Characterization and Transcriptome Analysis of Maize Small-Kernel Mutant smk7a in Different Development Stages

The kernel serves as a storage organ for various nutrients and determines the yield and quality of maize. Understanding the mechanisms regulating kernel development is important for maize production. In this study, a small-kernel mutant smk7a of maize was characterized. Cytological observation suggested that the development of the endosperm and embryo was arrested in smk7a in the early development stage. Biochemical tests revealed that the starch, zein protein, and indole-3-acetic acid (IAA) contents were significantly lower in smk7a compared with wild-type (WT). Consistent with the defective development phenotype, transcriptome analysis of the kernels 12 and 20 days after pollination (DAP) revealed that the starch, zein, and auxin biosynthesis-related genes were dramatically downregulated in smk7a. Genetic mapping indicated that the mutant was controlled by a recessive gene located on chromosome 2. Our results suggest that disrupted nutrition accumulation and auxin synthesis cause the defective endosperm and embryo development of smk7a.


Introduction
Maize (Zea mays) is one of the most important crops in the world. The maize kernel is the storage organ that contains the essential components for plant growth and reproduction. The maize kernel consists of the embryo, endosperm, and surrounding pericarp. The embryo consists of the shoot apical meristem, root apical meristem, scutellum, and leaf primordium and has the potential to develop into a new plant [1]. The endosperm accounts for 85% of the kernel's weight and stores about 80% of the starch and 8-15% of the protein.
The development of the endosperm begins with the fertilized central cells, followed by cellularization and differentiation into four main types of cells: the basal endosperm transfer layer (BETL), the aleurone layer (AL), the starchy endosperm (SE), and the embryosurrounding region (ESR) [2][3][4][5][6][7]. Maize kernel development is a complex and dynamic process and is regulated by a large number of genes. With the development of molecular techniques, many genes that are crucial for kernel development have been identified, and the regulatory mechanisms have also been uncovered [8].
Regarding embryo development, a number of embryo-specific (emb) genes have been studied, such as LEM1, EMB8516, EMB14, EMB12, and EMB16 [9][10][11][12][13], all of which have been implicated in plastid function. These studies have demonstrated that plastid protein translation is essential to embryogenesis. Regarding endosperm development, the functional genes involved in the development of the major endosperm cell types have also been elucidated. The AL is a cuboidal cell layer that forms on the epidermal layer of maize endosperm cells. Several genes have been reported to control the fate of the AL cells via positional signaling, including CR4 [14], DEK1 [15,16], SAL1 [17], and THK1 [18,19]. The

smk7a Mutant Exhibited Defective Embryo and Endosperm Development
Compared with WT kernels, smk7a exhibited defective kernel development and a lower 100-kernel weight (Figures 1A-C and S1). To observe the structural and cytological changes, we carried out paraffin sectioning on WT and smk7a kernels collected at 12,16,20,25, and 30 DAP. Longitudinal sections of the whole seed revealed that smk7a showed a small and shrunken kernel phenotype, with severely arrested embryogenesis and endosperm development. The WT embryo developed to the late embryogenesis stage, with established scutellum, coleoptiles, and shoot apical meristem (SAM) at 12 DAP; differentiated primary leaves at 16 DAP; established root apical meristem (RAM) at 20 DAP; and good differentiation at 30 DAP. However, the development of the smk7a embryos was severely delayed at 12 DAP, and it did not reach the coleoptilar stage with differentiated scutellum, coleoptile, SAM, and RAM until 30 DAP ( Figure 1D). The WT endosperm was well developed and starch-filled from 12 to 30 DAP, while the endosperm of smk7a was small and underdeveloped, and a gap was observed between the endosperm and epidermis ( Figure 1D). Further observations revealed that the transfer cells in the BETL of the WT were large and exhibited clear cell wall ingrowths at 12 DAP, and the flange wall ingrowth became more dense at 20 DAP. In contrast, the BETL cells in the smk7a kernels were mostly small and lacked extensive cell wall ingrowth ( Figure 1E).
We also found that smk7a had a 56.3% reduction in the endosperm cell number compared to the WT, and the endosperm cell size was 188.7% larger than that of the WT at 20 DAP (Figure 2A-C and Table 1). Although the endosperm cell size was larger, the starch granules of smk7a appeared to be less filled than the endosperm of the WT (Figure 2A). Scanning electron microscopy (SEM) was used to investigate the starch accumulation in the endosperm at 34 DAP. The results revealed that the endosperm cells of the WT were filled with starch granules, while there were few granules in the endosperm cells of smk7a, and irregular polygonal granules were observed ( Figure 2E,F). We also found that smk7a had a 56.3% reduction in the endosperm cell number compared to the WT, and the endosperm cell size was 188.7% larger than that of the WT at 20 DAP (Figure 2A-C and Table 1). Although the endosperm cell size was larger, the starch granules of smk7a appeared to be less filled than the endosperm of the WT ( Figure  2A). Scanning electron microscopy (SEM) was used to investigate the starch accumulation in the endosperm at 34 DAP. The results revealed that the endosperm cells of the WT were filled with starch granules, while there were few granules in the endosperm cells of smk7a, and irregular polygonal granules were observed ( Figure 2E,F).    To further determine the underlying biochemical basis for the defective phenotype of the smk7a mutant, we measured the soluble sugar, total starch, and protein contents of the WT and smk7a kernels. Based on our results, the content of the soluble sugars (including glucose, fructose, and sucrose) of the smk7a kernels was 1.40-2.70-fold higher than that of the WT kernels at 12 DAP, and it was 3.72-4.07-fold higher at 20 DAP ( Figure 3A-C). In contrast, the starch content was significantly lower, i.e., 4.25-fold and 4.67-fold at 12 and 20 DAP, respectively ( Figure 3D). In addition, the starch content of smk7a was only 87.8% that of the WT in the mature stage ( Figure 3D). We also noticed that the total soluble sugar content of the WT decreased by 40.3% from 12 DAP to 20 DAP, while the starch content increased, which represents the balanced conversion of sugar to starch. However, for smk7a, when the starch content increased, the soluble sugar content also increased by 22.3% from 12 to 20 DAP. These results suggest that the sugar metabolism and starch synthesis were disturbed in the smk7a kernels in the filling stage. Quantitative analysis of the protein contents revealed that the zein protein was also significantly lower in smk7a and was 71.7%, 76.7%, and 94.0% that of the WT at 12 DAP, 20 DAP, and in the mature stage, respectively ( Figure 3E).   IAA has been reported to play an important role in kernel development. We compared the free IAA contents of the WT and smk7a at 12 and 20 DAP. Based on our UHPLC-ESI-MS/MS analyses, the free IAA content of smk7a was 58.3% and 83.2% that of the WT at 12 DAP and 20 DAP, respectively ( Figure 3F). These results suggest that IAA synthesis was suppressed in smk7a in the early development stages.

Transcriptome Comparison of Developing Kernels from WT and smk7a
To reveal the gene expression changes between the WT and the smk7a mutant, we conducted RNA sequencing (RNA-seq) using the kernels collected at 12 and 20 DAP. On average, 48.2 million clean reads were obtained for each sample, and 89.2-94.9% of the reads were uniquely mapped to the maize B73 reference genome (Table S1). With a threshold fold change of >2 and an adjusted p value of <0.05, a total of 3405 and 2747 differentially expressed genes (DEGs) were identified between smk7a and WT for the 12 and 20 DAP kernels, respectively. Further classification of the DEGs revealed that there were 1522 downregulated genes and 1883 upregulated genes at 12 DAP, and there were 1124 downregulated genes and 1623 upregulated genes at 20 DAP (Tables S2-S3). Since the BETL development was impaired in smk7a, we checked the expression levels of the BETL-specific genes. At 12 DAP, six genes were significantly downregulated in smk7a, namely, BETL3, BETL4, BETL9, TCRR1, TCRR2, and SWEET4C, while only the SWEET4C gene was significantly downregulated at 20 DAP. The expression levels of BETL4, BETL9, TCRR1, and SWEET4C in smk7a and WT were further verified using the quantitative realtime polymerase chain reaction (qRT-PCR) method ( Figure S3A). The reduced expression of the BETL-specific genes in smk7a validated the development defects of the BETL cells.
To associate functions with the DEGs, we performed gene ontology (GO) enrichment analysis. The 20 most significantly (p < 0.05) enriched GO terms in the biological process, molecular function, and cellular component categories are presented in Figure S3. At 12 DAP, the downregulated genes were mainly enriched in the BP term "starch biosynthetic process", the MF term "nutrient reservoir activity", and the CC terms "chloroplast and amyloplast", while the upregulated genes were mainly enriched in the BP terms "cell differentiation", "lipid catabolic process", and "hydrogen peroxide catabolic processes", the MF term "heme binding", and the CC term "extracellular region" ( Figure S2A). At 20 DAP, the downregulated genes were significantly enriched in the BP term "secondary metabolite biosynthetic process", the MF term "nutrient reservoir activity", and the CC term "extracellular region", while the upregulated genes were mainly enriched in the BP terms "defense response" and "hydrogen peroxide catabolic processes", the MF term "sequence-specific DNA binding", and the CC terms "extracellular region" and "plant type cell wall" ( Figure S2B).
We further conducted Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of the DEGs. The results revealed that the "starch and sucrose metabolism" and "plant hormone signal transduction" were the most significantly enriched pathways for the downregulated genes in both development stages ( Figure 4A,B), while the "fatty acid biosynthesis" and "glycolysis/gluconeogenesis pathways" were only enriched at 12 DAP ( Figure 4A). For the upregulated genes, the "phenylpropanoid biosynthesis" pathway was significantly enriched in the two development stages ( Figure 4A,B).
These results suggest that the expression of the genes with starch synthesis and nutrient accumulation functions was suppressed in smk7a compared to that in the WT in both development stages. The suppressed transcriptional activity of these genes could be the major reason for the defective development of the smk7a endosperm and embryo. acid biosynthesis" and "glycolysis/gluconeogenesis pathways" were only enriched at 12 DAP ( Figure 4A). For the upregulated genes, the "phenylpropanoid biosynthesis" pathway was significantly enriched in the two development stages ( Figure 4A,B).
These results suggest that the expression of the genes with starch synthesis and nutrient accumulation functions was suppressed in smk7a compared to that in the WT in both development stages. The suppressed transcriptional activity of these genes could be the major reason for the defective development of the smk7a endosperm and embryo.

Suppressed Expression of Genes Involved in Starch Synthesis and Protein Storage in smk7a
Starch and storage protein accumulation is essential for kernel development and yield formation. Our RNA-seq results revealed that the genes that participated in the "starch and sucrose metabolism" pathway were significantly downregulated in smk7a compared to that in the WT (Figure 4). In the starch synthesis process, sucrose, glucose, and fructose are first imported into the endosperm cell and are metabolized to hexose phosphates. Then, these hexose phosphates serve as substrates for starch biosynthesis in the amyloplast [32,33]. The sucrose synthase SH1, SUS2, and SUS3 functions for converting sucrose into fructose and uridine-diphosphoglucose (UDPG) were downregulated in the smk7a by 6.9-13.9-fold at 12 DAP and 0.9-1.9-fold at 20 DAP. The hexokinase (HEX) phosphorylates hexoses forming hexose phosphate, HEX1, and HEX5 genes were downregulated in smk7a by 2.3-5.0-fold and 1.0-1.8-fold at 12 and 20 DAP, respectively ( Figure 5A). The interconversions among fructose 6-phosphate (F6P), glucose 6-phosphate (G6P), glucose 1-phosphate (G1P), and UDPG were catalyzed by phosphohexose isomerase (PHI), phosphoglucomutase (PGM), and UDP-glucose pyrophosphorylase (UGP), and four related genes were also found to have a reduced expression in smk7a ( Figure 5A). The ADP-glucose pyrophosphorylase genes (APGs) catalyze the first step in starch biosynthesis, which is also the limiting step. Five APGs were significantly downregulated in smk7a, including SH2 and BT2, the expressions of which were decreased by 9.2-and 26-fold, respectively, at 12 DAP ( Figure 5A). During the starch syn-

Suppressed Expression of Genes Involved in Starch Synthesis and Protein Storage in smk7a
Starch and storage protein accumulation is essential for kernel development and yield formation. Our RNA-seq results revealed that the genes that participated in the "starch and sucrose metabolism" pathway were significantly downregulated in smk7a compared to that in the WT (Figure 4). In the starch synthesis process, sucrose, glucose, and fructose are first imported into the endosperm cell and are metabolized to hexose phosphates. Then, these hexose phosphates serve as substrates for starch biosynthesis in the amyloplast [32,33]. The sucrose synthase SH1, SUS2, and SUS3 functions for converting sucrose into fructose and uridine-diphosphoglucose (UDPG) were downregulated in the smk7a by 6.9-13.9-fold at 12 DAP and 0.9-1.9-fold at 20 DAP. The hexokinase (HEX) phosphorylates hexoses forming hexose phosphate, HEX1, and HEX5 genes were downregulated in smk7a by 2.3-5.0-fold and 1.0-1.8-fold at 12 and 20 DAP, respectively ( Figure 5A). The interconversions among fructose 6-phosphate (F6P), glucose 6-phosphate (G6P), glucose 1-phosphate (G1P), and UDPG were catalyzed by phosphohexose isomerase (PHI), phosphoglucomutase (PGM), and UDP-glucose pyrophosphorylase (UGP), and four related genes were also found to have a reduced expression in smk7a ( Figure 5A). The ADP-glucose pyrophosphorylase genes (APGs) catalyze the first step in starch biosynthesis, which is also the limiting step. Five APGs were significantly downregulated in smk7a, including SH2 and BT2, the expressions of which were decreased by 9.2-and 26-fold, respectively, at 12 DAP ( Figure 5A). During the starch synthesis in the amyloplast, the expressions of the genes with amylose synthase (such as WX1, GBSSI, and GBSSIb), amylopectin synthase (SSIV, SSV, and SSI), branching enzyme (SBE), and starch debranching enzyme (SU1) functions were all significantly reduced ( Figure 5A and Table S4). To further verify these results, the expression levels of SH1, SH2, WX1, BT2, AGP, and SSI in smk7a and the WT were examined using the qRT-PCR method ( Figure S3B). The overall downregulation of the expression of the genes with starch and sucrose metabolism pathway functions was in agreement with the lower starch content in smk7a at 12 and 20 DAP and in the mature stage.
We further checked the expression of the sugar transporter genes since the soluble sugar content was higher in smk7a compared to that in the WT. We detected 19 transporters, including SWEETs, sugar transporter proteins (STPs), and sucrose transporters (SUTs), which were expressed differently in smk7a and the WT, and 14 of these genes were upregulated in at least one development stage ( Figure 5B and Table S5). These results suggest that suppressed starch synthesis may act as a feedback signal for stimulating sugar transport from the maternal tissue to the kernel endosperm, even when the development of the BETL cells is defective. thesis in the amyloplast, the expressions of the genes with amylose synthase (such as WX1, GBSSI, and GBSSIb), amylopectin synthase (SSIV, SSV, and SSI), branching enzyme (SBE), and starch debranching enzyme (SU1) functions were all significantly reduced ( Figure 5A and Table S4). To further verify these results, the expression levels of SH1, SH2, WX1, BT2, AGP, and SSI in smk7a and the WT were examined using the qRT-PCR method ( Figure S3B). The overall downregulation of the expression of the genes with starch and sucrose metabolism pathway functions was in agreement with the lower starch content in smk7a at 12 and 20 DAP and in the mature stage. We further checked the expression of the sugar transporter genes since the soluble sugar content was higher in smk7a compared to that in the WT. We detected 19 transporters, including SWEETs, sugar transporter proteins (STPs), and sucrose transporters (SUTs), which were expressed differently in smk7a and the WT, and 14 of these genes were upregulated in at least one development stage ( Figure 5B and Table S5). These results suggest that suppressed starch synthesis may act as a feedback signal for stimulating sugar transport from the maternal tissue to the kernel endosperm, even when the development of the BETL cells is defective.
Moreover, we also found that 23 genes related to protein storage under the GO term "nutrient reservoir activity" (GO: 0045735) were significantly downregulated in smk7a. It should be noted that central player transcript factors O2, O11, and PBF were also downregulated in smk7a in both development stages ( Figure 5C and Table S6). These results provide a sufficient explanation for the reduced zein protein content in smk7a compared to that in the WT.

Disrupted Expression of Genes Related to IAA Synthesis and Signaling in smk7a
Studies have reported that phytohormones such as auxin, cytokinins, and brassinosteroids play an important role in plant seed development. In this study, the IAA content in smk7a was significantly lower, and our RNA-seq results also revealed that the expression of the key genes involved in auxin synthesis was decreased in smk7a ( Figure 6 and Table S7). The tryptophan aminotransferase (TAA1/TAR) genes convert tryptophan to indole-3-pyruvic acid (IPA). Mutations in the TAA1/TAR of Arabidopsis lead to a dramatic Moreover, we also found that 23 genes related to protein storage under the GO term "nutrient reservoir activity" (GO: 0045735) were significantly downregulated in smk7a. It should be noted that central player transcript factors O2, O11, and PBF were also downregulated in smk7a in both development stages ( Figure 5C and Table S6). These results provide a sufficient explanation for the reduced zein protein content in smk7a compared to that in the WT.

Disrupted Expression of Genes Related to IAA Synthesis and Signaling in smk7a
Studies have reported that phytohormones such as auxin, cytokinins, and brassinosteroids play an important role in plant seed development. In this study, the IAA content in smk7a was significantly lower, and our RNA-seq results also revealed that the expression of the key genes involved in auxin synthesis was decreased in smk7a ( Figure 6 and Table S7). The tryptophan aminotransferase (TAA1/TAR) genes convert tryptophan to indole-3-pyruvic acid (IPA). Mutations in the TAA1/TAR of Arabidopsis lead to a dramatic reduction in the free IAA level. Three TAR genes were found to be downregulated in smk7a at 12 DAP, while the expression level of TAR1 and TAR3 increased at 20 DAP. ZmYUC1, a flavin-monoxygenase, which is the key enzyme related to auxin biosynthesis in maize endosperm, was significantly downregulated (by 6.1-fold) in smk7a at 12 DAP. The suppressed IAA synthesis also caused a disturbance of the IAA signaling. The expression levels of the core auxin signaling factors, including 10 auxin/indole-3-acetc acid (Aux/IAA), five auxin responsive factors (ARFs), six Gretchen Hagen3 (GH3), and seven small auxin RNA (SAUR) were changed in smk7a ( Figure 6). In the presence of low-level intracellular auxin, AUX/IAA proteins inhibit the function of ARF proteins. Based on our RNA-seq, nine of the 10 AUX/IAAs were upregulated in smk7a, while all five ARFs were downregulated in smk7a ( Figure 6). These results suggest that auxin synthesis and signaling were suppressed in smk7a, which may be the major cause of the reduced number of cells in the smk7a endosperm.
(Aux/IAA), five auxin responsive factors (ARFs), six Gretchen Hagen3 (GH3), and seven small auxin RNA (SAUR) were changed in smk7a ( Figure 6). In the presence of low-level intracellular auxin, AUX/IAA proteins inhibit the function of ARF proteins. Based on our RNA-seq, nine of the 10 AUX/IAAs were upregulated in smk7a, while all five ARFs were downregulated in smk7a ( Figure 6). These results suggest that auxin synthesis and signaling were suppressed in smk7a, which may be the major cause of the reduced number of cells in the smk7a endosperm.

Genetic Analysis and Fine Mapping of the smk7a Mutant
Our previous study reported that smk7a mutant was controlled by a recessive gene located in the 120 kb region on chromosome 2 [34]. To further narrow down its location, we genotype another F2 population of 1056 homozygous mutant kernels using newly developed insertion and deletion markers (Table S8); however, no new recombination was detected at this interval. The RNA-seq results revealed that only three genes exhibited an expression level greater than 1 fragment per kilobase of the transcript per million fragments (FPKM) in at least one of the samples in this region. Zm00001d001819 encodes an N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase, Zm00001d001824 encodes a Dof zinc finger protein, and Zm00001d001825 encodes a transducin/WD40 repeat-like superfamily protein.

Discussion
The maize kernel has been used extensively as a valuable model for cereal kernel development studies. The development of the maize kernel can be divided into three stages: the early development, filling, and dehydration stages [1]. In the early development stage (about 0-15 DAP), the embryo and endosperm cells are quickly divided and differentiated into specific tissues. In the filling stage (about 15-45 DAP), the embryo and

Genetic Analysis and Fine Mapping of the smk7a Mutant
Our previous study reported that smk7a mutant was controlled by a recessive gene located in the 120 kb region on chromosome 2 [34]. To further narrow down its location, we genotype another F 2 population of 1056 homozygous mutant kernels using newly developed insertion and deletion markers (Table S8); however, no new recombination was detected at this interval. The RNA-seq results revealed that only three genes exhibited an expression level greater than 1 fragment per kilobase of the transcript per million fragments (FPKM) in at least one of the samples in this region. Zm00001d001819 encodes an N-acetylglucosaminyl-phosphatidylinositol de-N-acetylase, Zm00001d001824 encodes a Dof zinc finger protein, and Zm00001d001825 encodes a transducin/WD40 repeat-like superfamily protein.

Discussion
The maize kernel has been used extensively as a valuable model for cereal kernel development studies. The development of the maize kernel can be divided into three stages: the early development, filling, and dehydration stages [1]. In the early development stage (about 0-15 DAP), the embryo and endosperm cells are quickly divided and differentiated into specific tissues. In the filling stage (about 15-45 DAP), the embryo and endosperm start to accumulate starch and to store protein lipids and vitamins [35]. Then, the kernel enters the mature stage and begins to dehydrate (about 45-60 DAP) [36]. Kernel mutants with developmental defects in different stages have been characterized, and a great number of genes have been cloned [7,8]. These studies have provided valuable information for unveiling the gene regulation network of maize kernel development.
In this study, we characterized a small-kernel mutant smk7a, which exhibited arrested embryo and endosperm development compared to the WT. The small and pale smk7a kernels with irregular embryo and endosperm can be clearly distinguished from the WT at 12 DAP. These results suggest that the mutant phenotype of smk7a first occurred before 12 DAP. This phenotype of smk7a is quite similar to some kernel mutants with variant genes related to early maize kernel development. For example, the dek15 mutant exhibits a defective endosperm and embryo in the early stage, and the embryos appear to be more severely affected than the endosperm. The DEK15 encodes cohesin-loading complex subunit SCC4, and the loss of the SCC4 function leads to defects in the cell division and early kernel development [37]. Both embryo development and endosperm development are delayed in the smk4-1 mutant compared with that in the WT at 8 DAP. SMK4 encodes an E-subclass PPR protein, which is exclusively localized in mitochondria [38]. The smk1 mutation leads to defects in the assembly of complex I, which results in the arrest of the embryo and endosperm in the early development stage [39]. Thus, we speculate that the functional gene controlling the smk7a phenotype may play an important role in early kernel development.
Our cytological observations also revealed that the BETL morphology is impaired in smk7a. The smk7a BETL developed less-elongated cells and had very few cell wall ingrowths. The RNA-seq results revealed that the expression levels of several BETL-specific genes were significantly downregulated. Defective BETL development usually results in reduced sucrose transport from the maternal plant to the endosperm, and it ultimately leads to suppressed starch synthesis and endosperm development [6,40,41]. For example, ENB1 encodes a cellulose synthase 5 that directs the synthesis of cell wall ingrowths in maize BETL cells. The enb1 mutant exhibits a drastic reduction in the formation of flange cell wall ingrowths in the BETL cells and impaired nutrient uptake and starch synthesis [22]. ZmCTLP1 participates in lipid homeostasis in transfer cells. The loss of the function of ZmCTLP1 results in irregular-shaped BETL cells with few wall ingrowths and significantly reduced starch, sugar, and protein contents in the mutant [4]. Based on our biochemical results, the starch content of smk7a was significantly lower, while the soluble sugar content was higher in smk7a compared to WT at 12 and 20 DAP. Further analysis revealed that the starch accumulation in the WT increased from 12 to 20 DAP, which was accompanied by a reduction of the soluble sugar content, indicating balanced conversion from soluble sugar to starch. In contrast, the soluble sugar content in smk7a increased from 12 DAP to 20 DAP. These results suggest that the sugar and starch metabolism was disrupted in smk7a.
The RNA-seq data revealed that at 12 and 20 DAP, the DEGs enriched in the "starch and sucrose metabolism" pathway were significantly downregulated. Starch began accumulating in the endosperm of the maize kernels at approximately 10 DAP, starting with the transport of sucrose (Suc) from the maternal plant to the endosperm. Suc was then converted into fructose (Fru) and glucose (Glu) by neutral invertase (NINV) or into Fru and UDPG by sucrose synthase [42]. The resulting Glu and Fru were then phosphorylated to generate hexose phosphates, including G6P, F6P, and G1P. G1P was then further activated by AGPase to produce ADP-glucose [43], which was the substrate for starch synthesis [44,45]. In our RNA-seq results, almost all of the DEGs involved in this sucrose metabolism process exhibited a decreased expression level in smk7a, including the well-known genes SH1, SH2, and BT2. SH1 encodes sucrose synthase, and the loss of the function of SH1 caused a significant reduction in the carbohydrates that flow to starch synthesis, ultimately contributing to the defects in the starch granule development and reduction of the starch content [46]. SH2 and BT2 encode AGPase and are rate-limiting enzymes in starch biosynthesis. Mutations in SH2 and BT2 resulted in greatly reduced starch production and increased sugar in the endosperm [47][48][49]. In smk7a, at 12 DAP, the expressions of SH2 and BT2 were decreased by 9.2-and 26-fold, respectively, compared to that in the WT. These results explain the increased soluble sugar contents of the smk7a kernels compared to that of the WT kernels. After the sugar was metabolized into ADPglucose, starch was synthesized by multiple subunits or isoforms of five classes of enzymes, namely, AGP, SS, GBSS, SBE, and DBE [33]. In our RNA-seq, the expressions of amylose synthases WX1, GBSSI, and GBSSIb; amylopectin synthases SSIV, SSV, and SSI; branching enzyme SBE; and starch debranching enzyme SU1 were all significantly reduced in smk7a compared to those in the WT. All of the above results indicate that the process of starch synthesis is severely suppressed in smk7a, which could be the major reason for the defective phenotype of the smk7a kernel.
Our results also revealed that the accumulation of the major storage protein zein was significantly reduced in smk7a at 12 and 20 DAP and in the mature stage. Consistent with this, all of the zein genes in the "nutrient reservoir activity" term (GO: 0045735) were downregulated in smk7a. It should be noted that several key regulator factors, including O2, O11, and PBF, were also found to be downregulated in smk7a. O2 was the first bZIP transcription factor known to regulate the expression of zein genes by recognizing the O2 box in their promoters. Loss of the function of O2 results in a drastically decreased zein content [50]. Several studies subsequently reported that O2 not only acts as a master regulator of the zein-coding genes but can also physically interact with the prolamin-box binding factor (PBF) to regulate starch accumulation [51]. O11 is a central hub of the regulatory network for maize endosperm development and nutrient metabolism [52]. It works upstream to regulate carbohydrate metabolic enzymes and the key regulators (such as O2 and PBF). The decreased expression of these genes suggests that the mutation in smk7a has a noticeable effect on starch and nutrition accumulation. The candidate gene of smk7a may interact with these key regulatory factors and could play an important role in the complex regulatory network of maize kernel development.
IAA is the most abundant endogenous auxin in the maize kernel, and it plays an essential role throughout the entire kernel development process [53]. In smk7a, the free IAA content was only 58.3% and 85.0% that of the WT at 12 DAP and 20 DAP, respectively. These results suggest that IAA synthesis is suppressed in smk7a. The RNA-seq results revealed that the essential genes that participate in the IPA pathway for IAA biosynthesis, including ZmYUC1 [54], TAR1, TAR2, and TAR3 [55,56], are significantly downregulated in smk7a. Defective kernel (dek) mutants in maize have been shown to have altered levels of IAA. For example, DE18 encodes the auxin biosynthesis gene ZmYUC1, and de18 mutation causes a large reduction of the free IAA level compared with the WT, leading to a 40% reduction in the dry mass of the endosperm. Several studies have also reported that developing kernels may regulate growth by influencing the auxin level in response to the sugar concentrations. The MN1 gene encodes a cell wall invertase [20,57], which is localized in the BETL and catalyzes the conversion of sucrose into glucose and fructose. Loss of the function of MN1 results in pleiotropic changes, including reduced cell number and cell size, a lower kernel mass, and a lower free IAA level [58]. In addition, the mn1 mutant also exhibits retarded wall ingrowth formation in the BETL cells. Gene expression analysis suggests that the expression of the ZmYUC1 and TAR1 genes is greatly reduced in mn1 kernels. In smk7a, the BETL cells were impaired and wall ingrowth was reduced, which could be the result of disrupted sugar hormone signaling in the endosperm.

Plant Materials
In our previous study, a maize small-kernel mutant smk7 was isolated from the EMS mutant library produced in our laboratory on the B73 genetic background [34]. However, its name is repeated with another kernel mutant reported in a recent study [59], and therefore, we renamed it as smk7a. Since the homozygote smk7a cannot grow into a normal plant, the mutants were maintained in heterozygote by continuous self-crossing and phenotype selection. The developing kernels of WT and smk7a used for cytological, biochemical, and transcriptome analysis in this study were collected from the same segregating ears of the M8 generation. The heterozygous plant (smk7a/WT) was crossed with Mo17 and Chang7-2 and then self-pollinated to generate F 2 populations. The segregating F 2 ear were used for genetic mapping of smk7a. Maize plants were grown under natural conditions at the Changping experimental station of the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences, Beijing.

Light Microscopy and Scanning Electron Microscopy Observations of smk7a and WT Kernels
WT and smk7a kernels were acquired from the same heterozygous ear at 12, 16, 20, 25, and 30 DAP. Then, the kernels were cut along the longitudinal axis into half-sections and fixed at room temperature in formalin-aceto-alcohol (FAA) solution (90 mL of 70% ethanol, 5 mL of glacial acetic acid, and 5 mL of 37% formaldehyde). The fixed material was dehydrated in an ethanol gradient series (50%, 70%, 85%, 90%, and 100% ethanol), embedded in paraffin, and subsequently cut into 12 µm sections. The sections were stained with toluidine blue and observed under a microscope (BX53; Olympus, Tokyo, Japan).
The WT and smk7a kernels collected at 34 DAP were fixed in FAA. The fixed material was dehydrated in an ethanol gradient series and then treated with isoamyl acetate for 15 min twice to replace the remaining ethanol. Then, the material was subjected to critical point drying. The samples were then coated with Pt particles and analyzed under a scanning electron microscope (SU8020, Hitachi, Tokyo, Japan).

Measurement of Protein and Starch Contents
Mature WT and smk7a kernels were obtained from the same segregating ears at 12 and 20 DAP. After removing the pericarp and embryo, the endosperms of the WT and smk7a were each pooled and ground into powder. Three biological replicates were prepared for the starch and protein measurements. For the starch quantification, 100 mg of powder per sample was analyzed using an a-amylase/amyloglucosidase starch assay kit (K-TSTA, Megazyme, Ireland) according to the manufacturer's instructions. The starch content of the mature seeds was determined as has been previously described by [60]. For the protein quantification, 50 mg of powder was used for the protein extraction according to the method described by [61]. Then, the protein content was measured using a Bradford Protein Assay Kit (PC0010, Solarbio, Beijing, China).

Measurement of Soluble Sugar and Free IAA Contents
The kernels of WT and smk7a collected at 12 and 20 DAP from the same heterozygous ear were ground into power in the presence of liquid nitrogen. The soluble sugar and IAA contents were determined by Wuhan Greensword Creation Technology Co., Ltd. (Wuhan, China) according to previously reported methods [62,63]. The soluble sugars were analyzed via ultra-high performance liquid chromatography-electron spray ionizationtandem mass spectrometry (UHPLC-ESI-MS/MS, Thermo Fisher Scientific, Inc., Waltham, MA, USA). The IAA analysis was performed using a Thermo Scientific Ultimate 3000 UHPLC system equipped with a TSQ Quantiva-Stage Quadrupole Mass Spectrometer (Thermo Fisher Scientific, Sunnyvale, CA, USA). Three biological replicates were prepared for each measurement.

RNA Sequencing Analysis
The total ribonucleic acid (RNA) was isolated from the kernels of WT and smk7a collected at 12 and 20 DAP using a RNA Easy Fast Extraction Kit (DP452, Tiangen, Beijing, China). A total of 1 µg of RNA per sample was used to construct sequencing libraries using a TruSeq Stranded mRNA LTSample Prep Kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions, and index codes were added to attribute the sequences to each sample. Then, these libraries were sequenced using the Illumina HiSeqTM 2500 platform to generate 125 bp paired-end reads. Following the removal of low-quality reads and adaptor contaminants, the clean reads were mapped to the maize reference genome RefGen_V4 (ftp.ensemblgenomes.org/pub/plants/release-32/gff3/zea_mays/, accessed on 1 November 2018) using HISAT2 [64]. Only perfectly matching reads were further analyzed. The gene expression levels were estimated using the fragments per kilobase of the transcript per million fragments (FPKM) mapped. Differential expression analysis of the two groups was performed using DESeq2 [65]. The resulting p values were adjusted using Benjamini and Hochberg's approach for controlling the false discovery rate. The genes with an adjusted p-value of <0.05 and a fold change of >2 or <0.5 were assigned as differentially expressed.
Gene ontology (GO) enrichment analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis were conducted using R based on the hypergeometric distribution, and those with a q value of <0.05 were defined as significantly enriched terms or pathways.

RNA Extraction and Quantitative RT-PCR Verification
The quantitative real-time polymerase chain reaction (qRT-PCR) technique was used to verify the expression level of the differentially expressed genes (DEGs) related to the BETL development and starch synthesis process. The total RNA of the kernels from the WT and smk7a collected at 12 and 20 DAP was extracted using an RNA Easy Fast Extraction Kit (DP452, Tiangen, Beijing, China) and then transcribed into complementary deoxyribonucleic acid (cDNA) using qPCR RT Master Mix with gDNA Remover (KR116, TIANGEN, Beijing, China). The qRT-PCR reactions were carried out using Taq Pro Universal SYBR qPCR Master Mix (Q712-02, Vazyme, Nanjing, China). The qRT-PCR analysis was performed using a Bio-Rad CFX96 qRT-PCR system. The reaction was carried out using three biological replicates with three technical replicates, with tubulin as the endogenous control (Zm00001d013367). The relative expression values were calculated using the 2 −∆∆Ct method [66], and the t-test was conducted to analyze the significant differences. The specific primers for the qRT-PCR are listed in Table S9.

Conclusions
In summary, we characterized a small-kernel mutant smk7a of maize. Our results revealed that: (1) The early development of the embryo and endosperm of smk7a is seriously affected. (2) The starch and zein protein accumulation is lower in smk7a compared to in WT, and the expression levels of the DEGs involved in these processes are significantly suppressed in smk7a. (3) The auxin (IAA) content is lower in smk7a compared to in WT, and the DEGs that participate in IAA synthesis and signaling are downregulated in smk7a. (4) smk7a is a recessive gene located at chromosome 2. We speculate that smk7a may be related to the sugar and hormone signaling in the maize kernel and may play a crucial role in embryo and endosperm development.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/plants12020354/s1, Figure S1. Comparison of the kernels traits of WT and smk7a. Figure S2. GO enrichment analysis of the DEGs between WT and smk7a at 12 and 20 DAP. Figure S3. The qRT-PCR verification of the expression pattern of DEGs between WT and smk7a. Table S1. The genome mapping state of RNA-seq. Table S2. The differentially expressed genes between WT and smk7a at 12 DAP. Table S3. The differentially expressed genes between WT and smk7a at 20 DAP. Table S4. The differentially expressed genes enriched in starch and sucrose metabolism pathway. Table S5. The differentially expressed genes related to sugar transport. Table S6. The differentially expressed genes enriched in nutrient reservoir activity. Table S7. The differentially expressed genes related to IAA synthesis and signaling. Table S8. List of all primers for 14 Insertion/Deletion markers used in genetic mapping of smk7a. Table S9. Molecular marker primers used for qRT-PCR verification.