Maize Dek15 Encodes the Cohesin-Loading Complex Subunit SCC4 and Is Essential for Chromosome Segregation and Kernel Development

The classic maize mutant dek15 is mutated in a cohesin loader subunit SISTER CHROMATID COHESION PROTEIN 4 that is essential for mitotic chromosome segregation and maize kernel development.

Abstract

Cohesin complexes maintain sister chromatid cohesion to ensure proper chromosome segregation during mitosis and meiosis. In plants, the exact components and functions of the cohesin complex remain poorly understood. Here, we positionally cloned the classic maize (Zea mays) mutant defective kernel 15 (dek15), revealing that it encodes a homolog of SISTER CHROMATID COHESION PROTEIN 4 (SCC4), a loader subunit of the cohesin ring. Developing dek15 kernels contained fewer cells than the wild type, but had a highly variable cell size. The dek15 mutation was found to disrupt the mitotic cell cycle and endoreduplication, resulting in a reduced endosperm and embryo lethality. The cells in the dek15 endosperm and embryo exhibited precocious sister chromatid separation and other chromosome segregation errors, including misaligned chromosomes, lagging chromosomes, and micronuclei, resulting in a high percentage of aneuploid cells. The loss of Dek15/Scc4 function upregulated the expression of genes involved in cell cycle progression and stress responses, and downregulated key genes involved in organic synthesis during maize endosperm development. Our yeast two-hybrid screen identified the chromatin remodeling proteins chromatin remodeling factor 4, chromatin remodeling complex subunit B (CHB)102, CHB105, and CHB106 as SCC4-interacting proteins, suggesting a possible mechanism by which the cohesin ring is loaded onto chromatin in plant cells. This study revealed biological functions for DEK15/SCC4 in mitotic chromosome segregation and kernel development in maize.

INTRODUCTION

Plant development depends on the proper regulation of mitosis. The mitotic cell cycle contains interphase stages (G1, S, and G2) and a mitosis phase (the M-phase, comprising prophase, metaphase, anaphase, and telophase; McIntosh, 2016). Sister chromatid cohesion and segregation is a critical step for guaranteeing the equal distribution of genetic materials between daughter cells. From the G1/S phase to anaphase, the sister chromatids are linked together by cohesin, a ring-shaped SMC (structural maintenance of chromosomes) complex, comprising two heterodimeric ATPases (SMC1 and SMC3), an α-kleisin hinge (sister chromatid cohesion protein 1; SCC1), and an adaptor protein (SCC3; Uhlmann and Nasmyth, 1998; Uhlmann et al., 1999; Nasmyth and Haering, 2009; Uhlmann, 2016). Together, these proteins form a tetramer ring encircling chromatin (Haering et al., 2002; Gligoris et al., 2014). The cohesin complex proteins are highly conserved in microbes, plants, and animals (Nasmyth and Haering, 2009; Uhlmann, 2016; Bolaños-Villegas et al., 2017).

The localization of cohesin ring depends on a heterodimeric complex of SCC2 and SCC4 homologs (Ciosk et al., 2000; Chao et al., 2015). SCC4 is a small (624 amino acids in budding yeast; Saccharomyces cerevisiae) protein containing a multiple-tetratricopeptide-repeats (TPRs) superhelix, whereas SCC2 is a large (1493 amino acid in budding yeast) protein with multiple Huntingtin-elongation factor 3-protein phosphatase 2A-TOR1 repeats (Ciosk et al., 2000; Hinshaw et al., 2015). The N terminus of SCC2 is the “handle” that binds SCC4 to form a cohesin loader, and the C terminus of SCC2 forms a flexible “hook” to interact with cohesin (Chao et al., 2015, 2017; Hinshaw et al., 2015; Kikuchi et al., 2016). The SCC2/SCC4 complex is specifically required to promote cohesin linkage to chromatin in an ATP-dependent manner at G1/S phase (Bernard et al., 2006; Murayama and Uhlmann, 2014). SCC4 was first identified as a cohesin loader subunit for its role in recruiting cohesin to both the centromere and the chromosome arms of yeast, after which its conserved function was also detected in humans and animals (Ciosk et al., 2000; Seitan et al., 2006; Watrin et al., 2006).

SCC4 is indispensable for cell division and developmental processes. In yeast, scc4 mutant spores die after one or two divisions, whereas in the nematode Caenorhabditis elegans, the mau-2 mutant (lacking an ortholog of SCC4) shows partial unviability of larvae as well as defects in locomotion and egg laying (Ciosk et al., 2000; Bénard et al., 2004). The western clawed frog (Xenopus tropicalis) knockdown mutant mau-2 exhibited growth retardation and developmental defects in the early embryo (Seitan et al., 2006). In Arabidopsis (Arabidopsis thaliana), the scc4 mutation leads to endosperm defects and embryo lethality, similar to the effects of the scc2, smc1, and smc3 mutations (Liu Cm et al., 2002; Sebastian et al., 2009; Minina et al., 2017). SCC4 depletion leads to precocious sister chromatid separation (PSCS) during mitosis in yeast and animals (Ciosk et al., 2000; Seitan et al., 2006; Watrin et al., 2006); however, the function of SCC4 in plant cell mitosis remains unclear.

Maize (Zea mays) kernels contain an embryo and endosperm, both of which are products of double fertilization (Russell, 1992). Endosperm development begins with a triploid (3C) cell formed from the two polar nuclei of the ovule and one sperm cell. In the very early stages of endosperm development, a large number of nuclei are produced without subsequent cytokinesis, forming a syncytium (Olsen, 2001). These nuclei are then surrounded by cell walls to form cells, which continue to divide and differentiate to form the major endosperm cell types, such as the starchy endosperm (SE), basal endosperm transfer layer (BETL), aleurone layer (AL), and embryo-surrounding region (Olsen, 2001; Sabelli and Larkins, 2009; Doll et al., 2017). Endosperm cells have strong mitotic activity between 8 and 14 d after pollination (DAP), after which mitotic cell proliferation persists only in the peripheral cell layers (AL and the subaleurone layer) until ∼20 to 25 DAP (Kowles and Phillips, 1985, 1988; Schweizer et al., 1995). The SE cells in the central endosperm change from mitosis to endoreduplication at 10 DAP, increasing their DNA contents without undergoing sister chromatid segregation (Kowles and Phillips, 1985; Sabelli and Larkins, 2009). The embryo itself begins development as a diploid (2C) zygote, from which the mature embryo is formed by a continuous process of mitosis and cell differentiation (Doll et al., 2017). The development of the endosperm and the embryo therefore depend on the proper regulation of mitosis and cell differentiation, making maize kernel development a good model for studying these processes.

Maize has a long history of genetic study, during which a rich resource of mutants has been developed. The defective kernel (dek) mutants are affected in the development of both the embryo and the endosperm, and were initially generated through ethyl methanesulfonate (EMS) mutagenesis of the pollen (Neuffer and Sheridan, 1980). Only a fraction of the dek mutants have been cloned and functionally characterized (Lid et al., 2002; Qi et al., 2016b, 2017a, 2017b; Garcia et al., 2017; Wang et al., 2017; Dai et al., 2018; Li et al., 2018b). In this study, we analyzed the classic maize mutation dek15, which leads to a reduced endosperm and is embryo lethal. Positional cloning revealed that Dek15 encodes the maize homolog of SCC4. Our cytological analysis showed that the dek15 mutation causes defects in sister chromatid cohesion and aneuploidy, and we found that the transcriptome of the mutants was also dramatically altered. We conclude that Dek15/Scc4 is required to precisely regulate chromosome segregation, possibly by interacting with the chromatin remodeling complex to assist cohesin binding to chromatin.

RESULTS

dek15 Causes a Reduced Endosperm and Is Embryo Lethal

The classic dek15 mutant was previously generated using EMS mutagenesis in maize (Neuffer and Sheridan, 1980). This mutant was obtained from the Maize Genetics Cooperation Stock Center, then crossed to the W22 inbred line and selfed to obtain F₂ ears. The segregation ratio of wild-type (+/+ and dek15/+) and mutant (dek15/dek15) kernels on the F₂ ears was close to 3:1 (Supplemental Figure 1A), suggesting that dek15 plants contained a recessive mutation in a single gene.

Compared with the wild type, the mature dek15 kernels were pale and small but more variable in size (Figures 1A and 1B), with a 100-kernel weight only 42.0% of that of the wild type (Figure 1C). In the dek15 kernels, both the endosperm and the embryo were severely affected (Figure 1D); the dek15 embryos were difficult to observe in the mature kernels (Figure 1B), because only small embryo debris could be identified using an anatomical microscope (Figure 1D). The dek15 kernels were incapable of germinating (Figure 1E). All attempts to rescue the immature dek15 embryos at 18 DAP on Murashige and Skoog medium failed (Figure 1F).

Figure 1.

Phenotypic Features of dek15 Kernels.

(A) Mature F₂ ear of dek15 × W22. Arrows indicate the dek15 kernels. Bar = 1 cm.

(B) Mature wild-type (WT) and dek15 kernels from a segregated F₂ ear. Bar = 1 cm.

(C) Comparison of the 100-kernel weight of randomly selected mature wild-type and dek15 kernels in a segregated F₂ population. Values are means with ±se; n = 3 (***, P < 0.001, Student’s t test).

(D) Longitudinal sections of wild-type and dek15 mature kernels. En, endosperm; Em, embryo. Bar = 1 mm.

(E) Germination test of wild-type and dek15 mature kernels (7 DAG). Bar = 1 cm.

(F) Attempts to rescue the immature embryos (18 DAP) of wild type and dek15 on Murashige and Skoog medium. The result was observed after 6 d of cultivation. Bar = 1 cm.

(G) Longitudinal paraffin sections of developing wild-type and dek15 kernels at 15 DAP. En, endosperm; Em, embryo. Bars = 1 mm.

(H) Developing wild-type and dek15 embryos at 15 DAP. SC, scutellum; LP, leaf primordia; SAM, shoot apical meristem; RAM, root apical meristem. Bars = 100 μm.

(I) Comparison of the developing endosperm area and embryo area in wild-type and dek15 kernels at 15 DAP. Values are means ±se; n = 3 kernels for wild type; n = 5 kernels for dek15. (***, P < 0.001; ns, not significant; Student’s t test)

The developing kernels of dek15 were observed from 9 DAP to 24 DAP (Supplemental Figure 1B). The dek15 kernels could be clearly distinguished from the wild type by their lighter color as early as 12 DAP, and the immature mutant kernels were smaller than the wild type at all developmental stages (Supplemental Figure 1C). The dek15 embryos appeared to be more severely affected than the endosperm (Figures 1G to 1I; Supplemental Figure 2); while the wild type formed the typical embryonic structures, including the scutellum, leaf primordia, shoot apical meristem, and root apical meristem, the dek15 embryos arrested during early development, and were much smaller than those of the wild type (Figure 1H). In addition there was no relationship between defective embryo size and endosperm size in the dek15 kernels at 15 DAP (Supplemental Figure 2B).

Transmission electron microscopy observations revealed that the immature dek15 endosperm contained fewer and smaller starch grains and protein bodies than that of the wild type (Figure 2A). In the mature dek15 endosperm, scanning electron microscopy observation revealed that the starch grains were also smaller at this stage, and that the mature endosperm contained less of the proteinaceous matrix than the wild type (Figure 2B). The starch and protein contents of the dek15 endosperm were significantly lower than those of the wild-type endosperm, both as a percentage of kernel weight (Figure 2C) and as the average contents per kernel (Figure 2D).

Figure 2.

Cytological and Biochemical Analysis of Wild-Type (WT) and dek15 Endosperm.

(A) Transmission electron microscopy analysis of the third cell layer from the aleurone layer in the wild-type and dek15 endosperm at 15 DAP. Bars = 10 µm. SG, starch granule; PB, protein body; Nu, nucleus.

(B) Scanning electron microscopy analysis of the central regions of mature wild-type and dek15 endosperm. Bars = 10 µm. SG, starch granule; PM, proteinaceous matrix.

(C) Starch and total protein contents of mature wild-type and dek15 endosperm relative to the kernel weight. Values are means ±se; n = 3 (*, P < 0.05; **, P < 0.01, Student’s t test).

(D) Starch and total protein contents of the mature wild-type and dek15 endosperm of individual kernels. Values are means ±se; n = 3 (***, P < 0.001, Student’s t test).

Positional Cloning of Dek15

Dek15 was mapped to a 28.236-Mb interval on chromosome 4 (from 160,472,432 to 188,708,387) using the maize SNP3072 genotyping array. With use of additional markers, the mapping interval was narrowed down to 431 kb (from 182,738,064 to 183,169,294) based on the analysis of 979 mutant kernels from the F₂ population (Figure 3A). The Dek15 interval contained nine predicted protein-coding genes according to the B73 reference genome (Jiao et al., 2017). A comparison of the sequences of these predicted genes in the wild-type and dek15 plants revealed that only G3 (Zm00001d052192) and G7 (Zm00001d052197) contained mutations that would be expected to cause the loss of gene function. G3 was excluded as a Dek15 candidate because loss-of-function mutations of this gene were also present in other tested inbred lines (Zong31, Zheng58, W64a, W22, Mo17, 2674, Chang7-2, and Xun928) that lacked the dek15 phenotype. In dek15, G7 contained a typical EMS-induced G-to-A mutation 659 bp from the start codon, resulting in a codon change of TGG (tryptophan) to TAG (stop codon) and the premature termination of the open reading frame (ORF; Figure 3B). This point mutation meant that only the N-terminal 219 amino acids of the 727-amino acid wild-type protein were retained in the mature dek15 protein (Figure 3C), suggesting that G7 (Zm00001d052197) was the candidate Dek15 gene.

Figure 3.

Positional Cloning and Identification of Dek15.

(A) The dek15 mutant was crossed to the W22 inbred line, then selfed to obtain F₂ ears. A total of 979 mutant kernels from F₂ population were analyzed. The number below molecular marker indicates the ratio of the recombinant exchange kernels in the tested population. The dek15 locus was narrowed down to a 431-kb interval on chromosome 4, which contained nine candidate genes: Zm00001d052190 (G1), Zm00001d052191 (G2), Zm00001d052192 (G3), Zm00001d052193 (G4), Zm00001d052194 (G5), Zm00001d052195 (G6), Zm00001d052197 (G7), Zm00001d052198 (G8), and Zm00001d052200 (G9).

(B) Structure and mutation site of the Zm00001d052197 gene. Lines represent introns, black boxes represent exons, and white boxes represent the 5′ and 3′ untranslated regions.

(C) Schematic diagram of the Zm00001d052197 protein structure and mutation site. Zm00001d052197 was predicted to contains five tandem repeats of the TPR domain (gray squares) using TPRpred (https://toolkit.tuebingen.mpg.de/tprpred/). AA, amino acid.

(D) to (F) Functional complementation test of dek15. (D) Representative kernels with the wild-type (wild type [WT]; top) and mutant (bottom) phenotypes, from F₂ ears produced from a cross between the G7-ORF-expressing transgenic lines (T₀) and the dek15 heterozygous plant. (E) Homozygous dek15 kernels were identified using the marker AC190640.21, which is closely linked to the dek15 locus. dek15, homozygous dek15; -, water. (F) The transgene detection of the above kernels using the primers for the Bar gene on transgene construct. +, Zm00001d052197 transgene construct; -, water.

Functional Complementation and Allelic Confirmation of Dek15

To confirm that G7 (Zm00001d052197) was Dek15, a transgenic functional complementation test was performed. The genomic fragment of Zm00001d052197 is very long (20,084 bp); therefore, only the ORF of G7 (2181 bp) was used to construct the transgenic vector, and was transformed into maize (pApB, Hi-II hybrid) under the control of its native promoter (2030 bp upstream from the translational start codon). Three independent transgenic lines were obtained, two of which were crossed to the dek15 heterozygous plants and then selfed to obtain F₂ ears. With use of the molecular marker AC190640.21, which was tightly linked to the dek15 locus, the kernels containing the homozygous dek15 locus were identified (Figure 3D and 3E). These homozygous dek15 kernels were then genotyped using primers specific for the Bar gene on the transgene construct to identify the kernels containing the G7 transgene (Figure 3F). The homozygous dek15 kernels containing the G7 transgene had all reverted to the wild-type phenotype, whereas the kernels lacking the G7 transgene retained the mutant phenotype. This demonstrated that the transformed G7 ORF complemented the dek15 mutation, and indicated that Zm00001d052197 is Dek15.

Targeted mutation of Zm00001d052197 was also performed using the CRISPR/Cas9 system (Qi et al., 2016a). The guide RNA (gRNA) spacer sequences were designed to target the 2nd exon of Zm00001d052197 (1,000–1,019 bp of the G7 ORF). Six independent CRISPR/Cas9 transgenic lines were obtained, three of which contained codon-shifting mutations caused by a deletion or insertion at the target sequence (Figure 4A). The phenotypes of the kernels carrying these mutated alleles showed a similar defective phenotype to dek15 (Figures 4B to 4D; Supplemental Figure 3A). Allelism tests were performed by crossing two independent heterozygous mutated alleles (dek15-cas9-1 and dek15-cas9-3) with heterozygous dek15. The kernel phenotypes in the F₂ ears displayed a 3:1 segregation of wild-type and mutant kernels (Supplemental Figure 3B). The genotyping of randomly selected kernels from the allelism test ears indicated that all those with the mutant phenotype contained both the dek15 locus and the dek15-cas9 locus (Figures 4E to 4G). This result indicated that dek15-cas9 and dek15 cannot complement each other, and further confirmed that G7 (Zm00001d052197) is Dek15.

Figure 4.

CRISPR-Cas9-Based Mutation of Dek15 and Allelism Test with dek15.

(A) The sequence in the Zm00001d052197 locus targeted using CRISPR/Cas9. The gRNA target sequence and the protospacer-adjacent motif (PAM) are shown in green and blue, respectively. Alignments of mutant sequences from three independent transgenic plants are indicated. Red letters and dashes represent insertions and deletions, respectively.

(B) Mature F₂ ear of dek15-cas9-1 × W22. Arrows indicate the mutant kernels. Bar = 1 cm.

(C) Mature wild-type (WT) and dek15-cas9-1 kernels from (B). Bar = 1 mm.

(D) Longitudinal paraffin sections of developing wild-type and dek15-cas9-1 kernels at 18 DAP. En, endosperm; Em, embryo. Bars = 1 mm.

(E) to (G) An allelism test was performed using a cross between heterozygous dek15 (dek15/+) and heterozygous dek15-cas9-1 (dek15-cas9-1/+). (E) Randomly selected kernels from the allelism test ear. Bar = 1 cm. (F) Sequences of the kernels in (E) at the dek15 locus in the Zm00001d052197 genomic fragment. (G) Sequences of the kernels in (E) at the editing site in dek15-cas9-1. Red arrows indicate the mutation sites or editing site.

Dek15 Encodes a Homolog of SCC4

The genomic DNA sequence of Dek15 (Zm00001d052197) is 20,084 bp, comprising 10 exons and 9 introns (https://www.maizegdb.org/). The mature transcript of Dek15 has a 2,181-bp coding sequence, encoding an 80-kD protein comprising 727 amino acids. The Zm00001d052197 protein was previously annotated as a protein with unknown function in maizeGDB. Sequences homologous to this protein were identified using a Protein Basic Local Alignment Search Tool (BLASTp) search (https://blast.ncbi.nlm.nih.gov/Blast.cgi), the results of which indicated that DEK15 shared significant sequence similarity with a cohesin loader subunit, SCC4, which is conserved in a variety of other species. Both bioinformatics predictions and protein crystal structures have shown that SCC4 in other species contains multiple TPR domains, which are composed of 34 amino acid residues in tandem repeating units (Blatch and Lässle, 1999; Seitan et al., 2006; Watrin et al., 2006; Hinshaw et al., 2015). The TPRpred program (Zimmermann et al., 2018) predicted that the C terminus of DEK15 contained five TPRs (Figure 3C). In the maize B73 reference genome (Jiao et al., 2017), Dek15 was found to be the only SCC4 homolog. It was therefore named ZmSCC4.

Scc4 Is Highly Conserved and Constitutively Expressed

A phylogenetic tree was constructed using ZmSCC4 and its homologous proteins from other species (Figure 5A; Supplemental Data Set 1). ZmSCC4 was most closely related to the SCC4 proteins in plants, followed by those of animals and microbes. Among plants, ZmSCC4 was more closely related to the SCC4 proteins of monocots than dicots. The plant SCC4s contained an extra segment of ∼100 amino acids in the middle of the protein (Supplemental Figure 4), suggesting that plant SCC4s may have some unique features not present in their animal and microbe counterparts.

Figure 5.

Phylogenetic Analysis, Expression Pattern, and Subcellular Localization of SCC4.

(A) Neighbor-joining sequence similarity analysis of ZmSCC4 and its homologs in other organisms. SCC4 homologs were aligned using Muscle in the MEGA 7.0 software. The phylogeny reconstruction was conducted in MEGA 7.0. The numbers next to the branches represent the percentage of support from 1000 bootstraps. Scale bar = average number of amino acid substitutions per site.

(B) and (C) RT-quantitative PCR analysis of Scc4 in various tissues (B) and the developing kernels (C) of maize. Root, stem, leaf, silk, tassel, and ear tissues were collected from field-cultivated W22 plants at the V12 stage. The kernel sample in (B) was harvested at 15 DAP. The developing kernels in (C) were collected at different stages and labeled as DAP.

(D) SCC4 protein content patterns in wild-type (WT) kernels at different developmental stages. α-Tubulin was used as a loading control.

(E) Comparing the expression levels of Scc4 in wild-type and dek15 endosperm at 15, 18, and 21 DAP. Values are means ±se; n = 4 (***, P < 0.001, Student’s t test).

(F) Immunoblot comparing the accumulation of the SCC4 protein in wild-type and dek15 endosperm at 15, 18, and 21 DAP. α-Actin was used as a sample loading control.

(G) Immunoblot analysis of the SCC4 protein accumulation in nuclear and cytoplasmic fractions of 15-DAP wild-type and dek15 kernels. Bip is a cytoplasm marker, histones are a nuclear marker and nuclear sample loading control, and Actin is a cytoplasm sample loading control.

The results of RT-quantitative PCR analysis revealed that Scc4 was constitutively expressed in broad range of maize tissues. Its expression level was low in the tassels, but high in the ear, husk, and kernel (Figure 5B). During kernel development, the expression of Scc4 peaked at 6 and 9 DAP and slowly decreased thereafter (Figure 5C). A polyclonal antibody was raised against the C-terminal portion (500–727 amino acids) of SCC4 (Supplemental Figure 5). At the protein level, the SCC4 content was higher in the early to middle stages of kernel development (6 DAP to 18 DAP), and decreased gradually as development progressed (Figure 5D).

The Scc4 mRNA expression was significantly reduced in dek15 in comparison with the wild type during endosperm development (15, 18, and 21 DAP; Figure 5E). The total proteins from the dek15 and wild-type immature endosperms were analyzed using protein gel blotting with a SCC4 antibody, revealing that SCC4 was absent in the dek15 kernels (Figure 5F).

To examine the subcellular localization of SCC4, the full-length Scc4 ORF was fused to the N terminus of yellow fluorescent protein (YFP) and driven by the Cauliflower mosaic virus 35S promoter. The fusion construct was transiently expressed in onion (Allium cepa) mesophyll cells, and the YFP fluorescence signals were detected in both the cytoplasm and the nucleus (Supplemental Figure 6). The cytoplasmic and nuclear fractions were isolated from the developing endosperm cells (15 DAP) of dek15 and wild-type kernels and analyzed using an immunoblot. The SCC4 protein was detected in both the cytoplasmic and nuclear fractions of the wild-type endosperm, but not in the fractions derived from dek15 (Figure 5G). These results indicated that SCC4 was present in both the cytoplasmic and nuclear fractions of the wild-type endosperm.

Loss of Dek15/Scc4 Function Dramatically Reduces Cell Proliferation

To explore the cause of the dek15 phenotype, the cell morphologies in the developing wild-type and dek15 kernels were examined at 15 DAP. The wild-type embryo cells were uniformly patterned, except for the shoot apical meristem and root apical meristem where the cells divided more vigorously (Figure 6A). By contrast, the dek15 embryo typically comprised smaller cells with highly variable sizes, accompanied by a few larger vacuolated cells (Figure 6B). Consistent with the huge difference in embryo size between the wild type and dek15 (Figure 1I), the cell number in dek15 was drastically reduced in comparison with the wild type.

Figure 6.

Comparison of Cell Size, Cell Number, and Nucleus Size in Developing Wild-Type (WT) and dek15 Kernels.

(A) and (B) Longitudinal paraffin sections of wild-type (A) and dek15 (B) embryos from the same segregating ear at 15 DAP. The insets are a magnified image of the boxed regions. Bars = 100 μm.

(C) and (D) BETL cells on longitudinal paraffin sections of wild-type (C) and dek15 (D) endosperm at 15 DAP. The figures below provide a higher magnification of the boxed areas. Bar = 100 μm.

(E) and (F) Longitudinal paraffin sections of wild-type and dek15 endosperm from the same segregating ear at 15 DAP. The insets are a magnified image of the boxed regions. AL, aleurone layer; PE, pericarp; SE, starchy endosperm; SG, starch granule. Bars = 100 μm.

(G) and (H) Transmission electron microscopic (TEM) analysis of the wild-type and dek15 SE cells adjacent to the AL at 18 DAP. The inset shows a magnified image of the boxed region. Nu, nucleus; NL, nucleolus; NP, nucleoplasm; SG, starch granule; PB, protein body. Bars = 10 µm.

(I) and (J) DAPI-stained wild-type and dek15 SE cells adjacent to the AL at 15 DAP. Nu, nucleus; AL, aleurone layer. Bars = 10 µm.

(K) Comparison of the nucleus diameters of wild-type and dek15 endosperm cells in the second and third cell layers from the AL in the TEM (G and H) and DAPI-stained longitudinal sections (I and J). Values are means ±se; n = 12 nuclei in (G and H), n = 100 nuclei in (I and J).

The wild-type BETL cells were darkly stained with thick and extensive cell wall ingrowths (Figure 6C). In contrast, the dek15 BETL cells were less stained with much reduced cell wall ingrowths (Figure 6D). The wild-type SE cells were smaller and denser at the endosperm periphery, and gradually increased in size toward the interior of the endosperm (Figure 6E), whereas the dek15 SE cells were larger and irregularly shaped (Figure 6F). The wild-type AL cells comprised a single sheet layer of uniform cuboidal cells; however, the dek15 AL cells were variable in cell size and shape (Figures 6E and 6F). To quantify the cell numbers in the endosperm, the numbers of SE cells in a defined area (1 mm²) was counted at different regions of the wild-type and dek15 endosperm. In comparison with the wild type, fewer SE cells were observed in the dek15 endosperm (Supplemental Figure 7). These results, coupled with the smaller endosperm size in dek15 (Figure 1I), suggested that the dek15 endosperm contained fewer but larger cells.

The ultrastructural changes in the 18-DAP endosperm cells at the subaleurone region were further observed using transmission electron microscopy. The nuclei in the dek15 endosperm cells were much larger than those of the wild type at the same number of cells from the AL (Figures 6G, 6H, and 6K). Toward the center of the endosperm, the diameter of the SE nuclei increased gradually, consistent with the observed endoreduplication (Supplemental Figure 8B; Schweizer et al., 1995; Leiva-Neto et al., 2004; Sabelli et al., 2013). 4',6-Diamidino-2-phenylindole (DAPI)-stained paraffin sections also showed a statistically significant increase in the nucleus size of the dek15 endosperm in comparison with the wild-type endosperm at the same cell layer (Figures 6I to 6K; Supplemental Figure 8B). In addition, these larger nuclei were usually also highly misshapen, often containing intranuclear inclusion bodies (Supplemental Figure 8A). These results indicate that dek15 contains severe nuclear structural changes, suggesting that SCC4 may affect the function of the nucleus.

The Cell Cycle Is Disrupted in dek15 Endosperm

To detect whether the cell cycle was affected by the loss of Dek15/Scc4 function, the DNA contents of the wild-type and dek15 endosperm cells at 12 DAP and 15 DAP were examined using flow cytometry. In the wild-type endosperm, the peaks were sharp; the 3C peak was the tallest, followed by the peaks with progressively higher C numbers (6C, 12C, and so on; Figure 7A). In the dek15 endosperm cells, all peaks were substantially smaller and less pronounced than in the wild type. The positions of the peaks in dek15 were also shifted to lower C values in comparison with their counterpart peaks in the wild type.

Figure 7.

Flow Cytometric Profiles of Wild-Type (WT) and dek15 Endosperms at 12 DAP and 15 DAP.

(A) Histograms of the relative nuclear DNA contents (relative fluorescence intensities) obtained from the analysis of wild-type and dek15 endosperms at 12 DAP (left) and 15 DAP (right) using a flow cytometer. The C-value is indicated for each peak.

(B) Mean ploidy levels in the wild-type and dek15 endosperms at 12 DAP and 15 DAP. Mean ploidy was calculated by weighting the number of nuclei by their ploidy.

(C) and (D) Distribution of DNA contents in the wild-type and dek15 endosperms at 12 DAP (C) and 15 DAP (D).

Values are means ±se; n = 4 (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant; Student’s t test).

The mean ploidy of the dek15 endosperm nuclei was increased by up to 71.7% at 12 DAP, and by 65.7% at 15 DAP, in comparison with the wild type (Figure 7B). At 12 DAP, the abundance of 3C nuclei in dek15 were reduced by 81.65% compared with the wild type, whereas no difference was observed in the numbers of 6C nuclei (Figure 7C). At 15 DAP, the 3C nuclei in dek15 were 72.17% less abundant than in the wild type, whereas the abundance of the 6C nuclei was also decreased by 42.65% (Figure 7D). Similar relative decreases were also observed in peaks corresponding to 12C or greater in the dek15 nuclei in comparison with the wild type (Figures 7C and 7D). Significantly more intermediate DNA contents were observed in the dek15 nuclei than in the wild type (Figures 7C and 7D). These results indicated that the endoreduplication cell cycle was disrupted in dek15.

Loss of Dek15/Scc4 Function Causes Mitotic Disorder

To investigate the role of SCC4 in the nucleus, chromosomes spreads in the wild-type and dek15 endosperm cells were stained with DAPI and observed (Figures 8A to 8F). In the wild-type endosperm cells, deeply stained chromosomes condensed at prophase (Figure 8B), then aligned on the equatorial plate in the center of the cell during metaphase (Figure 8C). At anaphase, the sister chromatids were precisely separated to form two chromosome sets, which were then pulled to either side of the cell (Figure 8D).

Figure 8.

Mitotic Chromosome Behaviors and Chromosome Number Analysis in the Wild-Type (WT) and dek15 mutant.

(A) to (H) The chromosome behavior of endosperm cells in the wild type and dek15 at 15 DAP, stained with DAPI. The micronuclei interphase (E), the sister chromatid cohesion defects at prophase (F), the misaligned chromosomes at metaphase (G), and the lagging chromosomes at anaphase (H) in dek15 are indicated by arrows. Bars = 10 μm.

(I) The proportion of micronuclei at interphase in the endosperm cells (>2000) of wild type and dek15 at 15 DAP.

(J) The mitotic index was analyzed by counting the proportion of mitotic M-phase cells in the assessed cell population (>2000). Values are means ±se; n = 3 (*, P < 0.05, Student’s t test).

(K) The proportion of cells at different M-phases in the total assessed cell population (>2000). Values are means ±se; n = 3 (*, P < 0.05; ns, not significant, Student’s t test).

(L) and (M) FISH with 45S rRNA gene (red) probes in the 15-DAP endosperm (L) and embryo (M) of the wild type and dek15. In (L), the histogram represents the number of 45S signal detected in the wild-type (n = 220) and dek15 (n = 316) endosperm cells at interphase. In (M), the histogram represents the number of 45S signal detected in the wild-type (n = 164) and dek15 (n = 283) embryo cells at interphase. Blue indicates the chromosomes stained with DAPI. Bars = 10 μm.

(N) The chromosome number was identified in the wild-type and dek15 embryo cells at 15 DAP; n = 35 endosperm cells in the wild type, n = 149 endosperm cells in dek15.

In comparison with the wild type, the dek15 endosperm cells contained significantly more micronuclei (small dots surrounding the main nucleus; Figures 8A, 8E, and 8I). Nearly half of the dek15 endosperm cells (49.4%, n = 85) contained partially separated or completely separated sister chromatids at prophase (Figure 8F), whereas a large portion of dek15 endosperm cells (57.1%, n = 49) contained chromosomes that were not arranged on the equatorial plate at metaphase (Figure 8G). Some of the dek15 sister chromatids lagged behind when being pulled to the poles of the cells during anaphase (Figure 8H). Chromosome bridges and lagging chromosomes were easily observed in majority of dek15 endosperm cells (66.1%, n = 47). The mitotic index of the dek15 endosperm cells was significantly higher than that of the wild type (Figure 8J), because many cells were arrested in prophase and metaphase (Figure 8K).

Similar alterations in mitosis were also observed in the dek15 embryo cells, as well as in the endosperm and embryo cells of dek15-cas9-1 (Supplemental Figure 9). These results indicated that the loss of Dek15/Scc4 function caused mitotic defects, including precocious sister chromatid separation and various chromosome segregation errors.

dek15 Cells Had a High Frequency of Sister Chromatid Cohesion Defects

To monitor sister chromatid cohesion, fluorescence in situ hybridization (FISH) experiments were performed using the dek15 endosperm cells at 15 DAP. These experiments used 45S rDNA as a probe, which is located on the short arm of chromosome 6. Because maize endosperm cell are triploid, three signals were detected per cell in 93.1% (n = 164) of the wild-type endosperm cells (Figure 8L). In the 283 observed dek15 endosperm cells, only 12.4% (35) nuclei had three signal spots. About 8.5% (24) nuclei contained fewer than three signal points, whereas the vast majority of dek15 endosperm cell nuclei (224; 79.1%) contained four or more signal spots.

The 45S probe was also used to monitor the number of chromosome 6s in the embryo cells at 15 DAP (Figure 8M). In the wild type, 94.5% (n = 220) of the embryo cells were observed to have two signal spots, indicating that these cells contained a pair of chromosome 6s. In contrast, out of the 316 observed dek15 embryo cells, only 37.6% (119) contained two signal spots. A small but significant proportion of cells (53; 16.8%) contained no or only one signal spot per cell, and nearly half of cells (144; 45.6%) had more than three signal spots per cell. Similar signal changes were also observed for chromosome 4 using centromere 4 (Cent4) as a chromosome 4-specific FISH probe (Supplemental Figure 10). These FISH results showed that appropriate sister chromatid cohesion was present in almost all wild-type cells, whereas the dek15 cells contained high frequencies of sister chromatids with cohesion defects.

The intertwined chromosomes of the dek15 endosperm cells were hard to count, so the chromosome numbers were counted in the wild-type and dek15 embryos at 15 DAP. The chromosome number in the wild-type cells was 20 (n = 35); however, the average chromosome number in the dek15 cells was 22.03 (n = 149; ranging from 12 to 43). Of the 149 observed dek15 embryo cells, only 22.1% (33) cells had a normal chromosome number of 20, whereas 35.6% (53) cells contained fewer than 20 chromosomes and 42.3% (63) had too many chromosomes (Figure 8N). The loss of Dek15/Scc4 function therefore causes sister chromatid cohesion defects, and eventually leads to chromosome gain or loss in the maize kernel cells.

Dek15/Scc4 Influences the Expression of Genes Involved in the Cell Cycle and Nutrient Metabolism

To investigate the impact of the dek15 mutation on global gene expression during maize endosperm development, an RNA-sequencing (RNA-seq) analysis was performed using the dek15 and wild-type endosperm at 15 DAP. Significantly differentially expressed genes (DEGs) were identified with a threshold fold change greater than two times and P < 0.01. Under these criteria, a total of 1457 DEGs were identified, including 1080 and 377 that were upregulated and downregulated in dek15, respectively (Supplemental Data Set 2). Within these DEGs, 1173 genes could be functionally annotated using a Gene Ontology (GO) analysis (AgriGO, http://bioinfo.cau.edu.cn/agriGO/analysis.php). The significantly enriched terms were related to nutrient reservoir activity, transporter activity, hydrolase activity, DNA binding, single-organism biosynthetic processes, mitotic cell cycle processes, microtubule binding, and responses to stimuli (Figure 9A; Supplemental Data Set 3).

Figure 9.

GO Classification of Differentially Expressed Genes (DEGs) Based on RNA-seq Data and a Quantitative RT-PCR Confirmation in the 15-DAP Endosperms of Wild-type (WT) and dek15.

(A) The most significantly enriched GO terms in the DEGs, based on an RNA-seq analysis of 15-DAP wild-type and dek15 endosperms. A total of 1173 genes were functionally annotated classified in the GO analysis. The number of genes and the P-value for each GO term are shown. E indicates 10 raised to a power in scientific notation.

(B) RT-quantitative PCR confirmation of 24 selected DEGs associated with the DNA binding, cell cycle, and zein and starch biosynthesis in 15-DAP wild-type and dek15 endosperms. Values are means ±se; n = 3 (**, P < 0.01; ***, P < 0.001, Student’s t test).

Among the DEGs involved in mitotic cell cycle processes (GO:1903047), the expression levels of Sister chromatid cohesion protein dcc1, Condensin complex subunit 2, Budding uninhibited by benzimidazole-related 1, and Mitotic arrest-deficient 2 were increased by more than 111%. In addition, some genes encoding cyclins and cyclin-dependent kinases were also upregulated. Regarding microtubule binding (GO:0008017), 15 of the 19 DEGs encoding kinesins and microtubule-associated proteins were upregulated. We identified 19 upregulated genes encoding histones that are involved in the formation of DNA packaging complexes (DNA binding; GO:0003677). The expression changes of these genes suggested that the loss of Scc4 function affected the cell cycle and DNA structure.

GO terms associated with nutrient reservoir activity (GO:0045735), transporter activity (GO:0005215), and single-organism biosynthetic processes (GO:0044711) were also found to be enriched in the DEGs between the wild type and dek15. Multiple key genes involved in protein and starch biosynthesis were significantly downregulated in dek15 (Supplemental Data Set 3). The genes encoding Opaque1, Opaque10, and 19- and 22-kD zein proteins were downregulated by more than 50% in dek15, whereas some of the key genes in the starch biosynthetic pathway were also significantly downregulated, including Shrunken2, Starch synthase I (SSI), SSIIa, and other starch synthase genes.

In addition, most genes involved in the specific process of hydrolase activity acting on glycosyl bonds (GO:0016798) were upregulated, which may affect the accumulation of storage compounds in dek15. The dek15 mutation also affected the expression of genes that respond to stimuli (GO:0050896), suggesting that the mutant cells likely experienced an unfavorable intracellular environment during kernel development.

The expression levels of 24 genes involved in the DNA binding, cell cycle, and zein and starch biosynthesis in the wild-type and dek15 endosperm were analyzed using RT-quantitative PCR. These results validated those of the RNA-seq analysis of the wild-type and dek15 transcriptomes (Figure 9B).

SCC4 Interacts with SCC2, and the Chromosome Remodeling Proteins CHR4 and Chromatin Remodeling Complex Subunits B 102, 105, and 106

SCC4 is known to form a complex with SCC2 through its interaction with the SCC2 N terminus (Ciosk et al., 2000; Chao et al., 2015). The maize SCC2 homolog (Zm00001d018657) was identified using a BLASTp search with the SCC2 sequence from Arabidopsis. The cDNA coding for the N terminus of ZmSCC2 (SCC2-N; 1–214 amino acids) was constructed into pGADK7, whereas the full-length ORF or truncated segments of Scc4 (encoding SCC4-N, 1–219 amino acids; SCC4-M, 220–473 amino acids; or SCC4-N, 474–727 amino acids) were individually constructed into pGBKT7. Yeast two-hybrid (Y2H) assays were used to show that SCC2-N interacts with SCC4, and that the interacting segment was contained within the N terminus of SCC4 (1–474 amino acids; Figure 10A). The results of luciferase complementation image (LCI) assays showed that the co-expression of SCC4-NLUC (N-terminal domains of LUCIFERASE) and SCC2-N-CLUC (C-terminal domains of LUCIFERASE) resulted in strong luciferase activity, indicating that SCC4 and SCC2-N can interact in Nicotiana benthamiana cells (Figure 10B). The results indicated that SCC4 interacts with SCC2.

Figure 10.

Interaction Analysis of SCC4 with SCC2, CHR4, CHB102, CHB105, or CHB106, Revealed Using Yeast Two-hybrid and Luciferase Complementation Image Assays.

(A) Yeast two-hybrid analysis of the interaction between SCC4 and the N terminus of SCC2 (SCC2-N). The interaction between the T-antigen and Human P53 was used as a positive control. The interaction between the empty pGBKT7 and pGADT7 vectors was used as a negative control.

(B) Luciferase complementation image assay of the interaction between SCC4 and the SCC2-N. Fluorescence signal intensities represent their interaction activities.

(C) Yeast two-hybrid analysis of the interaction between SCC4 and CHR4-C, CHB102, CHB105, or CHB106.

(D) Luciferase complementation image assay of the interaction between SCC4 and CHR4-C, CHB102, CHB105, or CHB106 in Nicotiana benthamiana cells. Fluorescence signal intensities represent their interaction activities.

(E) Model depicting the function of SCC4 in cohesin loading and chromosome segregation.

In the SCC2/SCC4 complex, SCC4 may be responsible for identifying chromatin loci for cohesin loading (Chao et al., 2015; Hinshaw et al., 2015; Uhlmann, 2016). To identify other potential SCC4-interacting proteins involved in cohesin loading, the full-length Scc4 ORF was constructed into pGBKT7 as bait, then used to screen a Y2H cDNA library constructed from maize developing kernels (Zhang et al., 2012). After 250 SCC4-interacting clones were sequenced, a list of putative SCC4-interacting proteins was obtained (Supplemental Data Set 4), which included four chromatin remodeling proteins (CHR4, chromatin remodeling complex subunit B [CHB]102, CHB105, and CHB106) as potential SCC4 interaction partners. In yeast, the RSC (remodels the structure of chromatin) complex has the potential function of recruiting the SCC2/SCC4 complex (Lopez-Serra et al., 2014). CHR4 is a homolog of Chromodomain helicase DNA binding protein 3, an ATPase involved in NURD (nucleosome-remodeling and histone deacetylase) complexes (Ho and Crabtree, 2010; Hu et al., 2012). CHB102, CHB105, and CHB106 are SWI3-type proteins in maize, and are highly similar to SWI3B (CHB102) or SWI3C (CHB105 and CHB106) in Arabidopsis. SWI3B and SWI3C are the core subunits of the SWI/SNF (SWITCH/SUCROSE NONFERMENTING) chromatin remodeling complex (Sarnowski et al., 2005). To verify their interaction with SCC4, the full-length ORFs of Chb102, Chb105, and Chb106 were cloned into pGADT7 and were cotransformed into yeast cells alongside pGBKT7-Scc4. For CHR4, the full-length ORF was too long to include (6570 bp); therefore, the fragment encoding the C terminus of CHR4 (CHR4-C; 406 amino acids) was selected for these experiments. These proteins were found to interact with SCC4 (Figure 10C). A LCI assay was performed by cotransforming N. benthamiana cells with SCC4-CLUC and either CHR4-C-NLUC, CHB102-NLUC, CHB105-NLUC, or CHB106-NLUC. The transformed cells had a strong luciferase activity (Figure 10D), which taken together with the Y2H results indicated that SCC4 interacts with CHR4, CHB102, CHB105, and CHB106.

DISCUSSION

Dek15 Encodes SCC4 and Is Required for Cohesion Function during Mitosis in Maize Kernels

In this study, the classic maize kernel mutant dek15 was cloned and found to contain a mutation in a gene encoding a cohesin loader, SCC4. In yeast and animals, SCC4 forms a loader complex with SCC2 to facilitate the loading of the cohesin complex onto chromatin during interphase (Ciosk et al., 2000; Seitan et al., 2006; Watrin et al., 2006). Multiple pieces of evidence indicated that Dek15 encodes a maize homolog of SCC4; for example, ZmSCC4 has a high similarity to AtSCC4 (Figure 5A; Supplemental Figure 4) and contains multiple tandem repeats of TPR, which is a common pattern of SCC4 protein (Watrin et al., 2006; Hinshaw et al., 2015). ZmSCC4 could also be detected in both the nucleus and cytoplasm, similar to the distribution of SCC4-homologous proteins and cohesin subunits in other organisms (Ciosk et al., 2000; Lam et al., 2005; Seitan et al., 2006; Minina et al., 2017). Previous studies have shown that SCC4 interacts with the N terminus of SCC2 (Ciosk et al., 2000; Chao et al., 2015; Minina et al., 2017), which was also demonstrated for the N terminus of ZmSCC4 in the present study using Y2H and LCI assays (Figures 10A and 10B). PSCS is a typical feature of plants with a functional deficiency of SCC4 (Ciosk et al., 2000; Seitan et al., 2006; Watrin et al., 2006), and was observed in both the embryo and endosperm cells of the dek15 and dek15-cas9 mutants (Figure 8F; Supplemental Figure 9), suggesting that SCC4 is necessary for sister chromatid cohesion in maize.

Although SCC4 is well studied in yeast and animals, it remains unknown whether it has additional functions in plants. Our phylogenetic analysis revealed that ZmSCC4 has a distant relationship and low sequence similarity with its homologs in yeast and animals (Figure 5A; Supplemental Figure 4). In yeast scc4 mutants, the only cell cycle error observed was PSCS, whereas in cultured human cells, misaligned chromosomes were observed after transfection with Scc4 short interfering RNAs (Ciosk et al., 2000; Watrin et al., 2006). In this study, much more severe mitotic alterations were observed in the dek15 mutants, including misaligned chromosomes, lagging chromosomes, and the presence of micronuclei (Figure 8; Supplemental Figure 9), which are typical features of chromosome segregation errors (Solomon et al., 2011; Siegel and Amon, 2012; Haarhuis et al., 2013). Highly variable chromosome number (aneuploidy) in dek15 was demonstrated using the chromosome counts and flow cytometry analyses (Figures 7 and 8). In humans, deficiency in the cohesin subunits (such as SA1 and SA2; SCC3-homologous protein) causes aneuploidy (Solomon et al., 2011; Remeseiro et al., 2012; Losada, 2014). The aneuploidy resulting from the loss of Scc4 function in maize may be associated with an aberrant cohesin function.

The FISH results and observations of the chromosome behavior at prophase confirmed the presence of disassociated single sister chromatids in the dek15 mutant cells (Figures 8L and 8M; Supplemental Figure 10). In yeast and humans, the loss of Scc4 function is known to result in reduced amounts of cohesin onto chromatin, but does not affect cohesin assembly (Ciosk et al., 2000; Watrin et al., 2006). This suggests that cohesin is formed normally in scc4 mutants but cannot bind to chromatin, thereby causing cohesion defects. The deficient cohesion of dek15 was likely the main cause of its mitotic defects.

Scc4 Affects Cell Division and Cell Function during Maize Kernel Development

Many genes involved in sister chromatid cohesion also play a role in developmental regulation (Nasmyth and Haering, 2009; Sebastian et al., 2009; Bolaños-Villegas et al., 2017). The dek15 and dek15-cas9 mutant alleles had small but variably sized kernels, with structurally deficient embryos and a reduced endosperm (Figures 1 and 4). The loss of Scc4 function was demonstrated to be responsible for these kernel-defective phenotypes using a transgenic functional complementation and allelism tests (Figures 3 and 4). The seed defects were similar to phenotypes caused by the loss of the cohesin loader (SCC2 and SCC4) or cohesin subunits (SMC1 and SMC3) in Arabidopsis (Liu Cm et al., 2002; Sebastian et al., 2009; Minina et al., 2017), which implied that Scc4 could affect plant development through its function in cohesion regulation.

Our results indicated that the loss of Scc4 function caused the defects in cohesin loading and function, which eventually lead to defects in cell division and plant development. The development of the dek15 embryo was strongly inhibited, whereas the effect on its endosperm during kernel development was less severe (Figure 1; Supplemental Figure 2). Embryonic growth relies on continuous cell division, which means mutations affecting mitosis result in a much stronger inhibition of embryonic than endosperm development. The endosperm undergoes multiple developmental stages in maize (Sabelli and Larkins, 2009), and the relatively short duration of mitotic cell division likely reduced the negative effects of the mitotic defects during dek15 endosperm development. In the central endosperm cells, an endoreduplication stage is followed by mitosis in maize. At this stage, even though the dek15 endosperm could contain a mixture of aneuploid cells (Figures 7 and 8), these cells would still survive because they no longer need to divide.

The dek15 kernels exhibited a drastic reduction in their total starch and protein contents (Figure 2), which in turn reduced their kernel weights. The decrease in nutrient storage is presumed to result from the functional failure of the endosperm cells. Nutrients are transported from the maternal vascular tissue through the BETL into the endosperm (Gómez et al., 2009); however, the reduced BETL cell wall ingrowth in dek15 was likely an obstacle for nutrition transportation into the developing kernels (Figures 6C and 6D). In addition, the decreased cell number and abnormal morphology of the dek15 SE cells may also affect grain filling (Figures 6E and 6F; Supplemental Figure 7).

The Chromatin Remodeling Complex Plays a Potential Role in Recruiting the Cohesin Ring to Chromatin

In yeast and animals, SCC4 and DNA binding proteins are thought to be responsible for determining the localization of cohesin on chromatin (Chao et al., 2015; Hinshaw et al., 2015, 2017). In the African clawed frog (Xenopus laevis), CDC7-DRF1 protein kinase was found to recruit SCC2/SCC4 to chromatin (Takahashi et al., 2008). In yeast, the phosphorylation of the Ctf19 kinetochore protein by Dbf4-dependent kinase provides a binding site for SCC4 to the centromeres (Hinshaw et al., 2017). Other studies have also suggested that cohesin and the chromatin remodeling complex might interact directly (Hakimi et al., 2002; Baetz et al., 2004; Huang et al., 2004; Clapier and Cairns, 2009) and that the chromatin remodeling complex may recruit the SCC2/SCC4 complex to a specific location on the chromatin (Lopez-Serra et al., 2014; Hinshaw et al., 2015).

In this study, four chromatin remodeling proteins, CHR4, CHB102, CHB105, and CHB106, were found to interact with SCC4 (Figures 10C and 10D; Supplemental Data Set 4). The chromatin remodeling complexes play a vital role in regulating chromatin structure and assembly, the dynamic nature of chromatin, DNA methylation, and histone modification (Ho and Crabtree, 2010). In maize, CHB101 is involved in maintaining nucleosome density and chromosome structure (Yu et al., 2016). In Arabidopsis, the SWI3 proteins may alter the interaction between the histones and DNA to affect plant development and transcription (Sarnowski et al., 2005; Zhu et al., 2013). In rice (Oryza sativa), the CHR4 homolog OsCHR729 recognizes and modulates the methylation of the H3K4 and H3K27 histones to regulate gene expression (Hu et al., 2012). The loss of function of any Chr4-, Chb102-, Chb105-, or Chb106-homologous gene causes severe development defects in plants, including developmental retardation and dwarfing (Sarnowski et al., 2005; Hu et al., 2012). The identification of these SCC4-interacting chromatin remodeling proteins enabled us to propose a model by which SCC4 can facilitate the recruitment of cohesin to chromatin (Figure 10E), which is consistent with the hypothesis that cohesin loading is dependent on the chromatin remodeling complex in yeast (Lopez-Serra et al., 2014; Hinshaw et al., 2015).

The manipulation of cohesin has potential applications in plant breeding, such as the production of clonal seeds using apomixis (Bolaños-Villegas et al., 2017). In two recent reports, genes associated with meiosis were knocked out using CRISPR/Cas9 to produce of clonal seeds in rice (Khanday et al., 2019; Wang et al., 2019). One of these genes encoded the cohesin subunit RECOMBINATION8 (REC8, homologous to SCC1 in mitosis). Considering its function in regulating cohesion in plants, SCC4 could be a potential target for engineering apomixis in plants.

METHODS

Plant Materials

The maize (Zea mays) EMS-generated dek15 mutant (dek15-N1427A) was obtained from the Maize Genetics Cooperation Stock Center (http://maizecoop.cropsci.uiuc.edu/). The dek15 stock was crossed into the W22 inbred line, and kernels were collected from a self-pollinated dek15/+ ear in a predominantly W22 genetic background.

Root, stem, leaf, silk, tassel, and ear tissues were collected from at least three field-cultivated W22 plants at the V12 stage, which were grown in the Experimental Station in Shangzhuang, China Agricultural University, Beijing. Immature kernels were harvested at 6, 9, 12, 15, 18, 21, 24, 27, and 30 DAP. Nicotiana benthamiana plants were grown in growth chambers under a 16-h:8-h light (white fluorescent lamp, 20,000 LUX):dark photoperiod at 25°C.

Histological Analysis

The histological analysis was performed as previously described (Leiva-Neto et al., 2004; Wang et al., 2011; Feng et al., 2018). Five developing wild-type and dek15 kernels were each harvested from the same segregating well-filled ear. Three independent ears were used for subsequent analyses.

For the paraffin sections, fresh dek15 and wild-type kernels were fixed informalin-acetic acid-alcohol fixative (50% [v/v] ethanol, 5% [v/v] acetic acid, and 3.7% [v/v] formaldehyde), after which they were evacuated three times for 5 min with a vacuum pump. The fixed materials were dehydrated in a gradient of ethanol (50%, 60%, 70%, 85%, 95%, and 100% ethanol in water [v/v] ) and a gradient of xylene solution (25%, 50%, 75%, and 100% xylene in ethanol [v/v] ). The samples were then soaked three times in paraffin at 58°C for 12 h, after which they were embedded in a paraffin block. Thin sections (10 μm) were obtained using a microtome (RM2265; Leica), which were then dewaxed in xylene and stained with fuchsin or toluidine blue. To detect the nuclei, the sections were stained with DAPI (H-1200; Vector Laboratories). Fluorescent images were observed using a Nikon Ci-S fluorescence microscope with a DS-Qi2 CCD camera attached to a Epi-fluorescence attachment (Nikon).

For the resin sections, the endosperm tissues were fixed overnight in formalin-acetic acid-alcohol fixative at 4°C. The samples were sequentially soaked in a concentration gradient of ethanol, acetone, and resin, then embedded in Spurr’s epoxy resin at 70°C for 8 h. Thin sections (4 μm) were mounted on glass slides and stained with fuchsin. The dyed sections were rinsed sequentially with deionized water, 50% ethanol, and absolute ethanol, then air-dried. The slices were sealed with cover slips over gum and imaged in a bright field using a Leica microscope (DM2000LED).

For the transmission electron microscopy, endosperm tissues were fixed in 4% paraformaldehyde then transferred to osmium tetraoxide. The fixed samples were dehydrated in an ethanol gradient then transferred to propylene oxide solution, after which they were slowly polymerized in acrylic resin (London Resin Company) for at least 48 h. Thin sections (70 nm) were obtained using a diamond knife microtome (Ultracut E; Reichert Technologies). The sections were placed on 100-mesh copper grids, stained with uranyl acetate for 30 min, then stained with lead citrate for 15 min. The sections were observed using a Hitachi H7600 transmission electron microscope (Japan).

Analysis of Total Protein and Starch

The total protein and starch contents of the kernels were analyzed as previously described (Wang et al., 2011; Feng et al., 2018; Li et al., 2018a). The mature wild-type or dek15 kernels were collected from the same segregating well-filled ears. The seed coats and embryos were removed from the kernels after being soaked in water (30 min). A total of 20 wild-type or dek15 endosperms from the same ear were pooled as a single replicate, and ground into a fine powder in liquid nitrogen. Three biological replicates were used for the subsequent analysis.

A 50-mg sample was incubated overnight in 1 mL lysis buffer (12.5 mM sodium borate, 1% SDS, 2% β-mercaptoethanol, 1% cocktail [Merck], and 1% phenylmethylsulfonyl fluoride) in a 37°C shaker. The mixture was centrifuged for 10 min at 12,000 rpm, after which the supernatant was carefully transferred into a new 1.5-mL centrifuge tube. The total protein was measured according to abicinchoninic acid standard kit (Thermo Fisher Scientific).

For total starch measurements, starch quantification was performed followed the method described by instructions of amyloglucosidase/α-amylase starch assay kit (Megazyme).

Genetic Mapping of dek15 Locus

Initial mapping of dek15 was using maize SNP3072 genotyping array (Tian et al., 2015) with the pooled DNA samples extracted from endosperm tissue of 40 mutant kernels and 40 wild-type kernels from a segregated F2 ear. The dek15 was mapped to a 28.236 Mb interval from 160,472,432 to 188,708,387 on chromosome 4 (maize B73 RefGen_v4). The markers were developed in the loci interval, and 252 kernels were used to narrow the candidate interval between 181.433 (13 recombinants) M and 183.711 M (5 recombinants). Additional markers AC190640.21, AC185627.2, AC185638.10, and AC185456.15 were developed within this interval (Supplemental Data Set 5), and a total of 979 mutant kernels from the F2 population were used to narrow down the dek15 to a 43-kb interval from 182.738 M (7 recombinants) to 183.169 M (3 recombinants) on chromosome 4.

Vector Construction

For the targeting sequences from the cDNA of Dek15 (1000 bp – 1019 bp), 20 bp (GGC​ACC​ACC​GCT​GAT​GCA​TG) was chosen. Oligonucleotides including targeting sequences and gRNA was synthesized as primer, and cloned into pCAMBIA3301 vector using a simplex editing strategy (Qi et al., 2016a). The construct was transformed into Agrobacterium tumefaciens (EHA105).

The full-length ORF of Dek15 with the restriction enzyme sites for BamHI and XbaI was cloned from mRNA extracted from 15 DAP B73 endosperm using primers 5′-CGG​GAT​CCA​TGT​CCA​TCG​CCG​CCG​TG-3′ and 5′-GCT​CTA​GAC​TAC​CGC​CGC​CTC​CTG​GTA​C-3′. The fragment was cloned into a pHB expression vector, in which Cauliflower mosaic virus 35S promoter was replaced by SCC4-promoter (2030 bp upstream DNA sequence from initiation codon of SCC4) using primers 5′-CGG​AAT​TCC​GGC​CTT​CGA​CAG​TTT​TTG​C-3′ and 5′-CGG​GAT​CCG​GCG​GCG​AGG​AGG​G-3′. The construct harbors a Basta selection marker. The construct was transformed into A. tumefaciens (EHA105).

Maize Transformation

Transgenic plants were generated by A. tumefaciens–mediated maize transformation (Frame et al., 2002). A cross between pB and pA lines (pBpA) was used as the recipient for maize transformation experiments. This Bar primer was used for molecular characterization of T₀ transgenic plants. For CRISPR/Cas9 editing plants, sequencing was used to identify editing sites near the target position. Three transgene lines were obtained after crossed to W22. Two independent transgenic lines were selected for functional complementation tests. Six independent transgenic lines with Zm00001d052197 knockout were obtained via CRISPR/Cas9, and two transgenic lines were selected for allelism tests.

Subcellular Localization of SCC4

The full-length ORF (2181 bp) of Scc4 was cloned into pENTR/D-TOPO (primers listed in Supplemental Data Set 5) using a Gateway TOPO cloning kit (Thermo Fisher Scientific). The DNA fragments were fused into the pB7CWG2 plant expression vector through the LR reaction of the Gateway system (Thermo Fisher Scientific), after which the fusion constructs were transformed into onion (Allium cepa) epidermal cells, as previously described (Qiao et al., 2016). The YFP signal was captured using a confocal microscope (A1; Nikon). To examine the subcellular localization of SCC4 in maize kernels, the proteins of the cytoplasm and nucleus were carefully extracted, as previously described (Qi et al., 2016b).

Polyclonal Antibodies

To get the antibody against DEK15/SCC4, a specific cDNA fragments of Scc4 (1500 to 2181 sequence site, representing 500 to 727 amino acids) was cloned into pGEX-4T-1 (Amersham Biosciences). Glutathione S-transferase-tagged SCC4 fusion protein was purified using a GSTrap FF column in the ÄKTA purification system (GE Healthcare). Protein expression, purification, and subsequent production of antibodies with rabbits were performed by Shanghai ImmunoGen Biological Technology according to standard protocol.

Immunoblot Analysis

The total proteins were separated using SDS-PAGE and then transferred onto a polyvinylidene difluoride membrane (0.45 μm; Millipore Sigma, MA, USA) using the Mini-transblot system (Bio-Rad Laboratories). The membranes were blocked using 5% skim milk inTris-buffered saline with Tween 20 (TBST) (20 mM Tris-HCl, pH 7.5; 150 mM NaCl; and 0.05% Tween 20) for 1 h at 25°C, then incubated with the primary antibodies (anti-SCC4 [1:500], anti-actin [1:5,000; Bioeasytech], anti-tubulin [1:5,000; Bioeasytech], anti-bip [1:1,000; Santa Cruz Biotechnology], anti-histone [1:1,000; Cell Signaling]) in 5% milk in TBST for 1 h. These membranes were washed five times for 5 min using TBST, after which the secondary antibody was applied at a 1:5,000 dilution in 5% milk in TBST for 1 h. The SCC4-, bip-, and actin-specific antibodies were detected using goat anti-rabbit IgG conjugated to horseradish peroxidase (Bioeasytech). The tubulin- and histone-specific antibodies were detected using goat anti-rat IgG conjugated to horseradish peroxidase (Bioeasytech). After the samples were washed five times with TBST, the secondary antibodies were visualized using the Super Signal West Pico chemiluminescent substrate kit (Thermo Fisher Scientific) and the Tanon-5200 imaging system (Biotanon, China).

Flow Cytometry

The flow cytometry procedure was performed as previously described (Dolezel et al., 2007). Five developing wild-type or dek15 mutant kernels were each collected from the same ears at 12 and 15 DAP. The endosperm was obtained by removing the seed coat and embryos from each kernel. A single wild-type or dek15 endosperm was analyzed for each repeat. Three individual ears were used for the flow cytometry analysis.

The endosperm was rapidly chopped using a new razor blade in 1 mL ice-cold Galbraith’s buffer (45 mM MgCl₂, 20 mM MOPS, 30 mM sodium citrate, 0.1% [v/v] Triton X-100, adjusted to pH 7.0 using 1 M NaOH, and filtered through a 0.22-μm filter) on a glass Petri dish. The homogenate was filtered through a 42-µm nylon mesh into a 1.5-mL sample tube. Propidium iodide and RNase were added until their final concentration reached 50 μg mL⁻¹ and 50 μg mL⁻¹, respectively, after which the tubes were shaken gently. The homogenate was measured immediately using a flow cytometer (BD FACSCalibur), with an argon-ion laser tuned to 488 nm. A total of 15,000 particles were collected and analyzed using Summit V5.0.1.3804 software (Dako Colorado, Inc.).

Chromosome Preparation

The chromosome preparation procedure was performed as previously described (Kato et al., 2004). Developing wild-type or dek15 mutant kernels were collected from the same segregating ears at 15 DAP, and the embryo and endosperm tissues were obtained by removing the seed coat. Three individual ears were used for subsequent analyses. The embryos or endosperms were fixed overnight in Carnoy's Fluid (ethanol: acetic acid = 3:1, [v/v]), and stored in 70% ethanol at −20°C. After being washed three times with ddH₂O, the tissue sample was transferred to a 20-μL solution of 1% pectolyase and 2% cellulose and hydrolyzed at 37°C for 2 h. After being washed three times with 70% ethanol, the sample was homogenized using a blunt dissecting needle. After the ethanol had evaporated, 100% acetic acid was added to form a mixed-cell suspension. The mixture placed onto glass slides for cell rupture in a humidity chamber, then stained with 10 μL DAPI (Vector Laboratories). The stained homogenate was covered with a cover slip and observed on a fluorescence microscope (Nikon).

Probes and FISH Assay

Plasmids containing the maize tandem repeat 45S rDNA and Cent4-specific sequences were reported previously (Zhao et al., 2013). The 45S and Cent4 probes were nick-translated and labeled with digoxigenin-11-dUTP (Roche). The in situ hybridization protocol was slightly modified from a previously described method (Rayburn and Gill, 1985; Kato et al., 2004). The embryos or endosperms were obtained from three 15-DAP wild-type or dek15 kernels in a segregating ear and independently pooled as individual replicates. Two individual ears were used for the FISH analysis. A 5-μL mixture containing 0.4 μg labeled probe and 40 μg salmon sperm DNA was incubated at 65°C for 5 min. Subsequently, the probe was denatured at 80°C for 10 min, after which it was transferred into a 20-μL mixture of 50% deionized formamide, 10% dextran sulfate, and 2×SSC (0.6 M sodium chloride, 0.06 M sodium citrate). Slides were prepared by dripping 100-μL of 70% deionized formamide and 2×SSC onto their surfaces and incubating them at 85°C for 2 min. The slides were rapidly dehydrated in an alcohol series (70%, 95%, and 100%) at −20°C, after which a 20-μL aliquot of the probe mixture was applied to the slides. The slides were then incubated for 16–24 h at 37°C in a humidity chamber. These probes were detected using an antidigoxigenin antibody conjugated with Rhodamin (Vector Laboratories). The samples were stained using 10 μL DAPI (Vector Laboratories), then imaged using a fluorescence microscope (Nikon).

Yeast Two-Hybrid Assay

The Y2H library was constructed previously (Zhang et al., 2012). The ORFs of Scc4 and its potentially interacting proteins were cloned into the pGADT7 vector at the EcoRI and BamHI restriction sites using a ClonExpress II One Step Cloning Kit (Vazyme Biotech). These ORFs were amplified from B73 cDNA using the primers listed in Supplemental Data Set 5. These constructs were co-transfected into AH109 and pGBKT7-Scc4. The interaction of these constructs with pGBKT7-empty, as well as the interaction between pGBKT7-Scc4 and pGADT7-empty, was used as negative controls. The interaction between the T-antigen and human P53 was used as a positive control. Subsequently, the resulting transformants were spotted onto SD/-Leu/-Trp medium and SD/-Ade/-His/-Leu/-Trp medium.

Luciferase Complementation Image Assay

The N-terminal sequence of SCC2 (642 bp) and the full-length ORF of Scc4 and four genes encoding putative SCC4-interacting proteins were cloned into JW772 (CLUC) and JW771 (NLUC; Zhang et al., 2015) using ClonExpress II One Step Cloning Kit (Vazyme Biotech). These constructs were transfected into A. tumefaciens (strain GV3101), after which the Agrobacterium cells were cultured to OD₆₀₀ = 0.8, pelleted, and suspended in a buffer (10 mM methylester sulfonate, 10 mM MgCl₂, and 150 μM acetosyringone, pH 5.7). The suspended cells were infiltrated into 5-week-old N. benthamiana leaves in different combinations using a needleless syringe. After incubation for 48 h in a growth chamber (16-h:8-h light:dark), the leaves were injected with 1 mM luciferin (Promega Corporation). The luciferase signals were imaged using a Tanon-5200 imaging system. These experiments were independently repeated at least three times.

RNA-seq and qPCR

RNA samples were collected from pooled wild-type or dek15 mutant kernels obtained from the same F₂ ear at 15 DAP (15 kernels per sample). Three biological replicates were collected from three independent ears. The total RNA was extracted from each sample using an RNAprep Pure Plant Kit (Tiangen Biotech, China). The integrity and concentration of RNA samples were tested using a 2100 RNA Nano 6000 Assay Kit (Agilent Technologies). The libraries were constructed and sequenced using an Illumina HiSeq 2500 with Annoroad Gene Technology, which eventually produced ∼48 million reads per sample. The clean reads were obtained by excluding reads containing a poly-N sequence, as well as adapter-polluted and low-quality reads using fqtools_plus (Annoroad). The clean reads were mapped to maize B73 RefGen_v4.37 using TopHat (version: 2.0.13; Langmead et al., 2009). The gene expression levels were estimated using Cufflinks (version: 2.2.1) and cuffdiff2 (version: 2.2.1; Trapnell et al., 2013), and the data were normalized as fragments per kilobase of exon per million fragments mapped (FPKM) after excluding rRNA and tRNA. Genes with expression fold changes > 2 (P < 0.01) were considered significant DEGs.

For the qPCR analysis, cDNA was synthesized from 1 μg total RNA using an oligo-dT andMoloney Murine Leukemia Virus reverse transcriptase (Promega), following the manufacturer's instructions. Gene fragments were amplified using SYBR Select Master Mix (Tiangen Biotech) on an ABI 7500 Real-Time PCR system (Thermo Fisher Scientific). The gene expression levels were assessed using the ΔCt (threshold cycle) method, with Ubiquitin expression as the internal control. The primers used are listed in Supplemental Data Set 5.

Statistical Analysis

All Student’s t tests are shown in the Supplemental Table.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL databases under the following accession numbers: ZmSCC4/DEK15, NP_001352441.1. Sequences used for phylogenetic analysis are as follows: Sorghum bicolor, XP_002446219.2; Oryza sativa Japonica Group, CAD40351.2; Brachypodium distachyon, XP_003581104.1; Arabidopsis (Arabidopsis thaliana), NP_199947.1; Vitis vinifera, XP_010650792.1; Glycine max, XP_003519302.1; Medicago truncatula, XP_003616084.1; Physcomitrella patens, XP_024387203.1; Drosophila melanogaster, NP_650428.1; Xenopus laevis, NP_001124425.1; Homo sapiens, NP_056144.3; Mus musculus, NP_083269.4; Saccharomyces cerevisiae S288C, NP_011074.3; and Eremothecium gossypii, NP_986799.1. RNA-Seq data are available from the National Center for Biotechnology Information Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo) under the series entry GSE120674.

Supplemental Data

• Supplemental Figure 1. Kernel phenotype inheritance and a comparison of developing wild-type (WT) and dek15 kernels at different developmental stages.

• Supplemental Figure 2. Longitudinal paraffin sections of developing wild-type (WT) and dek15 kernels at different developmental stages.

• Supplemental Figure 3. Ears of the Zm00001d052197 knockout alleles and those used for the allelism test with dek15.

• Supplemental Figure 4. Protein sequence alignment of SCC4 and its homologs in other organisms.

• Supplemental Figure 5. Full, scanned images of the immunoblots.

• Supplemental Figure 6. Subcellular localization of SCC4-YFP in onion (Allium cepa) mesophyll cells.

• Supplemental Figure 7. Comparison of the cell morphology of wild-type (WT) and dek15 endosperm at 15 DAP.

• Supplemental Figure 8. Analysis of the nuclear morphology of wild-type (WT) and dek15 endosperm cells using transmission electron microscopy (TEM) and DAPI-stained paraffin sections.

• Supplemental Figure 9. The chromosome behavior of embryo cells in dek15, and endosperm cells and embryo cells in dek15-cas9-1.

• Supplemental Figure 10. FISH with chromosome 4-specific sequences (Cent4) probe in 15-DAP wild-type (WT) and dek15 embryos.

• Supplemental Table. Student’s t test Tables.

• Supplemental Data Set 1. Text file of the alignment used for the phylogenetic analysis shown in Figure 3A.

• Supplemental Data Set 2. Significantly differentially expressed genes of dek15 compared with WT at 15 DAP.

• Supplemental Data Set 3. GO terms classifications in dek15 DEGs with functional annotations.

• Supplemental Data Set 4. Summary of putative ZmSCC4-interacting proteins by a yeast two-hybrid assay.

• Supplemental Data Set 5. List of primers.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• CHR4 Gramene: Zm00001d006428

• CHR4 Araport: Zm00001d006428

• SCC4 Gramene: Zm00001d052197

• SCC4 Araport: Zm00001d052197

• CHB105 Gramene: Zm00001d000110

• CHB105 Araport: Zm00001d000110

• CHB106 Gramene: Zm00001d041847

• CHB106 Araport: Zm00001d041847

• CHB102 Gramene: Zm00001d053687

• CHB102 Araport: Zm00001d053687

• DAPI CHEBI: CHEBI:51231

• PAM Gramene: protospacer adjacent motif

• PAM Araport: protospacer adjacent motif

• SCC2 Gramene: Zm00001d018657

• SCC2 Araport: Zm00001d018657