shrunken4 is a mutant allele of ZmYSL2 that affects aleurone development and starch synthesis in maize

Abstract

Minerals are stored in the aleurone layer and embryo during maize seed development, but how they affect endosperm development and activity is unclear. Here, we cloned the gene underlying the classic maize kernel mutant shrunken4 (sh4) and found that it encodes the YELLOW STRIPE-LIKE oligopeptide metal transporter ZmYSL2. sh4 kernels had a shrunken phenotype with developmental defects in the aleurone layer and starchy endosperm cells. ZmYSL2 showed iron and zinc transporter activity in Xenopus laevis oocytes. Analysis using a specific antibody indicated that ZmYSL2 predominately accumulated in the aleurone and sub-aleurone layers in endosperm and the scutellum in embryos. Specific iron deposition was observed in the aleurone layer in wild-type kernels. In sh4, however, the outermost monolayer of endosperm cells failed to accumulate iron and lost aleurone cell characteristics, indicating that proper functioning of ZmYSL2 and iron accumulation are essential for aleurone cell development. Transcriptome analysis of sh4 endosperm revealed that loss of ZmYSL2 function affects the expression of genes involved in starch synthesis and degradation processes, which is consistent with the delayed development and premature degradation of starch grains in sh4 kernels. Therefore, ZmYSL2 is critical for aleurone cell development and starchy endosperm cell activity during maize seed development.

maize, iron transporter, seed development, aleurone, starch synthesis

Introduction

Maize (Zea mays L.) seeds are economically important, providing a significant source of metabolizable energy for humans and livestock. The maize seed is composed of an embryo and endosperm (the products of double fertilization), together with peripheral maternal tissue (Sabelli and Larkins 2009; Olsen 2020). Through rapid mitosis and differentiation, the zygote develops into an embryo containing the scutellum, leaf primordia, shoot apical, meristem, and root apical meristem. At the same time, the initial triploid endosperm cell becomes a typical endosperm, containing a basal endosperm transfer layer (BETL), embryo-surrounding region (ESR), starchy endosperm, and aleurone layer (Sabelli and Larkins 2009; Gontarek and Becraft 2017; Olsen 2020). The BETL and ESR provide pathways for nutrient transport to the endosperm and embryo, respectively. The starchy endosperm is the primary region for the storage of starch and protein, which accumulate and provide nutrients for seed development and germination. The aleurone layer, the outermost monolayer of cells, stores vitamins and minerals, and provides signals to starchy endosperm cells.

The aleurone layer is associated with the expansion of the entire endosperm (Olsen 2004). Aleurone cell differentiation in the periphery of maize endosperm was detected as early as 5 days after pollination (DAP) using the aleurone layer marker BETL9-like (formerly known as AL9) (Royo et al. 2014). Aleurone cells are rectangular and highly vacuolated at 8 DAP and begin to accumulate dense cytoplasmic inclusions (including aleurone grains and protein-carbohydrate bodies) at 12 DAP (Becraft and Asuncion-Crabb 2000; Gontarek and Becraft 2017).

Multiple genes involved in aleurone cell identity have been identified, including Crinkly4 (CR4), Defective kernel1 (DEK1), and Supernumerary aleurone1 (SAL1), which encode a receptor-like kinase, calpain, and vacuolar sorting protein, respectively (Becraft et al. 1996; Lid et al. 2002; Shen et al. 2003). dek1 and cr4 mutants exhibit a loss of aleurone cells, indicating that DEK1 and CR4 encode proteins with positive regulatory functions as receptors (Becraft et al. 1996, 2002; Lid et al. 2002; Tian et al. 2007). Analysis of the sal1 mutant, which shows multilayered aleurone cells, revealed that SAL1 negatively regulates DEK1 or CR4 activity by affecting membrane vesicle trafficking (Shen et al. 2003). The multilayered aleurone cell mutant thick aleurone1 (thk1) harbors a deletion of NEGATIVE ON TATA-LESS1, indicating that Thk1 functions as a negative regulator of cell patterning (Wu et al. 2020). Transcription factors, such as VIVIPAROUS1 (VP1) and Naked endosperm (NKD), are also involved in regulating aleurone cell development (Cao et al. 2007; Yi et al. 2015). However, the molecular mechanism and regulatory pathway of aleurone cell differentiation are not fully understood.

The aleurone layer is the major site for the storage of minerals (including iron) in the maize endosperm (Cheah et al. 2019). Iron is widely involved in photosynthesis, respiration, and the regulation of enzyme activity (Kobayashi and Nishizawa 2012). Iron uptake in graminaceous plants primarily involves a chelation-based strategy (Morrissey and Guerinot 2009; Kobayashi and Nishizawa 2012). Plants secrete ferric acid carriers (i.e., mugineic acid family phytosiderophores) into the rhizosphere soil. These carriers then chelate iron and are transported to the root by transporters such as YS1 (Yellow Stripe1) and YSLs (Yellow Stripe-Like) (Curie et al. 2001; Koike et al. 2004). OsYSL2 in rice (Oryza sativa) and AtOPT3 in Arabidopsis (Arabidopsis thaliana) are involved in the long-distance transport of iron in vegetative organs and influence iron content in seeds (Ishimaru et al. 2010; Zhai et al. 2014). ZmYSL2 is required for iron accumulation and distribution in maize seeds, which likely involves the transfer of iron from the endosperm to embryo (Zang et al. 2020). However, the physiological functions of YSLs in cell development and identity remain unknown.

Starch, accounting for ∼70% of kernel weight, provides carbohydrates as a source of energy for germinating seeds (Hannah and Boehlein 2017). Starch biosynthesis requires a series of metabolic enzymes, including adenosine 5’ diphosphate-glucose (ADP-Glc) pyrophosphorylase (AGPase), starch synthase, starch branching enzyme, and starch debranching enzyme (Keeling and Myers 2010; Zhang et al. 2019; He et al. 2020). The starch degradation process, another key process affecting starch accumulation, is regulated by amylases, glucosidases, and debranching enzymes (Smith et al. 2005).

Maize shrunken (sh) mutants exhibit shrunken and collapsed endosperm due to substantially reduced starch content. The sh1 mutant produces a defective sucrose synthase, and sh2 contains a defective AGPase subunit (Chourey and Nelson 1976; Bhave et al. 1990). In shrunken4 (sh4), another classic mutant with impaired starch accumulation, the endosperm shows reduced activities of multiple enzymes involved in starch synthesis and phosphate metabolism (Tsai and Nelson 1969; Ozbun et al. 1973; Doehlert and Kuo 1990). However, the gene underlying sh4 has not yet been cloned, and the role of Sh4 in regulating starch metabolism is still unknown.

In this study, we performed map-based cloning of the gene underlying the sh4 mutant and found that it is a mutant allele of ZmYSL2. ZmYSL2 preferentially accumulates in aleurone, sub-aleurone, and embryo cells and is essential for proper iron distribution and accumulation in the aleurone layer and embryo. Cytological and biochemical analyses of sh4 revealed that ZmYSL2 is essential for aleurone cell development and starch accumulation in starchy endosperm cells. In addition to its role in regulating iron distribution between the endosperm and embryo (Zang et al. 2020), we demonstrate that ZmYSL2 is involved in aleurone cell fate specification and the proper functioning of starchy endosperm cells. Our findings suggest that iron accumulation controlled by the ZmYSL2 transporter is involved in the determination of aleurone cell identity and starch synthesis in maize.

Materials and methods

Plant materials

The maize sh4 mutant (sh4-518A) was obtained from the Maize Genetics Cooperation Stock Center (http://maizecoop.cropsci.uiuc.edu/ (Accessed: 2021 May 8)). The sh4 stock crossed to the W22 inbred line and subsequently self-pollinated to produce an F₂ population. Developing kernels were collected from three W22 plants that grew well in the Experimental Station in Shangzhuang, China Agricultural University, Beijing. Tobacco (Nicotiana benthamiana) plants were grown in a culture chamber with a 16-h: 8-h light: dark photoperiod at 25°C.

Vector construction and maize transformation were reported previously (Qi et al. 2016). The 20 (GGACGAGATCGACAAGTGCG) bp on the second exon of Zm00001d017427 was selected as the target sequence. The hybrid (pBpA) of pB and pA lines was taken as the recipient for transformation experiments.

Measurement of starch, protein, and lipid contents

The mature kernels of wild type and sh4 were obtained from the same segregating ears. After removing the pericarp and embryo, 20 endosperms of the wild type and sh4 were pooled and ground into powder as a single replicate, respectively. Three biological replicates were prepared for the following analysis as described previously (Wang et al. 2011).

After drying to constant weight, a 100-mg sample was used for starch quantification following the manufacturer’s protocol of amyloglucosidase/a-amylase starch assay kit (Megazyme). For protein quantification, a 50-mg powder was used for total protein, zein, and nonzein extraction and then measured with an abicinchoninic acid standard kit (Thermo Fisher Scientific). The lipid content was measured as described previously (Wang et al. 2011).

Histological analysis

Five developing wild-type and sh4 kernels were used for paraffin section or transmission electron microscopy analysis. Three well-filled ears were harvested as biological repeats. Detailed experimental procedures were referred to a previous report (He et al. 2019).

Genetic mapping of sh4 locus

The genetic mapping of the sh4 locus used 6000 kernels from an F₂ population of the cross between the sh4 and the B73 inbred line. The Sh4 was mapped between the molecular marker SSR045-3 (15 recombinants) and SSR873-1 (24 recombinants) on chromosome 5 (from 194.938 to 196.069 M based on the maize B73 RefGen_v4). Additional markers were used to narrow the interval to a ∼43-Kb (from 195.723 to 195.766 M) region between the marker Indel385-1 (1 recombinant) and SSR592-3 (2 recombinants). The primers for the molecular markers are listed in Supplementary Data S4.

Immunoblot analysis

A specific cDNA sequence of ZmYSL2 (1–270 bp from ATG, encoding the 1–90 amino acids) was cloned into pGEX-4T-1. The production and purification of the antibody were performed by Shanghai ImmunoGen Biological Technology. Total proteins were separated by SDS-PAGE and then transferred to polyvinylidene difluoride membrane (0.45 μm; Millipore Sigma, USA). After blocked using 5% skim milk, the membrane was incubate with primary [anti-ZmYSL2 (1:1000) or anti-actin (1:5000; Abclonal Technology)] and secondary [anti-rabbit (or anti-rat) LgG conjugated to horseradish peroxidase (1:5,000; Abclonal Technology)] antibodies.

Voltage-clamp experiment in Xenopus laevis oocytes

The voltage-clamp experiment was performed as described previously (Zhang et al. 2016). Full-length CDS of maize ZmYSL2 gene was constructed into pGEMHE vector by BamHI and EcoRI restriction sites. ZmYSL2 cRNA was microinjected into Xenopus oocytes. Oocytes injected with an equal amount of RNAase-free H₂O were used as a negative control. After 2–3 days in the ND96 incubation buffer, the whole-oocyte current-voltage recordings were measured using an Axoclamp 900 A patch-clamp equipment (Axon Instruments, USA) with a 1440 A digitizer. The glass microelectrode filled with 3 M KCl is submerged in ND96 bath buffer (containing 96 mM KCl, 1.8 mM MgCl₂, 1.8 mM CaCl₂, and 10 mM HEPES-NaOH, pH 7.2) to record Fe²⁺ and Zn²⁺ current. The whole-cell current filtered by 20 MHz low-pass filter was analyzed by Clampex10.2 software (Axon Instruments).

Immunocytochemical analysis

The in situ ZmYSL2 protein localization analysis was slightly modified from previously described (Paciorek et al. 2006). Developing kernels were harvested from an F₂ segregating ear. The kernels were trimmed and fixed in 4% paraformaldehyde for 60 minutes at room temperature. The samples were dehydrated in a gradient of ethanol [50, 60, 75, 95, and 100% ethanol in 1× PBS (v/v)] and xylene solution [25, 50, 75, and 100% xylene in ethanol (v/v)]. The material was then embedded in a paraffin block. Paraffin sections (10 μm) were gained from a microtome (RM2265; Leica). After dewaxing with xylene, the sections were blocked in 2% BSA solution, followed by incubation with primary antibodies [anti-ZmYSL2 (1:300) or anti-actin (1:300; Abclonal Technology)] in a humid chamber overnight at 4°C. Then, it was incubated with anti-rabbit IgG (Fc) alkaline phosphatase conjugate (1:1250; Promega) as a secondary antibody. NBT/BCIP (1:50, Roche) was used to visualize the signal.

Elemental analysis

Elemental composition was analyzed in the endosperm, embryo, and whole kernel from wild type and sh4 mature kernels. The sample from 10 kernels was pooled and ground into the powder as a single replicate. Three biological replicates were used for the elemental analysis. After being digested in HNO₃ and diluted to an appropriate concentration, the elemental composition was analyzed on a PerkinElmer NexION 350 D inductively coupled plasma-mass spectrometry.

Iron localization

The longitudinally bisected kernels were incubated in the 1% H₂O₂ in ddH₂O [v/v] for 10 minutes. The sample was washed three times with ddH₂O and transferred in 2% HCl and 2% potassium hexacyano ferrate II trihydrate (Solarbio) for 20 minutes. After being washed with ddH₂O, the stained kernel was imaged using a Leica microscope (DM2000LED).

RNA-seq and qPCR

RNA sample pools (12 kernels per sample) of wild-type or sh4 endosperms were collected from an F₂ segregating ear, respectively. Three independent ears were used for transcript profiles. An RNAprep Pure Plant Kit (Tiangen Biotech, China) was used to extract total RNA. Libraries are prepared and sequenced using an Illumina NovaSeq 6000 (berrygenomics), which produced ≥20 million reads per sample. The clean reads were mapped to maize B73 RefGen_v4.47 using STAR (version: 2.7.1a). The expression level of genes was estimated using cufflinks (version: v2.2.1) and cuffdiff (version: 2.2.1).

For qRT, cDNA was produced from 1 μg total RNA using a FastKing gDNA Dispelling RT SuperMix (Tiangen Biotech, China). Primer pairs were designed in Quantiprime (https://quantprime.mpimp-golm.mpg.de/?page=home (accessed: 2021 May 8)). Gene fragments were amplified using Hieff qPCR SYBR Green Master Mix (Yeasen Biotech, China) on a CFX Connect Real-Time System (Bio-Rad). Gene expression data were evaluated by ΔCt (threshold cycle) method, based on a maize Ubiquitin expression level.

Data availability

Sequence data can be found in MaizeGDB (https://www.maizegdb.org/ (accessed: 2021 May 8)) or GenBank databases (https://www.ncbi.nlm.nih.gov/genbank/ (accessed: 2021 May 8)) under following accession numbers: Sh4/ZmYSL2, Zm00001d017427; ZmYS1, Zm00001d017429; OsYSL2, XP_015625410.1; AtYSL2, NP_197826.2. RNA-Seq data are available from the National Center for Biotechnology Information Sequence Read Archive (http://www.ncbi.nlm.nih.gov/sra/ (accessed: 2021 May 8)) under accession PRJNA649607. Supplementary files are available at figshare: https://doi.org/10.25386/genetics.14498529 (accessed: 2021 May 8).

Results

The maize kernel mutant sh4 exhibits a starch-deficient phenotype

We obtained the classic sh4 mutant from the Maize Genetics Cooperation Stock Center. We then crossed the mutant to the inbred line W22 and self-pollinated it to generate F₂ ears (Figure 1A). The ratio of wild-type to mutant seeds in F₂ ears was approximately 3:1, indicating that sh4 is controlled by a recessive gene. Mature sh4 homozygous kernels had thin, shrunken, opaque endosperm compared to wild-type kernels (Figure 1, A and B). The upper part of the sh4 endosperm was more severely impaired than the lower part (Figure 1B). sh4 embryos were also shorter and thinner than the wild type, with a lower germination rate (∼30%). After germination, sh4 seedlings were developmentally delayed (Figure 1C). Only a small percentage of immature sh4 embryos could be rescued by in vitro culture on Murashige and Skoog medium (Supplementary Figure S1A), suggesting that the sh4 mutation is partially lethal.

Figure 1

The maize kernel mutant shrunken4 (sh4) exhibits a starch-deficient phenotype. (A) Mature F₂ ear from a cross between sh4 and W22. Arrows indicate sh4 kernels. Bar = 1 cm. (B) Phenotypic observation and longitudinal sections of wild-type (WT) and sh4 mature kernels from a segregated F₂ ear. The mature kernels were viewed under natural light (left), under transmission light (middle), and as longitudinal sections (right). En, endosperm; Em, embryo. Bars = 1 mm. (C) Germination test of WT and sh4 mature kernels (7 DAG). Bar = 1 cm. (D) Developing F₂ ear of sh4 × W22. Arrows indicate sh4 kernels. Bar = 1 cm. (E) Longitudinal sections of developing wild-type and sh4 kernels at 15 DAP. Bar = 1 mm. (F,G) Comparison of the 100-kernel weight of randomly selected mature WT and sh4 kernels (F) and the starch contents of mature WT and sh4 endosperm from individual kernels (G). Values are means ± SE; n = 3 (***, P < 0.001; Student’s t-test).

We observed developing kernels of the wild-type and sh4 mutant plants from 9 to 21 DAP (Figure 1D and Supplementary Figure S1B). The sh4 kernels did not differ from the wild type in appearance and size during the early stage of development (9 DAP). However, homozygous sh4 kernels became distinguishable from the wild type on segregating ears due to their light pigmentation beginning at 12 DAP. After 18 DAP, sh4 kernels were smaller than those of the wild type, and this difference increased with time (Supplementary Figure S1, B and C). During kernel development, sh4 embryos were smaller than wild-type embryos at all time points (Figure 1E).

The 100-kernel weight of sh4 was only 22.0% that of the wild type (Figure 1F). Biochemical analysis of mature endosperm showed that the starch, protein, and lipid contents per kernel of sh4 endosperm were more than 60% lower than in the wild type (Supplementary Figure S2, A and B). The levels of these storage nutrients were also significantly reduced in sh4 endosperm when measured per weight (Supplementary Figure S2, C–E). These results indicate that sh4 kernels contain starch-deficient endosperm.

Positional cloning and genetic confirmation of Sh4

To clone Sh4, we performed map-based cloning using 6000 sh4 homozygous kernels from the F₂ mapping population. We initially mapped Sh4 to a 1.13-Mb interval between molecular markers SSR045-3 and SSR873-1 on chromosome 5 (Figure 2A). After developing additional markers, we narrowed the mapping interval down to a ∼43 Kb interval (from 195.723 to 195.766 M based on the B73 RefGen_v4 reference genome) between markers Indel385-1 and SSR592-3. This interval contained only two predicted protein-coding genes, Zm00001d017426 and Zm00001d017427 (Figure 2A). A comparison of the sequences of these two genes between the wild type and sh4 revealed that Zm00001d017427 contains a single-base C insertion 1945 bp downstream from the start codon in the coding sequence in sh4, resulting in a frameshift and premature stopping of translation (Figure 2B). We excluded Zm00001d017426 from further analysis because no loss-of-function mutations in this gene were identified in sh4 compared to the wild type. Therefore, we designated Zm00001d017427 as the candidate gene for Sh4.

Figure 2

Positional cloning and genetic confirmation of Sh4. (A) The sh4 mutant was crossed to the B73 inbred line, and the progeny were selfed to obtain F₂ ears. A total of 6000 sh4 homozygous kernels from the F₂ population were analyzed. The numbers below each molecular marker indicate the ratio of recombinant kernels in the tested population. The sh4 locus was narrowed down to a ∼43-Kb interval on chromosome 5 containing two candidate genes: Zm00001d017426 and Zm00001d017427. (B) Schematic of the structure and mutation site of the Zm00001d017427 gene. Lines and black boxes indicate introns and exons, respectively. The open boxes indicate the 3’ and 5’ untranslated regions. (C–E) Functional complementation of sh4 via transformation. (C) Characteristic kernels with wild type (1–10) and shrunken and opaque mutant (11–20) phenotypes from F₂ ears produced from a cross between a Zm00001d017427-ORF-expressing transgenic plants (T₀) and an sh4 heterozygous plant. The kernels were observed under natural light (1–5 and 11–15) and transmission light (6–10 and 16–20). (D) Detection of homozygous sh4 kernels using the molecular marker SSR873-1. (E) Identification of transgenic kernels using primers for the Bar gene. −, water control; +, Zm00001d017427 transgene construct. (F,H) CRISPR-Cas9-based mutation of Zm00001d017427 and allelism test with sh4. (F) The targeted sequence in the second exon of Zm00001d017427 using CRISPR/Cas9. gRNA, guide RNA; PAM, protospacer-adjacent motif. (G) Allelism test ear produced from a cross between sh4-cas9-1/+ and sh4/+. Arrows indicate mutant kernels. Bar = 1 cm. (H) Randomly selected kernels from (G). Bar = 1 cm.

We generated transgenic lines expressing the open reading frame (ORF) of Zm00001d017427 driven by its native promoter. We performed genetic complementation of sh4 by crossing a heterozygous sh4 plant with two independent transgenic lines (Figure 2, C–E). We used a molecular marker (SSR873-1) linked to the sh4 locus to identify kernels with homozygous sh4 alleles (Figure 2D). Introducing a copy of Zm00001d017427 via transformation rescued the mutant phenotype of sh4 (Figure 2E), indicating that sh4 was genetically complemented by Zm00001d017427.

We generated knockout alleles of Zm00001d017427 via CRISPR/Cas9-mediated gene editing (Qi et al. 2016) using guide RNA (gRNA) targeted the second exon of Zm00001d017427 (Figure 2F). Two independent knockout lines were generated: sh4-cas9-1, containing a single-nucleotide “T” insertion; and sh4-cas9-5, harboring a single-nucleotide “A” deletion and an eight-“A” insertion. Both mutations caused frameshifts at the target sequence and disrupted Sh4 function (Figure 2F). These homozygous knockout kernels showed similar phenotypes to sh4 (Supplementary Figure S3). We performed allelism tests by crossing the knockout lines (sh4-cas9-1 and sh4-cas9-5) with sh4 heterozygous plants (Figure 2G). Randomly selected kernels on the resulting ears with a shrunken, opaque phenotype contained both the sh4 and sh4-cas9 alleles (Figure 2H), indicating that sh4-cas9 cannot complement sh4. Taken together, these results indicate that Zm00001d017427 is Sh4.

Sh4 encodes ZmYSL2, which shows iron transporter activity in Xenopus oocytes

Sh4 (Zm00001d017427) was predicted to contain eight exons and seven introns, with a 2181-bp coding sequence of the canonical transcript according to the maize B73 RefGen_v3 (GRMZM2G135291_T01) and RefGen_v5 (Zm00001eb248990_T004) genomes (Figure 2B and Supplementary Figure S4). Sh4 was predicted to encode a membrane protein containing 14 putative transmembrane domains according to TMHMM 2.0 (http://www.cbs.dtu.dk/services/TMHMM/ (accessed: 2021 May 8)). A protein Basic Local Alignment Search Tool (BLASTp) search revealed that Sh4 belongs to the Yellow Stripe-Like subfamily and was named ZmYSL2 (Zang et al. 2020). Phylogenetic analysis indicated that ZmYSL2 homologs are highly conserved in plants and form two separate clades in angiosperms, one in monocots and one in dicots (Figure 3A and Supplementary Data S1).

Figure 3

Transport activity of Sh4/ZmYSL2. (A) Phylogenetic analysis of Sh4/ZmYSL2 and its homologs. The neighbor-joining tree was constructed with MEGA 7.0. The numbers at the branches indicate the percentage of 1000 bootstraps. Scale bar indicates the average number of amino acid substitutions per site. (B) Comparison of iron concentration in fet3fet4 yeast cells expressing empty vector, ZmYSL2, or YS1. Values are means ± SE; n = 3 (**, P < 0.01; ***, P < 0.001; Student’s t-test). (C) Results of voltage-clamp experiments analyzing ZmYSL2 in X. laevis oocytes. The oocytes were sequentially examined in three different buffers [ND96 buffer as a control, Zn²⁺-NA (100 μM) in ND96 buffer, and Fe²⁺-NA (100 μM) in ND96 buffer] after 15 minutes of washing with ND96 buffer. The current-voltage curves were recorded in oocytes expressing ZmYSL2 in control, Zn²⁺-NA, and Fe²⁺-NA buffer.

Because YS1/YSL proteins were predicted to function as iron transporters (Curie et al. 2001; Koike et al. 2004), we investigated the iron transport activity of ZmYSL2 in the yeast (Saccharomyces cerevisiae) mutant strain fet3fet4 (Figure 3B). We measured iron contents in yeast cells by inductively coupled plasma mass spectrometry (ICP-MS). fet3fet4 cells expressing ZmYSL2 or YS1 had significantly higher iron contents than those expressing the empty vector.

To further verify the metal transport activity of ZmYSL2, we performed voltage-clamp experiments using X. laevis oocytes. The full-length ORF of ZmYSL2 fused to the N-terminus of GFP was cloned into the pGEMHE expression vector. We generated ZmYSL2 complementary ribonucleic acid (cRNA) using an in vitro transcription system, injected it into oocytes, and measured ion transport activity in ND96 recording buffer using the two-electrode voltage-clamp method (Zhang et al. 2016). ZmYSL2-expressing cells generated larger negative (i.e., inward) currents in Fe²⁺-nicotinamide (NA) or Zn²⁺-NA than in ND96 buffer (negative control). In addition, oocytes expressing ZmYSL2 had higher Fe²⁺ (∼2.63 times) and Zn²⁺ (∼1.96 times) uptake rates in NA than the negative control (Figure 3C). These results demonstrate that ZmYSL2 is a functional metal transporter for iron and zinc in oocytes.

ZmYSL2 is concentrated in the aleurone layer

We raised a polyclonal antibody against the N-terminal portion (amino acids 1–90) of ZmYSL2 and measured ZmYSL2 protein accumulation (Supplementary Figure S5A). ZmYSL2 accumulated throughout kernel development (6–30 DAP), with a peak at ∼21 DAP (Figure 4A). In kernels, ZmYSL2 was much more highly expressed in the endosperm than in the embryo and pericarp (Figure 4B).

Figure 4

Sh4/ZmYSL2 is concentrated in the aleurone layer. (A,B) ZmYSL2 protein content in developing kernels over time (A) and in different tissues (B). α-Actin was used as an internal standard. The developing kernels in (A) were collected at different stages and labeled according to DAP. Pericarp (PE), embryo (Em), and endosperm (En) samples in (B) were collected at 15 DAP. (C) Immunoblot comparing the ZmYSL2 protein content in WT, sh4, and sh4-cas9-1 endosperm. α-Actin was used as an internal standard. (D) Immunostaining with anti-ZmYSL2 antibody in a longitudinal section of a 15 DAP developing WT kernel. Bar = 1 mm. The inset shows an enlarged image of the boxed region; bar = 100 μm. SC, scutellum; LP, leaf primordia; SAM, shoot apical meristem; RAM, root apical meristem; AL, aleurone layer; SAl, sub-aleurone layer.

ZmYSL2 protein levels were dramatically reduced in sh4 endosperm, and this protein was absent from the sh4-cas line, as determined by immunoblotting using an antibody highly specific to ZmYSL2 (Figure 4C and Supplementary Figure S5A). To examine the spatial expression pattern of ZmYSL2 in developing kernels, we performed immunohistology using anti-ZmYSL2 antibody. No obvious signal was detected in sh4-cas9 kernels, confirming the specificity of the anti-ZmYSL2 antibody (Supplementary Figure S5B). In wild-type kernels, the ZmYSL2 signal was highly enriched in the aleurone and sub-aleurone cells in the endosperm and in the scutellum cells in the embryo (Figure 4D). These results suggested that ZmYSL2 may play a role in aleurone cells.

ZmYSL2 is essential for iron accumulation in the aleurone layer

To examine the transport activity of ZmYSL2 for specific metals, we measured the levels of several metals in mature wild-type and sh4 kernels by ICP-MS. The contents of iron and zinc in the embryo, endosperm, and whole kernels were dramatically lower in sh4 than in the wild type (Figure 5A and Supplementary Figure S6). The copper contents were also significantly reduced in sh4 endosperm and whole kernels, but not in embryos. The contents of other metals, such as manganese, did not differ between the two genotypes (Supplementary Figure S6). These results indicate that ZmYSL2 is responsible for iron and zinc accumulation in maize kernels.

Figure 5

The sh4 mutation alters aleurone cell fate in maize endosperm. (A) Comparison of iron concentrations in mature WT and sh4 whole kernels, endosperm, and embryos. Values are means ± SE; n = 3 (***, P < 0.001, Student’s t-test). (B) Iron localization in the aleurone layer of 15 DAP WT and sh4 kernels by Perl’s staining. AL, aleurone layer; En, endosperm. Bars = 100 μm. (C) Cell morphology in the aleurone layer in wild type and sh4 endosperm at 15 DAP. Bars = 100 μm. (D) TEM showing the aleurone layer and three endosperm cell layers near the aleurone layer. Bars = 10 μm. (E) Cell morphology of the aleurone layer in wild type and sh4 endosperm under TEM. AG, aleurone granule; SG, starch granule. Bars = 5 µm.

We examined the iron distribution patterns in wild-type and sh4 (or sh4-cas9) kernels at 15 DAP using Perl’s staining (Ishimaru et al. 2010). High iron accumulation was also detected in the scutellum of the embryo in wild-type kernels, as revealed by examining longitudinal hand-cut sections of developing kernels. However, no such iron accumulation pattern was observed in the embryos of sh4 or sh4-cas9 kernels; instead, iron accumulated in the base of the endosperm (Supplementary Figure S7). Iron also strongly accumulated in the aleurone cells of wild-type kernels but was absent from the corresponding peripheral cells in sh4 endosperm (Figure 5B). These results indicate that the loss of ZmYSL2 function disrupted the proper iron distribution pattern in the aleurone layer.

Loss of Sh4/ZmYSL2 function affects aleurone cell identity

To investigate the role of ZmYSL2 in kernel development, we examined histological sections of kernels at different stages of development. Although the endosperm size of sh4 was similar to that of the wild type, sh4 embryo development was delayed by approximately 3 days compared to the wild type (Supplementary Figure S8A). Typical embryonic structures, including the scutellum and leaf primordia, were shorter and smaller in sh4 than in the wild type (Supplementary Figure S8B). Furthermore, wild-type kernels displayed a visible space between the scutellum and the surrounding endosperm, whereas the lower part of the sh4 scutellum protruded outwards into the gradually decaying endosperm (Supplementary Figure S8, A–C). Unlike the fully filled endosperm of the wild type, sh4 endosperm had a large central cavity and poorly filled margins. When we measured the cell number and cell size in two comparable areas (1 mm²) of wild-type and sh4 starchy endosperm (Supplementary Figure S8D), we found that sh4 starchy endosperm contained fewer but larger cells compared to the wild type.

The aleurone cells in maize endosperm form a highly differentiated single layer in the peripheral endosperm (Sabelli and Larkins, 2009). From 9 to 21 DAP, the aleurone cells in wild-type endosperm formed a single, uniformly rectangular layer in the peripheral endosperm. However, the cells in the peripheral layer of sh4 endosperm were variable in shape and loosely arranged and became indistinguishable from the adjacent sub-aleurone cells based on morphology (Figure 5C and Supplementary Figure S9). We examined the ultrastructures of wild-type and sh4 aleurone cells by transmission electron microscopy. In the wild-type, aleurone cells were filled with densely packed aleurone grains. However, sh4 aleurone cells contained far fewer aleurone grains but, unexpectedly, they contained many starch grains (Figure 5, D and E), which is typical of starchy endosperm cells (Becraft et al. 1996; Sabelli and Larkins 2009; Gontarek and Becraft 2017). These results indicate that the differentiation of aleurone cells is defective in sh4 endosperm.

Loss of Sh4/ZmYSL2 function affects the expression of genes involved in storage protein and starch synthesis

We obtained the transcriptome profiles of developing wild-type and sh4 endosperm at 15 DAP by RNA-sequencing. A total of 3115 significant differentially expressed genes (DEGs) were identified using the thresholds fold change >2 and P < 0.001. Among these, 1881 genes were upregulated and 1234 were downregulated in sh4 (Supplementary Data S2). Of these DEGs, 2702 could be functionally annotated and classified using the Gene Ontology (GO; http://systemsbiology.cau.edu.cn/agriGOv2/ (accessed: 2021 May 8)) database. Loss of ZmYSL2 function significantly affected the expression of genes with enriched GO terms related to nutrient metabolism (nutrient reservoir activity and carbohydrate metabolic process), metal transport (metal ion binding and transmembrane transport), and cell proliferation (DNA packaging complex and cell cycle) (Figure 6A and Supplementary Data S3).

Figure 6

Loss of Sh4/ZmYSL2 function alters the expression of genes involved in storage protein and starch synthesis. (A) The most significant GO terms of the DEGs based on RNA-seq analysis of 15 DAP wild type and sh4 endosperm. (B) Heatmap showing the changes in the expression levels of storage-protein-related genes in WT and sh4 endosperm. (C) Immunoblot comparing the storage protein contents in WT and sh4 endosperm. α-Tubulin was used as an internal standard. (D) Heatmap showing the changes in expression levels of starch-synthesis-related genes in WT and sh4 endosperm. The expression levels of these genes are listed in Data S3. UDPG, Uridine 5’ diphosphate-glucose; G1P, glucose 1-phosphate; G6P, glucose 6-phosphate; ADPG, Adenosine 5’ diphosphate-glucose; Sus, sucrose synthase; Ugp, UDP-glucose pyrophosphorylases; Pgm, phosphoglucomutase; Gpt, glucose-6-phosphate/phosphate translocator; AGPL, ADP-glucose pyrophosphorylase large subunit; AGPS, ADP-glucose pyrophosphorylase small subunit; SS, starch synthases; SBE, starch branching enzyme; DBE, starch debranching enzyme; Ae1, amylose extender1; Su1, Sugary1; ISA, Isoamylase; Zpu1, Z. mays pullulanase-type starch debranching enzyme1; Wx1, Waxy1; Pho, Phosphorylase.

Most of the DEGs involved in nutrient reservoir activity (GO: 0045735) were zein genes, which were downregulated by more than 92.8%. In contrast, five genes encoding germin-like protein (Wang et al. 2020) were upregulated in sh4, suggesting that additional protein storage pathways were activated in the mutant (Figure 6B and Supplementary Data S3). Immunoblotting revealed a significant reduction in storage protein levels in sh4 endosperm compared to the wild type (Figure 6C). Loss of ZmYSL2 function also dramatically altered the expression levels of genes involved in the carbohydrate metabolic process (GO: 0005975), with 198 DEGs enriched in this GO term (Figure 6A and Supplementary Data S3).

Because sh4 was originally identified on the basis of altered starch accumulation (Tsai and Nelson 1969; Doehlert and Kuo 1990), we investigated the impact of this mutation on starch accumulation by analyzing the changes in the expression of genes involved in the starch biosynthetic pathway (Figure 6D). The majority of starch biosynthetic genes were downregulated in the mutant, including genes encoding sucrose synthases (Sh1, Sus2, and Sus3), AGPase subunit Sh2 and Brittle2 (Bt2), ADP-Glc transporter Bt1, starch synthases [SS, e.g., Waxy1 (Wx1), SSIIb, and SSV], starch branching enzymes [SBE, e.g., Amylose extender1 (Ae1)], and starch debranching enzymes [DBE, e.g., Sugary1 (Su1), Pullulanase-type starch debranching enzyme1 (Zpu1)] (Figure 6D). ADPG synthesis is the rate-limiting step in starch synthesis in maize endosperm (Hannah and Boehlein 2017). The expression levels of key genes involved in ADPG synthesis (e.g., Sh1, Sh2, and Bt2) and transport into the amyloplast (Bt1 and Gpt1) were downregulated in sh4. By contrast, several genes encoding amylases and glucosidases with hydrolase activity (GO: 0016787) were upregulated in sh4 (Supplementary Data S3). These results suggest that the sh4 mutation affects starch metabolism by inhibiting starch synthesis and promoting starch degradation.

Loss of Sh4/ZmYSL2 function affects starch grain formation

Compared to the wild type, the glucose, fructose, and sucrose contents in developing sh4 endosperm were elevated by 2.87-, 6.51-, and 2.47-fold, respectively (Figure 7A). The increased sugar content in sh4 endosperm is similar to that of other starch-deficient mutants (Keeling and Myers 2010). To further confirm the starch accumulation defect in sh4, we examined the distribution of starch grains in starchy endosperm cells by performing Iodine staining of paraffin sections at 12 and 15 DAP (Figure 7B). The upper part of wild-type endosperm was filled with deeply stained, dense starch granules, representing abundant starch accumulation. By contrast, sh4 endosperm contained fewer, scattered, weakly stained starch grains, indicative of inadequate starch accumulation and altered starch distribution.

Figure 7

Loss of Sh4/ZmYSL2 function leads to delayed formation and premature degradation of starch grains in maize endosperm. (A) Comparison of glucose, fructose, and sucrose contents of developing WT and sh4 kernels. Values are means ± SE; n = 3 (***, P < 0.001, Student’s t-test). (B) Iodine stained longitudinal paraffin sections of developing kernels. Bar = 1 mm. (C) Magnified images of starchy endosperm cells at 15 DAP. Bars = 100 µm. (D) Images of the fourth starchy endosperm cell layer from the aleurone layer under a transmission electron microscope (TEM). Bars = 10 μm. (E) Scanning electron microscopy of WT and sh4 developing endosperm. En, endosperm; PB, protein body; SG, starch granule. Bars = 5 µm.

Wild-type starchy endosperm cells were filled with densely packed starch grains, but the central endosperm cells of sh4 contained dramatically smaller and fewer starch grains (Figure 7C and Supplementary Figure S8A). Furthermore, developing sh4 endosperm contained dramatically smaller starch grains than wild-type endosperm when viewed under a transmission electron microscope (Figure 7D).

We observed the morphology of starch grains in developing kernels at 12, 15, and 18 DAP and at maturity via scanning electron microscopy (Figure 7E). The developing starch grains of sh4 were normal in appearance but smaller than those of the wild type. However, at the mature stage, wild-type endosperm contained fully developed starch grains with a uniform spherical shape and smooth surface, whereas sh4 endosperm contained starch grains of variable size with hollow surfaces. Bisected wild-type starch grains contained nearly solid cores at their center. However, sh4 starch grains displayed an irregular hollow interior (Figure 7E), likely due to starch degradation (Smith et al. 2005). These results suggest that the loss of ZmYSL2 function affects the formation and integrity of starch grains.

Discussion

Previous studies have focused on the starch biosynthetic pathway in sh4, a classic shrunken mutant (Tsai and Nelson 1969; Doehlert and Kuo 1990). Our findings demonstrate that the sh4 phenotype is caused by a mutation in the YELLOW STRIPE-LIKE metal transporter gene ZmYSL2, which is involved in the accumulation and distribution of iron in the aleurone layer, starchy endosperm, and scutellum. The altered iron distribution in the mutant causes cellular dysfunction, which is responsible for the shrunken endosperm and small embryos of sh4 (Figure 8). These findings provide evidence that iron acts not only as a nutrient element, but also as a co-factor in cell fate determination and the accumulation of nutrient reserves.

Figure 8

A model of the role of Sh4/ZmYSL2 in maize kernel development. AL, aleurone layer; AG, aleurone granule; SG, starch granule; PB, protein body.

Sh4/ZmYSL2 is specifically enriched in the aleurone layer

We showed here that the maize gene Sh4, which has been known as a Mendelian factor since the 1950s (Richardson 1955), is the genetic element encoding the NA-iron transporter ZmYSL2. zmysl2 is a small-kernel mutant characterized by collapsed endosperm and small embryos (Zang et al. 2020), while sh4 has shrunken endosperm characterized by starch deficiency (Tsai and Nelson 1969; Doehlert and Kuo 1990). Our findings demonstrate that sh4 has an additional phenotypic feature, namely, the absence of an aleurone layer (Figure 5). The morphological characteristics and mutation site of sh4 are different from those of zmysl2 (Zang et al. 2020). sh4 thus represents a new mutant allele of ZmYSL2. The abnormal iron distribution pattern of sh4 is consistent with that of another zmysl2 allelic line, which has a similar phenotype to sh4 (Zang et al. 2020). ZmYSL2 was previously reported to affect iron accumulation and distribution in maize seeds (Zang et al. 2020). Furthermore, our data indicate that iron accumulation is defective in the sh4 aleurone layer (Figure 5B).

Immunohistochemical staining showed that ZmYSL2 was predominantly present in the aleurone layer, sub-aleurone layer, and embryo (Figure 4D), which is similar to the localization pattern of OsYSL2 in developing rice seeds (Koike et al. 2004). The localization of ZmYSL2 in different tissues is also consistent with the distribution pattern of ZmYSL2 mRNA in developing kernels revealed by transcriptome analysis (Doll et al. 2020). Iron was predominantly present in the scutellum of wild-type seeds, whereas in sh4 seeds its transport was blocked and it only accumulated near the base of the endosperm. Our findings suggest that ZmYSL2 plays a key role in mediating the transport of iron from the basal endosperm to the central starchy endosperm and aleurone layer.

Sh4/ZmYSL2 is essential for aleurone cell fate specification

Maize aleurone cell identity is determined by multiple genes (Becraft and Yi 2011). A positional signal in the outer endosperm layer is involved in cell fate specification in the aleurone layer (Becraft and Asuncion-Crabb 2000; Geisler-Lee and Gallie 2005). It was proposed that DEK1 senses and transmits this positional signal, CR4 promotes the flow of this signal between cells in the aleurone layer, and SAL1 maintains the proper concentrations of DEK1 and CR4 in the plasma membrane (Tian et al. 2007). Our observations here indicate that the loss of ZmYSL2 function results in the conversion of the outermost layer of endosperm cells to starchy-endosperm-like cells (Figure 5), mimicking the phenotypes of dek1 and cr4 mutants lacking aleurone cells (Becraft et al. 1996; Becraft et al. 2002; Lid et al. 2002). These findings suggest that ZmYSL2 plays a positive role in the determination of aleurone cell identity.

The strong accumulation of iron in the aleurone layer is a common feature of cereal seeds (Cheah et al. 2019; Mari et al. 2020). Iron co-localizes with phytate-containing globoids in the protein storage vacuoles of wheat aleurone cells (Regvar et al. 2011), indicating that iron is mainly stored in the aleurone layer in grains. The dramatically reduced formation of aleurone cells in sh4 grains could be a result of iron deficiency (Figures 4D and 5B). The failure to accumulate a sufficient amount of iron (an essential nutrient) could block the normal development of aleurone cells.

A properly developed aleurone layer is also required for the development of the sub-aleurone layer and starchy endosperm (Olsen 2004, 2020). In sh4 endosperm, deficient aleurone cells were also associated with abnormal sub-aleurone and starchy endosperm cells. We detected a gradient in cell size, starch grain size, and protein body size from sub-aleurone cells toward the central starchy endosperm. It appears that a signal from the aleurone layer towards the starchy endosperm guides the proper development of a gradient of starchy endosperm cells from sub-aleurone cells. The potential signals from the aleurone layer and the mechanism for gradient formation in starchy endosperm deserve further study.

Sh4/ZmYSL2 is required for starch synthesis and accumulation in endosperm

Previous studies of maize starch mutants identified metabolic enzymes in the starch biosynthetic pathway, including AGPase subunits Sh2 and Bt2, ADP-Glc transporter Bt1, starch synthases (e.g., Wx1), branching enzymes (e.g., Ae1), and debranching enzymes (e.g., Su1 and Zpu1) (Figure 6D; Keeling and Myers 2010; Hannah and Boehlein 2017). These studies also identified proteins that regulate starch accumulation by influencing the starch biosynthetic pathway, such as Sugary enhancer1 (Se1), a recessive modifier of Su1 (Zhang et al. 2019). Early reports suggested that the phenotype of sh4 may be caused by the reduced enzyme activity of AGPase, starch synthetase, or phosphorylase (Tsai and Nelson 1969; Ozbun et al. 1973; Doehlert and Kuo 1990). However, this study revealed that Sh4 is a metal transporter that is unrelated to the starch biosynthetic pathway. Therefore, phosphorylase deficiency is likely to be a pleiotropic effect resulting from altered gene expression in the Sh4 mutant. This would explain why earlier attempts using metabolic analysis and enzyme activity assays failed to identify the Sh4 gene.

As a nutrient storage tissue, the starchy endosperm is densely packed with starch grains (Sabelli and Larkins 2009). Starchy endosperm cells in sh4 maize showed a dramatically reduced number of starch grains, resulting in reduced kernel weight (Figures 1 and 7). The activities of enzymes involved in protein synthesis were also reduced (Doehlert and Kuo 1990), leading to inhibited starch accumulation, in sh4 endosperm. Interestingly, sh4 had higher sucrose levels than the wild type (Figure 7A), likely due to reduced starch synthesis and enhanced starch degradation. Thus, ZmYSL2 affects starch accumulation by influencing the expression of genes or the activity of enzymes in both the starch biosynthetic and starch degradation pathways.

Iron functions in many physiological and biochemical processes in plant cells (Kobayashi and Nishizawa 2012). Therefore, ZmYSL2 might also be involved in maintaining many cellular processes in maize endosperm by regulating iron accumulation and distribution. Although the iron content of starchy endosperm is much lower than that in other parts of the seed (Cheah et al. 2019), iron is still essential for the normal functioning of starchy endosperm cells. Our results indicate that the loss of ZmYSL2 function disrupted iron accumulation and distribution, which may be responsible for the abnormal cell cycle and reduced cell number in the mutant (Figure 6A and Supplementary Figure S8D). Starch synthesis occurs primarily in the amyloplast, a potentially iron-rich organelle (Roschzttardtz et al. 2013; Hannah and Boehlein 2017). Thus, maintaining the normal activity of amyloplasts in starchy endosperm cells might require a high level of iron accumulation. Iron is also involved in the respiratory electron transport chain in the mitochondria, which function as energy factories in the cell (Mari et al. 2020). Therefore, it is not surprising that iron deficiency led to the failed development of amyloplast and starchy endosperm cells. However, further investigation is needed to reveal the underlying mechanism.

Acknowledgments

The authors thank Liping Yin (Capital Normal University) and Shinan Dong (Nanjing Agricultural University) for their generous sharing of yeast expression plasmid and fet3fet4 strain, and Hongying Yi (Science of Plant Physiology and Ecology, CAS) for ICP-MS analysis.

R.S. designed and supervised the project; Y.H. and Q.Y. performed most of the experiments; X.S., S.W., J.Y., and W.Q. performed partial experiments on map-based cloning and maize transformation; Y.W. performed partial experiments on voltage-clamp experiment in X. laevis oocytes; all authors analyzed the data; Y.H. and R.S. wrote the article.

Funding

This work was supported by the National Natural Science Foundation of China (91935305 and 31730065 to R.S., 31871629 to W.Q.).

Conflicts of interest: The authors have no competing interests.