Rice EARLY SENESCENCE 2, encoding an inositol polyphosphate kinase, is involved in leaf senescence

Background Early leaf senescence influences yield and yield quality by affecting plant growth and development. A series of leaf senescence-associated molecular mechanisms have been reported in rice. However, the complex genetic regulatory networks that control leaf senescence need to be elucidated. Results In this study, an early senescence 2 (es2) mutant was obtained from ethyl methanesulfonate mutagenesis (EMS)-induced mutational library for the Japonica rice cultivar Wuyugeng 7 (WYG7). Leaves of es2 showed early senescence at the seedling stage and became severe at the tillering stage. The contents of reactive oxygen species (ROS) significantly increased, while chlorophyll content, photosynthetic rate, catalase (CAT) activity significantly decreased in the es2 mutant. Moreover, genes which related to senescence, ROS and chlorophyll degradation were up-regulated, while those associated with photosynthesis and chlorophyll synthesis were down-regulated in es2 mutant compared to WYG7. The ES2 gene, which encodes an inositol polyphosphate kinase (OsIPK2), was fine mapped to a 116.73-kb region on chromosome 2. DNA sequencing of ES2 in the mutant revealed a missense mutation, ES2 was localized to nucleus and plasma membrane of cells, and expressed in various tissues of rice. Complementation test and overexpression experiment confirmed that ES2 completely restored the normal phenotype, with chlorophyll contents and photosynthetic rate increased comparable with the wild type. These results reveal the new role of OsIPK2 in regulating leaf senescence in rice and therefore will provide additional genetic evidence on the molecular mechanisms controlling early leaf senescence. Conclusions The ES2 gene, encoding an inositol polyphosphate kinase localized in the nucleus and plasma membrane of cells, is essential for leaf senescence in rice. Further study of ES2 will facilitate the dissection of the genetic mechanisms underlying early leaf senescence and plant growth.


Background
Leaf senescence is an internally programmed degeneration phase in the life span of leaf. During growth and development, plants have both stress-induced and agerelated senescence. From seedling to maturity, leaves are the main source of photosynthesis. At maturity stage, leaf cells undergo drastic physiological and metabolic changes in an orderly manner, such as chlorophyll degradation, oxidation and hydrolysis of proteins, lipids and nucleic acids [1]. These hydrolyzed metabolites remobilize to new bud, stem or root in perennial plants or to seed in annual plants [2]. In case of early leaf senescence, chloroplasts are the first organelles to be degraded. As a consequence, chlorophyll pigments are disrupted, leaf color changes gradually from green to yellow, brown and finally wither [3]. Simultaneously, decreased photosynthesis and lower assimilates accumulation in reproductive organs are regulated by genetic factors and often triggered by environmental stress which leads to yield losses [4][5][6]. In some hybrid rice varieties, early senescence leads to impaired leaf function and reduced accumulation of photosynthetic products, which ultimately impact rice yield [7,8]. As leaves with delayed senescence stayed photosynthetically active and the flowering period extended, plants were shown to set more seeds and accumulate more biomass than the wild type [9]. Therefore, it is important to understand the mechanism underlying early leaf senescence for rice breeding.
The most common physiological alteration during early senescence includes chlorophyll degradation, reactive oxygen species (ROS) scavenging, carbon and nitrogen imbalances, and hormone responses as a coordinated action at the cellular, tissue, organ and organism levels [10]. In the last few decades, series of genes associated with leaf senescence have been isolated and characterized in rice, including transcription factors, receptors and signaling components of hormones or stress responses, and metabolic regulators [4,[11][12][13][14][15][16][17][18]. According to the metabolic pathway involved, senescence genes can be classified into five categories. The first type are also involved in chlorophyll degradation, such as the mutation of RLS1 and Osh69 causes, chlorophyll degradation and accelerate leaf senescence [19,20]. In addition, functional and non-functional stagnant green type gene that hinders the degradation of chlorophyll in late growth, including NON-YELLOW COLOR-ING 1 (NYC1), NON-YELLOW COLORING 3 (NYC3), NYC1-LIKE (NOL) and STAY GREEN RICE (SGR) [21][22][23][24]. The second type of senescence-related genes are associated with phytohormones and transcription factors that regulate leaf senescence. For example, abscisic acid (ABA) induces OsNAP to regulate chlorophyll degradation, which affects nutrient transport and expression of senescence-related genes [25]. OsFBK12 was reported to be involved in ethylene (ETH) signaling pathways promoting leaf senescence [26]. ZOG1 gene encodes zeatin glucosyltransferase, a Cytokinin (CTK) synthesis-related gene, and mediates senescence of flag leaf in rice [27]. In Arabidopsis, SAUR is an Indole-3-acetic acid (IAA) response factor. In order to study the cellular function of plant SAUR, knock-out gene was utilized to generate mutants with non-expression in Arabidopsis, and resulted to delayed leaf senescence phenotype [28]. The third type of genes related to senescence, which are involve in energy metabolism pathway, including OsNaPRT1/LTS1 gene. Mutation of LTS1 led to reduction of NAD content, inhibiting the deacetylation ability of OsSRTs. This activates senescence-related genes by increasing the acetylation of H3K9, which ultimately lead to senescence of rice leaves [12]. The fourth type are nitrogen remobilization related, such as Osl2 and OsFd-GOGAT [29,30]. Other genes, such as SPOTTED LEAF 29 (SPL29), OsSWEET5, SENESCENCE 1 (ES1/ TUTOU1), OsGDCH and DWARF AND EARLY-SENE SCENCE 1 (DEL1) [3,[31][32][33][34]. Most of early senescence mutants exhibited defects in plant growth and development, such as dwarfism or semi-dwarf, withered leaf tip and early heading date [3,12,17,35]. Therefore, identification of premature senescence genes is of great significance to explore the mechanism for early senescence and improve rice yield.
Inositol polyphosphates are important class of organic phosphorus compounds and signaling molecules, which are widely distributed in diverse organisms [36,37]. Inositol polyphosphate kinase (IPMK/IPK2) is a key enzyme in inositol phosphate metabolism, which converts inositol 1,4,5-trisphosphate (IP3) to inositol tetrakisphosphate (IP4) and inositol pentakisphosphate (IP5) [38][39][40]. In eukaryotic cells, IPK2-mediated production of IP4 and IP5 for cellular growth, yeast strain bearing deletion of ScIPK2 displays a temperature-sensitive growth defect [41]. ScIPK2 is also an indispensable component of the ArgR-Mcm1 transcriptional complex, which regulates arginine metabolism [42]. In eukaryotes, they are involved in multiple biological functions, such as programmed cell death, hormone signal transduction, ion channels regulation and sensitivity to oxidative stress [37,[43][44][45][46]. For instance, the auxin biosynthesis, transport and mediation of axillary shoot branching were also regulated by inositol polyphosphate kinase gene (AtIPK2α and AtIpk2β) [47]. The overexpression of AtIPK2α enhanced root growth through the regulation of inositol trisphosphate (IP3)-mediated calcium accumulation, and overexpression of AtIpk2β led to more axillary shoot branches in Arabidopsis [47,48]. AtIPK2β is participated in the synthesis of myo-inositol 1,2,3,4,5,6hexakisphosphate (IP6, phytate). AtIPK2b interacts with sucrose non-fermenting-1-related protein kinase (SnRK1.1), which is involved in glucose suppression of seed germination, vegetative growth, flowering and senescence [49][50][51][52][53]. Inositol polyphosphate kinase gene (OsIPK2) has been previously isolated and identified as a candidate phytic acid biosynthetic gene in rice. Upregulation of OsIPK2 in anthers suggests that a phospholipase C (PLC)-mediated pathway is active in addition to a lipid-independent pathway in anthers [54]. The inositol polyphosphate multikinase (IPMK) acts as a transcriptional activator that binds to the tumor suppressor P53 in mammals, thereby promoting p53-mediated cell death [55]. Although the function of IPK2 has been reported in mammals and plants, no genetic study has been published on its role in early leaf senescence of rice.
To study molecular mechanism underlying leaf senescence in rice, a new leaf senescence mutant, early senescence 2 (es2), was isolated from ethyl methanesulfonate (EMS)-treated Oryza sativa japonica cv. Wuyugeng 7 (WYG7). From the seedling stage, yellow spots appeared in leaves and gradually withered until the tillering stage. The es2 mutant showed lower chlorophyll content, photosynthetic rate and higher ROS compared to WYG7. Map-based cloning and sequence analysis of the ES2 gene revealed that it encodes inositol polyphosphate kinase (OsIPK2). Complementation test and overexpression analysis demonstrated that ES2/OsIPK2 mutations causes leaf early senescence, and influences yield-related traits in rice.

Results
Early leaf senescence and decreased agronomic traits in the es2 mutant We identified a mutant, es2, with an early leaf senescence phenotype from the EMS mutagenesis mutant population of Japonica variety, WYG7. The es2 mutant showed leaf senescence from the seedling stage until the maturity stage (45 days after pollination). At the seedling stage, yellow spots distributed throughout the leaves, and the aging leaves gradually withered ( Fig. 1a-f). In order to analyze the leaf senescence phenotype more precisely, the detached leaves of WYG7 and the es2 mutant at 2-leaf stage were kept in the dark at 28°C. Compared to WYG7, the leaves of es2 almost turned more yellower after 5 days in darkness. This indicates that mutation in ES2 accelerates darkinduced leaf senescence (Additional file 5: Figure S1).
In addition, the es2 mutant showed significant decrease in some agronomic traits including plant height, internode length, panicle length, number of primary branches, number of secondary branches, grain width and 1000-grain weight compared to wild-type plants. Moreover, the number of tillers increased significantly compared to the wild-type plants (Additional file 6: Figure S2). These results indicates that the early senescent leaves in es2 negatively affects yield.
Decreased chlorophyll content and photosynthetic ability with abnormal chloroplast ultrastructure in the es2 mutant Compared to WYG7, chlorophyll a, b, and Carotenoid (Car) contents significantly decreased in the es2 mutant at seedling, tillering and heading stages (Fig. 1g-i). Photosynthetic parameters were examined at the tillering stage in WYG7 and the es2 mutant. Net photosynthetic rate (P n ), transpiration rate (T r ) and stomatal conductance (G s ) were significantly declined respectively in the es2 mutant compared to the wild type ( Fig. 1j-l). Transmission electron microscopy (TEM) analysis showed that the number of chloroplasts dramatically reduced in es2 mutant compared with in WYG7. The thylakoids and stroma lamellae structure in leaves were disorderly arranged in the es2 mutant. Simultaneously, more osmiophilic granules (OG) were found in es2 compared to WYG7. Therefore, mutation in ES2 led to an abnormal chloroplast development ( Fig. 2a-d).
The es2 mutant exhibited cell death and more ROS accumulation in leaves To determine whether early senescence in es2 resulted from concomitant accumulation of superoxide radicals, nitroblue tetrazolium (NBT) staining was performed and more blue formazan precipitates were found in es2 leaves than in WYG7 at the tillering stage (Fig. 2e). We also examined hydrogen peroxide (H 2 O 2 ) levels by 3, 3diaminobenzidine (DAB) staining in both WYG7 and es2. The formation of reddish-brown formazan precipitates in the es2 leaves at the tillering stage indicated the accumulation of H 2 O 2 (Fig. 2f). Furthermore, the contents of H 2 O 2 and malondialdehyde (MDA), activities of catalase (CAT), peroxidase (POD), superoxide dismutase (SOD) and ascorbate peroxidase (APX) were measured at the tillering stage in WYG7 and es2, respectively. The results showed that H 2 O 2 and MDA contents were significantly higher in es2 than in WYG7 (Fig. 2g, i). The activities of POD, SOD and APX increased, while CAT activity decreased in the es2 mutant. Therefore, the es2 mutant exhibited more ROS accumulation than the wild type (Fig. 2h, j-l). In order to examine cell death in es2, leaves of the es2 mutant and WYG7 at the tillering stage were tested for TUNEL assay. Stronger green fluorescence was observed in es2 mesophyll cells than that of WYG7, indicating that a large number of DNA fragments accumulated and cell apoptosis occurred in es2 ( Fig. 3a-d).

Genetic analysis and fine mapping of the ES2 gene
For genetic analysis of the es2 mutant, we crossed the mutant with japonica variety WYG7. All F 1 plants showed the wild-type phenotype, suggesting that the mutation is recessive. Among 784 randomly selected F 2 plants, the number of wild-type plants and the mutant plants were 604 and 180, respectively, consistent with the ratio of 3:1 (χ 2 = 1.7361). Therefore, the mutant was controlled by a recessive nuclear gene, designated as ES2 gene (Additional file 1: Table S1).
In order to fine map the ES2 gene, an F 2 segregating population was derived from the crossing between the es2 mutant and indica variety 93-11. Chromosomal linkage analysis was performed using 94 mutant plants with 225 polymorphic SSR markers distributed on 12 chromosomes. Three SSR/STS markers on chromosome 2, B2-9, B2-10 and B2-11 were linked to ES2 (Fig. 5a). Total of 94 mutant plants were tested using these 3 markers and the gene was initially mapped between B2-10 and B2-11. To fine map ES2, the mapping population was expanded to 521 F 2 mutant individuals. Based on the comparison between genomic sequences of Nipponbare and 93-11, 7 polymorphic Indel markers were developed between B2-10 and B2-11. Using these markers, the ES2 gene was fine mapped between ID2-3 and ID2-4 within a 116.73 kb region (Fig.  5b). According to Nipponbare genomic sequence in the region (https://rapdb.dna.affrc.go.jp/), there were 18 annotated ORFs, among which a G to A substitution in exon of the gene LOC_Os02g32370 in es2 was detected, which caused replacement of glycine with glutamic acid (Fig. 5c-e). Ultrastructure of the chloroplast, DAB and NBT staining, and senescence related-indices determination in WYG7 and es2 at the tillering stage. a, b Ultrastructure of the chloroplast at the tillering stage of WYG7. a Scale bar = 1 μm, (b) Scale bar = 0.5 μm. c, d Ultrastructure of the chloroplast at the tillering stage of es2. c Scale bar = 1 μm, (d) Scale bar = 0.5 μm. CP, chloroplast; G, granum; OG, osmiophilic granule. e NBT straining. Scale bar = 2 cm. f DAB straining. Scale bar = 2 cm. g H 2 O 2 content. h Catalase (CAT) activity. i malondialdehyde (MDA) content. j Superoxide dismutase (SOD) activity. k Peroxidase (POD) activity. l ascorbate peroxidase (APX) activity. Mean ± SD, n = 3. * significance at P < 5%, ** extremely significance at P < 1% (Student's t-test) Sequence and phylogenetic analysis of the OsIPK2 gene The ES2 gene encodes an inositol polyphosphate kinase (OsIPK2). Protein sequence alignment showed that Gly 42 in OsES2 was highly conserved in plants (Fig. 6a). Phylogenetic tree constructed with homolog sequences of the OsES2 protein from different species showed that they can be divided into two groups: monocots and dicots. Among these species, rice ES2 exhibited the highest similarity (76.57%) to the ortholog in Brachypodium distachyon (Fig. 6b).

OsIPK2/ES2 gene was responsible for leaf early senescence in rice
In order to confirm OsIPK2 the ES2 gene, complementation test was conducted and all of the 13 transgenic plants restored the wild-type phenotype (Fig. 7a, b). As expected, gene expression level, chlorophyll contents and photosynthetic parameters were restored to those of the wild type (Fig. 7c, e; Additional file 7: Figure S3). Furthermore, overexpression experiment showed all the 9 transgenic plants restored the wild-type phenotype (Fig. 7a, b). With dramatic increase in expression levels of OsIPK2, chlorophyll contents and photosynthetic parameters were also elevated (Fig. 7d, e, Additional file 7: Figure S3).

ES2 was localized to nucleus and plasma membrane and expressed in various tissues of rice
To determine subcellular localization of the ES2 protein, the p35S::ES2::GFP vector was transferred into rice protoplasts and tobacco (N. benthamiana), respectively, with the p35S:: GFP empty vector as control (Additional file 8: Figure S4a-c, g-i). The p35S::ES2::GFP vector was transferred into rice protoplasts, and fluorescence confocal observation revealed that ES2 was localized to nucleus and plasma membrane (Additional file 8: Figure S4d-f). Consistently, fluorescent signals were also observed in nucleus and plasma membrane in epidermal leaf cells of tobacco (N. benthamiana) transformed with the p35S::ES2::GFP vector (Additional file 8: Figure S4jl). Besides, the p35S::ES2::YFP vector was transferred into rice protoplasts, and fluorescence confocal observation revealed that ES2 was also localized to nucleus and plasma membrane (Additional file 9: Figure S5).
The β-glucuronidase (GUS) staining observations detected ES2 expressed in various tissues, including roots, stems, leaves, sheaths and panicles with strongest in leaves (Additional file 10: Figure S6a-e). Consistently, qRT-PCR results showed that ES2 were expressed in roots, stems, leaves, sheaths and panicles, with highest expression level in leaves followed by stems, panicles, roots and sheaths (Additional file 10: Figure S6f).

Discussion
Leaf senescence is an essential developmental stage of programmed procedure, and accompanied by the emergence of characteristics in plants, such as changes in leaf color, degradation of chloroplast, and reduction in chlorophyll content and photosynthetic efficiency.
Eventually, early leaf senescence may cause plant growth and development retardation, and yield reduction [72]. Recently, a series of leaf senescence-associated genes have been cloned and characterized in rice [3,12,17,33,35,73]. The lower leaves of premature leaf senescence 3 mutant (pls3) firstly turned yellow at tillering stage, and Relative expression levels of senescence-related genes. b Relative expression levels of reactive oxygen scavenging system (ROS) related genes. c Relative expression levels of chlorophyll synthesis and degradation related genes. d Relative expression levels of photosynthesis and chloroplast development related genes. A Histone gene was used as the reference. Mean ± SD, n = 3. * significance at P < 5%, ** extremely significance at P < 1% (Student's t-test) senescence becomes severe at the mature stage. The pls3 mutant also showed decreased chlorophyll and melatonin contents, shorter plant height, and 1000-grain weight [17]. The tips and margins of the lower leaves of es4 mutant became yellow at tillering stage, and senesce at the grain-filling stage. The es4 mutant also showed reduced chlorophyll content and photosynthetic rate, shorter plant height, and 1000-grain weight [73]. In this study, the es2 mutant also showed rapid leaf senescence, different from previously reported mutants. Firstly, the early senescence of plants is accompanied by changes in chlorophyll content. The leaves color changed from green to yellow spot at seedling stage, and alteration of withered at tillering stage until maturity (Fig. 1a-f), while the degradation of chlorophyll content and lower photosynthetic rate (Fig. 1g-l). Secondly, the plant height, internode length, panicle length and thousand-grain weight were significantly reduced in es2 mutant (Additional file 6: Figure S2). These results showed that the senescence process in the es2 mutant began earlier than in previously reported mutants, therefore, es2 is a novel mutant to dissect the mechanism of leaf senescence.
In higher plants, there were some reports about inositol polyphosphate kinase involved in multiple biological functions. There are two IPK2 genes in Arabidopsis, AtIPK2α and AtIpk2β, and only one in rice [54]. In Arabidopsis, IPK2 plays important role in axillary shoot branching, axillary shoot branching, root growth, Fig. 5 Map-based cloning of ES2 and identification of mutated site in es2. a ES2 was preliminarily mapped between markers B2-10 and B2-11 on chromosome 2. b Fine mapping of ES2. The ES2 locus was fine mapped to a 116.7 kb region between markers ID2-3 and ID2-4. c Eighteen putative ORFs are located in the 116.7 kb region. d Gene structure of the candidate gene LOC_Os02g32370. Scale bar =200 bp. The black rectangle represents the exon. The point mutation of G to A on the exon led to amino acid replacement of Gly with Glu. e Comparison of sequence chromatograms around the mutated site between the wild-type WYG7 and es2 synthesis of phytic acid, abiotic stress responses, auxin response, seed germination, vegetative growth, flowering and senescence [47-53, 74, 75]. OsIPK2 has been previously isolated and identified as a candidate biosynthetic gene for phytic acid in rice [54]. However, their physiological functions have not been reported. Recently, it was reported that inositol polyphosphate kinase directly interacts with OsIAA11 to regulate lateral root formation and was involved in gibberellic acid signaling modulation to affect shoot elongation and fertility [76,77]. In this study, different from previously reported genes, the early senescence phenotype was caused by a mutation in the ES2/OsIPK2 gene for inositol polyphosphate kinase. ES2 expression was detected in all tissues surveyed, but predominantly in leaf mesophyll cells. The qRT-PCR results showed that the expression level of OsIPK2 in es2 was significantly up-regulated at seedling stage, while down-regulated at tillering and heading stage compared with WYG7 (Additional file 11: Figure S7). Additionally, the mutated ES2 gene was transformed into the es2 mutant plants and could not restore the wild-type phenotype, although expression level of the gene was significantly increased compared with es2 mutants (Additional file 12: Figure S8). Considering the occurrence of leaf senescence in the es2 mutant from seedling stage until maturity, phenotypic change caused by a single nucleotide mutation in ES2/OsIPK2 might not be correlated with gene expression level. In order to find whether the mutation impairs kinase activity towards the phytic acid production level, we measured phytic acid content in the wild-type WYG7 and es2 mutants. The contents of phytic acid were significantly reduced in es2, indicating that the mutation in IPK2 hampers phytic acid biosynthesis (Additional file 13: Figure S9).
Previous studies showed that early leaf senescence was associated with the accumulation of ROS [17,32,78], especially H 2 O 2 accumulation [79][80][81]. ROS accumulation led to oxidative damage in the thylakoid membranes and other cellular components [82]. Here, NBT and DAB staining showed that O 2 − and H 2 O 2 accumulated in the es2 mutant compared with the wild-type WYG7 (Fig. 2e,  f). We speculate that reduction in chlorophyll content and abnormality in chloroplast ultrastructure in es2 are due to oxidative damage caused by ROS. The qRT-PCR analysis showed that expression levels of senescenceassociated genes were up-regulated in the es2 mutant (Fig. 4a). TUNEL assay also showed that a large number of DNA fragments appeared in cells of es2 (Fig. 3). Furthermore, the ES2 protein localized to the nucleus and plasma membrane (Fig. S3), wherein large amounts of ROS were produced. These results suggests that leaf senescence may lead to by ROS accumulation.
During early senescence, plant leaves undergo a series of physiological changes, such as alteration in contents of H 2 O 2 and MDA, activities of CAT, SOD, POD and APX, and cell death [73,[83][84][85]. In early senescence, SOD and CAT can remove ROS [86,87]. SOD catalyzes O 2− dismutase to produce H 2 O 2. CAT is the major H 2 O 2 -scavenging enzyme. APX also plays an important role in control of H 2 O 2 level in cells [73,85,88]. In our study, activities of SOD, POD and APX in es2 were significantly higher than those in WYG7, while activity of CAT was significantly lower than that of WYG7 (Fig. 2h,  j-l). This is consistent with previous reports concerning early senescence mutants in rice [17,73,[89][90][91]. Therefore, it is indicates that the increased SOD activity in es2 was due to high O 2 − production and reducedCAT activity accelerated accumulation of H 2 O 2 . The increased MDA content in leaves provided further evidence of lipid peroxidation and ROS accumulation in es2 (Fig. 2i). Leaf senescence is mediated by a large number of ROS related genes, such as AOX1a, AOX1b, APX1, APX2, SODB, SODA1, CATA and CATB. The qRT-PCR results showed that their expression levels were significantly up-regulated in es2 compared to WYG7 (Fig. 4b).

Conclusions
In conclusion, the mutation of ES2/OsIPK2 gene resulted to increased H 2 O 2 and MDA contentand CAT, SOD, POD and APX activity, and reduced and chlorophyll and P n content, which eventually leads to leaf senescence and reduced rice yield.

Plant materials, growth conditions and dark treatment
The japonica variety Wuyugeng 7 (WYG7) and indica variety 93-11, were conventionally cultivated in China. These seeds were provided by the China National Rice Research Institute (CNRRI) in Hangzhou, Zhejiang Province, China. Early senescence leaf mutant es2 was obtained by EMS mutagenesis of WYG7. The mutant was crossed with indica variety 93-11, which was used to construct F 2 mapping population. The es2 mutant and progenies showed stable inheritance and exhibited premature senescence. All rice plants were grown in the fields of CNRRI in Hangzhou, Zhejiang Province, China. For the dark treatment, the detached leaves from 2-leaf stage plants (before yellow spots appearing) grown in an incubator were incubated on ddH 2 O at 28°C in complete darkness for 5 days.

Detection of chlorophyll content
The fresh leaves from wild-type plants and the es2 mutant were sampled in different growth periods in field conditions. The fresh leaves were cut into small pieces of about 0.5 cm length and 0.05 g weight after removing the main leaf veins, then placed into 5.0 ml 80% acetone and soaked for 24 h in the dark, with shaking every 8 to 10 h until the photosynthetic pigment was fully dissolved. The 1 ml sample solution was measured at 470 nm, 645 nm and 663 nm using visible spectrophotometer (BACKMAN COULTER DU800, United States). The experiment was performed with three biological replicates per group, and t-test was conducted in statistical analysis. The formulas used to calculate chlorophyll a (Chl a), chlorophyll b (Chl b) and carotenoid (Car) contents are listed as following [17]: Photosynthetic rate measurement In case of sunshine day at 9:00, the net photosynthetic rate (P n ), transpiration rate (T r ), and stomatal conductance (G s ) of fully expanded leaves from WYG7 and the es2 mutant were measured by portable photosynthesis measurement device LI-6400 (Li-Cor, Lincoln, NE, United States) with 28°C, 1200 μmol photons m − 2 s − 1 irradiance and 400 μmol mol − 1 CO 2 concentration under field conditions at tillering stage. Three biological replicates were used and t-test was conducted for statistical analysis.

Transmission electron microscopy analysis of chloroplast ultrastructure
Fresh leaves of the wild-type WYG7 and the es2 plants at the seedling stage and the tillering stage in field were sampled, and the main vein was removed. A small section cut into about 0.5-1 cm segments and immediately placed in 2 ml containing 2.5% glutaraldehyde fixative, air was removed by a vacuum for about 2 h until completely sinked to the bottom of the tube and kept at 4°C for more than 16 h. The segments were washed 3 times in the phosphate buffer (0.1 M, PH 7.0) for 15 min at each step, and treated with 1% (w/v) OsO 4 in phosphate buffer for 1-2 h. After washing 3 times in the phosphate buffer for 15 min at each step, the samples were dehydrated using a gradient of ethanol solutions from 30 to 100% for 15 min at each step. The samples were placed in 1:1 mixture of alcohol and 90% acetone for 20 min at room temperature. Next, the samples were transferred into 90% acetone for 20 min and then into 100% acetone for dehydration treatment 3 times for 15 min at each step. After dehydration treatment, the samples were transferred into a final Spurr resin mixture overnight. The specimens were then placed in capsules with embedding medium and heated at 70°C for 9 h. The specimen sections were stained using uranyl acetate and alkaline lead citrate for 15 min each, then observed by electron microscopy (Hitachi H − 7650 , Tokyo, Japan).

Determination of senescence-related physiological index and phytic acid
The contents of hydrogen peroxide (H 2 O 2 ) and malondialdehyde (MDA), activities of catalase (CAT), superoxide dismutase (SOD), peroxidase (POD) and ascorbate peroxidase (APX) were measured using an Assay Kit (Suzhou Keming Biotechnology Co, Ltd. In). The leaves of WYG7 and es2 were sampled from plants grown in the paddy field at tillering stage. The contents of phytic acid were measured using an Assay Kit (Suzhou Keming Biotechnology Co, Ltd. In). The leaves of WYG7 and es2 were sampled from plants grown in the paddy field at seedling, tillering and heading stage. Three biological replicates were used and t-test was conducted.

Detection of apoptosis by TUNEL assay
The leaves were fixed with FAA fixative and embedded in paraffin at the tillering stage. The sections were microscopically examined to select a suitable slide and dewaxed by gradient alcohol. TUNEL staining was performed according to Huang et al. (2001) [94], and apoptosis was detected by DeadEnd™ Fluorometric TUNEL system kit (Promega, Wisconsin, USA). The localized green fluorescence (520 nm) of apoptotic cells (fluorescein-12-dUTP) in a red (620 nm) background (Propidium Iodide, PI) was detected by laser-scanning confocal microscope (Zeiss LSM700, Carl Zeiss, Inc., Thornwood, NY, USA).

Genetic analysis and fine mapping
A reciprocal cross between es2 and the japonica WYG7 was performed for genetic analysis. The F 2 segregating populations were used for the χ 2 test. The genetic analysis information used was listed in Additional file 1: Table S1. For fine mapping, F 2 segregating populations were derived from the cross between es2 mutant and the indica 93-11. F 2 plants were grown in the paddy fields in Hangzhou, which were used for segregation analysis. 94 F 2 individual plants phenotyped as mutants were used for the initial localization of es2. 521 F 2 mutant individual plants were used for fine mapping. Total DNA samples were extracted from the leaves using the cetyltriethyl ammonium bromide (CTAB) method. The initial mapping was conducted by 225 SSR and Indel markers, which distributed across 12 chromosomes in rice (http:// www.Gramene.org). For fine mapping of the ES2 gene, the new Indel markers were designed by Primer Premier 5.0 after comparing the sequence between Nipponbare and 93-11 in Gramene website. The marker information used were listed in Additional file 2: Table S2.

Sequence and phylogenetic analysis
The ES2 protein sequence composed of 295 amino acids was obtained from Gramene website (http://www.gramene.org/). A homologous analysis of ES2 was performed using the BLASTP program on National Center for Biotechnology Information website (NCBI, https:// www.ncbi.nlm.nih.gov/). DNAMAN was used to aligned sequences. MEGA 7.0 software with the bootstrap method and 1000 replicates was used to construct a phylogenetic tree of ES2 and homologous proteins.

Plasmid construction and plant transformation
To verify whether LOC_Os02g32370 was the ES2 gene, a 6030 bp genome sequence that includes the ES2 coding region along with the upstream sequence and downstream sequences was amplified from wild-type WYG7 by PCR using the ES2-COM primer, and inserted into pCAMBIA1300 vector. In order to construct an overexpression vector, 888 bp CDS was inserted into pCA-MIA1300s vector. All vectors were transformed into the es2 mutants by Agrobacterium (EHA105)-mediated transformation. Sequences of primers used were listed in Additional file 3: Table S3.

Subcellular localization and GUS assay
Primers with Sal I linker were designed to amplify the CDS sequence of WYG7, which was ligated to GFP vector pCAMBIA1301-35S-S65T-GFP by recombinant method, transformed into rice protoplasts [95], and transformed into tobacco (N. benthamiana) leaf epidermal cells with the protocol described by Ruan et al. (2019) [96]. Besides, YFP vector pCAMBIA1300-35S-YFP by recombinant method was also transformed into rice protoplasts. The transient expression of es2 was analyzed. GFP and YFP signals were observed by laserscanning confocal microscopy, respectively (Zeiss LSM700, Carl Zeiss, Inc., Thornwood, NY, USA).
To verify the tissue-specific expression of ES2, the promoter of ES2 (2103 bp upstream of ATG) was amplified from WYG7 genomic DNA and inserted into the binary vector pCAMBIA1305 with the GUS reporter gene. The recombinant vector is then introduced into the callus of WYG7 to produce a transgenic plant. GUS assays were performed on different tissues of transgenic plants, including roots, stems, leaves, sheaths, and panicle.

RNA extraction and quantitative real-time PCR (qRT-PCR) analysis
Total RNA samples were extracted from roots, stems, leaves, sheaths and panicles of wild-type WYG7 and es2 plants using AxyPrep™ total RNA Miniprep Kit (Axygen) at the tillering stage. Then the total RNA was reverse transcribed into cDNA using the ReverTra Ace® quantitative PCR RT Master Mix gDNA remover Kit (Toyobo Co. Ltd.; http://www.toyobo.cn/). qRT-PCR analysis was conducted with CFX96 Touch™ Real-Time PCR (Bio-Rad). All target genes were tested for expression levels using the rice reference gene Histone (LOC_Os06g04030) as a standard. Three biological replicates were performed in this experiment and t-test was used for statistical analysis. The primers used for qRT-PCR are listed in Additional file 4: Table S4.