated with the mutant phenotype. Regions exhibiting significantly higher average ∆(SNP-index) values than surrounding regions, with sliding-window p values < 0.05, were considered candidate loci.

Based on the initial BSA-seq results, public simple sequence repeat (SSR) markers and newly developed InDel markers within the primary mapping interval were used to screen for polymorphic markers between the parental lines. F₂ mutant individuals were genotyped with polymorphic markers for linkage analysis and gene mapping. Genomic DNA of F₂ mutant individuals was isolated from fresh leaf tissue using the previously described modified CTAB method. PCR amplification was performed with PowerPol 2× PCR Mix with Dye V2 (ABclonal, Wuhan, China) following the manufacturer’s instructions (Supplementary Methods). Amplified fragments were separated on a 3.5% agarose gel by electrophoresis and visualized using GoldView (ELK Biotechnology, Sugar Land, TX, USA).

2.5. Starch Properties

2.5.1. Starch Isolation

Starch was isolated from mature kernels and developing kernels at 20 DAP based on a previously reported method with modifications [40]. Briefly, dehulled maize kernels were soaked overnight in 0.45% (w/v) sodium metabisulfite solution at room temperature, followed by homogenization using a blender. The homogenate was filtered through a 200-mesh sieve, centrifuged at 2500 rpm for 5 min, and the supernatant was discarded. The precipitates were rinsed with deionized water to remove residual impurities. The resulting wet starch was dried in an oven at 40 °C.

2.5.2. Starch Granule Morphology Observation

Starch samples isolated from 20 DAP developing kernels were suspended in anhydrous ethanol and mounted onto copper stubs using conductive double-sided adhesive tape. After gold sputter-coating, the samples were imaged at 1000× and 5000× using a Zeiss Merlin VP Compact field emission (FE) SEM (Carl Zeiss, Oberkochen, Germany).

2.5.3. Starch Granule Size Distribution

The granule size distribution of starch from mature kernels was measured using a Mastersizer 3000 laser diffraction particle size analyzer (Malvern Panalytical, Malvern, UK) following the method of Li et al. [41]. Measurements were performed at an obscuration level of 5–8%, with the dispersant refractive index set to 1.3330 and the particle refractive index of starch set to 1.50.

2.5.4. Gel Permeation Chromatography (GPC)

Purified starch from mature kernels was digested with isoamylase and subsequently dissolved in a DMSO/LiBr solution. The homogeneity and molecular weight of various fractions were determined by size-exclusion chromatography with refractive index detection (SEC-RI). The molecular weight distribution of the debranched starch was analyzed using a differential refractive index detector (Optilab T-rEX, Wyatt Technology, Santa Barbara, CA, USA) coupled with two tandem columns (300 × 7.5 mm; PLgel 10 μm MIXED-B and PLgel 5 μm MIXED-D; Agilent Technologies, Santa Clara, CA, USA) maintained at 80 °C using a column heater. The flow rate was 0.8 mL/min. Data acquisition and processing were conducted using ASTRA 6.1 (Wyatt Technology).

2.5.5. High-Performance Anion-Exchange Chromatography (HPAEC)

Purified starch from mature kernels was digested with isoamylase and subsequently dissolved in distilled water. The samples were analyzed by HPAEC using an ICS5000+ system (Thermo Scientific Dionex, Sunnyvale, CA, USA) with a CarboPac PA-200 column (4 mm × 250 mm) coupled to a pulsed amperometric detector (PAD). The flow rate was 0.4 mL/min. Data acquisition and processing were performed on the ICS5000 system using Chromeleon 7.2 CDS (Thermo Scientific).

2.6. RNA-seq Analysis

Kernels at 15 DAP, representing the onset of rapid starch accumulation, were collected for RNA-seq analysis to investigate transcriptomic differences between the wild-type and du1-2018. For each genotype, 20 kernels were randomly collected from the middle portion of the ear and pooled as a single replicate, with three biological replicates for each genotype. Total RNA from kernels was extracted using TRIzol reagent, followed by assessment of RNA concentration and quality. Six cDNA libraries were constructed and sequenced on an Illumina HiSeq 4000 platform, yielding 150 bp paired-end reads. The clean reads were aligned against the maize B73 reference genome using STAR (v2.7.7a) [42]. The expression level of each gene was quantified as fragments per kilobase of transcript per million mapped reads (FPKM). Differential expression analysis was performed using DESeq2 based on raw read counts, and differentially expressed genes (DEGs) were defined as those with a false discovery rate (FDR) < 0.05 and |log₂(fold change)| > 1. All DEGs were subjected to Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses using the clusterProfiler package in R [43].

2.7. Amplification and Sequencing Analysis of Candidate Gene

Genes located within the mapping interval were extracted from the Ensembl database (https://plants.ensembl.org/index.html, accessed on 5 January 2025). Candidate genes were identified by integrating preliminary mapping results with RNA-seq data. Genomic sequences of the candidate genes were amplified with KOD FX DNA polymerase (Toyobo, Osaka, Japan) according to the manufacturer’s instructions (Supplementary Methods). The amplification products were detected on a 1% agarose gel and subsequently submitted to Sangon Biotech Co., Ltd. (Shanghai, China) for sequencing.

3. Results

3.1. Phenotypic Characteristics of the du1-2018 Mutant

The du1-2018 mutant produces kernels that are dull, glassy, slightly tarnished, and mildly shrunken, with a collapsed crown (Figure 1A). Observations of longitudinal and transverse kernel sections revealed that the du1-2018 mutant exhibited reduced endosperm size and a higher proportion of vitreous endosperm relative to the wild-type (WT). The hundred-kernel weight (HKW) of the du1-2018 mutant was significantly reduced by 7.94% compared with the WT. Scanning electron microscopy (SEM) revealed a marked increase in starch granules of reduced size and irregular shape in the floury endosperm of mature du1-2018 kernels compared with the WT (Figure 1B), similar to patterns reported for several previously characterized starch biosynthetic mutants [19,44,45,46]. We subsequently quantified the levels of starch, soluble sugars, and protein in mature kernels. The du1-2018 mutant exhibited a statistically significant 23.04% decrease in starch content compared with the WT (Figure 1C). Amylose content in the du1-2018 mutant was slightly higher than in the WT, but the difference did not reach statistical significance (Figure 1D). Soluble sugar and protein levels in the du1-2018 mutant were significantly higher than in the WT, by 28.60% and 17.80%, respectively (Figure 1E–F). These results collectively demonstrate that the du1-2018 mutation affects kernel development and storage reserve accumulation in the endosperm.

Figure 1.

Phenotypes and carbohydrate and protein contents of WT and the du1-2018 mutant. (A) Ear and kernel phenotypes. (B) SEM images of starch granules in the floury endosperm of mature kernels. (C) Total starch, (D) amylose, (E) soluble sugar, and (F) protein contents in mature kernels. Each value represents the mean  ±  SD of three independent biological replicates. Statistical significance was assessed using Student’s t-test (** p < 0.01).

3.2. Genetic Analysis and Preliminary Mapping of du1-2018

All F₁ kernels exhibited a normal phenotype. Chi-square analysis showed that the segregation ratios of WT and mutant phenotypes fit the expected 3:1 ratio in the F₂ population and the 1:1 ratio in the BC₁F₁ population (χ² < 3.84; Table S1), implying that a single recessive gene controls the du1-2018 phenotype. BSA-seq was conducted to identify the genomic region responsible for the du1-2018 mutation. Table S2 summarizes the sequencing statistics of the two bulks and their parental lines. A total of 1.11 billion clean reads were generated for the two bulks. These reads were subsequently aligned to the maize B73 reference genome, yielding mapping rates of 98.54% and 96.58% for the WT and mutant bulks, respectively, with average coverage depths of 43.8× and 32.3×. After integrating the parental sequencing data and filtering out low-coverage or inconsistent homozygous SNPs, 15,968,769 high-quality SNPs were retained for downstream analysis. Allele frequency differences between the two bulks were assessed using Δ(SNP-index) analysis. Based on the average ∆(SNP-index) values across the ten chromosomes, a significant peak was detected on chromosome 10 within the 17.35–75.32 Mb interval at the 95% confidence level (Figure 2A). To validate the BSA-seq results, we screened 15 publicly available SSR markers within the candidate region and identified two polymorphic markers, bnlg210 and umc1345, between the parental lines. Genotyping of 114 F₂ mutant individuals with the two polymorphic SSR markers showed that both markers were linked to the du1-2018 mutation. Subsequently, seven additional InDel markers within the candidate interval were designed from the sequencing data and used to genotype 230 F₂ mutant individuals. The sequences of these polymorphic markers are listed in Table 1. Linkage analysis revealed that du1-2018 was delimited to a 5.73 Mb region between markers InDel5516 and InDel5915, with three and two recombinants, respectively (Figure 2B).

Figure 2.

Mapping of the du1-2018 mutation. (A) Plot of the average ∆(SNP-index) values (Y-axis) across the ten maize chromosomes (X-axis). Blue and red lines represent the 95% and 99% confidence intervals, respectively. (B) Mapping of du1-2018 using a (du1-2018 × B73)F₂ population comprising 230 mutant individuals. The du1-2018 locus was mapped to the interval between InDel5516 and InDel5915 on chromosome 10.

Table 1.

Polymorphic markers used for genetic linkage analysis of the du1-2018 mutation.

Primer Name	Forward Primer Sequence	Reverse Primer Sequence
bnlg210	GCCTCGCACCAAGACATAATA	TGCCCCATTTGAGTAGACTTC
umc1345	ACCGCAGCAGCAGAGAAGAG	GAACATCTGGGTCACCTCTGTCAT
InDel5068	CAAGTTGGAGCCTCTCACCC	TTTGAGCTGTGCTCTCGCTG
InDel5075	TCTAGGGTGTGTTCAAGGCA	TGCACTTGGGGGAAATTCAGA
InDel5516	CCACGTTATTATTATTTTCCAGTCA	AATCGTAGCAATGTACGGGC
InDel5737	GGGCCACCTAATCTAAATCTCG	ATGAAATGTTGGCTATCATGAAC
InDel5915	GCCATGAGTGATCGAATGGCTG	TTGAGCTCCTCTATCGGGACG
InDel5927	GAATCGCCTATCAACAGCAAATCA	TTCTCACAATATGCATGCTGCC
InDel5978	TTCCATGTCATTTGCATGTCCCT	CATAGTTAGCCTGGCCTTGACT

3.3. Characterization of Starch Properties in the du1-2018 Mutant

Paraffin-embedded sections of 20 DAP kernels were stained with Safranin O/fast green to assess the starch accumulation. In the WT endosperm cells, amyloplasts were well developed and starch granules were densely arranged. In contrast, du1-2018 endosperm cells, particularly in the central region, contained fewer starch granules that were more loosely arranged (Figure 3A). SEM was employed to further examine the ultrastructure and surface morphology of starch granules from 20 DAP kernels. WT starch granules exhibited relatively uniform size and shape, being predominantly elliptical or rounded, with smooth, flat surfaces and occasional shallow ridges. In contrast, the majority of starch granules in du1-2018 were smaller, exhibiting an irregular or polyhedral shape with well-defined edges and corners, while a small fraction were spherical. The surfaces of some granules were rough, featuring shallow depressions and pores of varying sizes (Figure 3B). This increased heterogeneity in starch granules likely reflects delayed granule development and may account for their loosely packed arrangement. These observations indicate that starch accumulation in the du1-2018 endosperm is abnormal from the early stages of kernel development. We compared the distributions of starch granule number, volume, and surface area in mature kernels. All three distributions showed a single, unimodal peak in both genotypes. Starch granule diameters ranged from 5.2–35.3 μm in WT and 3.1–27.4 μm in du1-2018, with mean volume diameters of 15.4 μm and 10.9 μm, respectively (Figure 3C). These results suggest that starch granules in du1-2018 were considerably smaller than those in the WT, consistent with SEM observations of the mature endosperm (Figure 1B). Overall, the du1-2018 mutation exerts a profound effect on starch granule formation during kernel development, likely contributing to the markedly reduced starch content observed in the mutant.

Figure 3.

Comparison of starch granules in developing and mature kernels of WT and du1-2018 mutant. (A) Paraffin-embedded sections of developing endosperms from WT and du1-2018 kernels at 20 DAP. Starch granules are stained pink. (B) SEM images of purified starch granules from 20 DAP developing kernels. (C) Distributions of starch granule number, volume, and surface area in purified starch granules from mature kernels.

Isoamylase-debranched starches from mature WT and du1-2018 kernels were analyzed using GPC to determine their molecular weight distributions. The starch chain length distribution (CLD) pattern of debranched starches comprised three components: short amylopectin (Peak 1), medium–long and long amylopectin (Peak 2), and amylose (Peak 3). Distinct differences were observed in the CLD patterns of debranched starches between WT and du1-2018 (Figure 4A). In the du1-2018 mutant, Peak 2 was markedly reduced and barely detectable in the chromatograms. In contrast, Peak 2 exhibited a substantially larger area in the WT than in the mutant. The Peak 1/Peak 2 area ratio has been used as an indicator of amylopectin branching, with larger ratios indicating a higher degree of branching [19]. The Peak 1/Peak 2 area ratio was higher in the du1-2018 mutant (4.02) than in the WT (2.94), indicating increased amylopectin branching. Peak 3 was smaller in the WT than in the mutant, consistent with the differences in amylose content between the two genotypes. Thus, the du1-2018 mutation significantly altered the CLD pattern of starch. The CLD of amylopectin from mature WT and du1-2018 kernels was further analyzed using HPAEC-PAD. Amylopectin chains were fractionated into four categories: fa (A chains, DP 6–12), fb1 (B1 chains, DP 13–24), fb2 (B2 chains, DP 25–36), and fb3 (B3 chains, DP ≥ 37) [47,48]. In WT starch, A, B1, B2, and B3 chains accounted for 27.95%, 44.36%, 14.19%, and 13.50%, respectively (Figure 4B). In the du1-2018 mutant, B1 chains decreased by 8.32%, whereas A chains increased by 9.56%. B2 and B3 chains abundance remained largely stable, with slight increases relative to the WT. The average degree of polymerization (DP) of all chains was slightly higher in du1-2018 (21.28) than in the WT (21.16). The most notable alterations in du1-2018 were an increase in A chains with DP < 10 and a decrease in B chains with DP 13–24, compared with the WT (Figure 4C). Collectively, our results highlight the role of du1-2018 in regulating both starch accumulation and its fine structure in the maize endosperm.

Figure 4.

Starch molecular characteristics and amylopectin fine structure of mature WT and du1-2018 kernels. (A) GPC profiles of debranched starch. DP, degree of polymerization. (B) CLDs of debranched amylopectin as determined by HPAEC-PAD. (C) Differences in amylopectin CLDs between WT and du1-2018, calculated for each chain length as WT minus du1-2018 normalized values.

3.4. Transcriptome Alterations in du1-2018 Kernels

RNA-seq was conducted on 15 DAP kernels to compare transcriptome profiles between WT and du1-2018 and evaluate the genome-wide impact of the du1-2018 mutation. A total of 361.11 million clean reads were generated from six libraries. Each library generated more than 51.75 million clean reads, of which over 77.53% were uniquely aligned to the reference genome assembly (Table S3). In total, 326 differentially expressed genes (DEGs) were detected, with 150 upregulated and 176 downregulated in the WT relative to du1-2018 (Figure 5A and Table S4). Cluster heatmap analysis revealed that biological replicates within each genotype clustered tightly, indicating high consistency among replicates (Figure 5B). GO enrichment analysis indicated significant enrichment of the upregulated DEGs in eight GO terms (Figure 5C): proline dehydrogenase activity (GO:0004657); oxidoreductase activity, acting on the CH-NH group of donors (GO:0016645); aspartic-type peptidase activity (GO:0070001); aspartic-type endopeptidase activity (GO:0004190); proline catabolic process (GO:0006562); proline catabolic process to glutamate (GO:0010133); cytokinin dehydrogenase activity (GO:0019139); and glutamine family amino acid catabolic process (GO:0009065). KEGG pathway analysis revealed that among all pathways, only protein processing in the endoplasmic reticulum (zma04141) was significantly enriched for the upregulated DEGs, exhibiting the highest richness factor and lowest Padj value (Figure 5D). However, the 176 downregulated DEGs did not exhibit significant enrichment in any GO categories or KEGG pathways. Additionally, DEGs were enriched in pathways such as sesquiterpenoid and triterpenoid biosynthesis (zma00909) and zeatin biosynthesis (zma00908), although the enrichment was not statistically significant. Notably, no significant changes were observed in the expression of key starch biosynthetic genes except for Du1/ZmSSIII. Du1 exhibited very high expression in WT kernels, with an average FPKM of 36379.5, whereas the average FPKM in du1-2018 was only 2.3, with no detectable expression in one biological replicate. Interestingly, a large proportion of the DEGs were associated with amino acid and protein metabolism (Table S5). More specifically, we screened 13 DEGs across nine amino acid metabolic pathways, nine of which were upregulated in du1-2018.

Figure 5.

Transcriptome analysis of WT and du1-2018 kernels at 15 DAP. (A) Volcano plot showing DEGs identified between WT and du1-2018. (B) Heatmap showing hierarchical clustering of 326 DEGs based on expression levels. (C) GO enrichment analysis of the 150 upregulated DEGs. (D) KEGG pathway enrichment analysis of the 150 upregulated DEGs.

3.5. Identification of the Candidate Genes Underlying du1-2018

Functional annotations of the 44 genes located within the candidate interval were retrieved from the Ensembl database. RNA-seq data revealed that only three genes within the candidate interval, Zm00001eb412970, Zm00001eb412980, and Zm00001eb413290, exhibited significant differential expression between WT and du1-2018 kernels. An analysis of gene expression across different tissues based on the qTeller database showed that Zm00001eb413290 was highly expressed in developing kernels, whereas Zm00001eb412970 and Zm00001eb412980 were predominantly expressed in internodes and leaves. RNA-seq analysis further revealed that Zm00001eb413290 (Du1) was highly expressed in developing kernels of the WT, whereas its expression was almost undetectable in du1-2018. Moreover, du1 mutations have been reported to cause mature kernels to exhibit a “dull” phenotype [14,25], similar to the phenotype observed in du1-2018. Consequently, Zm00001eb413290 was considered the most likely candidate gene responsible for the du1-2018 mutation. The wild-type Du1 gene was sequenced using PCR-amplified genomic fragments, uncovering a 10.9 kb transcript region composed of 16 exons and 15 introns (Figure S1). Analysis against the B73 reference sequence revealed just a single SNP in the coding region of the WT. The corresponding locus from a homozygous du1-2018 plant was amplified and found to contain an aberrant sequence, likely an uncharacterized insertion, within the first 685 bp of the gene. Apart from this aberrant sequence, no differences were observed between WT and du1-2018 across the remainder of the gene. Several PCR-based approaches were employed to characterize the molecular structure of the mutation site; however, all attempts were unsuccessful. The aberration in the 5′ region of the gene may account for the failure of the du1-2018 allele to produce a detectable transcript.

4. Discussion

Starch biosynthesis is a multifaceted biological process requiring the concerted action of a series of key enzymes [49]. Studies of maize mutants with abnormal endosperm phenotypes have been instrumental in identifying key enzymes and regulatory proteins contributing to starch biosynthesis and in elucidating the molecular mechanisms underlying this process. In this study, we characterized a novel naturally occurring du1-2018 mutant exhibiting slightly tarnished, glassy, and dull kernels. The du1-2018 mutation is likely a severe loss-of-function allele of the du1 gene, given that its transcript is virtually undetectable in developing kernels. The precise molecular nature of the mutation in du1-2018 has not yet been determined. The phenotypic manifestations of du1 mutations appear to be modulated by the genetic background. du1 mutants typically exhibit a mild phenotype, characterized by slightly glassy and dull kernels. However, when pyramided with other maize mutations such as su1 [14] and isa2 [21], the resulting double mutants exhibit a much more severe starch biosynthetic phenotype than either corresponding single mutants, indicating strong genetic interactions. In the (du1-2018 × B73)F₂ population, mutant kernels exhibited a slightly altered and less severe phenotype compared with du1-2018 in its original genetic background (Figure S2). This observation further supports the influence of genetic background. In addition to kernel phenotypes, du1-2018 mutants exhibited germination defects and consistently displayed lower germination rates than the WT under both greenhouse and field conditions. Further analyses indicated that the du1-2018 mutation adversely affected starch granule formation and additionally modulated the accumulation of storage compounds.

Accumulating evidence indicates that different SS isoforms perform specific roles in starch biosynthesis [50]. SSI and SSIII are the principal SS isoforms responsible for most soluble SS activity in the developing endosperms of maize and rice [12,13]. SSIIa has been detected in the soluble fraction of developing maize endosperm, together with SSI and SSIII [51]. In contrast to SSI and SSII, the functional roles of SSIII remain less clearly defined [26,52]. In the present study, the du1-2018 mutant showed a modest increase in amylose content relative to the WT, together with a pronounced decrease in amylopectin synthesis, consistent with previous reports that du1 mutations are generally associated with increased amylose accumulation across diverse maize genetic backgrounds [17,18,19,22]. This elevated amylose content is likely attributable to reduced amylopectin synthesis, which may lead to a redirection of carbon flux toward amylose biosynthesis [53]. Elevated GBSSI protein abundance and enzymatic activity have been consistently reported in SSIIIa-deficient rice endosperm [54], further supporting this proposed model. With regard to amylopectin, studies of SSIII-deficient lines in maize and rice have demonstrated a reduction in long, cluster-spanning B chains relative to the wild-type, indicating a specific role for SSIII in the elongation of B2 and B3 chains, as well as longer amylopectin chains [21,22,55]. Previous research has reported that SSIII deficiency in maize also alters the abundance of short amylopectin chains, suggesting the involvement of SSIII in the synthesis of short A and B chains. Zhu et al. demonstrated that du1-Ref and du1-M3 mutants have an increased abundance of short amylopectin chains (DP < 25) in comparison with the wild-type line W64A [22]. Lin et al. showed that the effects of du1 mutations on amylopectin structure were discontinuous. Specifically, DP7–9 chains were consistently decreased, DP11–15 chains were consistently increased, DP17–18 chains were generally decreased, and DP21–33 chains were increased [21]. In this study, the du1-2018 mutation led to a decreased abundance of intermediate-length chains (DP13–24) and an increased frequency of very short chains (DP6–9). Interestingly, these alterations more closely resemble those observed in SSIIa-deficient mutants [51]. SSII and SSIII are thought to exhibit partially overlapping functions in rice grains [56] and Arabidopsis leaves [57]; however, whether a similar functional relationship exists in maize remains unclear. These observed discrepancies indicate that SSIII deficiency in maize may lead to distinct changes in amylopectin fine structure across different genetic backgrounds.

The pleiotropic effects of SSIII deficiency on the composition and molecular structure of starch cannot be fully accounted for by its catalytic role alone. The functional scope of SSIII appears to extend beyond its enzymatic activity, encompassing distinct regulatory interactions with other starch biosynthetic enzymes [58]. The du1 mutation significantly diminished SSIII activity and concomitantly reduced SBEIIa activity in maize kernels [28]. In maize du1 mutants, total SS activity in soluble endosperm extracts is increased despite the loss of SSIII function, primarily due to a specific enhancement of SSI activity [12]. The identification of a high-molecular-weight enzyme complex comprising SSIII, SSIIa, SBEIIa, and SBEIIb in maize endosperm provides additional evidence of regulatory interactions among specific starch biosynthetic enzymes [29,30]. Direct interactions between SSI and SSIII have been confirmed through multiple protein–protein interaction methods [29,30]. Furthermore, in vitro studies of recombinant maize SSs indicate that SSIII negatively regulates SSI activity and may also exert a weaker negative effect on SSIIa [58]. Together, these results add to the growing evidence that SSIII plays a central role in regulating or coordinating other starch biosynthetic enzymes, thereby influencing starch biosynthesis. The indirect and potentially pleiotropic effects of SSIII deficiency remain to be elucidated. However, the situation is likely more complex due to functional overlap among different SS isoforms, the formation of enzyme complexes, and interactions among starch biosynthetic enzymes [26,52]. Consequently, the molecular basis of SSIII’s functional roles in starch biosynthesis is not yet fully understood.

RNA-seq analysis indicated that, except for Du1, the expression of key starch biosynthetic genes was largely unchanged between WT and du1-2018. This suggests that their expression may be primarily regulated at the post-transcriptional level. A substantial proportion of the DEGs in du1-2018 were associated with amino acid and protein metabolism, potentially contributing to the observed changes in protein content. Most of the DEGs related to amino acid metabolism were upregulated in du1-2018, notably including glutamate synthase 2 (Zm00001eb295220) and alanine aminotransferase 2 (Zm00001eb313780), two key enzymes in the central pathway for glutamate biosynthesis. Collectively, these results indicate that the du1-2018 mutation alters nitrogen-related metabolic pathways. Previous studies have reported that DEGs in starch mutants such as Shrunken2 (Sh2) and Brittle2 (Bt2) are predominantly involved in carbohydrate and amino acid metabolic pathways, reflecting the coordinated regulation of carbon and nitrogen pathways essential for maize endosperm development [59,60,61]. In these starch mutants, impaired starch biosynthesis appears to redirect carbon flux from starch accumulation toward sugar metabolism, indicating a shift in carbon partitioning. A transcriptional regulatory network is thought to temporally coordinate starch and storage protein synthesis during maize endosperm development [8,62]. However, the molecular mechanisms underlying this coordination remain poorly understood. Pyruvate orthophosphate dikinase (PPDK) has been suggested to act as a regulator of starch/protein balance, likely due to the inhibitory effect of inorganic pyrophosphate (PPi) on ADP-glucose pyrophosphorylase (AGPase) [5,63]. Notably, a small portion of PPDK localized in amyloplasts can stably associate with AGPase, SSIII, SSIIa, SBEIIa, and SBEIIb in maize [30]. Integrating enzyme activity assays with proteomic and metabolomic analyses could provide comprehensive insights and help clarify the potential role of SSIII in regulating carbon/nitrogen balance in developing endosperm.

SSIII deficiency leads to pronounced changes in the physicochemical properties of maize starch, which are likely attributable to alterations in its composition and fine structure [18,20]. These unique properties of SSIII make it a valuable genetic resource for developing specialty maize, including high-amylose varieties. In addition, SSIII acts as a genetic modifier influencing sugar accumulation, highlighting its potential for improving the flavor of sweet corn [14,21,64]. SSIII is a major starch synthase catalyzing glucan chain elongation and also serves as a scaffolding protein coordinating various components of the starch biosynthesis pathway [30,65]. Further research is necessary to clarify the broader regulatory roles of SSIII, encompassing both starch biosynthesis and additional metabolic pathways. Genome-editing technologies are expected to enable systematic studies of different combinations of starch biosynthetic genes, thereby advancing our understanding of the regulatory mechanisms of starch biosynthesis and facilitating targeted improvements in maize grain quality.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/genes17020250/s1, Supplementary Methods; Table S1: Chi-square test of segregation of normal and mutant kernels in the F₂ and BC₁F₁ populations; Table S2: Summary of Illumina sequencing data from parental lines and two bulks for BSA-seq analysis; Table S3: Summary of RNA-seq data generated in this study; Table S4: List of the DEGs between WT and du1-2018; Table S5: List of the DEGs associated with amino acid and protein metabolism; Figure S1: Sequence of the transcribed region of Du1 in the WT; Figure S2: Kernel phenotypes in the (du1-2018 × B73)F₂ population.

Author Contributions

M.Z. conceived and designed the research, conducted the experiments and data analysis, drafted the manuscript, and acquired funding. X.L. performed the bioinformatics analysis. Z.S., X.Y., Y.G., and X.Z. conducted experiments and participated in data analysis. Q.H. conceived the research and revised the manuscript. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

All data are available in the article and its Supplementary Materials. Further inquiries can be directed to the corresponding authors.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

This research was funded by the Natural Science Foundation of Sichuan Province (Grant No. 2024NSFSC1210) and the High-Level Talent Introduction Program of Chengdu Normal University (Grant No. YJRC2020-23).