OsCOMT, encoding a caffeic acid O-methyltransferase in melatonin biosynthesis, increases rice grain yield through dual regulation of leaf senescence and vascular development.

Melatonin, a natural phytohormone in plants, plays multiple critical roles in plant growth and stress responses. Although melatonin biosynthesis-related genes have been suggested to possess diverse biological functions, their roles and functional mechanisms in regulating rice grain yield remain largely unexplored. Here, we uncovered the roles of a caffeic acid O-methyltransferase (OsCOMT) gene in mediating rice grain yield through dual regulation of leaf senescence and vascular development. In vitro and in vivo evidence revealed that OsCOMT is involved in melatonin biosynthesis. Transgenic assays suggested that OsCOMT significantly delays leaf senescence at the grain filling stage by inhibiting degradation of chlorophyll and chloroplast, which, in turn, improves photosynthesis efficiency. In addition, the number and size of vascular bundles in the culms and leaves were significantly increased in the OsCOMT-overexpressing plants, while decreased in the knockout plants, suggesting that OsCOMT plays a positive role in vascular development of rice. Further evidence indicated that OsCOMT-mediated vascular development might owe to the crosstalk between melatonin and cytokinin. More importantly, we found that OsCOMT is a positive regulator of grain yield, and overexpression of OsCOMT increase grain yield per plant even in a high-yield variety background, suggesting that OsCOMT can be used as an important target for enhancing rice yield. Our findings shed novel insights into melatonin-mediated leaf senescence and vascular development and provide a possible strategy for genetic improvement of rice grain yield.

Introduction

Melatonin, also known as N‐acetyl‐5‐methoxytryptamine, has been characterized as an important animal hormone with a broad spectrum of biological functions (Calvo et al., 2013; Garcia et al., 2014; Hardeland et al., 2012). Recent investigations revealed that melatonin plays essential roles in plant growth and development, such as root formation, seed germination, floral transition, fruit ripening and photosynthesis regulation (Arnao and Hernandez‐Ruiz 2020; Hardeland, 2016; Lee et al., 2019; Sharma et al., 2020; Sun et al., 2021; Yang et al., 2021). Melatonin also have anti‐senescence capacity in ageing leaves (Sun et al., 2021; Wang et al., 2018). In addition, melatonin has been shown to protect plants from multiple abiotic and biotic stresses, including cold, drought, salt, heat and pathogens (Arnao and Hernandez‐Ruiz 2019; Sun et al., 2021; Wang et al., 2018; Zhang et al., 2015; Zhao et al., 2021a). Thus far, several mechanisms have been proposed to understand the functions of melatonin. For example, given its structure similar to indole‐3‐acetic acid (IAA, auxin), melatonin has been considered as an important plant hormone regulator that influences root organogenesis and other plant developmental processes (Liang et al., 2017; Wang et al., 2016). Melatonin is also believed to have the ability to scavenge reactive oxygen species (ROS) by stimulating antioxidant enzymes and increasing the levels of several antioxidants in plants under various stress conditions (Li et al., 2020; Liang et al., 2018; Manchester et al., 2015). In addition, melatonin functions as a defence signalling molecule, which directly regulates the expression of some defence genes or interacts with other signalling elements to mediate stress responses (Lee et al., 2015; Shi et al., 2015). Nevertheless, the molecular mechanisms underpinning the diverse roles of melatonin in plants remain largely unexplored.

The melatonin biosynthetic pathway in plants begins with a precursor tryptophan and involves four consecutive enzymatic reactions (Kanwar et al., 2018). Several key enzymes involved in the biosynthesis of melatonin in plants have been identified, including tryptophan decarboxylase (TDC), tryptamine 5‐hydroxylase (T5H), tryptophan hydroxylase (TPH), serotonin N‐acetyltransferase (SNAT) and N‐acetylserotonin O‐methyltransferase (ASMT) (Kanwar et al., 2018; Sun et al., 2021). SNAT and ASMT play essential roles in determining the melatonin levels in plants, because their enzymatic activities are much lower than those of TDC and T5H (Byeon et al., 2014b). Similar to ASMT, caffeic acid O‐methyltransferase (COMT) also belongs to the O‐methyltransferase (OMT) family, which has been found to catalyse the core O‐methylation reaction in melatonin biosynthesis (Byeon et al., 2014a). Recent phylogenetic and protein structural evidence revealed that COMT likely evolved from ASMT through gene duplication and subsequent functional divergence (Zhao et al., 2021b). Newly emergent COMT not only has a significantly higher ASMT activity in melatonin synthesis but also gains some new functions to catalyse the production of lignin and flavonoid, suggestive of the potential roles of COMT in adaptation to stressful environments (Byeon et al., 2014a; Lee et al., 2014; Zhao et al., 2021b). Indeed, the roles of several melatonin biosynthesis‐associated genes have been illustrated in certain plant species. For instance, down‐regulation of SNAT2 in rice led to a semi‐dwarf phenotype (Lee and Back, 2019), while knockout of SNAT2 in Arabidopsis delayed flowering and reduced biomass (Lee et al., 2019). Knocking out SNAT1 and/or COMT significantly affected the accumulation of oil and anthocyanin in mature seeds of Arabidopsis (Zhao et al., 2020). In addition, overexpression of SlCOMT from Solanum lycopersicum increased the melatonin accumulation in tomato and improved the tolerance to salt stress (Sun et al., 2020). Transgenic Arabidopsis overexpressing MzASMT from apple (Malus zumi Mats) increased the tolerance to drought stress compared to the wild‐type (WT) (Zuo et al., 2014). Despite the roles of melatonin biosynthesis‐related genes in plant growth regulation and stress tolerance, their molecular mechanisms and potential for agricultural applications especially in staple crops, are poorly understood.

Rice (Oryza sativa L.) is one of the most important crops in the world, providing nearly 50% of the calories consumed by humankind (Fukagawa and Ziska 2019; Zhao et al., 2020). Rice yield needs to increase up to 50% by 2030 to meet the increasing demand (Ahmadi et al., 2014). Although it was suggested that enhancing the accumulation of melatonin might have massive potential for improving crop production (Sun et al., 2021), it remains largely unclear whether and how melatonin biosynthesis‐associated gene affect grain yield in rice. In this study, we investigated that rice OsCOMT is involved in melatonin biosynthesis through both evidence of in vitro and in vivo. Transgenic assays showed that OsCOMT regulates leaf senescence and vascular bundle development. OsCOMT also has large effects on yield and its related components. More importantly, overexpression of OsCOMT significantly increased grain yield even in a high‐yield variety background. Our findings show new insights into melatonin‐mediated leaf senescence and vascular development and provide a novel possible strategy to enhance rice yield production.

Results

Expression pattern and subcellular localization of OsCOMT

qRT‐PCR analysis was performed to explore the expression profile of OsCOMT. The result showed that OsCOMT was constitutively expressed in all tested rice tissues, including the root, stem, leaf, sheath and panicle (Figure 1a). In addition, we noticed that the expression of OsCOMT was gradually increased in the flag leaf from 70 days post‐sowing (dps) to 110 dps, implying a possible role of OsCOMT during leaf senescence. The transgenic lines expressing GUS (β‐glucuronidase) reporter gene driven by the promoter of OsCOMT were also generated. GUS staining showed that OsCOMT was abundantly expressed in the sheath, node, root, stem and panicle (Figure 1b‐h). Notably, the GUS activities increased in the leaves from 70 dps to 110 dps, which was consistent with the qRT‐PCR results (Figure 1h). Moreover, the vascular tissues in the stems and leaves showed high GUS activities (Figure 1f, h). To visualize the subcellular localization of OsCOMT, we performed transient protoplast transformation of both the green fluorescent protein (GFP) alone and the OsCOMT‐GFP fusion protein under the control of the CaMV 35S promoter. GFP fluorescence observation showed that OsCOMT protein is predominantly localized to the cytoplasm (Figure 1i). Similar results were observed in transiently expressing the OsCOMT‐GFP fusion protein in tobacco leaf cells (Figure 1j).

Figure 1

Expression pattern of OsCOMT and subcellular localization of OsCOMT protein. (a) qRT‐PCR analyses of the expression pattern of OsCOMT in rice different tissues. Total RNA was isolated from the root, stem, sheath and panicle at the heading stage and the flag leaf at the 70 days post‐sowing (dps) to 110 dps. (b‐h) GUS staining in the sheath (b, c), node (d), root (e), stem (f), panicle (g), leaf (h). Scale Bar, 2 mm in (c‐g). Scale bar, 2 cm in (b). (i) Confocal micrographs of rice protoplasts transformed with 35S::GFP (upper) and 35S::OsCOMT‐GFP (lower). (j) Confocal micrographs of tobacco leaf cells transformed with 35S::GFP (upper) and 35S::OsCOMT‐GFP (lower). Scale bar, 10 μm in (i) and (j).

OsCOMT is involved in melatonin biosynthesis

To investigate the enzyme activity of OsCOMT, we expressed and purified OsCOMT‐GST fusion protein in an Escherichia coli expression system. The purified recombinant protein yielded a single band of about 67 KD on the sodium dodecyl sulphate‐polyacrylamide gel electrophoresis (SDS‐PAGE), which was consistent with the predicted molecular weight of GST‐tagged OsCOMT protein (Figure 2a). Western blot analysis was further performed to confirm the purified protein. As expected, the single band size was also consistent with the molecular weight of recombinant OsCOMT protein (Figure 2b). To determine whether the OsCOMT protein has the ASMT activity, enzyme‐linked immunosorbent assay (ELISA) was employed to detect the ASMT activities with different amount of purified OsCOMT‐GST proteins. We found that the enzymatic activities of ASMT significantly correlated with the increasing concentrations of OsCOMT protein (Figure 2c). Given that ASMT enzymes in Arabidopsis thaliana catalyse N‐acetylserotonin (NAS) into melatonin (Byeon et al., 2014a), we further examined OsCOMT enzyme activity using NAS substrate dose response. With the raise of NAS concentrations, the contents of melatonin significantly increased (Figure 2d). This indicated that the recombinant OsCOMT protein possessed the ASMT activity and catalysed the NAS into melatonin in vitro.

Figure 2

Purification of the OsCOMT protein and methyltransferase activity. (a) Purification of N‐terminal GST‐tagged OsCOMT protein in Escherichia coli. Lanes 1‐2: GST‐tag; M: molecular protein standard; Lanes 4‐10: purified GST‐OsCOMT proteins. (b) Western blot test of GST‐OsCOMT protein by anti‐GST antibody. Lanes 1‐2: GST‐tag; M: molecular protein standard; lanes 4‐9: purified GST‐OsCOMT proteins. (c) Acetylserotonin O‐methyltransferase (ASMT) activity of purified OsCOMT protein by ELISA. (d) Methyltransferase activity of purified OsCOMT performed in the presence of various concentrations of N‐acetylserotonin (NAS). Values are means ± SD (n = 3). Different letters indicate significant difference (Tukey’s HSD test at P < 0.05).

To further evaluate the OsCOMT activity in vivo, we generated five knockout mutants of OsCOMT using the CRISPR/Cas9 system and 15 independent OsCOMT‐overexpression lines in the Nipponbare (NIP) background. Two independent overexpression transgenic lines (OE1 and OE2) and two independent knockout mutants (KO1 and KO2), respectively, were chosen for further analysis. The KO1 mutant had 1‐bp insertion and the KO2 mutant had a 2‐bp insertion (Figure 3a and S1a). Both of the mutations led to the truncated OsCOMT protein (Figure S1b). The expression levels of OsCOMT in OE1 and OE2 were 8.89‐fold and 5.18‐fold higher than that of WT respectively (Figure 3b). We then measured the endogenous melatonin content of flag leaves at 85 dps and found that the melatonin contents were increased by 51.65% and 45.61% in the OE1 and OE2 respectively. By contrast, the contents of melatonin decreased by 59.01% and 57.21% in the KO1 and KO2 plants (Figure 3c). These data revealed that OsCOMT is involved in melatonin biosynthesis in vivo.

Figure 3

Comparison of endogenous melatonin contents and phenotypes between the Nipponbare and OsCOMT transgenic lines. (a) The mutation information of two independent OsCOMT knockout lines generated by CRISPR/Cas9 system. Blue font labels the target site, and the PAM is labelled in green. Red font indicated the mutation sites. The truncated amino acid sequence of two OsCOMT‐KO lines can be found in Figure S1. (b) Relative expression levels of OsCOMT in two independent OsCOMT‐overexpression lines tested by qRT‐PCR. OsActin was used as the internal control. (c) Comparison of endogenous melatonin content between WT and OsCOMT transgenic plants. Three biological replicates were performed. Data show means ± SD. Asterisks indicate significant differences as determined by Student’s t‐test (**, P < 0.01). (d‐g) Plant phenotype of Nipponbare (NIP), two overexpression lines (OE1 and OE2) and two knockout mutants (KO1 and KO2) at the mature stage. (e) Panicles, (f) grains, (g) grains from the whole plants of the NIP, OE1, OE2, KO1 and KO2. Scale bar, 10 cm in (d), 5 cm in (e) and (g), 1 cm in (F). (H‐I) Comparison of the 1,000‐grain weight and grain yield per plant between the NIP and OsCOMT transgenic lines. Values are means ± SD (n = 15). Asterisks indicate significant differences as determined by Student’s t‐test (**, P < 0.01).

OsCOMT has large effects on multiple agronomic traits and regulates rice grain yield

To understand the biological function of OsCOMT, we performed phenotypic observations of the OsCOMT transgenic lines (Table S1). Compared with WT, the OE lines exhibited increased plant height and panicle length, while the KO mutants showed the opposite phenotypes (Figure 3d, e). In addition, we observed that the grain length, grain width and thickness increased in the OE lines (Figure 3f and Table S1). As a result, the 1,000‐grain weights of the OE1 and OE2 lines were increased by 7.68% and 5.50% respectively (Figure 3h). By contrast, the length, width and thickness of grains were decreased in the KO lines, which result in 12.61% and 11.74% decrease in the 1,000‐grain weight respectively (Figure 3f, h). Moreover, panicle number per plant and grain number per panicle were increased in the overexpression lines, while decreased in the mutants (Table S1). Subsequently, the grain yield per plant of OE1 and OE2 were increased by 21.37% and 10.98%, respectively, while the grain yield per plant of KO1 and KO2 decreased by 30.69% and 19.42% respectively (Figure 3g, i). These data suggested that OsCOMT has large effects on multiple agronomic traits and plays a positive role in grain yield improvement in rice.

OsCOMT delays leaf senescence and improves photosynthesis

No significant differences in leaf colour were observed between WT and transgenic plants during the vegetative growth period (Figure 4a). After heading stage, we observed that the KO mutants exhibited premature leaf senescence when compared with WT. By contrast, the overexpression lines of OsCOMT turned yellow later than the WT (Figure 4a, b). We further measured the chlorophyll content in the flag leaves. The OsCOMT transgenic lines and the WT plants showed similar chlorophyll content at the heading stage (85 dps) (Figure 4d). The chlorophyll content of WT leaves declined with the senescence process. Notably, the decreases of chlorophyll content of KO mutants were significantly faster than that of WT. By contrast, the OE lines had higher chlorophyll content than WT even at the end of grain filling stage (123 dps) (Figure 4d).

Figure 4

Leaf characterization of the OsCOMT transgenic lines. (a) Plant phenotypes of NIP and OsCOMT transgenic plants at 79 to 123 dps. Scale bar, 20 cm. (b) Magnified views of flag leaves from NIP and OsCOMT transgenic plants at 85 to 123 stages. (c) Comparison of ROS levels and cell death in the flag leaves between NIP and OsCOMT transgenic plants. DAB, NBT and TB staining were used to detect the levels of H2O2, O2 − and cell death in the flag leaves at 110 dps between NIP and OsCOMT transgenic plants respectively. (d‐e) Comparison of the total chlorophyll content (d) and effective photochemical quantum yield of photosystem‐II (Y (II)) (e) in the flag leaves at 85 to 123 dps between NIP and OsCOMT transgenic plants. Data show means ± SD (n = 3). Asterisks indicate significant differences as determined by Student’s t‐test (*, P < 0.05; **, P < 0.01).

Diaminobenzidine tetrahydrochloride (DAB) and nitroblue tetrazolium (NBT) staining revealed that the ROS levels in leaves, including H2O2 and O2 ‐, were lower in the OE1 than the WT. By contrast, higher ROS levels were detected in the leaves of KO1 mutant (Figure 4c). Leaf premature senescence is usually accompanied with cell death (Lee and Chen, 2002). Trypan blue (TB) staining was performed to examine the cell death in the leaves of the WT and OsCOMT transgenic lines. We found that the cell death was obviously reduced in the OE1 leaves, but increased in the KO1 leaves (Figure 4c). We further measured the fluorescence parameter of effective quantum yield of photosystem‐II (Y (II)) in flag leaf from 85 to 123 dps. The OsCOMT transgenic lines and the WT plants showed similar Y (II) at the heading stage (85 dps) (Figure 4e). During the whole grain filling stage, the Y (II) of OE lines were significantly higher than that of WT. However, the Y (II) of KO lines were lower (Figure 4d‐e). In addition, the other photosystem‐II chlorophyll fluorescence parameters, including the net photosynthetic rate (Pn), electron transfer rate (ETR), photochemical fluorescence quenching (qP), and non‐photochemical fluorescence quenching (NPQ), were measured to estimate the photosynthetic activity of flag leaf. The Pn, ETR and qP in the whole grain filling stage were significantly greater in OE lines than WT, while the Pn, ETR and qP of KO lines showed the opposite trend (Figure S2a‐c). By contrast, the NPQ was consistently higher in the KO lines, but lower in the OE lines than in the WT (Figure S2d). These results indicated that up‐regulation of OsCOMT delays leaf senescence and improves photosynthetic activity.

Several reports have suggested that exogenous application of melatonin delays leaf senescence in plants (Ahmad et al., 2020; Liang et al., 2015). In this study, we found that the endogenous melatonin content was significantly reduced in the KO1 mutant, while increased in the OE1 line (Figure 3c). 100 μM melatonin was applied to the KO leaves at 85 dps for 10 days in paddy field conditions (Figure S3a). At 110 dps, the premature leaf senescence of the KO plants was rescued, and no obviously phenotypic difference was observed between the melatonin‐treated KO leaves and WT leaves (Figure 5a and Figure S3b). After melatonin treatment, both the chlorophyll a and chlorophyll b contents were significantly increased in the leaves of KO mutants (Figure 5c, d). The total chlorophyll contents in the leaves of melatonin‐treated KO mutants were significantly higher (approximately 2‐fold) than in the non‐treated plants (Figure 5e). These results indicated that premature leaf senescence of the OsCOMT mutants mainly result from the decreased melatonin level.

Figure 5

Exogenous application of melatonin delayed leaf senescence and increased grain yield. (a) Comparison of the flag leaves among non‐treated NIP, melatonin‐treated and non‐treated OsCOMT‐KO plants at 110 dps. 100 μM melatonin solution was applied to the leaves of KO plants at 85 dps for 10 days. Non‐treated NIP and KO plants were applied with mock as the control. (b) Grains and brown rice from the main panicle (upper) and grains from the whole plants (Lower) of the non‐treated NIP and melatonin‐treated and non‐treated KO plants. Scale bar, 1 cm. (c‐e) Comparison of the content of chlorophyll a (c), chlorophyll b (d) and total chlorophyll (e) among the non‐treated NIP and melatonin‐treated and non‐treated KO plants. Data show means ± SD (n = 3). (f‐k) Comparison of the grain length (f), grain width (g), grain thickness (h), 1,000‐grain weight (i), grain numbers per plant (j) and grain yield per plant (k) among the non‐treated NIP and melatonin‐treated and non‐treated KO mutants. Data show means ± SD (n = 15). Different letters indicate significant difference (Tukey’s HSD test at P < 0.05).

In addition, we found that exogenous melatonin application significantly increased grain size of KO mutants (Figure 5b, f‐h). As a result, exogenous melatonin rescued the decreased grain weight of KO mutants (Figure 5i). However, application of melatonin had no obvious effect on grain number per plant (Figure 5j). Subsequently, the grain yield per plant of the melatonin‐treated KO plants increased by 13.44% and 10.01%, respectively, when compared with the non‐treated control (Figure 5k). Our findings suggested that melatonin application promotes grain size and grain weight and has the potential to enhance yield production in rice.

OsCOMT represses chloroplast and chlorophyll degradation during leaf senescence

Chloroplast degradation occurs during leaf senescence, which results in the reduce of chlorophyll content (Dominguez and Cejudo, 2021). We compared the chloroplast structure of senescent leaves in the WT and OsCOMT transgenic plants at the grain maturation stage. Transmission electron microscopy (TEM) observation showed that the number of chloroplasts dramatically reduced in the KO1 leaf cells compared with that in the WT and OE1 (Figure 6a). In addition, the thylakoid and granum were disorderly arranged in KO1 leaf cells, while intact thylakoid and granum structure with more tightly stacked lamellae in the grana were observed in the OE1 leaf cells (Figure 6a). Generally, the number of osmiophilic granules (OG) increased with leaf senescence (He et al., 2018). The KO1 leaf cells produced more OG than WT. However, the OG number of OE1 leaf cells was less than that in WT (Figure 6a). These findings suggested that OsCOMT inhibits the chloroplast degradation during leaf senescence.

Figure 6

OsCOMT delayed leaf senescence by regulating chloroplast development and chlorophyll degradation. (a) Ultrastructure of the chloroplast in the leaf cells of NIP and OsCOMT transgenic lines. CP, chloroplast; G, granum; OG, osmiophilic granule. (b) Transcriptional profile from RNA‐seq data for a subset of genes related to chloroplast development, chlorophyll synthesis, chlorophyll degradation and senescence in OE1 and KO1 plants for OsCOMT. (c‐f) Transcript levels of the selected genes were confirmed in two OE and KO lines through qRT‐PCR analysis with three biological replicates. OsActin was used as a control. Data shown are means ± SD. Different letters indicate significant difference (Tukey’s HSD test at P < 0.05).

RNA‐seq analysis was further performed to investigate the possible regulatory mechanism of OsCOMT in rice. The expression levels of the genes involved in chloroplast development and chlorophyll synthesis were significantly declined in the KO1 leaves than in WT leaves (Figure 6b). By contrast, their expression levels were up‐regulated in the OE1 leaves. Moreover, the transcripts of the genes responsible for chlorophyll degradation and leaf senescence were increased in the KO1 leaves, while decreased in the OE1 leaves (Figure 6b). qRT‐PCR analysis was carried out to validate the expression levels of several selected genes (Figure 6c‐f), which is consistent with the RNA‐seq data. Taken together, we propose that OsCOMT delays leaf senescence through melatonin‐mediated repression of chloroplast and chlorophyll degradation.

OsCOMT affects vascular bundle development

We also observed that the flag leaves of KO mutants were narrower than that of WT, whereas overexpression of OsCOMT significantly increased the width of flag leaves (Table S1). Histological analysis showed that the area and number of large vascular bundles (LVB) were significantly reduced in the KO1 leaves than in the WT leaves, while the OE1 leaves exhibited larger and more LVB (Figure 7a‐c). We also found that the culm diameter of KO1 mutant was thinner than that in WT. By contrast, the OE1 plant produced much thicker culms than WT (Figure 7d, e). In addition, the OE1 plant displayed larger LVB areas, increased numbers of LVB and small vascular bundle (SVB) than those in WT (Figure 7d, f‐h). By contrast, the KO1 mutant exhibited smaller LVB and decreased numbers of LVB and SVB, when compared with the WT plant (Figure 7d, f‐h). These findings suggested that OsCOMT plays an important role in the development of vascular bundle of rice leaf and stem.

Figure 7

OsCOMT affected vascular development of rice. (a) The cross section of flag leaves from NIP and OsCOMT transgenic lines. LVB, large vascular bundles. Scale bar, 50 μm. (b‐c) The areas (b) and numbers (c) of LVB in the flag leaves from NIP and OsCOMT transgenic lines. Data shown are means ± SD (n ≥ 5). (d) The cross section of culm from NIP and OsCOMT transgenic lines. SVB, small vascular bundles. Scale bar, 1 mm (upper) and 100 μm (lower). (e) Diameter of the culm from NIP and OsCOMT transgenic lines. (f‐g) The area (f) and number (g) of LVB in the culm from NIP and OsCOMT transgenic lines. (h) The number of SVB in the culm from NIP and OsCOMT transgenic lines. Data shown are means ± SD (n ≥ 5). (I‐J) The levels of melatonin (i) and cytokinin (trans‐zeatin, tZ) (j) in the stems from NIP and OsCOMT transgenic lines before rice heading date. Data shown are means ± SD (n = 3). (k) The expression of several cytokinin biosynthesis‐related genes tested by qRT‐PCR analysis. (l) The expression of several cytokinin degradation‐related genes tested by qRT‐PCR analysis. OsActin was used as a control. Data shown are means ± SD (n = 3). Asterisks indicate significant differences as determined by Student’s t‐test (*, P < 0.05; **, P < 0.01). Different letters indicate significant difference (Tukey’s HSD test at P < 0.05).

The plant hormone cytokinin has been suggested to be an important regulator for vascular development (De Rybel et al., 2016). We quantified the level of cytokinin (trans‐zeatin, tZ) and melatonin of the stems before heading date. The melatonin level was higher in the OE plants, but lower in the KO plants than that in the WT (Figure 7i). In addition, the content of tZ was increased by 12.21% and 17.03% in the OE1 and OE2 plants, but decreased by 3.53% and 7.69% in the KO1 and KO2 plants, when compared with that in WT (Figure 7j). These results suggested that the increase in melatonin content may lead to the accumulation of cytokinin. We investigated the expression levels of several key genes involved in the biosynthesis and degradation of cytokinin (Chen et al., 2020; Kudo et al., 2010). Compared with the WT, the transcripts of the cytokinin biosynthetic genes, including OsLOGL1, OsLOGL3 and OsLOGL10, were significantly up‐regulated in the OE plants, but down‐regulated in the KO plants respectively (Figure 7k). By contrast, the expression of degrading genes for cytokinin, including OsCKX2, OsCKX4 and OsCKX10 were declined in the OE plants, while increased in the KO plants (Figure 7l). These results suggested a possible crosstalk between melatonin and cytokinin.

To further investigate the interaction between melatonin and cytokinin, we detected endogenous tZ content of rice seedlings treated by exogenous melatonin. 20 μM melatonin was applied to the 14‐day‐old rice seedlings. After 2 h of melatonin treatment, the tZ content of melatonin‐treated leaves were significantly higher than the mock (Figure 8a). Similar results were also observed after 6, 24 and 48 h of melatonin application (Figure 8a). These data indicated that exogenous application of melatonin significantly induced the accumulation of endogenous cytokinin. We also observed that exogenous melatonin treatment significantly up‐regulated the expression levels of OsLOGL1, OsLOGL3 and OsLOGL10, while the transcripts of OsCKX2, OsCKX4 and OsCKX10 were suppressed (Figure 8b‐e). These findings revealed the crosstalk between melatonin and cytokinin in rice.

Figure 8

Exogenous application of melatonin affected endogenous cytokinin (trans‐zeatin, tZ) levels of rice. (a) Effect of exogenous application of melatonin on endogenous tZ content. (b‐e) Effect of exogenous application of melatonin on the expression levels of several genes for cytokinin biosynthesis and degradation. 20 μM melatonin was applied to the 14‐day‐old rice (NIP) seedlings. Non‐treated NIP plants were applied with mock as the control. Data shown are means ± SD (n = 3). Asterisks indicate significant differences as determined by Student’s t‐test (*, P < 0.05; **, P < 0.01).

Overexpression of OsCOMT increases grain yield in a high‐yield rice variety

Given that overexpression of OsCOMT increased the grain yield in Nipponbare, we were interested in whether this gene could further increase grain yield under a high‐yield rice variety. We introduced the overexpression construct of OsCOMT into Suken118 (SK118), a high‐yield variety released in 2016 Jiangsu Province, China. Two homozygous lines (SK118‐OE1 and SK118‐OE2) with significantly elevated OsCOMT transcription levels were grown for phenotypic evaluation (Figure 9a). Our results showed that the plant height, flag leaf length and width, panicle length and panicle number per plant were significantly increased in the two OsCOMT‐overexpressing SK118 lines compared with that in the SK118 (Figure 9b, c, f‐j). In addition, the grain number per main panicle was increased by 16.75% and 12.48% in the SK118‐OE1 and SK118‐OE2 respectively (Figure 9k). The OsCOMT‐overexpressing SK118 lines also exhibited larger grain size than SK118 (Figure 9d, l‐m), which result in 8.99% and 4.52% increases in the 1,000‐grain weight of the SK118‐OE1 and SK118‐OE2 respectively (Figure 9n). The seed setting rates, however, were significantly decreased in two SK118‐OE lines (Figure 9o). Compared with the SK118, the grain yield per plant was increased by 15.27% and 11.87% in the SK118‐OE1 and SK118‐OE2 respectively (Figure 9e, p). The field plot trial was further performed, and out results showed that two SK118‐OE lines had a 13.10% and 9.91% advantage in plot grain yield respectively (Figure 9q). These findings indicated that OsCOMT has the potential for increasing rice yield in high‐yield varieties.

Figure 9

Identification and phenotypic analysis of the OsCOMT‐overexpression lines in the high‐yield variety Suken118 (SK118). (a) The expression levels of OsCOMT in SK118 and two overexpression lines (SK118‐OE1 and SK118‐OE2) tested by qRT‐PCR. OsActin was used as a control. Data shown are means ± SD (n = 3). (b) Plant phenotypes of SK118 and two overexpression lines at the mature stage. Scale bar, 15 cm. (c‐e) Panicles (c), grains and brown rice (d) and grains from the whole plants (e) of SK118 and two overexpression lines. Scale bar, 5 cm in (c), 1 cm in (d), 5 cm in (e). (f‐q) Comparison of the plant height (f), flag leaf length (g), flag leaf width (h), panicle length (i), panicle number per plant (j), grain number per main panicle (k), grain length (l), grain width (m), 1,000‐grain weight (n), seed setting rate (o), grain yield per plant (p) and grain yield per plot (q) between SK118 and two overexpression lines. All agronomic traits of OsCOMT transgenic lines in the SK118 background were measured in 2021. Data are given as means ± SD (n = 15 in f‐p; n = 5 in q). Asterisks indicate significant differences as determined by Student’s t‐test (*, P < 0.05; **, P < 0.01).

Discussion

Melatonin has long been considered as an important hormone in animals with diverse biological functions, such as circadian rhythms and anti‐ageing agent (Hardeland et al., 2012; Karasek, 2004). Recent evidence showed that melatonin has essential roles in many aspects of plant growth and stress tolerance (Arnao and Hernandez‐Ruiz 2019; Huangfu et al., 2021; Nawaz et al., 2015; Sun et al., 2021; Wang et al., 2018; Zhang et al., 2015; Zhao et al., 2021a). Although several genes in melatonin synthesis were characterized in rice, the molecular mechanism of melatonin in rice have rarely been illustrated. Moreover, the roles of melatonin synthesis related genes in regulating rice grain yield improvement remains controversial (Byeon and Back 2014; Hong et al., 2018; Lee and Back 2017). For instance, overexpression of the sheep SNAT gene in rice reduced grain yield (Byeon and Back 2014). However, the transgenic plants overexpressing rice OsSANT1 exhibited increased grain yield due to the increasing panicle number per plant (Lee and Back 2017). This contradiction promoted us to investigate the function of melatonin synthesis related genes on plant growth and grain yield in rice. In this study, we focused on the role of rice caffeic acid O‐methyltransferase (OsCOMT). We found that OsCOMT possesses activity in biosynthesis of melatonin through in vitro and in vivo evidence. Moreover, OsCOMT has large effects on multiple traits, including leaf senescence, photosynthesis efficiency, vascular development and grain yield.

Leaf senescence, the final stage of leaf life cycle, is controlled by various external and internal factors (Lim et al., 2007; Zhang et al., 2021a). Although the physiological roles of melatonin on leaf senescence have been investigated (Hong et al., 2018; Lee and Back 2017; Liang et al., 2015), the molecular mechanism remains unclear. In this study, we found that knockout of OsCOMT led to premature leaf senescence, while overexpression of OsCOMT prevented leaf senescence. These results indicated that OsCOMT is a positive regulator for delaying leaf senescence in rice. qRT‐PCR and GUS staining analyses revealed that the expression of OsCOMT increased with leaf senescence. These data imply that OsCOMT plays essential roles in maintaining green leaf. However, we also noticed that the transcript of OsCOMT was higher in old leaves, which seems contradictory to the role of OsCOMT in delaying leaf senescence. A possible explanation is that the expression of OsCOMT is controlled by a leaf senescence‐dependent feedback regulation in rice, although the mechanism remains unknown.

In general, leaf senescence is manifested by irreversible yellowing of the leaves due to the rapid breakdown of chlorophyll and photosynthetic organelles (Dominguez and Cejudo, 2021). Compared to the WT, chlorophyll content of the flag leaves significantly increased in the OsCOMT‐OE plants, but decreased in the OsCOMT‐KO mutants. In addition, exogenous application of melatonin recovered the chlorophyll content of OsCOMT‐KO leaves. These results suggested that both endogenous and exogenous melatonin delay leaf senescence in rice. TEM observations showed fewer chloroplasts with more accumulation of osmiophilic granules in the leaves of the OsCOMT‐KO plant. RNA‐seq data and qRT‐PCR analyses also showed that the expression of genes related to chlorophyll degradation and senescence were down‐regulated in the OsCOMT‐OE plant, but up‐regulated in the OsCOMT‐KO plant. These findings indicated that melatonin delays leaf senescence by inhibiting degradation of chlorophyll and chloroplast.

Leaf photosynthesis is the most important source for crop yield (Yamori et al., 2016). Delay of leaf senescence at the grain filling stage improves photosynthesis efficiency, which can promote the accumulation of photosynthetic‐assimilates in reproductive organs and then enhance crop yield (Yang et al., 2016). In the present study, we found that OsCOMT positively regulates photosynthesis efficiency in rice. As a result, the plant growth of OE lines was significantly promoted. Three yield components, including panicle number per plant, grain number per panicle and grain weight of the OE lines significantly increased, which consequently resulted in the enhancement of rice grain yield production. Notably, exogenous melatonin application also increased grain yield of KO plants, implying the potential role of melatonin in promoting rice grain yield. Although several genes involved in melatonin synthesis have been suggested to possess roles in grain yield (Hong et al., 2018; Lee and Back 2017), our results firstly revealed that the rice gene OsCOMT is a positive regulator of grain yield. More importantly, field tests showed that overexpression of OsCOMT in a high‐yield rice variety SK118 can further increase grain yield, suggestive its great potential in genetic improvement of rice yield. All these results provide a possible strategy for increasing yield potential through manipulating OsCOMT in rice.

Plant vascular system is composed of two functionally distinct aspects, including phloem and xylem, which are specialized for nutrient and water transport respectively (Ruonala et al., 2017). Thus, the vascular bundle system in the culms and leaf veins is crucial for plant growth and development (Fu et al., 2021; Zhai et al., 2018; Zhang et al., 2021b). Vascular development in plants is controlled by complex genetic networks. Cytokinin has been suggested to be an important regulator in plant vascular growth (Ruonala et al., 2017; De Rybel et al., 2016). For instance, the Arabidopsis mutant of the adenosine phosphate isopentenyltransferase gene (IPT) involved in the rate‐limiting step of cytokinin biosynthesis results in a severe reduction in vascular cambium size (Matsumoto‐Kitano et al., 2008). Overexpression of a cytokinin‐degrading gene cytokinin oxidase 2 (CXK2) leads to a reduction of the size of vascular cambium in poplar (Nieminen et al., 2008). However, whether and how melatonin regulates vascular development in rice remain unclear. OsCOMT expressed primarily in the vascular tissues in stems and leaves. The number and size of large vascular bundles in the culms and leaf veins were significantly increased in the OE1 plant, but decreased in the KO1 plant, suggesting that OsCOMT regulates the development of vascular bundles in rice. Cytokinin contents in the OE and KO culms were higher and lower, respectively, than that in WT. In addition, the transcript levels of several key genes for cytokinin biosynthesis, including OsLOGL1, OsLOGL3 and OsLOGL10, were significantly increased in the OE plants. Meanwhile the expression of cytokinin degradation‐related genes, such as OsCKX2, OsCKX4 and OsCKX10, were suppressed in the OE plants. These results suggested that increased melatonin leads to cytokinin accumulation. In addition, we found that OsCOMT positively regulated grain number per panicle. Previous reports suggested that down‐regulation or knockout of the genes for cytokinin degradation, such as OsCKX2 and OsCKX11, significantly increased grain number per panicle (Ashikari et al., 2005; Zhang et al., 2021a). Given that cytokinin content is closely associated with vascular bundle development and grain number per panicle in rice (Ashikari et al., 2005; Campbell and Turner 2017; Matsumoto‐Kitano et al., 2008), we speculated that OsCOMT‐mediated grain number per panicle and vascular development at least partially result from the crosstalk between melatonin and cytokinin. In addition, cytokinin is suggested to be the most effective phytohormone for suppressing leaf senescence and decelerating photosynthetic deterioration (Zhang et al., 2021a; Zwack and Rashotte 2013). The crosstalk between melatonin and cytokinin may also play roles in OsCOMT‐mediated leaf senescence. Thus far, many studies have showed the cooperation between melatonin and other phytohormones, such as ethylene, gibberellin and abscisic acid (Sun et al., 2021). Our findings uncovered the interaction between cytokinin and melatonin in rice. However, the molecular mechanism remains elusive.

In summary, we proposed a possible model that OsCOMT increases yield production through dual roles in melatonin‐mediated delay of leaf senescence and enhancement of vascular tissues in rice (Figure 10). Genetic manipulation of OsCOMT provides a potential strategy in rice high‐yield breeding.

Figure 10

Proposed model for OsCOMT in melatonin‐mediated leaf senescence, vascular development and grain yield. OsCOMT catalyses melatonin biosynthesis. Overaccumulated melatonin level delays leaf senescence and then improves photosynthesis efficiency (source). Overexpression of OsCOMT increases panicle number per plant, grain number and grain size, which form a larger sink. In addition, OsCOMT promotes cytokinin accumulation by inducing the expression of cytokinin biosynthesis‐related gene OsLOGLs and suppressing the expression of cytokinin degradation‐associated gene OsCKXs. Overexpression of OsCOMT also increases vascular bundle number and size and forms an efficient flow. Thus, OsCOMT coordinates source‐flow‐sink relationship by delaying leaf senescence, improving photosynthesis and promoting vascular development to obtain high grain yield production.

Materials and methods

Plant materials and growth conditions

Two rice WT strains, Nipponbare (NIP, Oryza sativa L. ssp. Japonica) and Suken118 (SK118, a high‐yield japonica variety released in 2016 in Jiangsu Province, China), were used in this study. Rice organs/tissues, including the root, stem, sheath, panicle, were obtained from NIP at the heading stage. The leaves were collected from NIP at 70 days post‐sowing (dps) to 110 dps. All samples were kept frozen in liquid nitrogen for further analyses. To construct the overexpression vector of OsCOMT (Gene locus: LOC_Os08g06100), a 1107‐bp cDNA fragment containing the entire coding sequence of OsCOMT was amplified and then inserted into the plant binary vector pCAMBIA1301UbiNOS, which carries a maize (Zea mays) ubiquitin promoter (Zhou et al., 2009). The overexpression constructs were then transferred into NIP and SK118 respectively. To analyse gene expression patterns, an approximate 2‐kb OsCOMT promoter region was inserted into the vector pCAMBIA1301 to drive the expression of a β‐glucuronidase (GUS) gene. To generate the CRISPR/Cas9 mutant, a sgRNA targeting the first exon of OsCOMT was selected and then inserted into the pC1300‐Cas9 vector. These constructs were transformed by Agrobacterium tumefaciens‐mediated transformation into NIP (Hiei et al., 1994).

Before harvest, several agronomic traits were measured, including plant height, flag leaf length and width and panicle number per plant from different rice lines under field conditions. 15 main panicles from each line were chosen for measuring the panicle length and grain number per panicle. Grain‐related traits, including grain length, grain width, grain thickness, 1,000‐grain weight and grain yield per plant, were measured after harvesting and stored at 37°C for 1 week. The field plot trial was performed at the experimental field in Yangzhou, Jiangsu Province, China (119°40′ E, 32°40′ N) during the standard growing season in 2021. The distance between the plants within a row was 16.5 cm, and the distance between the rows was 25 cm. At maturity, the grains from all the plants in each field plot (1 m2) were harvested and naturally dried for yield test. Five replications of each plot were arranged in a randomized block design.

RNA isolation and gene expression analysis

Total RNA was isolated using the RNA extraction kit (Tiangen Biotech, Beijing, China). First‐strand cDNA was synthesized using a Revert Aid First Strand cDNA Synthesis kit (Vazyme Biotech, Nanjing, China). The expression levels were measured by quantitative reverse transcription‐PCR (qPCR) using the ABI ViiATM 7 real‐time PCR system (Applied Biosystems, Waltham, MA, USA). The OsActin gene (Gene locus: LOC_Os03g50885) was used as an internal control. Three replicates were performed for each analysis. Relative expression levels of the examined genes were determined according to the 2−∆∆ method (Livak and Schmittgen 2001).

GUS staining and subcellular localization

To determine the expression pattern of OsCOMT gene, various tissues were collected from OsCOMT promoter‐GUS transgenic plants during the heading stage for GUS staining, which was performed as described previously (Yamamoto et al., 2007). Samples were observed using a DM1000 microscope (Leica, Wetzlar, Germany). For subcellular localization analyses, the full length of OsCOMT coding region was amplified and then inserted into the pAN580 and pCAMBIA1300 vector carrying the GFP respectively. These constructs were introduced into rice protoplasts (Page et al., 2019) and tobacco leaf cells (Sparkes et al., 2006) for transient expression respectively. The GFP fluorescence signals were detected using a LSM 710 confocal microscope (Zeiss, Oberkochen, Germany).

Protein expression and purification of OsCOMT in Escherichia coli

The cDNA fragment of OsCOMT was amplified and inserted into the BamHI and EcoRI digested pGEX‐6p‐1 vector using the In‐Fusion HD cloning kit (Takara biotech, Beijing, China). The pGEX‐6p‐1‐OsCOMT was then introduced into the E. coli expression strain BL21. Protein expression was induced by adding 0.4 mM isopropyl‐β‐thiogalactopyranoside (IPTG) for 18 h at 37°C. The glutathione S‐transferase‐fused (GST‐fused) OsCOMT recombinant protein was purified using a glutathione resin column (Sangon biotech, Shanghai, China) according to the manufacturer’s instructions.

Western blotting analysis

For western blotting, the purified GST‐OsCOMT protein was separated using SDS‐PAGE and then transferred to PVDF membranes to detect GST protein after electrophoresis. The procedure of western blotting was performed according to previously described (Kobayashi et al., 2019) and analysed by immunoblotting with anti‐GST antibody (Abcam, Shanghai, China).

Assays of OsCOMT enzyme activity

The N‐acetylserotonin methyltransferase (ASMT) activity of purified recombinant proteins with different concentration were determined by ELISA using an ELISA method (EIAab Science Co., Ltd., Hubei, China). Meanwhile, the purified recombinant proteins were incubated using NAS (Sigma‐Aldrich, Shanghai, China) substrate dose responses (Byeon et al., 2015). A 10 μL aliquot was subjected to high‐performance liquid chromatography (HPLC) analysis for melatonin test. Quantification of endogenous melatonin in rice was performed according to previously described (Park et al., 2013). Samples were analysed using the HPLC system (Rigol, Beijing, China).

Histochemical staining of ROS and cell death

The accumulation of the superoxide anion (O2 ‐) and H2O2 was monitored with NBT (Coolaber Biotech, Beijing, China) and DAB (Coolaber Biotech, Beijing, China) histochemical staining respectively. The staining and bleaching of the samples were performed as previously described (Ke et al., 2019). To visualize the dead cells, detached leaves were stained using a TB (Solarbio Biotech, Beijing, China) staining method as reported previously (Yin et al., 2000).

Measurements of chlorophyll contents and chlorophyll fluorescence parameters

The chlorophyll a, chlorophyll b and total chlorophyll contents were measured as described previously (Gao et al., 2018). Pn was measured using a portable LI‐6400 photosynthesis measurement system (Li‐Cor Inc., USA). Chlorophyll fluorescence parameters were determined using a portable chlorophyll fluorometer PAM‐2000 (WALZ, Germany). The effective quantum yield of photosystem‐II (Y (II)), ETR, qP and NPQ were measured as described previously (Gao et al., 2015). All experiments were repeated with three biological replicates. Each measurement was performed in the sunny day between 10:00 a.m. and 11:00 a.m. at grain filling to maturity stages.

Transmission electron microscopy

Flag leaf sections were cut into 0.5‐ to 1‐cm segments and immediately placed in precooled 2.5% glutaraldehyde in phosphate buffer (100 mM, PH 7.4) and fixed for at least 6 h. After three washes with phosphate buffer (100 mM, PH 7.4), the fixed segments were treated overnight with 1% OsO4. Then, the segments were processed as previously described, embedded in paraffin wax, mounted and observed under an HT7800 transmission electron microscope (Hitachi, Japan).

Transcriptome analysis

Extraction of RNA was performed using a TRIzol reagent (Invitrogen). A total of 3 μg RNA per sample was used to build sequencing libraries by the Illumina TruSeq RNA Sample Preparation Kit and sequenced on Illumina HiSeq 2000. The generated reads were firstly filtered and then mapped onto the reference genome of Oryza sative (MSU7.0; http://rice.plantbiology.msu.edu/) using HISAT2. The featureCounts v1.5.0‐p3 software was used to count the read numbers mapped to each gene. Fragments per Kilobase of transcript sequence per Million mapped tags (FPKM) was used for transcription level quantification.

Histological analysis of vascular bundles

The flag leaves and third internodes from the uppermost internodes were fixed in 2.5% glutaraldehyde, dehydrated in a graded ethanol series and embedded in Spurr resin. The cross‐sections of the leaves and internodes were produced using an ultramicrotome (EM UC7, Leica, Germany), and then the sections were stained with 0.5% toluidine blue (Solarbio Biotech, Beijing, China) and observed using a microscope (DM1000, Leica, Germany). The areas of vascular bundles in the sections of leaves and internodes were measured by ImageJ software (National Institutes of Health, Maryland, USA).

Quantification of endogenous cytokinin

The extraction and quantification of cytokinin and auxin were as follows: the samples (~2 g) were homogenized in 1 mL potassium phosphate buffer (0.05 M, pH 8.0). The homogenates were centrifuged at 4 °C and 3000 rpm for 5 min. Then, 500 μL of the supernatant was mixed with 25 μL of chloroform, and then the samples were horizontally shaken for 10 min. The water phase was discarded and the chloroform phase was dried under vacuum. The residues were dissolved in 120 μL of the high‐performance liquid chromatography (HPLC) mobile phase and 30 μL was injected into the HPLC‐EC system. The analysis was following a method previously described (Zhang et al., 2019).

Primers and statistical analyses

The nucleotide sequences of all primers used in this study are listed in Table S2. All data are presented as the means ± standard deviation (SD), shown by error bars. Statistical analysis was performed by Tukey’s HSD test (P < 0.05) and independent‐samples t‐test (*P < 0.05; **P < 0.01) to detect significant differences using IBM SPSS software.

Conflicts of interest

The authors declare no competing interests.

Author contributions

Z.Y., Y.Z. and C.X. conceived the idea and supervised this study. L.H., R.C., Y.L., E.Z., J.M., Z.Z., Y.Z., M.Z. and Z.Z. performed the experiments. L.H., R.C., Y.L., P.L., Y.X., G.L. and C.X. participated in the result analysis. L.H., R.C., Y.L., Y.Z. and Z.Y. wrote the manuscript.

Figure S1 Identification of the OsCOMT knockout mutants.

Figure S2 Comparison of photochemical efficiency between NIP and OsCOMT transgenic lines.

Figure S3 Plant phenotypes of NIP and OsCOMT‐KO plants before and after melatonin treatment.

Table S1 Agronomic traits of OsCOMT transgenic lines in the Nipponbare background.

Table S2 List of primers used in this study.

Click here for additional data file.