Association of the molecular regulation of ear leaf senescence/stress response and photosynthesis/metabolism with heterosis at the reproductive stage in maize

Abstract

Maize exhibits a wide range of heterotic traits, but the molecular basis of heterosis at the reproductive stage has seldom been exploited. Leaf senescence is a degenerative process which affects crop yield and quality. In this study, we observed significantly delayed ear leaf senescence in the reciprocal hybrids of B73/Mo17 and Zheng58/Chang7-2 after silking, and all the hybrids displayed larger leaf areas and higher stems with higher yields. Our time-course transcriptome analysis identified 2,826 differentially expressed genes (DEGs) between two parental lines (PP-DEGs) and 2,328 DEGs between parental lines and the hybrid (PH-DEGs) after silking. Notably, several senescence promoting genes (ZmNYE1, ZmORE1, ZmWRKY53 and ZmPIFs) exhibited underdominant expression patterns in the hybrid, whereas putative photosynthesis and carbon-fixation (ZmPEPC)-associated, starch biosynthetic (ZmAPS1, ZmAPL), gibberellin biosynthetic genes (ZmGA20OX, ZmGA3OX) expressed overdominantly. We also identified 86 transcription factors from PH-DEGs, some of which were known to regulate senescence, stress and metabolic processes. Collectively, we demonstrate a molecular association of the regulations of both ear leaf senescence/stress response and photosynthesis/metabolism with heterosis at the late developmental stage. This finding not only extends our understanding to the molecular basis of maize heterosis but also provides basic information for molecular breeding.

Association of the molecular regulation of ear leaf senescence/stress response and photosynthesis/metabolism with heterosis at the reproductive stage in maize

R

A

P

Yi Song,*, Zhe Zhang,,*, XianjieTan, Yufeng Jiang, Jiong Gao, Li Lin, ZhenhuaWang, Jun Ren, XiaoleiWang, Lanqiu Qin, Weidong Cheng, Ji Qi, & Benke Kuai,

Maize exhibits a wide range of heterotic traits, but the molecular basis of heterosis at the reproductive and ZmPIFs ZmPEPC , senescence, stress and metabolic processes. Collectively, we demonstrate a molecular associationof the regulations of both ear leaf senescence/stress response and photosynthesis/metabolism with molecular basis of maize heterosis but also provides basic information for molecular breeding.

Heterosis (hybrid vigor) refers to the phenomenon that the hybrid progeny is superior to both parents in many aspects1, including biomass, photosynthetic capacity, biotic/abiotic stress tolerance and grain yield2. Heterosis is widely exploited in the breeding of major crops, such as maize, rice, sorghum and cotton. It was estimated that the commercial grain yield was increased by about six-fold during the maize hybrid era3.

Heterosis has been intensively investigated in both crops and model plants. There are three classical theories aiming to explain the genetic basis of heterosis: 1) the dominance theory which assumes that the complementation of deleterious alleles contributes to heterosis4,5, 2) the overdominance theory which postulates that the favorable allelic interaction of heterozygous alleles leads to a phenotypically superior performance of F1 hybrid6,

3) the epistasis theory assumes that epistasis between dierent loci contributes to heterosis7. Notably, the heterotic performance seems highly variable depending on the combination of parental lines and trait(s) examined8. To eectively exploit heterosis in breeding, maize elite inbred lines were empirically classied into heterotic groups based on their capability in enhancing their hybrids grain yield. Analysis of the restriction fragment length

Ministry of Education Key Laboratory for Biodiversity Science and Ecological Engineering, Institute of Biodiversity Maize Research Institute, Guangxi Academy of Agricultural *

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

polymorphism (RFLP) revealed a correlation between the hybrids grain yield and its parents genetic diversity9.

It was reported that there were diverse polymorphism in SSR (simple sequence repeats) loci between Zheng58 and Chang7-2, which belong to dierent heterotic groups10. Besides, there is also a correlation between genetic diversity and their transcriptional variation8.

Recently, both molecular and physiological evidence suggested that photosynthesis and carbohydrate metabolism were correlated with heterosis. A physiological study in Arabidopsis showed that, while the rate of photosyn-thesis was constant per unit leaf area in both parents and hybrids, hybrids exhibited higher total photosynthesis capacity because of more chloroplasts per cell, larger cell size and leaf area11. Furthermore, the heterosis could be eliminated by an inhibitor of photosynthesis, and promoted by enhanced photosynthetic activity due to higher light intensity11. Chlorophyll (Chl) content represents a physiological marker of photosynthetic capability. Both hybrids and allopolyploids in Arabidopsis exhibited higher expressions of Chl biosynthesis- and starch metabolism-associated genes, which produced more Chl and starch than their parents in the same environment12.

It was reported that dierentially expressed genes (DEGs) between a super-hybrid rice and its parental lines were signicantly enriched in photosynthesis and carbon xation-associated functions13. Recent transcriptome studies revealed that the transcriptomic prole as well as phenotypic features in the hybrid rice was biased to those of the parent with better performance, and the epigenetic modication also contributed to the regulation of heterotic genes expression14. Other studies suggested that gene expression discrimination in the hybrid was correlated with heterotic phenotypes13,15,16. Genes with expression levels between those in its parents are referred as additive genes, while genes with expression levels above or below those in both of its parents are termed as non-additive genes (overdominant or underdominant, respectively)8. Global gene expression studies indicated that about 20% of the DEGs between parents and the hybrid exhibited non-additive modes15,17,18.

Maize exhibits signicant heterosis in many traits, e.g. biomass, height, root growth, photosynthesis and starch metabolism, grain yield and biotic/abiotic stress resistances, and has been a model species for studying heterosis1,2. Although maize shows several heterotic traits at both vegetative and reproductive stages2, previous studies mainly focused on elucidating morphological and molecular phenotypes of heterosis at the early developmental stage. The heterotic phenotype and its molecular basis at the late developmental stage (post-silking and senescence stage) remain largely unexplored. Importantly, maize is sensitive to several abiotic stresses during silking and grain lling period (reproductive stage), e.g. low or high temperature and drought1921, and post-silking stresses can hasten leaf senescence and reduce yield. Leaf senescence terminates leaf lifespan in association with massive degradation of macromolecules and redistribution of nutrients, and therefore the dynamics of the process is critical to the formation of grain yield and quality22. Leaf senescence is an integral part of plant development and regulated by both genetic and environment factors. Thousands of senescence-associated genes (SAGs) have been identied in Arabidopsis23, with NAC092 (ORE1) and NAC029 (NAP) being typical positive senescence regulators24,25. Several phytochrome-interacting factors (PIFs) were found to mediate both age-triggered and dark-induced leaf senescence26,27. Ethylene, abscisic acid (ABA), and jasmonic acid (JA) promote leaf senescence while cytokinin inhibits leaf senescence28. Previously, physiological observations showed that hybrid breeding signicantly delayed maize leaf senescence during reproductive stage and extended photosynthetic period3; however, the molecular relationship between heterosis and leaf senescence remains elusive.

In this study, we investigated the heterotic eect on senescence regulation in maize ear leaf at the reproductive stage, using two well-known maize hybrids: B73 Mo17 and Zheng58 Chang7-2. All hybrids showed signicantly delayed leaf senescence, larger leaf areas, higher stem heights and enhanced yields. We analyzed the dynamic changes of transcriptomes in the ear leaves of B73, Mo17 and their hybrid to explore the molecular basis of the heterosis aer silking. It was found that some putative senescence regulatory genes (ZmNYE1, ZmORE1, ZmWRKY53 and ZmPIFs) exhibited underdominant expression patterns in the hybrid, while those putative photosynthesis- and starch biosynthesis-associated genes (ZmAPS1, ZmAPL) expressed overdominantly. Several cytokinin responsive marker genes and gibberellin synthetic genes showed higher expressions in the hybrid, whereas some ethylene biosynthetic genes displayed lower expressions. Furthermore, we identied 86 dierentially expressed transcription factors, including known senescence regulators, implying that they may be involved in both senescence and heterosis regulatory networks. These ndings demonstrate an obvious association of delayed ear leaf senescence and enhanced photosynthesis/metabolism with reproductive heterosis in maize, which sheds light on our global understanding of the molecular basis of heterosis, and importantly provides fundamental information for developing strategies and techniques of molecular breeding in maize.

Results

Maize B73 and Mo17 inbred lines belong to dierent heterotic groups and their hybrids are much taller and produce signicantly higher grain yields15. Zheng58 and Chang7-2 are elite inbred lines developed by Chinese maize breeders in Henan Academy of Agricultural Sciences and their hybrid represents a milestone variety in China10. To study their heterotic performance at the reproductive stage, we generated the reciprocal hybrids of B73/Mo17 and Zheng58/Chang7-2, respectively. It was found that the reciprocal hybrids, compared with their parental lines, were signicantly more vigorous, growing much taller and forming larger ear leaves (Fig.1).

Ear leaf is the largest leaf in maize, and contributes a substantial proportion of photosynthetic products to the reproductive growth29. Our initial interest was therefore to investigate the phenotypic and physiological features of the ear leaves of the hybrids in comparison with those features of their parental lines at the reproductive stage (aer silking). We found that the senescence process of the ear leaves was signicantly delayed in the hybrids relative to their parental lines. Besides, both Zheng58 and Chang7-2 showed delayed leaf senescence relative to B73 and Mo17, and Mo17 showed an obvious delayed senescence in the ear leaves than B73. These observations were validated by measurements of the Chl content, a physiological marker of senescence as well as an index of photosynthetic capacity (Fig.1c). Furthermore, the leaf width, leaf length and mature stem height of the hybrids all

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

Figure 1. Heterotic phenotypes of maize hybrids at the reproductive stage. (a) The photograph shows four inbred lines and their reciprocal hybrid lines at silking stage. Z58 and C7-2 represent Zheng58 and Chang7-2, respectively. (b) Phenotypes of ear leaves from dierent lines at the indicated time points aer silking (weeks aer silking, WAS). (c) Chlorophyll contents of ear leaves from dierent genotypes at the indicated time points. (d,e) Leaf widths and lengths of dierent genotypes. Third down and Third up refer to the third leaves down and up from the ear leaves, respectively. (f) Grain yields per plant and thousand-kernel weights of dierent genotypes. All the analyses involve three biological replicates and error bars represent the standard deviation.

signicantly exceeded those of their parents at silking stage. As expected, the reciprocal hybrids of B73 and Mo17 also had enhanced grain yields, as demonstrated by measurements of grain yields per plant and thousand-kernel weights (Fig.1f). Similarly, the reciprocal hybrids of Zheng58 and Chang7-2 also exhibited delayed ear leaf senescence (Fig.1ac), as well as larger leaf areas (Fig.1b,d,e) and higher grain yields (Fig.1f) relative to both parents. Since the phenotypic characteristics of F1 progenies seemed not signicantly inuenced by the direction of the cross between male and female parents, we therefore chose the F1 hybrid of B73 (male parent) Mo17 (female parent) and its parental lines for transcriptome proling due to the well-studied genome of B73.

To explore the molecular basis of the observed heterotic traits, we performed a high throughput RNA sequencing to investigate gene expression in the ear leaves of B73, Mo17 and B73 Mo17 (shortened as BM). Total mRNA was extracted from the upper half part of ear leaves in all the genotypes, with samples of three time-points: silking (S0), one week (S1) and two weeks (S2) aer silking, respectively. We obtained about 303 million high quality raw reads, 86.1% of which were mapped to the B73 reference genome APGV3, and 77.16% of mapped reads were unique mapping reads (Table1). Gene expression levels were measured as reads per kilobase per million reads (RPKM)30. Each sample contained a group of genes with very low RPKM values, probably due to their low expression levels or background. We took a conservative approach and a RPKM cut-o of 0.18 as a minimal expression value to avoid false positive estimation of gene expression, producing a unimodal distribution of genes expressed in each sample (Supplementary Fig. 1). Under this criterion, we detected a total of 30,000 (75.7%) expressed genes in our samples among 39,621 reference genes, with an average of 24,999 genes (63.1% of 39,621) detected in each sample (Supplementary Table 1). We then performed pairwise comparisons to identify DEGs among the samples by GFOLD31, which helps to reveal the molecular mechanisms underlying the observed heterotic phenotypes.

It has been reported that there is a correlation between parental lines transcriptional variation and their genetic diversity8, which may constitute the genetic basis for

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Genotype/stage Total reads Aligned reads Unique reads Expressed transcripts

B73_S0 34332584 30930562 23969506 25258 Mo17_S0 21066676 15709945 12662858 25093 BM_S0 40568672 35369740 26687000 25555 B73_S1 42803834 38669844 29418648 25105 Mo17_S1 37178274 31036284 23806834 24426 BM_S1 35771342 31063894 23659647 24635 B73_S2 38372646 34332615 26807905 25085 Mo17_S2 26244384 21075808 16785809 24653 BM_S2 26887434 22890669 17659278 25179

Table 1. Overview of mapping quality.

Figure 2. Gene ontology enrichment analysis of dierentially expressed genes (DEGs) between B73 and Mo17. (a) 1205 genes showed higher expressions in Mo17 than B73, 147 of them were shared by all three stages. Top 5 enriched GO categories were listed in the gure, while 147 genes in (a) only showed enrichment of three GO categories. (b) 1576 genes showed higher expression in B73 than Mo17, 231 of them were shared by three stages. Top 5 enriched GO categories are listed in the gure.

heterosis. We identied a total of 2,746 DEGs between B73 and Mo17 (PP-DEGs) (Supplementary Table 2), with 1,205 genes showing higher expressions in Mo17 at one or multiple stages. An examination of GO annotations showed that these DEGs mainly enriched in photosynthetic process, lipid metabolic process and cellular nitrogen compound metabolic process32, suggesting that Mo17 may have higher photosynthetic and metabolic activities. Among the 1,205 genes, 147 were shared by all three stages discussed in our study (Fig.2a), being enriched in three GO categories: protein folding, unfolded protein binding and heat shock protein binding. As maize plants are most sensitive to abiotic stresses at silking stage, the increased expression of these genes may imply that Mo17 is more responsive and presumably more tolerant to stresses than B73. On the other hand, 1,576 genes showed higher expression in B73 than in Mo17 at one or multiple stages (Supplementary Table 2), which were enriched in pollen-pistil interaction, pollination, programmed cell death, and apoptosis. Of those, 231 DEGs shared by three stages were enriched in nucleoside/nucleotide binding and protein serine/threonine kinase activity (Fig.2b). This nding suggests that senescence and cell death may be initiated earlier in B73 than in Mo17 aer pollination, which is consistent with the early senescence phenotype observed in B73 (Fig.1). In summary, Mo17s ear leaves exhibited higher photosynthetic and metabolic activities compared with B73s aer silking, which explains its relatively delayed senescence.

To manifest the transcriptome characteristics between the parental lines and the hybrid, we further compared the transcriptome of B73Mo17 hybrid to those of B73 and Mo17, respectively. We identied a total of 2,328 DEGs in the hybrid relative to at least one of its parents (PH-DEGs), representing 6.02% (2,328 out of 39,621) of the reference genes (Supplementary Table 3). Of these, 1,297, 1,198 and 924 DEGs were identied at S0, S1 or S2, respectively (Supplementary Table 4). Interestingly, among the 1,297 DEGs at S0, 211 (16.3%) were overdominant genes and 28 (2.2%) underdominant

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 3. Clustering and functional category enrichment analysis of DEGs between parental lines and the hybrid (PH-DEGs). (a) The percentages (%) in the graph represent the ratios of additive genes and non-additive genes during S0 to S2. (b) Six clusters were identied along the three developmental stages using the K-Means clustering algorithm. (c) Functional categories enrichment (modied MapMan bins) among six clusters according to the MapMan annotations, the color key represents (log10 P).

genes (Fig.3a and Supplementary Table 4), whereas at S1 and S2, the ratio of overdominant genes dramatically decreased to ~2% (20 and 21 overdominant genes at S1 and S2, respectively, Fig.3a and Supplementary Table 4). We further analyzed the functional enrichment of 211 overdominant genes at S0, and found that these genes were enriched in cell wall, sexual reproduction and enzyme-associated functions (Supplementary Table 5). This result, along with our previous data, demonstrated that these growth-associated and reproductive genes acted as overdominant genes at the silking stage but gradually became additive genes at S1 and S2 in the hybrid. The nding reveals that, even before phenotypic appearance of senescence initiation, the metabolism has undergone a signicant decline.

To further examine the gene expression modes among dierent genotypes, we performed a k-means cluster analysis of the PH-DEGs (Fig.3b and Supplementary Table 3). Among the 2,328 genes, 589 (Cluster 1, 25.3%) displayed similar expression levels between the hybrid and B73 but lower expressions in Mo17 across all stages. Numerous genes in Cluster 1 encode the proteins related to tetrapyrrole synthesis, stress, signaling, redox and amino acid metabolism according to MapMan annotation33. Likewise, 361 genes (15.5%) in Cluster 2 showed similar expressions in Mo17 and the hybrid, suggesting that these genes have an expression bias to Mo17 in the hybrid. These genes were enriched in growth-associated processes, such as photosynthesis, cell growth, DNA metabolism, polyamine metabolism, and N metabolism. Both C1 and C2 represent genes with similar expression levels to one parent, so PH-DEGs in C1 and C2 may also be PP-DEGs. Consistently, we found 1,528 shared genes between 2,328 PH-DEGs and 2,781 PP-DEGs (data not shown). A high similarity in the number and identity between PP-DEGs and PH-DEGs may imply that only a few or dozens of dierentially regulated genes are critically responsible for the dierent phenotype between the two parents or between the parents and the hybrid, while majority of the genes functioning downstream are shared, possibly involved in similar pathways, in dierent genotypes although likely exhibiting dierent expression dynamics. Indeed, among the 1528 genes, 763 and 551 genes showed similar expressions in the hybrid as in B73 and Mo17 (data not shown), respectively, among the three stages, which are mainly represented by C1 and C2 according to our clustering analysis (Fig.3). Furthermore, the clustering analysis also identied 438 genes (18.3%, Cluster 3)

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

Figure 4. Expression levels of Chlorophyll biosynthesis- and degradation- associated genes. (a) Chlorophyll synthetic pathway in Arabidopsis and putative maize homologous genes. (b) Expression patterns of several putative maize chlorophyll synthetic genes. (c) A diagram illustrating the chlorophyll degradation pathwayin Arabidopsis. NYE1: Non-Yellowing 1; PPH: pheophytin pheophorbide hydrolase; PAO: pheophorbide a oxygenase; RCCR: red chlorophyll catabolite reductase; pFCC: primiary uorescent catabolites. (d) Expression patterns of putative ZmNYE1, ZmPAO, and ZmPPH.

with higher expressions in the hybrid than in both parental lines at S1, which mainly represented the overdominant genes17. These genes were enriched in several growth- and metabolism-associated functions, including photosynthesis, N metabolism and amino acid metabolism, cell wall (cellulose synthesis) and redox state. We found a cell wall synthesis-associated gene GRMZM2G089699 (homologue of AtEXPB4, AT2G45110), a pollen tube growth-associated gene GRMZM2G125356 (homologue of AtVGD1, AT2G47040), and a putative glutathione peroxidase gene GRMZM2G012479 (homologue of AtGPX7, AT4G31870) in Cluster 3, and all of them were overdominant genes at S0. On the other hand, Cluster 4 (346 genes, 14.8%) mainly represented the underdominant genes at S0, which were enriched in stress and secondary metabolism-associated functions. Collectively, the higher expression of growth- and metabolism-associated genes and the lower expression of stress-associated genes aer silking well explain the heterotic phenotype of the hybrid.

To explore the molecular basis of the stay-green phenotype in the hybrid aer silking (Fig.1), we examined the expression of Chl biosynthesis-11 and degradation-associated genes34 (Fig.4). We found that nine putative Chl bio-synthetic genes, including two homologous genes of protochlorophyllide oxidoreductase B (GRMZM2G036455, GRMZM2G084958), exhibited overdominant expression patterns at one or multiple stages (Fig.4b). Conversely, GRMZM2G091837 (putative homologue of AtNYE1 or AtSGR1, At4G22920), a critical Chl degradation regulator, and GRMZM2G339563 (putative homologue of AtPAO, AT3G44880) and GRMZM2G109070 (putative homo-logue of AtPPH, AT5G13800), two putative genes encoding Chl catabolic enzymes35, displayed underdominant expression patterns at S1 and/or S2 (Fig.4d). Association of the elevated expression of Chl biosynthetic genes with the reduced expression of Chl catabolic genes (CCGs) well explains the stay-green phenotype of the hybrid.

Besides the higher Chl retention in the hybrid before leaf senescence, our analyses also showed that genes in the overdominant cluster (cluster 3) were enriched in photosynthesis-associated process, and many of them encoded photosynthetic proteins (Fig.3c), including photosynthesis-associated polypeptide subunits and enzymes (Supplementary Table 6). For example, a gene encoding phosphoenolpyruvate carboxylase (PEPC) (Fig.5), responsible for the higher efficiency of CO2 xation in C4 plants36, showed a higher expression in the hybrid than in both parental lines at S0 and S2. Additionally, the genes coding for 3-phosphoglycerate kinase (PGK), fructose-bisphospate aldolase (FBA), fructose-1, 6-bisphosphate kinase (FBP), ribulose-5-phosphate

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

Figure 5. Expression levels of photosynthesis-associated enzyme genes. (a) A diagram shows the intermediates and key enzymes involved in calvin cycle in maize. (b) Overdominant expression patterns ofsix genes encoding carbon xation- and calvin cycle-associated enzymes. PEP: phosphoenolpyruvate; OAA: oxaloacetate; Asp: aspartic acid; Asn: Asparagine; RuBP: ribulose-1,5-bisphosphate; PGA: phosphoglycerate; G-3-P: glyceraldehyde-3-phosphate; DHAP: Dihydroxyacetone phosphate; F-1,6-P: fructose-1,6- phosphate; F-6-P: fructose-6-phosphate; Xu5P: Xylulose 5-phosphate; AST: aspartate transaminase; Rubisco: Ribulose bisphosphate carboxylase oxygenase; PGK: 3-phosphoglycerate kinase; RPE: ribulose-5-phosphate epimerase; PEPC: phosphoenolpyruvate carboxylase; TRK: Transketolase; FBA: fructose-bisphospate aldolase; PRK: phosphoribulokinase.

epimerase (RPE), and phosphoribulokinase (PRK), the ve enzymes required for calvin cycle (Fig.5), also exhibited overdominant expression patterns at one or multiple stages. These transcriptomic features reveal an outline of the molecular basis for the hybrid to accumulate more dry matters during silking and grain lling period.

Starch is the main storage carbohydrate in vascular plants, and starch biosynthetic activity is conventionally considered as a measurement for photosynthetic capability. We therefore examined the expression pattern of starch synthetic genes in the hybrid. ADP-glucose pyrophosphorylase (AGP), consisting of one small subunit (APS) and one large subunit (APL), is responsible for the rate-limiting conversion of glucose-1-phosphate and ATP to inorganic pyrophosphate (PPi) and ADP-glucose37. The activity of AGP is allosterically activated by 3-phosphoglycerate, and inhibited by orthophosphate38. Recent studies revealed that overexpression of allosterically insensitive form of AGP homologous gene in potatoes successfully enhanced starch accumulation in tubers39,

while silencing AGP gene in Nicotiana benthamiana reduced the starch content in leaves to 60% of those in the wild type40. In this study, we found that maize putative APS showed overdominant expression at all the stages while APL only showed overdominant expression at S1 (Supplementary Fig. 2). It has been reported that APL is highly unstable in the absence of APS in Arabidopsis41, and APS1 T-DNA null mutants contain neither APL nor APS proteins42. It is possible that the higher expression of APS in the hybrid could efficiently maintain AGP complex stability and activity. AtSnRK1 encodes the sucrose non-fermenting-1-related protein kinase, a positive regulator of starch biosynthesis, and its overexpression could lead to 30% more starch accumulation in the tubers of transgenic potato43. Consistently, we found that the expression level of AtSnRK1 homologue (AT5G21170) in maize was signicantly higher in the hybrid than in both parents at S0 (Supplementary Fig. 2), indicating that the starch biosynthetic capacity in the hybrid may be enhanced during silking and grain lling period.

A recent study reported that the ear leaf initiates senescence approximately 15 days aer pollination44. Considering that maize could nish its pollination shortly aer silking, our time point S2 (two weeks aer silking), may represent the time point of senescence initiation. To analyze the heterotic eect on senescence-associated genes expression, we compared the PH-DEGs with 4,552 previously identied senescence-induced genes in maize44, and identied 680 shared genes (data not shown), accounting for 29.2% of PH-DEGs. The relatively high overlap between the PH-DEGs and senescence-induced genes indicates that heterosis may aect senescence regulation or vice versa. To investigate the molecular basis of the heterosis at pre-senescence stage, we analyzed the expression of some known senescence-associated genes (SAGs) in dierent genotypes through S0 to S2. Two putative senescence marker genes ZmWRKY53 (GRMZM2G411766) and ZmORE1 (GRMZM2G114850), together with ZmORE9 (GRMZM2G405203)44,45, were found to exhibit a significantly underdominant pattern in the hybrid at S0

SCIENTIFIC REPORTS

7

www.nature.com/scientificreports/

Figure 6. Expression levels of senescence-associated genes. (a) Expression levels of six putative senescence-associated positive regulators in maize. (b) Expression levels of putative cytokinin-related negative regulators of senescence. (c) Expression levels of ethylene biosynthesis- and signaling-related genes.

(Fig.6a). We further examined the expression of a few recently identied positive regulatory genes of senescence, e.g. phytochrome interacting factors (PIFs)26,27, and found that three homologous genes (GRMZM2G065374, GRMZM2G016756, and GRMZM2G165042) of AtPIF4 (AT2G43010) displayed lower expressions in the hybrid than in both of the parents at one or multiple stages (Fig.6a), indicating that the occurrence of heterosis in the hybrid is likely associated with the delay of its ear leaf senescence.

In plants, cytokinin and ethylene are two important phytohormones known to be involved in inhibiting and promoting leaf senescence28, respectively. As expected, three putative cytokinin response marker genes, ZmARR2, ZmARR7, and ZmAHK3, whose homologues known to inhibit leaf senescence in Arabidopsis46, showed overdominant expressions at one or multiple stages (Fig.6b). Conversely, several ethylene synthetic and signaling genes exhibited lower expressions in the hybrid at one or multiple stages, including putative 1-aminocyclopropane-1 -carboxylic acid synthase (ACS) and 1-aminocyclopropane-1-carboxylic acid oxidase (ACO) genes (Fig.6c).

Although the relationship between gibberellin and senescence has not been clearly elucidated, recent studies demonstrated that gibberellins are mainly responsible for the control of leaf growth and cell division in maize. Ectopic expression of gibberellin 20-oxidase1 (GA20OX-1) could promote growth in maize47. In this study, we did nd that two putative ZmGA20OX genes (GRMZM2G368411, GRMZM2G049418) and a putative ZmGA3OX gene (GRMZM2G036340) exhibited overdominant expression patterns at S0 (Supplementary Fig. 3), plausibly explaining the enhanced expression of growth-associated genes in the hybrid and the observed growth vigor. Collectively, the association of the higher expression of senescence inhibitory and growth promoting genes with the lower expression of senescence regulatory genes molecularly validates the delayed senescence phenotype observed in the hybrid.

Transcriptional regulation is one of the major regulatory modes during plant development and senescence. By referring to 3316 maize TFs recorded previously, we initially identied 1197 TFs expressed in our samples48. We also detected 86 TFs belonging to 26 families from PH-DEGs and they

SCIENTIFIC REPORTS

8

www.nature.com/scientificreports/

Figure 7. Dynamic changing inventories of dierentially expressed transcription factors. (a) Distribution of dierentially expressed transcription factor families at S0, S1 and S2. Fishers exact test was used to compare the ratios of TF families in our list to their ratios of total identied TF families, *p<0.05. (b) Representative transcription factors with non-additive expression patterns in the hybrid.

might be involved in heterosis regulation (Supplementary Table 7). By statistical signicance analysis (Fishers exact test), we found that NAC, HSF and CAMTA families exhibited signicant enrichment. NACs are widely involved in senescence regulation, and HSFs are generally regarded as stress responsive TFs. Besides, many genes of bHLH (11, 12.8%) were dierentially expressed among the three stages (Fig.7), presumably because they are involved in the regulation of leaf senescence and stress tolerance26,49. For example, a putative senescence regulatory bHLH gene GRMZM2G016756 (homologue of AtPIF4) and a putative senescence regulatory NAC gene GRMZM2G430849 (homologue of AtNAP) exhibited extraordinarily low expressions in the hybrid aer silking (Fig.7). In addition, genes of WRKY (8, 9.3%) and MYB (12, 13.9%) families were also dierentially expressed (Fig.7a). Of them was a putative WRKY54 (GRMZM2G004060) that exhibited an underdominant expression (Fig.7b). WRKY54 was reported to show a strong but transient induction at the onset of senescence50. Consistently, its expression was signicantly induced during S0 to S1 in all the three genotypes, but dramatically decreased at S2 (Fig.7b). In contrast, a MYB family gene GRMZM2G084583, the homologue of which (MYB32) could respond to diverse stresses in Arabidopsis51, exhibited an overdominant expression pattern in the hybrid at S0 and S2 (Fig.7b). We also identied four TFs with most signicantly overdominant expression patterns in the hybrid: GRMZM2G161512 (MYB), GRMZM2G083504 (bHLH), GRMZM2G381441 (ERF) and GRMZM2G159094 (NAC), and ve with most signicantly underdominant expression patterns: GRMZM2G003489 (HSF), GRMZM2G159119 (MYB), GRMZM2G301485 (HSF), GRMZM2G164909 (HSF), and GRMZM2G368491 (bZIP) (Supplementary Table 7). We assumed that these TFs with signicantly non-additive expression patterns are likely related to heterosis. A further GO analysis showed that 86 TFs were enriched in the regulation of multiple pathways, including cellular biosynthetic processes, nitrogen compound metabolic processes and macromolecule metabolic processes (Supplementary Table 7), suggesting that the dierentially expressed TFs may regulate diverse metabolic processes. Nevertheless, solid genetic evidence is required to validate whether these TFs are directly involved in the regulation of heterosis during silking and grain lling period.

Maize is not only a major crop but also a model species for studying plant heterosis. Eorts have been made to elucidate the molecular basis of heterosis at its early developmental stage8,15,17,18, but the molecular analysis of its heterotic performance at reproductive stage has not been reported. In this study, we conducted a time-course transcriptome analysis of ear leaves aer silking, and found that the heterosis was associated not only with the lower expression of SAGs but also with the higher expression of photosynthesis-, stress regulation- and metabolism-associated genes. This nding suggests that the mechanism of ear leaf heterosis involves both the enhancement of photosynthetic/metabolic activities, along with the stress responsiveness, and the delay of senescence. The kaleidoscopic revelation of molecular basis of the heterosis for the ear leaf helps to integrate diverse understandings of the heterotic mechanism proposed previously.

Studies in maize as well as in Arabidopsis and rice (Oryza sativa) have revealed that the yield heterosis was correlated with two critical physiological features: 1) expanded life span (delayed leaf senescence), which extends photosynthetic period22; 2) larger leaf area and cell size, which contain more chloroplasts and increase photo-synthetic capacity11,13. Consistently, we observed a signicantly delayed senescence of ear leaves in the reciprocal

SCIENTIFIC REPORTS

9

www.nature.com/scientificreports/

hybrids of both B73 vs Mo17 and Zheng58 vs Chang7-2 inbred lines, suggesting that the delay of leaf senescence may be a universal feature in maize hybrids. We also observed larger leaf areas and higher stem heights in the hybrids. Importantly, our transcriptome data revealed the lower expression of diverse SAGs (e.g. PIFs, OREs, CCGs and ACSs) and the enhanced expression of growth- (e.g. GA20OXs) and metabolism- (e.g. calvin cycle, chlorophyll synthesis) associated genes. The identication of heterosis-associated genes provides critical information about the molecular basis of heterotic development and maintenance at the reproductive stage.

Leaf senescence is regulated by a complex network with various regulatory nodes. Importantly, its initiation and progression facilitate massive nutrient remobilization from senescing leaves to developing organs, particularly reproductive organs22, the time and dynamics of which may signicantly aect crop yield and quality, i.e. precocious or overly delayed initiation of leaf senescence may result in either insufficient dry matter accumulation or incomplete nutrient remobilization52,53. Maize exhibits a dual direction pattern of leaf senescence: bottom-up and top-down, with ear leaves remaining functional to the last. This characteristic leaf senescence pattern is somehow shaped presumably due to its beneting the nutrient supply to cob development. This common feature implies that the heterosis for grain yield is likely acquired, at least in part, via optimizing the initiation and/or dynamics of leaf senescence and consequently maximizing both dry matter accumulation and nutrient remobilization during hybrid breeding. It is estimated that about 50% of total seasonal dry matter is xed during the grain lling period in maize54, and it is therefore of great importance to properly modulate leaf senescence to improve the grain yield of the hybrid. In this study, we identied diverse growth vigor- and senescence-associated genes, which may be exploited for molecular breeding. ZmNAP (GRMZM2G430849), a heterosis-associated TF gene shown in our study, would have a potential application since a reduced expression of its counterpart in rice delayed leaf senescence and enhanced yield55. Besides, putative GA biosynthetic genes (GA20OX, GA3OX) were also identied as overdominant genes in our study, and ectopic expression of GA20OX1 has been shown to promote growth and accumulate more vegetative biomass in transgenic maize plants, resulting in the improvement of feedstock for bioethanol production47. Taken together, the diverse heterosis-associated genes identied in our study may be exploitable in the molecular improvement of heterosis.

Methods

The maize B73 inbred line was originally obtained from American Germplasm Resources Information Network (www.ars-grin.gov). Mo17, Zheng58 and Chang7-2 inbred lines were kindly gied by Dr. Huiyong Li (Cereal Crop Research Institute, Henan Academy of Agricultural Sciences, Zhengzhou, China). Plants were hand planted in 1.0-m long rows with row and plant spacing of 70 and 40cm, respectively. The reciprocal crosspollination lines were conducted by our lab. Dierent genotypes were grown in the experimental eld (30.940495 N, 121.127765E) of the national center of plant gene research (Shanghai) during summer of 2011 (July 1th). The silking periods of Zheng 58/Chang7-2 hybrids were in middle August, while silking periods of B73/Mo17 hybrids and four parental lines were about 4 and 7 days later than Zheng 58/Chang7-2 hybrids, respectively. B73/Mo17 hybrids were harvested in early October and hybrids of Zheng58 and Chang7-2 were harvested in late October. Ear leaves without pathogen invasion were used for Chl content and gene expression analyses in dierent genotypes aer silking.

Chlorophyll content measurement. Upper half parts of ear leaves were used for Chl content quantication. Leaf samples were powdered in liquid nitrogen, and then incubated in 95% acetone/ethanol (v/v =2:1, 5% pure water) for 12h in darkness (20C). Extracts were centrifuged at 12,000g for 5min at 4C. Chl was quantied by measuring absorbance at 645 and 663 nm, and Chl content was calculated as a ratio of (20.23*A645+8.023*A663) g1 fresh weight. Three biological replicates were used for each measurement.

Upper half parts of ear leaves were ground in liquid nitrogen. We collected about 0.1 g powder in 1.5 mL tube and added 1 mL TRIzol reagent (Invitrogen) for RNA extraction according to the manufacturers protocol. Total RNA was treated with10U DNaseI (Ambion, USA) at 37C for 1hour, and the reaction mixture was then puried with Micropoly(A) PuristTM mRNA purication kit (Ambion, USA). cDNA was synthesized from 10g mRNA by Superscript II reverse transcriptase (Invitrogen, USA) and sonicated into 300500 bp fragments. The cDNA library was constructed by using TruSeqTM DNA Sample Prep Kit Set A (Illumina, USA) and amplied with TruSeq PE Cluster Kit (Illumina, USA). Sequencing was carried out on an Illumina HiSeq 2000 sequence analyzer at Chinese National Human Genomic Center at Shanghai (CHGCS). Double end 100-bp reads were generated. RNAseq data in this work have been deposited at the NCBI short read archive under accession number SRP064910.

Tophat v2.0.956 was used for reads-to-reference genome alignment. Sequence reads for each sample were aligned to the maize B73 AGPv3 reference genome (http://plants.ensembl.org/Zea_mays). We chose the highest mapping score for the reads with multiple mapped positions. Transcript abundance was normalized with GFOLD v1.1.231. We used 2 fold-change as cuto (adjusted values from GFOLD), and at least one genes RPKM value should be higher than 1.0 for dening candidates of DEGs.

To examine dierent gene expression modes among dierent stages, k-means cluster was adopted to cluster the DEGs in dierent genotypes. Fold change values (adjusted values from GFOLD) between the hybrid and parental lines were used as input for clustering. Functional annotation of maize genes was carried out according to MapMan annotation33. Gene function enrichment analysis of DEGs was performed by using Fishers exact test and referring to B73 genome data. The heat map graph was constructed in R environment. In some paragraphs, we also conducted GO analyses using the AgriGO tool (http://bioinfo.cau.edu.cn/agriGO/) to facilitate comparison of overrepresented GO terms32.

SCIENTIFIC REPORTS

10

www.nature.com/scientificreports/

Orthology Analysis of Annotation for Maize genes. Orthologous and paralogous gene relationships were identied by using online website putative orthologous groups database (http://pogs.uoregon.edu). We also used some orthologous or paralogous genes according to the published literatures.

References

1. Schnable, P. S. & Springer, N. M. Progress toward understanding heterosis in crop plants. Annu. Rev. Plant Biol. 64, 7188 (2013).2. Chen, Z. J. Genomic and epigenetic insights into the molecular bases of heterosis. Nat. Rev. Genet. 14, 471482, doi: 10.1038/ nrg3503 (2013).

3. Lee, E. & Tollenaar, M. Physiological basis of successful breeding strategies for maize grain yield. Crop Sci. 47, S-202S-215 (2007).4. Davenport, C. B. Degeneration, Albinism and Inbreeding. Science 28, 454455 (1908).5. Jones, D. F. Dominance of linked factors as a means of accounting for heterosis. Proc. Natl Acad. Sci. USA 3, 310312 (1917).6. Shull, G. H. The composition of a eld of maize. Journal of Heredity 296301 (1908).7. Yu, S. B. et al. Importance of epistasis as the genetic basis of heterosis in an elite rice hybrid. Proc. Natl Acad. Sci. USA 94, 92269231 (1997).

8. Stupar, R. M. et al. Gene expression analyses in maize inbreds and hybrids with varying levels of heterosis. BMC Plant Biol. 8, 33 (2008).

9. Marsan, P. A., Castiglioni, P., Fusari, F., Kuiper, M. & Motto, M. Genetic diversity and its relationship to hybrid performance in maize as revealed by RFLP and AFLP markers. Theor. Appl. Genet. 96, 219227 (1998).

10. Li, H., Wang, L., Tang, B. & Wang, Z. Research on the genetic structure and heterosis of Zhengdan958. J. Maize Sci. 17, 2831 (2009).11. Fujimoto, R., Taylor, J. M., Shirasawa, S., Peacock, W. J. & Dennis, E. S. Heterosis of Arabidopsis hybrids between C24 and Col is associated with increased photosynthesis capacity. Proc. Natl Acad. Sci. USA 109, 71097114 (2012).

12. Ni, Z. et al. Altered circadian rhythms regulate growth vigour in hybrids and allopolyploids. Nature 457, 327331 (2009).13. Song, G. S. et al. Comparative transcriptional proling and preliminary study on heterosis mechanism of super-hybrid rice. Mol. Plant 3, 10121025 (2010).

14. Song, G. Y. et al. The phenotypic predisposition of the parent in F1 hybrid is correlated with transcriptome preference of the positive general combining ability parent. BMC Genomics 15, 297 (2014).

15. Swanson-Wagner, R. A. et al. All possible modes of gene action are observed in a global comparison of gene expression in a maize F1 hybrid and its inbred parents. Proc. Natl Acad. Sci. USA 103, 68056810 (2006).

16. Zhang, H.-Y. et al. A Genome-Wide Transcription Analysis Reveals a Close Correlation of Promoter INDEL Polymorphism and Heterotic Gene Expression in Rice Hybrids. Mol. Plant 1, 720731 (2008).

17. Meyer, S., Pospisil, H. & Scholten, S. Heterosis associated gene expression in maize embryos 6 days aer fertilization exhibits additive, dominant and overdominant pattern. Plant Mol. Biol. 63, 381391 (2007).

18. Stupar, R. M. & Springer, N. M. Cis-transcriptional variation in maize inbred lines B73 and Mo17 leads to additive expression patterns in the F1 hybrid. Genetics 173, 21992210 (2006).

19. Li, Y. et al. Eect of post-silking drought on nitrogen partitioning and gene expression patterns of glutamine synthetase and asparagine synthetase in two maize (Zea mays L.) varieties. Plant Physiol. Bioch. 102, 6269 (2016).

20. Ordez, R. A., Savin, R., Cossani, C. M. & Slafer, G. A. Yield response to heat stress as aected by nitrogen availability in maize. Field Crop Res. 183, 184203 (2015).

21. Ying, J., Lee, E. A. & Tollenaar, M. Response of maize leaf photosynthesis to low temperature during the grain-lling period. Field Crop Res. 68, 8796 (2000).

22. Wu, X. Y., Kuai, B. K., Jia, J. Z. & Jing, H. C. Regulation of leaf senescence and crop genetic improvement. J. Integr. Plant Biol. 54, 936952 (2012).

23. Li, Z. et al. LSD 2.0: an update of the leaf senescence database. Nuc. Acids Res. 42, D1200D1205 (2014).24. Kim, J. H. et al. Trifurcate feed-forward regulation of age-dependent cell death involving miR164 in Arabidopsis. Science 323, 10531057 (2009).

25. Guo, Y. & Gan, S. AtNAP, a NAC family transcription factor, has an important role in leaf senescence. Plant J. 46, 601612 (2006).26. Song, Y. et al. Age-triggered and dark-induced leaf senescence require the bHLH transcription factors PIF3, 4, and 5. Mol. Plant 7, 17761787 (2014).

27. Sakuraba, Y. et al. Phytochrome-interacting transcription factors PIF4 and PIF5 induce leaf senescence in Arabidopsis. Nat. Commun. 5, 4636 (2014).

28. Sarwat, M., Naqvi, A. R., Ahmad, P., Ashraf, M. & Akram, N. A. Phytohormones and microRNAs as sensors and regulators of leaf senescence: Assigning macro roles to small molecules. Biotechnol. Adv. 31, 11531171 (2013).

29. Smart, C. M. et al. The Timing Of Maize Leaf Senescence And Characterization Of Senescence-Related Cdnas. Physiol. Plantarum. 93, 673682 (1995).

30. Mortazavi, A., Williams, B. A., McCue, K., Schaeer, L. & Wold, B. Mapping and quantifying mammalian transcriptomes by RNASeq. Nat. methods 5, 621628 (2008).

31. Feng, J. et al. GFOLD: a generalized fold change for ranking dierentially expressed genes from RNA-seq data. Bioinformatics 28, 27822788 (2012).

32. Du, Z., Zhou, X., Ling, Y., Zhang, Z. & Su, Z. agriGO: a GO analysis toolkit for the agricultural community. Nucleic Acids Res. 38, W64W70 (2010).

33. Zhang, W. Y. et al. Transcriptional analyses of natural leaf senescence in maize. PLoS One 9, e115617 (2014).34. Hortensteiner, S. & Krautler, B. Chlorophyll breakdown in higher plants. Biochim. Biophys. Acta. 1807, 977988 (2011).35. Hortensteiner, S. Stay-green regulates chlorophyll and chlorophyll-binding protein degradation during senescence. Trends Plant Sci. 14, 155162 (2009).

36. Raines, C. A. The Calvin cycle revisited. Photosynth Res. 75, 110 (2003).37. Bahaji, A. et al. Starch biosynthesis, its regulation and biotechnological approaches to improve crop yields. Biotechnol. Adv. 32, 87106 (2014).

38. Kleczkowski, L. A. A phosphoglycerate to inorganic phosphate ratio is the major factor in controlling starch levels in chloroplasts via ADP-glucose pyrophosphorylase regulation. FEBS letters 448, 153156 (1999).

39. Stark, D. M., Timmerman, K. P., Barry, G. F., Preiss, J. & Kishore, G. M. Regulation of the amount of starch in plant tissues by ADP glucose pyrophosphorylase. Science 258, 287292 (1992).

40. George, G. M. et al. Virus-Induced Gene Silencing of Plastidial Soluble Inorganic Pyrophosphatase Impairs Essential Leaf Anabolic Pathways and Reduces Drought Stress Tolerance in Nicotiana benthamiana. Plant Physiol. 154, 5566 (2010).

41. Wang, S. M. et al. Characterization of ADG1, an Arabidopsis locus encoding for ADPG pyrophosphorylase small subunit, demonstrates that the presence of the small subunit is required for large subunit stability. Plant J. 13, 6370 (1998).

42. Bahaji, A. et al. Arabidopsis thaliana mutants lacking ADP-glucose pyrophosphorylase accumulate starch and wild-type ADP-glucose content: further evidence for the occurrence of important sources, other than ADP-glucose pyrophosphorylase, of ADP-glucose linked to leaf starch biosynthesis. Plant Cell Physiol. 52, 11621176 (2011).

43. McKibbin, R. S. et al. Production of high-starch, low-glucose potatoes through over-expression of the metabolic regulator SnRK1. Plant Biotechnol. J. 4, 409418 (2006).

SCIENTIFIC REPORTS

11

www.nature.com/scientificreports/

44. Zhang, W. Y. et al. Transcriptional analyses of natural leaf senescence in maize. PLoS One 9, e115617 (2014).45. Woo, H. R. et al. ORE9, an F-box protein that regulates leaf senescence in Arabidopsis. Plant Cell 13, 17791790 (2001).46. Kim, H. J. et al. Cytokinin-mediated control of leaf longevity by AHK3 through phosphorylation of ARR2 in Arabidopsis. Proc. Natl Acad. Sci. USA 103, 814819 (2006).

47. Nelissen, H. et al. A Local Maximum in Gibberellin Levels Regulates Maize Leaf Growth by Spatial Control of Cell Division. Curr. Biol. 22, 11831187 (2012).

48. Guo, A. Y. et al. PlantTFDB: a comprehensive plant transcription factor database. Nucleic Acids Res. 36, D966D969 (2007).49. Nakashima, K., Takasaki, H., Mizoi, J., Shinozaki, K. & Yamaguchi-Shinozaki, K. NAC transcription factors in plant abiotic stress responses. Biochim. Biophys. Acta. 1819, 97103 (2012).

50. Besseau, S., Li, J. & Palva, E. T. WRKY54 and WRKY70 co-operate as negative regulators of leaf senescence in Arabidopsis thaliana. J. Exp. Bot. 63, 26672679 (2012).51. Zhang, L., Zhao, G., Jia, J., Liu, X. & Kong, X. Molecular characterization of 60 isolated wheat MYB genes and analysis of their expression during abiotic stress. J. Exp. Bot. 63, 203214 (2012).

52. Lobell, D. B., Sibley, A. & Ivan Ortiz-Monasterio, J. Extreme heat eects on wheat senescence in India. Nature Clim. Change 2, 186189 (2012).

53. Uauy, C., Distelfeld, A., Fahima, T., Blechl, A. & Dubcovsky, J. A. NAC Gene regulating senescence improves grain protein, zinc, and iron content in wheat. Science 314, 12981301 (2006).

54. Tollenaar, M., Ahmadzadeh, A. & Lee, E. A. Physiological basis of heterosis for grain yield in maize. Crop Sci 44, 20862094 (2004).55. Liang, C. et al. OsNAP connects abscisic acid and leaf senescence by ne-tuning abscisic acid biosynthesis and directly targeting senescence-associated genes in rice. Proc. Natl Acad. Sci. USA 111, 1001310018 (2014).

56. Trapnell, C., Pachter, L. & Salzberg, S. L. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics 25, 11051111 (2009).

Acknowledgements

This research was supported by grants from the National Program of Transgenic Variety Development of China (2016ZX08003004-009) and the National Natural Science Foundation of China (31371330). We thank the biological supercomputing server of Computing Center of Beijing Institutes of Life Science for its technical assistance and Ms Yamao Chen for her assistance in improving English.

Author Contributions

B.K. and J.Q. designed the experiments. X.T., Y.J., J.R. and Z.W. collected samples, Z.W. arranged for RNA sequencing. Y.S., Z.Z., L.L., L.Q., W.C. and X.W. analyzed the data. Y.S., Z.Z., J.G., B.K. and J.Q. wrote and edited the paper. All authors approved the manuscript.

Additional Information

Accession codes: RNAseq data in this work were deposited at the NCBI short read archive under accession number SRP064910.

Supplementary information accompanies this paper at http://www.nature.com/srep

Competing nancial interests: The authors declare no competing nancial interests.

How to cite this article: Song, Y. et al. Association of the molecular regulation of ear leaf senescence/stress response and photosynthesis/metabolism with heterosis at the reproductive stage in maize. Sci. Rep. 6, 29843; doi: 10.1038/srep29843 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

12