GENE-39710; No. of pages: 9; 4C: Gene xxx (2014) xxx–xxx

Contents lists available at ScienceDirect

Gene journal homepage: www.elsevier.com/locate/gene

Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant☆ Feng-peng Li a, Min-Young Yoon a, Gang Li a, Won-Hee Ra a, Jae-Wan Park a, Soon-Jae Kwon b, Soon-Wook Kwon c, Il-Pyung Ahn d, Yong-Jin Park a,⁎ a

Department of Plant Resources, College of Industrial Sciences, Kongju National University, Yesan 340-702, Republic of Korea Advanced Radiation Technology Institute, Atomic Energy Research Institute, Jeongeup 580-185, Republic of Korea Department of Plant Bioscience, College of Natural Resources and Life Science, Pusan National University, Milyang 627-706, Republic of Korea d National Academy of Agricultural Science, Rural Development Administration, Suwon 441-107, Republic of Korea b c

a r t i c l e

i n f o

Article history: Received 9 February 2014 Received in revised form 14 May 2014 Accepted 26 May 2014 Available online xxxx Keywords: Rice Transcriptome Starch-synthesis related genes sugary mutant ADP-glucose pyrophosphorylase

a b s t r a c t A sugary mutant with low total starch and high sugar contents was compared with its wild type Sindongjin for grain-filling caryopses. In the present study, developing seeds of Sindongjin and sugary mutant from the 11th day after flowering (DAF) were subjected to RNA sequencing (RNA-Seq). A total of 30,385 and 32,243 genes were identified in Sindongjin and sugary mutant. Transcriptomic change analysis showed that 7713 differentially expressed genes (DEGs) (log2 fold change ≥1, false discovery rate (FDR) ≤ 0.001) were identified based on our RNA-Seq data, with 7239 genes up-regulated and 474 down-regulated in the sugary mutant. A large number of DEGs were found related to metabolic, biosynthesis of secondary metabolites, plant-pathogen interaction, plant hormone signal transduction and starch/sugar metabolism. Detailed pathway dissection and quantitative real time PCR (qRT-PCR) demonstrated that most genes involved in sucrose to starch synthesis are upregulated, whereas the expression of the ADP-glucose pyrophosphorylase small subunit (OsAGPS2b) catalyzing the first committed step of starch biosynthesis was specifically inhibited during the grain-filling stage in sugary mutant. Further analysis suggested that the OsAGPS2b is a considerable candidate gene responsible for phenotype of sugary mutant. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Rice (Oryza sativa L.) is one member of Poaceae and this family also contains economically important cereal crops such as barley, wheat, maize, and sorghum supporting the global food supply (Zhao et al., 2013). Rice is also a representative model system for monocots, because of its various advantages as an experimental plant including a small

Abbreviations: RNA-Seq, RNA sequencing; FDR, false discovery rate; DAF, day after flowering; DEGs, differentially expressed genes; OsAGPS, Oryza sativa ADP-glucose pyrophosphorylase small subunit; OsAGPL, Oryza sativa ADP-glucose pyrophosphorylase large subunit; AGPase, ADP glucose pyrophosphorylase; GBSS, granule-bound starch synthase; SS, starch synthase; SBE, starch branching enzyme; DBE, debranching enzyme; DPE, disproportionating enzyme; PHO, phosphorylase; ISA1, isoamylase I; SAGE, serial analysis of gene expression; qRT-PCR, quantitative real-time PCR; DP, degree of polymerization; RPKM, reads per kb per million reads; FC, fold change; SSRGs, synthesis starch-related genes; GO, gene ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes; eEF1a, eukaryotic elongation factor 1-alpha; bt2, Brittle2; sh2, Shrunken; PPDK, Pyruvate orthophosphate dikinase; SAGE, Serial analysis of gene expression. ☆ The English in this document has been checked by at least two professional editors, both native speakers of English. For a certificate, please see: http://www.textcheck.com/ certificate/LS3obw. ⁎ Corresponding author. E-mail address: [email protected] (Y.-J. Park).

genome size and a known genome sequence (Sasaki and Burr, 2000). Consistent with other angiosperm species, seed development in rice is initiated by double fertilization and asymmetric zygote cell division, which produces a small apical cell that ultimately becomes the embryo and a large basal cell that develops into the endosperm. The classification of gene expression patterns associated with the specific stages of seed development and a functional characterization of the encoded genes are critical for understanding the molecular and biochemical events associated with endosperm development. Seed development is a major item of plant growth and development research, but most of the molecular mechanisms regulating this developmental process are still enigmatic. Starch is the major storage substance that accounts for over 80% of the total dry mass in rice grains and is stored as energy reserves in the sink tissues such as endosperm (Hoshikawa, 1968; Liu et al., 2010). Starch in rice endosperm is composed of relatively unbranched amylose (linear α-1, 4-polyglucans) and highly branched amylopectin (α-1, 6-branched polyglucans) and both starches are synthesized by adding glucose-1-phosphate (Glc-1-P) to the non-reducing ends of the αglucan acceptor molecules catalyzed by ADP glucose pyrophosphorylase (AGPase). Subsequent elongation reactions for the α-1,4chains of amylose and amylopectin are distinctively catalyzed by a

http://dx.doi.org/10.1016/j.gene.2014.05.059 0378-1119/© 2014 Elsevier B.V. All rights reserved.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

2

F. Li et al. / Gene xxx (2014) xxx–xxx

starch granule-bound form of starch synthase (GBSS) and a soluble form of starch synthase (SS), respectively. Amylopectin has a much more defined structure called ‘tandem-cluster structure’ than glycogen because it is composed of tandem-linked clusters (approximately 9–10 nm each in length) where linear α-1,4-glucan chains are regularly branched via α-1,6-glucosidic linkages (Ohdan et al., 2005). AGPase and GBSS synthesize amylose, whereas amylopectin is synthesized by the coordinated actions of AGPase, SS, starch branching enzyme (SBE), and starch debranching enzyme (DBE) (Fig. 1). Disproportionating enzyme (DPE) and phosphorylase (PHO) are involved in starch degradation, but several studies suggested that they may also play possible role(s) in starch biosynthesis (Ball and Morell, 2003; Colleoni et al., 1999; Lu and Park, 2012a; Tetlow et al., 2004). The α-1,4- and α-1,6-glucosidic linkages of amylopectin are formed by multiple types of SS (SSI, SSII, SSIII, and SSIV), SBE (SBEI and SBEII), and DBE (isoamylase and pullulanase) (Fig. 1). All these isoforms of starch-synthesizing enzymes coordinate a network that regulates starch synthesis in the rice endosperm, which finally affects flavor and taste of grain cooking. However, the detailed molecular mechanisms of starch synthesis remain largely unknown. With the availability of complete genome sequences (Sakai et al., 2013), critical materials such as mutants have been used to study gene function and genetic variations. For example, mutations in the waxy gene (encoding granule-bound sucrose synthase, or GBSSI) and its regulators du1 (encoding mRNA splicing factor) and du3 (encoding capbinding protein 20-kDa subunit) resulted in low amylose content (≤2%) and whole opaque endosperm (Dung et al., 2000; Isshiki et al., 2000, 2008). The amylose-extender mutation reduced activity of starch branching enzyme II (SBEIIb) and was culminated in the structural alterations of amylopectin (Nishi et al., 2001). The flo-2 and flo-5 floury endosperm mutations affected the activities of rice starch branching enzyme I (SBEI) and starch synthesis enzyme III (SSIIIa), respectively (Kawasaki et al., 1996; Ryoo et al., 2007). The floury endosperm-4 mutant and the sugary-1 mutant are defective in the activity of pyruvate orthophosphate dikinase (PPDK) and debranching enzyme isoamylase I (ISA1) (Kang et al., 2005; Nakamura et al., 1997). Similar with rice, several caryopsis-related mutations were described in maize. For example, maize sugary-1 and sugary-2 were defective in ISA1 and SSIIa (Kang et al., 2005; Nakamura et al., 1997; Zhang et al., 2004). In addition,

brittle-2 (bt2) and shrunken-2 (sh2) were resulted from the mutations in the small or large subunits of AGPase (Bhave et al., 1990; Hannah et al., 2001). In plants, the major cytosolic AGPase activity is prerequisite for normal starch synthesis in the seed endosperm among barley, maize and rice (Greene and Hannah, 1998; James et al., 2003; Johnson et al., 2003; Lee et al., 2007a). AGPase catalyzes the first committed step of starch biosynthesis and regulates the production of ADPglc and pyrophosphate (PPi) from glucose-1-phosphate (Glc-1-P) and adenosine 5′ triphosphate (ATP) (Lee et al., 2007b; Lu and Park, 2012b). The resulting ADPglc serves as an activated glucosyl donor during α-1,4-glucan synthesis (Lee et al., 2007b). Whereas the prokaryotic AGP is a homotetrameric structure composed of four identical subunits (α4) (Haugen et al., 1976; Lee et al., 2007a), the AGPases in higher plants exist as a heterotetramer (α2β2) containing two large and two small subunits with slightly different molecular weight (Okita et al., 1990; Smith-White and Preiss, 1992; Villand et al., 1993). Rice contains six AGPase genes; two of them encode small subunits OsAGPS1 and OsAGPS2 and remained four encode large subunits OsAGPL1, OsAGPL2, OsAGPL3, and OsAGPL4. The AGPS2 gene encodes the transcripts for AGPS2a and AGPS2b, which differ only in their first exons (the first exon of AGPS2a serves as the first intron of AGPS2b) and are either processed from the common pre-mRNA by alternative splicing or from different promoters. Previously reported gene expression results have also indicated that while OsAGPS2b is largely present in seed endosperm, OsAGPS2a is expressed in leaves (Akihiro et al., 2005; Hirose et al., 2006; Ohdan et al., 2005). Lee et al. suggested the complex formation of OsAGPS2a and OsAGPL3 during transitory starch in rice leaves (Lee et al., 2007a). In rice developing endosperm, at an early stage, the amyloplast-targeted OsAGPS1/OsAGPL1 heterotetramer has the main functional role and the cytosolic OsAGPS2b/OsAGPL2 complex plays a relatively minor role due to its low levels. As the endosperm matures, the cytosolic OsAGPS2b/OsAGPL2 complex confers the dominant enzyme activity in starch synthesis (Lee et al., 2007a). In maize and rice, mutations in AGPS2b and AGPL2 resulted in the bt2 and sh2 phenotypes due to the significant reduction of starch synthesis in grains (Bhave et al., 1990; Greene and Hannah, 1998; Lee et al., 2007a). Massively parallel sequencing technology is more sensitive for detection of transcripts expressed at low levels than traditional methods

Fig. 1. A simplified metabolic pathway from sucrose to starch in rice caryopsis. a, cell wall invertase; b, cytoplasmic invertase; c, sucrose synthase; d, hexokinase; e, phosphoglucose isomerase; f, UGPase; g, cytoplasmic and plastidial phosphoglucomutase; h, cytoplasmic and plastidial AGPase; i, ADPglc transporter.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

F. Li et al. / Gene xxx (2014) xxx–xxx

such as serial analysis of gene expression (SAGE) and microarrays (Mortazavi et al., 2008; Nagalakshmi et al., 2008; Sultan et al., 2008; Wang et al., 2008; Wilhelm et al., 2008). These genome-wide transcription profiling are important and useful tools allowing the generation of testable hypotheses for novel molecular processes. Their potentials were demonstrated in the investigations for a variety of developmental processes including flowering, endosperm/embryo formation, defense responses during plant-microbe interactions, and wounding (Cheong et al., 2002; Lan et al., 2004; Xiang et al., 2011). Especially, microarray analysis has been adopted to gain the initial clues for the selection and functional characterization of candidate genes based on their expression pattern. Plus, recently developed deep RNA sequencing (RNASeq) technologies such as digital gene expression and Solexa/Illumina RNA-Seq have dramatically enhanced the efficiency to identify seed development-related genes in plants because these technologies facilitate the investigation of the functional complexity of transcriptomes (Ozsolak et al., 2010; Wang et al., 2009). RNA-Seq allows wholetranscriptome shotgun sequencing, where mRNA or cDNA is mechanically fragmented, resulting in overlapping short fragments that cover the entire transcriptome. In addition, this massively parallel sequencing technology is more sensitive for detection of low-copy transcripts than traditional SAGE or microarray hybridizations (Anisimov, 2008; Cloonan et al., 2008; Garber et al., 2011). The expression of all virtual genes in a sample is measured by counting the number of individual complementary DNA (cDNA) molecules produced from each gene. RNA-Seq is more suitable and affordable for comparative gene expression studies because it verifies direct transcript profiling without compromise and potential bias, thus allowing for more sensitive and accurate profiling of a transcriptome that reflects the biology of the cell (Garber et al., 2011; Wang et al., 2009). This technology has been used in transcriptome profiling studies of various organisms, including maize, red pepper, rice, and soybean (Eveland et al., 2010; Liu et al., 2010; Lu et al., 2011; Luo et al., 2011; Severin et al., 2010; Zhai et al., 2013; Zhang et al., 2010). Rice grain development was divided into four stages; initiation stage (1–3 DAF), early developmental stage (3–5 DAF), middle stage (5–10 DAF) and the final stage (10 DAF onward). Subsequently, both endosperm starch and seed weight rapidly increase and reach to the maximum values (Ohdan et al., 2005). sugary mutant was obtained from the Sindongjin variety by artificial γ-ray irradiation, but the underlying molecular mechanisms are not yet characterized. In the present study, we comparatively analyzed the grain transcriptomes (11-DAF) of Sindongjin and its sugary mutant via RNA-Seq to gain insight for potential mechanisms of sugary mutation and enhancing our understanding for the molecular and cellular processes during the rice grain-filling caryopses. A comprehensive analysis of the transcription levels of genes encoding starch-synthesis enzymes was then carried out to assess the effects of sugary mutation on the regulation processes of starch biosynthesis in rice seeds. Quantitative real-time PCR was also used to validate the expression profiles of 27 rice genes known to be involved in starch synthesis during the grain-filling stages. The primary results provide a new perspective for the underlying mechanism of sugary mutation.

3

amylose and amylopectin contents, were compared between Sindongjin and sugary mutant (Table 1). No obvious differences were found in the relative content of the protein, amylose, and amylopectin between Sindongjin and the sugary mutant. However, the grain thicknesses were different significantly and the 1000 grain weight was decreased by 58.1% in sugary mutant (Table 1). Gel permeation chromatography revealed that sugary mutant grains had a higher content of short-chain amylopectin [DP (degree of polymerization) ≤ 12], but less medium (13 ≤ DP ≤ 24 and 25 ≤ DP ≤ 36) and long (37 ≤ DP) amylopectin chains. The lower ratio of medium- and long-chain amylopectin in sugary was previously described (Han and Hamaker, 2001) and will result in the serious breakdown of starch granules during heating. Normal fine amylopectin composition is thereby prerequisite for the starch granule formation and high quality rice grain. 2.2. Illumina sequencing and de novo assembly

2. Results

To analyze the transcriptome of Sindongjin and sugary mutant at the grain-filling stages, cDNA libraries were prepared from rice seeds at 11th DAF and analyzed by RNA-Seq using the Illumina HiSeq 2000 platform. Totals of 20,075,124 and 21,206,354 paired-end raw reads (101 bp) were generated from Sindongjin and sugary mutant. Transcript reads containing adaptor sequences, duplicate sequences, ambiguous reads, and low-quality reads were eliminated. The sequence reads were aligned to the rice reference genome database (Rice Genome Annotation Project) using SOAPaligner/soap2 software (set to allow five base mismatches). Among the total reads, 16,493,525 (82.16%) and 16,113,002 (75.98%) reads were mapped to genomic regions; the remaining 3,581,599 (17.84%) and 5,093,352 (24.02%) were not matched to the reference transcriptome (Table 2) because only reads aligning entirely inside exonic regions will be matched to the transcripts (reads from exon–exon junction regions will not match). One of the most elementary targets in RNA-Seq analysis is alignment of reads to the reference genome. By comparison with the rice genome and trimming, all reads were mapped into 33,674 genes. Between Sindongjin and sugary mutant, the number of genes found was 30,385 and 32,243, providing massive data for seed development. Heterogeneity and redundancy are two important characteristics for mRNA expression. While the majority of mRNA is expressed at low levels, a small proportion of mRNA is highly expressed. Therefore, distribution of genes' coverage was used to evaluate the quality of sequencing. As shown in Fig. 2b and c, the distribution of genes' coverage over two RNA-Seq datasets showed a reliable reproducibility and uniformity. The similarity distribution had a comparable pattern with more than 25% of the sequences having a similarity 90%, while approximately 90% of the hits had a similar range (Fig. 2b and c). The size distribution of these transcripts is shown in Table 3. The majority (94.83%) of the transcripts are distributed from 500 to 5000 bp (Table 3). Transcripts ranging between 2000 and 5000 bp accounted for 28.14% and followed by 1000–1500 bp (24.58%), 1500–2000 bp (21.08%), and 500–100 bp (21.03%). Comparative analyses of the transcriptome revealed that most (28,954) of the genes were transcribed commonly and 1431/3289 genes were specific in Sindongjin and sugary (Fig. 2d). Although there are 28,954 genes that were commonly expressed in two samples, many of them were quantitatively regulated.

2.1. Agronomic and physicochemical properties of Sindongjin and sugary mutant

2.3. RNA-Seq global data analysis and evaluation of differentially expressed genes

Grain shape of sugary mutant was evidently different compared to that of Sindonjin. The endosperm of this mutant was shrunken, opaquely whitish, and the contents within the hull were almost empty (Fig. 2a). Except these alterations, all other agronomic characteristics were remained unchanged. In the current study, six phenotypic traits and three physicochemical traits, culm length, panicle length, grain length, grain width, grain thickness, 1000 grain weight, protein content,

In this study, we employed RNA-Seq to investigate the transcriptome alteration underlying the sugary mutation. Transcript levels were described as reads per kb per million (RPKM) (Marioni et al., 2008). Their detailed gene expression level between Sindongjin and sugary mutant is shown in Supplementary Table 1. Based on this analysis, the average expression levels of genes in Sindongjin and sugary mutant were 26.7 and 25.2 RPKM, respectively. These gene expression

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

4

F. Li et al. / Gene xxx (2014) xxx–xxx

Fig. 2. RNA-Seq mapping data and comparison of genes expressed in the Sindongjin and sugary mutant transcriptomes. (a) Rice grains of Sindongjin and its sugary mutant. (b) Distribution of gene's coverage in sugary mutant. (c) Distribution of gene's coverage in Sindongjin. (d) Venn diagram of RNA-Seq genes unique and common to showing the number of gene expressed in Sindongjin and sugary mutant. (e) Number of transcripts whose expression levels differed between the Sindongjin and sugary mutant transcriptomes. (f) Scatter plot of the gene expression level pairwise between the two libraries.

levels in the two transcriptomes were classified into five categories (rare, low, moderate, high, and extremely high). The expression levels of the most genes were low (RPKM was between 3 and 10) or moderate (between 10 and 50). A small fraction (2.0–3.1%) of them was expressed at extremely high levels (RPKM N100) (Fig. 2e and Supplementary Table 1). There were clear linear relationships between the gene expression levels of Sindongjin and sugary mutant (Fig. 2f). To obtain statistical confirmation of the differences in gene expression between the two RNA-Seq datasets, we compared the normalized values using a likelihood ratio test. Differentially expressed genes (DEGs) were identified by calculating fold change (FC) and log2 values of FC. DEGs exhibiting an estimated absolute log2 (FC) ≥ 1 and FDR ≤ 0.001 were determined to be significantly differentially expressed. Table 1 Agronomic and physicochemical properties of the endosperm and grain shape of Sindongjin and the sugary mutant. Trait

Culm length (cm) Panicle length (cm) Grain length (mm) Grain width (mm) Grain thickness (mm) 1000 Grain weight (g) Protein content (%) Amylose content (%) DP ≤ 12 13 ≤ DP ≤ 24 25 ≤ DP ≤ 36 37 ≤ DP

A total of 7713 genes were differentially expressed between Sindongjin and sugary mutant; 7239 were up-regulated and 474 down-regulated during rice seed development (Fig. 3a, b, and Supplementary Table 2). All the DEGs were classified into four groups according to the fold change; 1–4 fold (1 ≤ log2 FC b 2), 4–8 fold (2 ≤ log2 FC b 3), 8–16 fold (3 ≤ log2 FC b4), and ≥ 16 fold (log2 FC ≥ 4) (Supplementary Table 2). Differences between two data points were marked, and many DEGs were classified into four groups: 1–4 fold (5348), 4–8 fold (1421), 8–16 fold (453), and ≥ 16 fold (491), indicating significant changes in expression levels during development of Sindongjin and sugary mutant seeds. To provide deep insights into the process of starch synthesis transcriptomes during the rice seed grain-filling stage, 27 starch synthesis-related genes (SSRGs) were investigated in our study. RPKM analyses showed that transcription of 25 SSRGs were relatively increased in sugary. However, 12 SSRGs' alterations were not significant because their increments were not reached to 2 folds. Subsequent

Plant Sindongjin

sugary mutant

82.3 ± 1.96 22.91 ± 1.11 5.96 ± 0.25 3.19 ± 0.12 2.13 ± 0.07 27.61 ± 0.17 7.93 18.9 36.7 53.78 8.47 1.04

75.8 ± 1.25 21.03 ± 1.33 6.02 ± 0.19 3.06 ± 0.11 1.23 ± 0.09 12.27 ± 1.04 8.9 17.7 40.05 52.88 6.79 0.28

Values are expressed as mean ± SD, DP: degree of polymerization.

Table 2 Mapping results of RNA-Seq reads from Sindongjin and sugary mutant at 11th DAF.

Total reads Mapped reads Perfect match reads ≤5 bp mismatch reads Unique match Muti-position match Unmapped reads

Sindongjin

sugary mutant

20,075,124 16,493,525 (82.16%) 11,791,187 (58.74%) 4,702,338 (23.42%) 14,992,225 (74.68%) 1,501,300 (7.48%) 3,581,599 (17.84%)

21,206,354 16,113,002 (75.98%) 11,234,113 (52.98%) 4,878,889 (23.01%) 15,596,351 (73.55%) 516,651 (2.44%) 5,093,352(24.02%)

Total reads which are filtered from dirty raw reads are mapped to reference sequences using SOAPaligner/soap2. Mismatches no more than 5 bases are allowed in the alignment. Mapped reads are the sum of perfect match reads and less than 5 bp mismatch reads. Numbers enclosed in parenthesis represent the percents of reads.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

F. Li et al. / Gene xxx (2014) xxx–xxx Table 3 Distribution of the 33,674 transcript sequences detected in developing rice seeds by RNASeq. Transcripts length (bp)

Number

Percentage (%)

100–500 500–1000 1000–1500 1500–2000 2000–5000 ≥5000 Total

1531 7083 8277 7097 9475 211 33,674

4.55 21.03 24.58 21.08 28.14 0.63 100

For 33,674 of these transcripts, the size distribution can be divided into six species. The total number and the percentage of all genes are presented in this table.

analyses for the transcripts exhibiting meaningful alterations indicated that 13 SSRGs including OsBEIIa, OsPHOL, OsDPE2, three AGPase, and seven starch synthase genes were up-regulated. Similarly, expressions of OsAGPL4 and OsAGPS2 (OsAGPS2a and OsAGPS2b) were decreased and OsAGPS2b's transcription was down-regulated by 33.5 times in sugary (Supplementary Table 1 and Supplementary Fig. 1).

2.4. Functional grouping of differentially expressed genes based on RNA-seq data A massive data was generated by RNA-Seq, which requires efficient tools for data visualization and meaningful analysis. To identify the putative functions of genes related to seed development, gene ontology (GO) terms were used to match annotated genes with known proteins. An in-house script was used to obtain GO terms for the annotated genes. As an international standardized gene functional classification system, GO offers both dynamically updated controlled vocabulary and strictly defined concepts to comprehensively describe the properties of genes and their proteins. It can be useful to understand the distribution of gene functions at the macrolevel. All DEGs were categorized into three major groups: biological process, molecular function, and cellular component. A total of 7713 genes were categorized into 42 functional subcategories based on sequence homology. In each of the three main categories of the GO classification, there were 11, 10, and 21 functional groups (Fig. 4). Cellular component is major in the GO annotations (50%), followed by biological process (26%) and molecular function (24%). Among these groups, the terms membrane-bounded organelle (GO: 0043227), catalytic activity (GO: 0003824), and small molecule metabolic process (GO: 0044281) were frequent in each of the three

5

main categories, respectively. All DEGs were associated with at least one GO term. These analyses demonstrated that the genes expressed in rice seeds encode diverse proteins involved in cellular, biological, and molecular processes. The relations among the genes are important for numerous biological functions and pathway-based analyses of these relations further enrich understanding for indwelled molecular functions. To perform functional classification and pathway assignment, all DEGs were analyzed against the KEGG (Kyoto Encyclopedia of Genes and Genomes) (http://www.genome.ad.jp/kegg/) database. KEGG is the major public pathway-related database providing classification that is valuable for research on genetically and biologically complex behaviors. It enables systematic analysis of inner-cell metabolic pathways and functions of gene products which will facilitate investigation of the complex biological behaviors of genes. By mapping to the reference canonical pathways, all DEGs were assigned to 125 KEGG pathways (Supplementary Table 3). A large number of DEGs were found related to metabolic (1082), biosynthesis of secondary metabolites (646), plant-pathogen interaction (347), plant hormone signal transduction (269) and starch and sucrose metabolism (122). These annotations provide a valuable resource for investigating specific processes, functions, and pathways during rice seed development.

2.5. Expression analysis of 27 SSRGs in Sindongjin and sugary mutant during grain-filling by real-time PCR (qRT-PCR) To confirm the gene expression patterns derived from our RNA-Seq data, the expression profiles of 27 SSRGs were assessed on the 11th DAF in Sindongjin and sugary mutant by qRT-PCR. The oligonucleotide primer sequences used to quantify the transcripts for individual genes of interest are listed in Supplementary Table 4. To discriminate between two transcript types of the OsAGPS2 gene (OsAGPS2a and OsAGPS2b transcripts), the primer pairs were designed from their variable first exon. To verify the specificity of each primer set, the amplification products were cloned and sequenced (data not shown). Relative transcription data from real-time RT-PCR and RPKM assays results showed that expression level of 25 SSRGs were up-regulated in sugary mutant (Supplementary Table 1 and Supplementary Fig. 1). Intriguingly, expression levels of OsAGPL4 and OsAGPS2 (OsAGPS2a and OsAGPS2b) were decreased and only OsAGPS2b was significantly inhibited (Fig. 5a and Supplementary Fig. 1). These results were consistent with the RNA-Seq data, although the calculated fold changes showed slight variation.

Fig. 3. Scatter plot of the expression levels of genes in Sindongjin and sugary mutant. (a) Gene expression profile. (b) Differentially expressed genes between Sindongjin and sugary mutant.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

6

F. Li et al. / Gene xxx (2014) xxx–xxx

Fig. 4. GO terms associated with DEG transcripts. The results are summarized in three main categories: biological processes, cellular component, and molecular function.

Further experiments showed that the expression profiles of OsAGPS2b exhibited a sharp decrease significantly from DAF 2 to 15 days in sugary mutant (Fig. 5b). 3. Discussion Starch is a group of polymers consisting of a large number of glucose and major energy reserve in sink tissues such as endosperm. A more complete understanding of the processes involved in starch biosynthesis is prerequisite for improving yields of cereal grains such as rice, maize, and wheat, and industrial needs (Lee et al., 2007a; Morell and Myers, 2005; Van Camp, 2005). Carbohydrate charges for 72.9–75.9% of rice grains and the starch contents are variable across the rice varieties. Research is being carried out using rice as a raw material for noodle preparation, cakes, and brewing, and the results have shown that diverse products can be developed from rice due to its amylose, sugar, and carbohydrate contents as well as its starch structure, in combination with its other qualities (Cho et al., 2008; Omura and Satoh, 1984; Satoh

and Omura, 1981). As sugar content in wild rice was only 0.7–1.3%, the sugar has not been a main investigation item for the rice quality enhancement. However, since the development of high-sugar-content rice (Koh and Heu, 1994), researchers now consider that sugar is one of the main factors in rice quality improvement. Omura and Satoh (1984) reported that variable species of “sugary and shrunken” rice contained higher sugar content as much as 1.6–5.1% and relatively lower starch. In addition, these rice mutants produced more soluble sugar. Several maize lines, including sugary-1, sugary-2, sugary enhancer, shrunken-1, sh2, shrunken-4, brittle-1, and bt2, also exhibited similar phenotypes. The different starch combinations within diverse sugary and shrunken varietal maize lines are well defined and are now being used in maize breeding (Boyer and Shannon, 1983; Yoon et al., 2009). However, the mechanism of rice lines with increased sugar content (including sugary-2 and bt2) are not fully understood. In this study, the sugary mutant was obtained from the Sindongjin variety by artificial γ-ray irradiation. The sugary mutation was really outstanding and distinguishable from its wild type with naked eye observation. There

Fig. 5. RPKM and qRT-PCR analysis. (a) Validation of RPKM data for the AGPase transcripts using relative real time PCR in Sindongjin and sugary mutant; (b) expression level of the OsAGPS2b during the development of rice seeds. Rice spikelets were harvested at 2, 5, 7, 10, 11, 13, and 15 days after flowering in Sindongjin and sugary mutant.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

F. Li et al. / Gene xxx (2014) xxx–xxx

were no differences except the grain shapes. Further agronomic and physicochemical analyses also went together with above grain-specific differences. Interestingly, this phenotypic variation was almost completely identical with that of the previously described rice mutation, osagps2-1 (Lee et al., 2007a). This mutation has resulted from the T-DNA insertion into the first intron region and resulted in the abolishment of OsAGPS2b expression. OsAGPS2b encodes the small subunit of the heterotrameric AGPase catalyzing the first step of starch biosynthesis. osagps2-1 mutant also showed the seed shrinking and opaque endosperm identical with phenotypic variation observed in our sugary mutant. In rice grains, starch is the major storage substance that accounts for over 80% of the total dry mass. There were also no significant differences in the relative contents of the protein, amylose and amylopectin. However, the 1000-grain weight in sugary mutant was decreased to 41.9%. Our previous research showed that the total tocopherol, free sugar, and fatty acid contents were elevated in sugary mutant; in particular, the free sugar content was 8.6 fold higher than the Sindongjin (Cho et al., 2008). Decrement of starch and increment of sugar were the common major features shared by AGPase mutants in rice and corn involving sh2 and osagps2-1, therefore, we hypothesized that our sugary phenotype is also responsible for mutation(s) in the starch/sugar metabolism. The transcriptome is the complete set of transcripts for certain types of cells or tissues at a specific developmental stage or environmental condition and its analysis can provide a comprehensive understanding of the molecular mechanisms involved in specific biological processes and diseases based on gene structure and functional information. Transcriptome analyses are no longer challenging to analyze due to the efficient and rapid RNA-Seq technique. Several recent studies have exploited this technology to generate transcriptome information for many plant species including Arabidopsis (Filichkin et al., 2010), red pepper (Lu et al., 2011), rice (Lu et al., 2010; Ma et al., 2012; Zhang et al., 2010), and soybean (Severin et al., 2010). In maize, RNA-Seq was used to obtain detailed transcriptome information for leaves (Li et al., 2010a) and inflorescence (Eveland et al., 2010; Li et al., 2010b; Sekhon et al., 2013). In the present study, developing seeds of Sindongjin and sugary mutant at 11th DAF were subjected to RNASeq. A greater number of genes were obtained using RNA-Seq technology from developing seeds than from endosperm (23,836) and embryos (27,190) (Gao et al., 2013; Xu et al., 2012). Totally, 28,954 genes were commonly expressed in Sindongjin and sugary, many of them are quantitatively regulated. While some of these genes exhibited little variation or a low expression level, and were thought to be housekeeping genes or expressed at a low-abundance during seed development. To obtain all the DEGs from RNA-Seq data, the expression of all genes was analyzed depending on the normalized RPKM after log transformation. At last, we detected 7713 DEGs composed with 7239 up- and 474 down-regulated genes in sugary mutant. The DEGs are excellent candidates for future functional genomic studies to better understand seed development. However, we cannot rule out the possibility that other biological differences may have led to the differences in transcript levels. Based on the comparative analyses of the RNA-Seq datasets and the public information for the metabolic pathway, we want to narrow down the number of candidate genes responsible for sugary mutation. RNA-Seq analyses revealed a large number of DEGs corresponding to metabolic (1082), biosynthesis of secondary metabolites (646), plantpathogen interaction (347), plant hormone-mediated signal transduction (269) and starch/sugar metabolism (122). In rice, a total 27 SSRGs were involved in rice starch biosynthesis. Based on qRT-PCR and RNA-Seq analysis results, almost all SSRGs (25 genes) transcribed more actively in the sugary mutant in spite of its poor starch accumulation during the endosperm development. Surprisingly, expression of OsAGPS2b has almost completely disappeared in the sugary mutant. This alteration is definitely reminiscent of osagps2-1 lack of OsAGPS2b expression. As described above, disruption of OsAGPS2b and subsequent disappearance of its expression resulted in the phenotypic alteration

7

identical with that in our sugary mutant. Therefore, our results here strongly indicated that our sugary should blow up normal OsAGPS2b's expression regulation required for starch accumulation in endosperm and higher transcription of other 25 SSRGs should be meaningless in the absence of the OsAGPS2b's expression. Lack of precursor catalyzed by AGPase might be culminated in the blackout of the whole starch biosynthesis. All the F1 plants between wild type Sindonjin and its sugary mutant produced normal seeds, and the F2 population segregated into 3 (normal seed producer):1 (producing shrunken and opaque seeds) ratio. These results indicated that sugary mutation was recessive and occurred in the single locus (unpublished results). Comparison of the nucleotide sequences for promoter region and coding DNA sequence region of OsAGPS2b didn't reveal any differences between Sindonjin and sugary (unpublished results). Therefore, we can speculate that our sugary mutation should be placed in a certain gene exerting its activity through the regulation of OsAGPS2b. In sum, our result presented here strongly implied the presence of a certain gene involved in starch biosynthesis and regulation of OsAGPS2b expression. This information could be important clues for the more intensified molecular and biochemical mechanisms during the grain-filling caryopses of rice. 4. Conclusion In this work, we comparatively analyzed the transcriptome alterations during the grain-filling stage in Sindonjin and sugary mutant through RNA-Seq. This study produced pregnant data for the rice seed development. RNA-Seq analyses enabled us to characterize gene expression profiles and identify DEGs. In addition to the previous molecular and biochemical studies for the grain-filling caryopses, our result here suggests an unknown gene's contribution or involvement of novel metabolic regulation involved in the OsAGPS2b's expression. This finding should be a cornerstone for the intensified investigation for endosperm formation and seed development in rice and other cereal crops. 5. Materials and methods 5.1. Plant materials and RNA extraction The rice cultivars Sindongjin (O. sativa L. ssp. japonica) and sugary mutant, which was induced from Sindongjin by artificial γ-irradiation, were grown in a field at Kongju National University. The seeds of the batches that flowered simultaneously for Sindongjin and sugary mutant were collected at 2, 5, 7, 10, 11, 13, and 15 DAF. The samples were immediately frozen in liquid nitrogen, and stored at − 80 °C. Total RNAs were isolated from seeds using TRIzol reagent following the manufacturer's instructions (Invitrogen). The yield and purity of the RNAs were assessed using the absorbance (Abs) at 260 and 280 nm. The ratio of the absorbance at 260 nm to that at 280 nm was used as an indication of sample purity, and values of 1.8–2.0 were considered indicative of relatively pure RNA. The integrity of RNA was verified on a 1.0% agarose gel using the RNA 6000 Nano Assay Kit and an Agilent Technologies 2100 Bioanalyzer. The extracted total RNA and harvested seeds were stored at −80 °C and 4 °C, respectively, for later use. 5.2. Agronomic and physicochemical properties of endosperm and grain shape analysis Amylose content was measured based on the method of Li et al. (2014), and Perez and Juliano (1978). Briefly, 100 mg rice flour was placed into a 100 mL volumetric flask, and 1 mL of 95% ethanol and 9 mL of 1 M aqueous sodium hydroxide were added. The contents were boiled for 8 min, cooled to room temperature, and distilled water was added. Five milliliters of the solution was placed in a 100 mL volumetric flask with 1 mL of 1 M aqueous acetic acid and

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

8

F. Li et al. / Gene xxx (2014) xxx–xxx

2 mL of 2% I2-KI solution, and the volume was increased to 100 mL with distilled water. The absorbance of the solution was measured at 620 nm using a spectrophotometer, and a standard curve was generated using rice samples of known amylose content and used to calculate the amylose content of each sample. Glucose chain length distributions were analyzed based on the method of Wong et al. (2003). Thirty milligrams of starch was dissolved in 2 N NaOH, solubilized at 4 °C overnight, and then neutralized using 0.5 N HCl. Acetate buffer (60 mM, pH 3.5) and isoamylase were added to the mixture, which was incubated with shaking at 40 °C for 24 h. Ethanol was then added, and the sample was evaporated in a vacuum at 30 °C. The dried sample was gelatinized again with 1 N NaOH, phosphoric acid was added, and it was then centrifuged twice. Next, the rice starches were debranched and the distribution of the chains was analyzed by gel-permeation chromatography (Breeze System, Waters, Milford, MA). The system is equipped with a differential refractive index (DRI) detector, and TSK-gel G2000PW and G3000PW columns (300 × 7.5 mm) connected in series. The mobile phase in the column was phosphate buffer with 20 mM NaNO3 at a flow rate of 1 mL/min. Crude proteins were analyzed using the association of official agricultural chemists' method (o.O.A.C. (AOAC), 2005). Briefly, 1 g of powdered white rice was mixed with concentrated sulfuric acid and digested for 45 min at 450 °C. After cooling to room temperature, the total nitrogen contents were measured using an automatic microKjeldahl system (FOSS: Kjeltec® 2300 Analyzer Unit, Foss Tecator AB, Höganas, Sweden). Protein content was calculated from the measured total nitrogen content using a conversion factor of 5.95. The other phenotypic traits, including culm length (cm, average of 20 plants), panicle length (cm, average of 20 plants), grain length (mm, average of 20 seeds), grain width (mm, average of 20 seeds), grain thickness (mm, average of 20 seeds), and 1000 grain weight (g), were also estimated (Satheeshkumar and Saravanan, 2012). 5.3. TruSeq mRNA library construction and sequencing The total RNA samples were pooled, and 10 μg total RNA from each pool was used to isolate poly(A) + mRNA and prepare a nondirectional Illumina RNA-Seq library using the reagents provided in the Illumina® TruSeq™ RNA Sample Preparation Kit. The first step in the workflow is the purification of poly-A-containing mRNAs using poly-T oligo-attached magnetic beads. Following purification, the mRNA is fragmented into small pieces using divalent cations under an elevated temperature. The cleaved RNA fragments were copied into first-strand cDNA using reverse transcriptase and random primers. The next stage was second-strand cDNA synthesis using DNA polymerase I and RNase H. The cDNA fragments were then subjected to an end-repair process, the addition of a single ‘A’ base, and then ligation of the adapters. To achieve the highest quality of data on Illumina sequencing platforms, it is important to create optimal clusters with densities across every lane of every flow cell. Two-hundred-and-two-base-pair cDNA fragments were isolated and enriched by PCR to create the final cDNA library. Pools of two samples per lane were sequenced on a HiSeq 2000 for 100 paired-end cycles (2 × 101 bp) following the manufacturer's instructions (Illumina, San Diego, CA). The processing of fluorescent images into sequences, base-calling, and quality value calculations were performed using the Illumina data processing pipeline (version 1.8). 5.4. Mapping RNA-Seq reads to the reference genome and annotated gene The rice genome and gene information were downloaded from the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu). Sequencing-received raw reads were transformed by base culling into sequence data. The raw reads were prior to mapping to the reference database and cleaned by removing adaptor sequences, empty reads and low quality reads (the percentage of the low quality bases of quality value ≤ 10 is more than 50% in a read). To correct sequence bias and

handle multi-mapped reads, “v” and “−b” options were used. We also used the “g” option for making the best use of known gene annotation information. Default options were used for all other parameters. The remaining reads were aligned to the rice genome using SOAPaligner/ soap2, allowing up to five base mismatches when mapping the reads to the reference. Reads that failed to be mapped were gradually trimmed off, one base at a time from the 30-end and mapped to the genome again until a match was found. 5.5. Analysis of differentially expressed genes (DEGs) and functional annotation during rice seed development Normalization is used to remove non-biological influences on biological data, and to make data comparable between experiments, runs, and lanes. It enables accurate comparisons of the expression levels between and within samples (Marioni et al., 2008; Mortazavi et al., 2008; Sultan et al., 2008). The expression levels of transcripts from RNA-Seq were normalized by RPKM method, which could eliminate the influence of different gene lengths and sequence discrepancies on the calculation of gene expression (Mortazavi et al., 2008). Therefore, all gene expression were calculated and directly used for comparing the difference of gene expression between Sindongjin and sugary mutant. The cutoff value for determining gene transcriptional activity was determined based on a 95% confidence interval for all RPKM values. In the study, we applied R package DEGseq to identify DEGs with the random sampling model based on the read count for each gene in two libraries from Sindongjin and sugary mutant (Trapnell et al., 2010). DEGs were assessed by fold-change (FC) and log2 values of FC. DEGs that exhibited an estimated absolute log2 (FC) ≥ 1 and FDR ≤ 0.001 were determined to be significantly differentially expressed. The Blast2GO (version 2.3.5) program was used to obtain GO annotations (http://www.blast2go.org/) for the all DEGs with the default parameters (Conesa et al., 2005). Blast2GO was also used for a GO functional enrichment analysis of all DEGs, by performing Fisher's exact test with a robust FDR correction to obtain an adjusted p-value between certain test gene groups and the whole genome annotation. For pathway analysis, we mapped all DEGs to terms in the KEGG (Kyoto Encyclopedia of Genes and Genomes) (http://www.genome.ad.jp/kegg/), and then looked for significantly enriched pathway terms compared with the genome background (Xie et al., 2011). 5.6. Quantitative real-time RT-PCR Total RNAs were isolated from developing grain-filling seeds using TRIzol reagent following the manufacturer's instructions (Invitrogen). First-strand cDNA was synthesized from 1 μg total RNA using SuperScript II RNase H− Reverse Transcriptase (Takara). Aliquots of the first-strand cDNA mixtures corresponding to 120 ng of cDNA served as the templates for quantitative real-time RT-PCR to analyze the expression profiles of 27 rice synthesis starch-related genes (SSRGs) with the Quantitect™ SYBR Green PCR Kit (SolGent). Reactions were carried out on an iCycler (Bio-Rad) according to the manufacturer's protocols. These results were confirmed using three biological replicates, with three technical repeats for each biological replicate. The eEF1a (eukaryotic elongation factor 1-alpha) gene from rice was used as an internal standard. To optimize the PCR conditions for each primer set, the annealing temperature, and PCR efficiency were assessed. The specificity of the PCR amplification was verified using melt curve analysis (from 55 °C to 94 °C) following the final cycle of PCR. Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.gene.2014.05.059. Conflict of interest There are no conflicts of interest.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059

F. Li et al. / Gene xxx (2014) xxx–xxx

Acknowledgments The Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Korean Ministry of Education, Science and Technology (No. 2010-0022616), and the BioGreen 21 Program (No. PJ009099), Rural Development Administration, Republic of Korea supported this research. References A.o.O.A.C. (AOAC) , 18th ed., AOAC Intl., Gaithersburg, MD, USA, (2005). Akihiro, T., Mizuno, K., Fujimura, T., 2005. Plant Cell Physiol. Plant. 46, 937–946. Anisimov, S.V., 2008. Curr. Pharm. Biotechnol. 9. Ball, S.G., Morell, M.K., 2003. Annu. Rev. Plant Biol. 54, 207–233. Bhave, M.R., Lawrence, S., Barton, C., Hannah, L.C., 1990. Plant Cell 2, 581–588 (Online). Boyer, C.D., Shannon, J.C., 1983. Plant Breed. 1, 139–161. Cheong, Y.H., Chang, H.S., Gupta, R., Wang, X., Zhu, T., Luan, S., 2002. Plant Physiol. 129, 661–677. Cho, Y.H., Lee, S.Y., Kim, S.M., Yu, J.M., Lee, J.R., Hong, H.C., Kim, J.B., Ma, K.H., Kwon, T.R., Kang, H.K., Lee, G.A., Gwag, J.G., Kim, T.S., Park, Y.J., 2008. J. Crop. Sci. Biotechnol. 11, 181–186. Cloonan, N., Forrest, A., Kolle, G., Gardiner, B., Faulkner, G.J., Brown, M.K., Taylor, D.F., Steptoe, A.L., Wani, S., Bethel, G., Alan, J.R., Andrew, C.P., Stephen, J.B., Clarence, C.L., Swati, S.R., Heather, E.P., Jonathan, M.M., Kevin, J.M., Sean, M.G., 2008. Nat. Methods 5, 613–619. Colleoni, C., Dauville'e, D., Mouille, G., 1999. Plant Physiol. 120, 993–1003. Conesa, A., Götz, S., García-Gómez, J.M., Terol, J., Talón, M., Robles, M., 2005. Bioinformatics 21, 3674–3676. Dung, L.V., Mikami, I., Amano, E., Sano, Y., 2000. Breed. Sci. 50, 215–219. Eveland, A.L., Satoh-Nagasawa, N., Goldshmidt, A., Meyer, S., Beatty, M., Sakai, H., Ware, D., Jackson, D., 2010. Plant Physiol. 154, 1024–1039. Filichkin, S.A., Priest, H.D., Givan, S.A., Shen, R.K., Bryant, D.W., Fox, S.E., Wong, W.K., Mockler, T.C., 2010. Genome Res. 20, 45–58. Gao, Y., Xu, H., Shen, Y.Y., Wang, J.B., 2013. Plant Mol. Biol. 81, 363–378. Garber, M., Grabherr, M.G., Guttman, M., Trapnell, C., 2011. Nat Methods 8, 469–477. Greene, T.W., Hannah, L.C., 1998. Plant Cell 10, 1295–1306. Han, X.Z., Hamaker, B.R., 2001. J. Cereal Sci. 34, 279–284. Hannah, L.C., Shaw, J.R., Giroux, M.J., Reyss, A., Prioul, J.L., Bae, J.M., Lee, J.Y., 2001. Plant Physiol. 127, 173–183. Haugen, T.H., Ishaque, A., Preiss, J., 1976. J. Biol. Chem. 251, 7880–7885. Hirose, T., Ohdan, T., Nakamurac, Y., Teraoa, T., 2006. Physiol. Plant. 128, 425–435. Hoshikawa, K., 1968. Proc. Crop Sci. Soc. Jpn. 37, 207–216. Isshiki, M., Nakajima, M., Satoh, H., Shimamoto, K., 2000. Plant J. 23, 451–460. Isshiki, M., Matsuda, Y., Takasaki, A., Wong, H.L., Satoh, H., Shimamoto, K., 2008. Plant Biotechnol. 25, 483–487. James, M., Denyer, K., Myres, A., 2003. Curr. Opin. Plant Biol. 6, 215–222. Johnson, P., Patron, N., Bottrill, A., Dinges, J., Fahy, B., Parker, M., Waite, D., Denyer, K., 2003. Plant Physiol. 131, 684–696. Kang, H.G., Park, S.H., Matsuoka, M., An, G.H., 2005. Plant J. 42, 901–911. Kawasaki, T., Mizuno, K., Shimada, H., Satoh, H., Kishimoto, N., Okumura, S., Ichikawa, N., Baba, T., 1996. Plant Physiol. 110, 89–96. Koh, H.J., Heu, M.H., 1994. Korean J. Crop Sci. 39, 1–6. Lan, L.F., Chen, W., Lai, Y., Suo, J.F., Kong, Z.S., Li, C., Lu, Y., Zhang, Y.J., Zhao, X.Y., Zhang, X.S., Zhang, Y.S., Han, B., Cheng, J., Xue, Y.B., 2004. Plant Mol. Biol. 54, 471–487. Lee, S.K., Hwang, S.K., Han, M., Eom, J.S., Kang, H.G., Han, Y., Choi, S.B., Cho, M.H., Bhoo, S.H., An, G., Hahn, T.R., Okita, T.W., Jeon, J.S., 2007a. Plant Mol. Biol. 65, 531–546. Lee, D.J., Kim, S., Ha, Y.M., Kim, J., 2007b. Planta 227, 577–587. Li, P., Ponnala, L., Gandotra, N., Wang, L., Si, Y., Tausta, S.L., Kebrom, T.H., Provart, N., Patel, R., Myers, C.R., Reidel, E.J., Turgeon, R., Liu, P., Sun, Q., Nelson, T., Brutnell, T.P., 2010a. Nat. Genet. 42, 1060–1067. Li, B., Ruotti, V., Stewart, R.M., Thomson, J.A., Dewey, C.N., 2010b. Bioinformatics 26, 493–500. Li, F.P., Lee, Y.S., Kwon, S.W., Li, G., Park, Y.J., 2014 Apr 14. Genet. Mol. Res. 13(AOP). (Epub ahead of print). Liu, X.L., Guo, T.L., Wan, X.Y., Wang, H.Y., Zhu, M.Z., Li, A.L., Su, N., Shen, Y.Y., Mao, B.G., Zhai, H.Q., Mao, L., Wan, J.M., 2010. BMC Genomics 11 (730), 1–15. Lu, F.H., Park, Y.J., 2012a. J. Cereal Sci. 56, 1–7. Lu, F.H., Park, Y.J., 2012b. Genet. Res. 94, 179–189. Lu, T.T., Lu, G.J., Fan, D.L., Zhu, C.R., Li, W., Zhao, Q., Feng, Q., Zhao, Y., Guo, Y.L., Li, W.J., Huang, X.H., Han, B., 2010. Genome Res. 20, 1238–1249.

9

Lu, F.H., Yoon, M.Y., Cho, Y.I., Chung, J.W., Kim, K.T., Cho, M.C., Cheong, S.R., Park, Y.J., 2011. Sci. Hortic. 129 (1), 38–45. Luo, M., Taylor, J.M., Spriggs, A., Zhang, H., Wu, X., 2011. PLoS Genet. 7, e1002125. Ma, T.L., Wu, W.H., Wang, Y., 2012. BMC Plant Biol. 12, 161. Marioni, J.C., Mason, C.E., Mane, S.M., Stephens, M., Gilad, Y., 2008. Genome Res. 18, 1509–1517. Morell, M.K., Myers, A.M., 2005. Curr. Opin. Plant Biol. 8, 204–210. Mortazavi, A., Williams, B.A., McCue, K., Schaeffer, L., Wold, B., 2008. Nat. Methods 5, 621–628. Nagalakshmi, U., Wang, Z., Waern, K., Shou, C., Raha, D., Gerstein, M., Snyder, M., 2008. Science 320, 1344–1349. Nakamura, Y., Kubo, A., Shimamune, T., Matsuda, T., Harada, K., Satoh, H., 1997. Plant J. 12, 143–153. Nishi, A., Nakamura, Y., Tanaka, N., Satoh, H., 2001. Plant Physiol. 127, 459–472. Ohdan, T., Francisco, P.B., Sawada, T., Hirose, T., Terao, T., Satoh, H., Nakamura, Y., 2005. J. Exp. Bot. 56, 3229–3244. Okita, T.W., Nakata, P.A., Anderson, J.M., Sowokinos, J., Morell, M., Preiss, J., 1990. Plant Physiol. 93, 785–790. Omura, T., Satoh, H., 1984. Biology of Rice, pp. 293–313. Ozsolak, F., Ting, D.T., Wittner, B.S., Brannigan, B.W., Paul, S., Bardeesy, N., Ramaswamy, S., Milos, P., Haber, D.A., 2010. Nat. Methods 7, 619–621. Perez, C.M., Juliano, B.O., 1978. Starch-Starke 30, 424–426. Ryoo, N., Yu, C., Park, C.S., Baik, M.Y., Park, I., Cho, M.H., Bhoo, S., An, G., Hahn, T.R., Jeon, J.S., 2007. Plant Cell Rep. 26, 1083–1095. Sakai, H., Lee, S.S., Tanaka, T., Numa, H., Kim, J., Kawahara, Y., Wakimoto, H., Yang, C.C., Iwamoto, M., Abe, T., Yamada, Y., Muto, A., Inokuchi, H., Ikemura, T., Matsumoto, T., Sasaki, T., Itoh, T., 2013. Plant Cell Physiol. 54, e6. Sasaki, T., Burr, B., 2000. Curr. Opin. Plant Biol. 3, 138–142. Satheeshkumar, P., Saravanan, K., 2012. Plant Arch. 12, 639–644. Satoh, H., Omura, T., 1981. Jpn. J. Breed. 31, 316–326. Sekhon, R.S., Briskine, R., Hirsch, C.N., Myers, C.L., Springer, N.M., 2013. PLoS ONE 8, e61005. Severin, A., Woody, J., Bolon, Y.T., Joseph, B., Diers, B., Farmer, A., Muehlbauer, G., Nelson, R., Grant, D., Specht, J., Graham, M., Cannon, S., May, G., Vance, C., Shoemaker, R., 2010. BMC Plant Biol. 10, 160. Smith-White, B.J., Preiss, J., 1992. J. Mol. Evol. 34, 449–464. Sultan, M., Schulz, M.H., Richard, H., Magen, A., Klingenhoff, A., Scherf, M., Seifert, M., Borodina, T., Soldatov, A., Parkhomchuk, D., Schmidt, D., O'Keeffe, S., Haas, S., Vingron, M., Lehrach, H., Yaspo, M.-L., 2008. Science 321, 956–960. Tetlow, I.J., Wait, R., Lu, Z.X., Akkasaeng, R., Bowsher, C.G., Esposito, S., Kosar-Hashemi, B., Morell, M.K., Emes, M.J., 2004. Plant Cell 16, 694–708 (Online). Trapnell, C., Williams, B., Pertea, G., Mortazavi, A., Kwan, G., van Baren, M., Salzberg, S., Wold, B., Pachter, L., 2010. Nat. Biotechnol. 28, 511–515. Van Camp, W., 2005. Curr. Opin. Biotechnol. 16, 147–153. Villand, P., Olsen, O., Kleczkowski, L.A., 1993. Plant Mol. Biol. 23, 1279–1284. Wang, J., Wang, W., Li, R., Li, Y., Tian, G., Goodman, L., Fan, W., Zhang, J., Li, J., Zhang, J., Guo, Y., Feng, B., Li, H., Lu, Y., Fang, X., Liang, H., Du, Z., Li, D., Zhao, Y., Hu, Y., Yang, Z., Zheng, H., Hellmann, I., Inouye, M., Pool, J., Yi, X., Zhao, J., Duan, J., Zhou, Y., Qin, J., 2008. Nature 456, 60–65. Wang, Z., Gerstein, M., Snyder, M., 2009. Nat. Rev. Genet. 10, 57–63. Wilhelm, B.T., Marguerat, S., Watt, S., Schubert, F., Wood, V., Goodhead, I., Penkett, C.J., Rogers, J., Bahler, J., 2008. Nature 453, 1239–1243. Wong, K.S., Kubo, A., Jane, J.L., Harada, K., Satoh, H., Nakamura, Y., 2003. J. Cereal Sci. 37, 139–149. Xiang, D.Q., Venglat, P., Tibiche, C., Yang, H., Risseeuw, E., Cao, Y., Babic, V., Cloutier, M., Keller, W., Wang, E., Selvaraj, G., Datla, R., 2011. Plant Physiol. 156, 346–356. Xie, C., Mao, X.Z., Huang, J.J., Ding, Y., Wu, J.M., Dong, S., Kong, L., Gao, G., Li, C.Y., Wei, L.P., 2011. Nucleic Acids Res. 39, 316–322. Xu, H., Gao, Y., Wang, J.B., 2012. PLoS ONE 7, e30646. Yoon, M.R., Koh, H.J., Lee, S.C., Kang, M.Y., 2009. J. Korean Soc. Appl. Biol. Chem. 52, 509–515. Zhai, R., Feng, Y., Wang, H., Zhan, X., Shen, X., Wu, W., Zhang, Y., Chen, D., Dai, G., Yang, Z., Cao, L., Cheng, S., 2013. BMC Genomics 2164, 14–19. Zhang, X., Colleoni, C., Ratushna, V., Sirghie-Colleoni, M., James, M., Myers, A., 2004. Plant Mol. Biol. 54, 865–879. Zhang, G., Guo, G., Hu, X., Zhang, Y., Li, Q., Li, R., Zhuang, R., Lu, Z., He, Z., Fang, X., Chen, L., Tian, W., Tao, Y., Kristiansen, K., Zhang, X., Li, S., Yang, H., Wang, J., Wang, J., 2010. Genome Res. 20, 646–654. Zhao, W.G., Chung, J.W., Kwon, S.W., Lee, J.H., Ma, K.H., Park, Y.J., 2013. Euphytica 191, 9–21.

Please cite this article as: Li, F., et al., Transcriptome analysis of grain-filling caryopses reveals the potential formation mechanism of the rice sugary mutant, Gene (2014), http://dx.doi.org/10.1016/j.gene.2014.05.059