Overexpression of OsACL5 triggers environmentally‐dependent leaf rolling and reduces grain size in rice

Summary

Major polyamines include putrescine, spermidine, spermine and thermospermine, which play vital roles in growth and adaptation against environmental changes in plants. Thermospermine (T‐Spm) is synthetised by ACL5. The function of ACL5 in rice is still unknown. In this study, we used a reverse genetic strategy to investigate the biological function of OsACL5. We generated several knockout mutants by pYLCRISPR/Cas9 system and overexpressing (OE) lines of OsACL5. Interestingly, the OE plants exhibited environmentally‐dependent leaf rolling, smaller grains, lighter 1000‐grain weight and reduction in yield per plot. The area of metaxylem vessels of roots and leaves of OE plants were significantly smaller than those of WT, which possibly caused reduction in leaf water potential, resulting in leaf rolling with rise in the environmental temperature and light intensity and decrease in humidity. Additionally, the T‐Spm contents were markedly increased by over ninefold whereas the ethylene evolution was reduced in OE plants, suggesting that T‐Spm signalling pathway interacts with ethylene pathway to regulate multiple agronomic characters. Moreover, the osacl5 exhibited an increase in grain length, 1000‐grain weight, and yield per plot. OsACL5 may affect grain size via mediating the expression of OsDEP1, OsGS3 and OsGW2. Furthermore, haplotypes analysis indicated that OsACL5 plays a conserved function on regulating T‐Spm levels during the domestication of rice. Our data demonstrated that identification of OsACL5 provides a theoretical basis for understanding the physiological mechanism of T‐Spm which may play roles in triggering environmentally dependent leaf rolling; OsACL5 will be an important gene resource for molecular breeding for higher yield.

Introduction

Polyamines (PAs) are low molecular weight aliphatic amines with strong biological activity that are involved in various biological processes (Gong et al., 2014; Handa and Mattoo, 2010; Lv et al., 2021; Masson et al., 2017; Shinohara et al., 2019; Tabor and Tabor, 1984; Takahashi, 2020). In plants, the putrescine (Put, diamine), spermidine (Spd, triamine), spermine (Spm, tetraamine) and thermospermine (T‐Spm, tetraamine) are major PAs (Hao et al., 2018; Michael, 2016). Plant PAs play vital roles in various physiological processes such as embryogenesis, cell division, organogenesis, germination, flowering, fruit setting and senescence, as well as in response to environmental stresses (Fujita et al., 2012; Gerlin et al., 2021; Moschou et al., 2008; Moschou and Roubelakis‐Angelakis, 2014; Navakoudis and Kotzabasis, 2022; Rossi et al., 2021).

The levels of PAs are precisely regulated by a dynamic balance between biosynthesis and catabolism. The catabolism of PAs, mainly catalysed by copper‐dependent diamine oxidases (DAOs, EC 1.4.3.6) and FAD‐associated polyamine oxidases (PAOs, EC 1.5.3.11), has been extensively studied in several plant species (Alabdallah et al., 2017; Cervelli et al., 2004; Liu et al., 2014a; Tavladoraki et al., 2016; Wang and Liu, 2016; Zhang et al., 2022). PA biosynthetic pathways are well studied in Arabidopsis (Alcázar et al., 2010; Liu et al., 2015; Pál et al., 2021), but still unclear in rice. The biosynthesis of PAs, which begins from ornithine or arginine. Arginine is converted to Put via agmatine in three sequential reactions. Put is converted to Spd by Spd synthases (SPDS, EC 2.5.1.16), and then the Spd is further converted to Spm or T‐Spm, which is catalysed by Spm synthase (SPMS, EC 2.5.1.22) and T‐Spm synthase (ACAULIS5 or ACL5), respectively (Kusano et al., 2015; Takano et al., 2012; Yu et al., 2019). An aminopropyl group, which is transferred from the decarboxylated S‐adenosylmethionine (dcSAM) produced from methionine catabolism catalysed by methionine adenosyltransferase and S‐adenosylmethionine decarboxylase (SAMDC) in two sequential reactions, participates in the biosynthesis processes of Spd, Spm and T‐Spm (Kusano et al., 2015; Michael, 2016; Yu et al., 2019).

S‐adenosylmethionine (SAM) is also the substrate for 1‐aminocyclopropane‐1‐carboxylic acid (ACC) biosynthesis to generate ACC which is the immediate precursor of ethylene (Fluhr et al., 1996; Yu et al., 2019). The aminopropyl group serves as a common precursor for both PA and ethylene biosynthesis, suggesting a competitive relationship between PA and ethylene biosynthesis (Bitrián et al., 2012; Tao et al., 2018). Overexpression of SPDS in tomato led to higher levels of Spd than in WT plants, and the gene transcripts involved in ethylene biosynthesis and signalling were suppressed, resulting in increased susceptibility to Botrytis cinerea. And this susceptibility response was reversed after exogenous application of S‐adenosyl‐Met and 1‐aminocyclopropane‐1‐carboxylic acid (Nambeesan et al., 2012). In wheat, an increase in ACC and Put concentration and a decrease in Spd concentration under water deficit conditions resulted in a reduction in the grain‐filling rate, suggesting that the balance between Spd and ACC is important for yield (Yang et al., 2017). OsSPMS1, the first functionally characterised PA synthase gene of rice, may function as a SPMS. The phenotypic analysis of OsSPMS1 RNAi and overexpressing transgenic lines indicated that OsSPMS1 negatively regulates seed germination, seed size and grain yield per plant. ACC treatment rescued the germination defects of the overexpressing lines, which had lower ethylene contents than WT, suggesting that OsSPMS1 may regulate seed germination and plant growth by mediating the ACC and ethylene pathways (Tao et al., 2018). Taken together, the PA pathway may interact with the ethylene pathway to mediate plant growth and help plants respond to various environmental stresses.

T‐Spm, a structural isomer of Spm, is not a minor PA in plants (Kakehi et al., 2008; Takahashi, 2018). T‐Spm is synthesised by the enzyme of ACL5. The encoding gene ACL5 was found to be specifically expressed from procabial cells to xylem vessels during xylem formation in Arabidopsis (Takano et al., 2012). acl5, the loss‐of‐function mutant of ACL5, showed overproliferation of xylem vessels and severe dwarfism phenotypes, indicating that T‐Spm represses xylem differentiation (Takano et al., 2012). Additionally, the dwarf phenotype of acl5 was suppressed by a mutation in the upstream open reading frame of SAC51 which encodes a basic helix–loop–helix‐type transcription factor (Imai et al., 2006). Genetic evidence suggested that SAC52, another semi‐dominant suppressor mutant of acl5‐1, recovered the stem elongation in acl5‐1 (Imai et al., 2008; Takahashi and Kakehi, 2010). The auxin signalling stimulated xylem differentiation which was suppressed by SAC51‐mediated T‐Spm signalling, but could be continually promoted by exogenous auxin in the absence of T‐Spm (Cai et al., 2016; Yoshimoto et al., 2012). Furthermore, exogenous application of Norspermine (Nor‐Spm) to acl5 mutant partially suppressed its dwarf phenotype, suggesting that Nor‐Spm could functionally substitute for T‐Spm (Kakehi et al., 2010). Taken together, the function of ACL5 of Arabidopsis has been extensively studied and various biological functions have been revealed. However, the homologous gene of ACL5 in rice is still unknown. OsACL5, the only known homologous gene of ACL5 in rice, was predicted in the rice genome. The predicted protein of OsACL5 showed high identity with AtACL5 of Arabidopsis and SlACL5 of tomato, and other predicted ACL5 proteins from eight different plant species (Liu et al., 2018a).

Here, we identified the biological functions of OsACL5 during the rice growth and development. We demonstrated that the overexpressing transgenic plants of OsACL5 contain over ninefold higher levels of T‐Spm and exhibit environmentally dependent leaf rolling, as well as smaller seeds, lighter 1000‐grain weight, and reduction in yield per plant and plot, compared to WT. Additionally, the area of metaxylem vessels of roots and leaves of OE plants was markedly smaller than those of WT. The endogenous higher T‐Spm accumulation in OE plants reduced ethylene contents, suggesting that the T‐Spm signalling pathway interacts with the ethylene pathway to regulate multiple agronomic characters in rice.

Results

OsACL5 serves as a thermospermine synthase in vitro

To identify the thermospermine synthase in rice, first, we analysed the phylogenetic relationship based on the amino acid sequences among three model species of rice, Arabidopsis and tomato. The results showed that OsACL5 shared high identity with Arabidopsis AtACL5 (65.49%), and three tomato SlACL5 (SlACL5‐1, 66.28%; SlACL5‐2, 62.54%; and SlACL5‐3, 55.49%; Figure S1a,c). Besides, the gene structure and amino acid sequence of these ACL5 showed very similar structures (Figure S1b,c), suggesting that OsACL5 possibly serves as a thermospermine synthase in rice. Furthermore, the results of enzymatic activity in vitro showed that the T‐Spm was produced after incubating Spd and dcSAM with purified recombinant OsACL5 proteins (Figure S2), indicating that OsACL5 functions as a thermospermine synthase in rice.

Overexpression of OsACL5 affects leaf development and causes environmentally dependent adaxial leaf rolling

To investigate the biological roles of OsACL5, we used the reverse genetic approaches. The OsACL5 gene was knocked out using CRISPR‐associated protein 9 nuclease (Cas9) system, or overexpressed under ubiquitin promoter cloned from maize. Two independent CRISPR‐Cas9 knockout mutants, osacl5‐1 and osacl5‐2, were selected from 10 individual lines. In two osacl5 lines, single base (G) was inserted or deleted in these two targets which caused shift mutation, resulting in premature termination of protein translation that only code 11 or 8 amino acids in osacl5‐1 and osacl5‐2, respectively (Figure 1a). Besides, we obtained eight independent overexpressing lines. The transcriptional level of OsACL5 was increased by over 200‐fold in OE‐1 to OE‐7, except the OE‐8 which only slightly increased less than twofold (Figure 1b). Two OE lines, OE‐1 and OE‐2, were used in this study. Interestingly, the OE plants showed adaxially rolled leaves when the temperature was rising up and light intensity was increasing from 8 am to 2 pm since about 40‐day‐old plants in the rice growing season. Especially, OE plants showed shallot‐like rolled leaves at midday at 2‐month‐old stage, while the osacl5 mutants did not show significant difference, compared to WT (Figure 1c–f,i). The leaf length and width of OE plants were significantly shorter or smaller than those of WT, while the osacl5 mutants did not show significant difference (Figure 1e–h). The leaf‐rolling index (LRI) was used to quantify the extent of leaf rolling. The LRI values of OE1 and OE2 were 6.2% and 18.8% at 8 am, which sharply rose to 36.9% and 71.8% at 2 pm, respectively; while the WT and osacl5 mutants only slightly increased from around 2% at 8 am to 5% at 2 pm (Figure 1j).

Figure 1.

Phenotypic characterization of overexpressing transgenic plants of OsACL5. (a) pYLCRISPR/Cas9 system‐mediated genome editing on OsACL5. The structure of OsACL5, and two target sites in the first exon in genomic sequence. Black boxes represent exons, grey lines between exons represent introns, and white boxes represent UTRs. The targeting sequences were highlighted in blue followed by the PAM sequences (AGG). The inserted single base was highlighted in red and the deletion base was shown as dot in red. (b) Relative expression level of OsACL5 in eight independent overexpressing (OE) lines, compared to that in WT (Zhonghua11, ZH11). Morphology of 60‐day‐old wild type, OE and osacl5 plants. Photos were taken at 8 am (temperature, 30.1 °C; humidity, 65%; light intensity, 460 μmol m⁻² s⁻¹) (c) and 2 pm (temperature, 40.1 °C; humidity, 32%; light intensity, 1230 μmol m⁻² s⁻¹) (d). The phenotypes of the blade leaves. At 8 am (e) and 2 pm (f). (g) Leaf length. (h) Leaf width. (i) Transverse sections at the middle of the leaves at 8 am and 2 pm. (j) Leaf rolling index (LRI). Cross sections of leaves in WT and OE lines at maturity. Midribs (k, l) and large veins (m, n) of leaves. (o) Metaxylem area of midribs and large veins indicated with red arrows. OE‐1 and OE‐2 represent two independent overexpressing transgenic lines, respectively. osacl5‐1 and osacl5‐2 represent two independent CRISPR/Cas9‐meditated knockout lines, respectively. Data are presented as mean values of three biological replicates with SD. (n > 40 for c to j; n > 8 for k to o). Significance of data is tested by Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). Scale bars: (c, d) 10 cm; (e, f) 5 cm; (i) 2 mm; (k, l) 100 μm; (m, n) 50 μm.

Generally, leaf rolling is regulated by the large, bubble‐shaped bulliform cells (BCs) arranged in groups between vascular bundles on adaxial epidermis of the leaf blade. Thus, we speculated that the morphology of BCs in OE plants might be altered. Unexpectedly, we could not find significant differences in the size and number of BCs of midribs, large and small veins of the second leaf from the top between WT and OE plants at three detected time points (8 am, 11 am, and 2 pm) (Figure S3a–i). Besides, we found the layers and numbers of sieve tube in large vein of OE plants were also unchanged compared to those of WT (Figure S3j–k). What's more, the phenotype of rolling leaf in OE plants suggested that leaf development or morphogenesis might be altered. We thus performed histological assay of the 6‐day‐old shoots (unexpanded young leaves) and the second leaf from the top of WT, OE and osacl5 plants at 2‐month‐old stage, respectively. The observation of the cross‐sections of 6‐day‐old shoots revealed changes in OE plants, including more underdeveloped vascular bundles compared to those of WT (Figure S4). Furthermore, examination of the cross‐sections of the second leaf from the top revealed that the area of metaxylem vessels of midribs and large veins of OE plants were markedly smaller than those of WT (Figure 1k–o). Taken together, the smaller metaxylem in midribs and large veins might be the reason for leaf rolling.

Overexpression of OsACL5 mediates temperature/light‐intensity/humidity dependent rolling leaf

To investigate the exact environmental factor for the phenotype of rolling leaf, we analysed the effect of temperature, light intensity and humidity. First, we set growth conditions with humidity 65% under light intensity 500 μmol m⁻² s⁻¹, with a series of temperature gradients. The results displayed that the LRI of OE plants were increased from around 8% at 25 °C to 18% at 40 °C (Figure 2a–d). Next, we set the growth conditions with humidity 65% at 25 °C, the LRI of OE plants were increased from around 8% to 15% when the light intensity increased from 500 to 1000 μmol m⁻² s⁻¹, respectively (Figure 2a,e,f). Besides, the LRI of OE plants were increased from around 8% to 14% when the humidity reduced from 65% to 35%, under light intensity 500 μmol m⁻² s⁻¹ at 25 °C (Figure 2a,g,h). Finally, in order to know whether the temperature, light intensity and humidity cooperatively act on causing rolling leaf, we designed three different experimental groups. The results showed that the LRI of OE plants at the conditions of light intensity 1000 μmol m⁻² s⁻¹, humidity 35% at 25 °C is lower than that at light intensity 1000 μmol m⁻² s⁻¹, humidity 65% at 40 °C (Figure 2i,j). Especially, the LRI of OE plants was sharply increased to about 65% at the conditions of light intensity 1000 μmol m⁻² s⁻¹, humidity 35% at 40 °C (Figure 2k). On the contrary, the LRI of WT only slightly increased with the increase in light intensity and temperature, or the decrease in humidity, while the LRI of osacl5 mutants more or less increased but did not show obvious difference like OE plants (Figure 2). Taken together, high temperature and light intensity, and low humidity cooperatively regulate rolling leaf in OE plants, and the temperature seems to be the main factor.

Figure 2.

Overexpression of OsACL5 exhibited temperature/light‐intensity/humility dependent rolling leaf. Measurement of the LRI under the condition of humidity 65% and light intensity 500 μmol m⁻² s⁻¹ at 25 °C (a), 30 °C (b), 35 °C (c) and 40 °C (d). Measurement of the LRI under the condition of humidity 65%, at 25 °C, under light intensity 750 μmol m⁻² s⁻¹ (e) and 1000 μmol m⁻² s⁻¹ (f). Measurement of the LRI under the condition of light intensity 500 μmol m⁻² s⁻¹ at 25 °C, with humidity 50% (g) and humidity 35% (h). (i) Measurement of the LRI under the condition of light intensity 1000 μmol m⁻² s⁻¹, humidity 35% at 25 °C. (j) Measurement of the LRI under the condition of light intensity 1000 μmol m⁻² s⁻¹, humidity 65% at 40 °C. (k) Measurement of the LRI under the condition of light intensity 1000 μmol m⁻² s⁻¹, humidity 35% at 40 °C. Data are presented as mean values of three biological replicates with SD. Significance of data is tested by Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001).

To further investigate the underling mechanisms in which overexpressing OsACL5 affects leaf rolling, the RNA‐seq analysis was performed. In detail, the blade leaves were sampled from OE, mutant and WT plants under midday conditions (temperature, 40 °C; humidity, 35%; light intensity, 1000 μmol m⁻² s⁻¹), compared to the morning conditions (temperature, 25 °C; humidity, 65%; light intensity, 500 μmol m⁻² s⁻¹). The data showed that 775 DEGs were upregulated under midday conditions in OE lines (Figure S5a,b), and those DEGs were mainly related to metabolic pathway and biosynthesis of secondary metabolites by KEGG enrichment analysis (Figure S5c), suggesting that a lot of genes were upregulated under midday conditions. In addition, we found that two known leaf development‐related genes, OsRL14 and OsZFP7, were significantly downregulated both in RNA‐seq and qRT‐PCR analyses (Figure S5d,e). rl14‐1, a knockout mutant of OsRL14 (Rolling‐leaf14), showed incurved leaves on the adaxial side in rice (Fang et al., 2012). When OsZFP7, a C2H2 transcription factor, was overexpressed, the transgenic plants exhibited curling leaves (Liu et al., 2018b). It suggested that overexpression of OsACL5 caused environmentally dependent leaf rolling possible via mediating the expression of OsRL14 and OsZFP7.

Manipulation of OsACL5 affects plant architecture and metaxylem vessels

To further investigate the role of OsACL5 in rice, we analysed the plant architecture of OE plants and osacl5 mutants. From the seedling to mature stages, the OE plants were shorter than WT, conversely, osacl5 mutant lines were taller than WT (Figure 3a,c; Figure S6). Besides, the length of panicle of OE lines was markedly shorter than WT, but that of the osacl5 mutants did not show significant difference compared to that of WT (Figure 3b,d). To investigate the elongation pattern of internodes, we analysed the lengths of the three uppermost internodes among WT, OE and osacl5 plants. The first and the third internodes from the top of OE lines were significantly shorter than those of WT, while the first internodes from the top of osacl5 mutants were longer than those of WT (Figure 3f). Additionally, we also compared the culm size of the three uppermost internodes. The culms of OE lines became thinner and the diameter became smaller than those of WT, while the knockout mutants having thicker and bigger culms in the uppermost two internodes (Figure 3e,g). To further investigate the developmental pattern of culm vessels, the first internodes from the top were analysed by paraffin cross section. The observation showed no significant difference among WT, OE and osacl5 plants, suggesting that OsACL5 does not affect the developmental pattern of culm vessels (Figure 3h–l). acl5 from Arabidopsis exhibited overproliferation of xylem vessels in roots and severe dwarfism phenotypes (Takano et al., 2012). We examined the development of root vascular. The observation of the cross‐sections of 6‐day‐old roots showed that the size and number of metaxylem vessels in OE roots were markedly smaller and reduced respectively, compared to those of WT (Figure 3m–t). Taken together, overexpression of OsACL5 affected the development of culms, and metaxylem vessels of roots.

Figure 3.

Overexpression of OsACL5 affected plant architecture and metaxylem vessels. (a) Plant architecture. (b) Panicle phenotype. (c) Plant height at mature stage. (d) Panicle length. (e) Cross sections by hand of the main culms. I‐III indicates the first, second and third internodes from the top, respectively. (f) The length of the internodes. (g) The diameter of the internodes. (n ≥ 21 for a to g). (h–l) Paraffin cross sections of the first internodes from the top, n ≥ 8. Magnification of the boxed area was displayed. (m–o) Paraffin cross sections of the 5‐day‐old roots, n ≥ 8. (p–r) Magnification of the boxed area in (m–o), respectively. Red stars indicate the metaxylem vessels. The total area of metaxylem (s) and the number of metaxylem vessels (t) in the image of p to r. Data are presented as mean values of three biological replicates with SD. Significance of data is tested by Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). Scale bars: (a) 10 cm; (b) 2 cm; (e) 1 mm; (h–l) 100, and 20 μm in the magnified images; (m–o) 50 μm; (p–r) 10 μm.

Overexpression of OsACL5 increased the stomatal density leading to reduction of leaf water potential

To further investigate the reason for leaf rolling, next we examined the stomatal density and size. The observation revealed that the stomatal density of OE plants was increased compared to that of WT and osacl5 (Figure S7a,c), whereas the stomatal size was not affected (Figure S7b). We also analysed the water potential. As expected, the OE‐1 showed markedly reduced water potential in leaves from morning to afternoon, while that of the WT and osacl5‐2 was not significantly changed (Figure S7d). Generally, the stomatal conductance and transpiration rate of leaves are closely related to water transport. Therefore, we analysed these two physiological indexes. Unexpectedly, the stomatal conductance and transpiration rate were not significantly changed from the morning to afternoon in WT, OE‐1 and osacl5‐2 plants (Figure S7e,f). Taken together, these data suggested that overexpression of OsACL5 caused an increment of stomatal density leading to the reduction of leaf water potential, subsequently resulting in leaf rolling when the environmental temperature was raised.

OsACL5 negatively regulates grain size

Grain size is a key factor for determining grain yield and is a target trait for genetic engineering and molecular breeding. The OE lines showed shorter panicle length than WT (Figure 3b,d). To investigate whether the OsACL5 affects the yield, we analysed the grain size. OsACL5 overexpressing lines exhibited obviously smaller grains. The grain length of OE‐1 and OE‐2, was significantly reduced by 12.3% and 10.0%, respectively, and the width reduced by around 2% (Figure 4a,b,h,i). In contrast, the grain length of osacl5‐1 and osacl5‐2 mutants was increased by 1.3% and 2.7%, respectively (Figure 4a,h), but their grain width was unchanged (Figure 4b,i). The grains after removing the glumes showed similar phenotype with seeds (Figure S8). As a result, the 1000‐grain weight of OE‐1 and OE‐2 reduced by 14% and 11%, respectively, and those of the osacl5‐2 increased by 2.1% (Figure 4j), compared to those of WT. The pie chart of 1000‐grain of OE lines showed visible differences from those of WT and osacl5 plants in size (Figure 4c–g). Besides, the data of yield per plant and per plot showed that the overexpression or knockout of OsACL5 resulted in reduce or increase in yield under field conditions (Figure 4k,l). In detail, the yields per plot were increased by 12.58% or 8.89% in osacl5‐1 and osacl5‐2, respectively. On the contrary, the yields per plot were significantly decreased by 27.38% or 23.08% in OE‐1 and OE‐2, respectively (Figure 4l). Furthermore, to reveal the possible reason for grain size alteration, the glume outer surfaces of the mature grains were observed by scanning electron microscopy (SEM; Figure 4m–o). The observation indicated that the cell length and width were significantly reduced in OE lines (Figure 4p–t). The cell length was increased in osacl5‐2 mutant, but the cell width was unchanged (Figure 4p–t). There was no significant difference in cell number in the longitudinal direction of glume outer surfaces (Figure 4u).

Figure 4.

OsACL5 affected seed size. (a, b) Seed morphology of WT, OE, and osacl5 plants. (c–g) The pie chart of 1000 seeds. (h) The length of grain. (i) The width of grain. (j) The weight of 1000 grains. (k) Yield per plant. (l) Yield per plot. The density was 25 cm × 25 cm, with one plant per hill. The area per plot was 5.06 m². (m–o) Scanning electron microscopy (SEM) observation of the glume outer surfaces of the mature grains. Images are generated by two field photographs (using the Photoshop6 software overlap function). (p–r) Magnification of the boxed area in (m–o), respectively. (s–u) Analysis of cell length, width and number in the longitudinal direction in (p–r). (v–x) Relative expression of OsDEP1, OsGS3, and OsGW2 in young seeds. Data are presented as mean values of three biological replicates with SD. (n ≥ 50 for a, b, h and i; n ≥ 5 for c–g, and j; n ≥ 30 for k; n = 5 for l; n ≥ 10 for m–r). Significance of data is tested by Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). Scale bars: (a, b) 5 mm; (c–g) 2 cm; (m–o) 400 μm; (p–r) 100 μm.

In order to investigate the possible mechanism of how OsACL5 functions on grain size, we examined the expression of five known size‐related genes in grains. The results showed that the expression of OsDEP1, OsGS3, and OsGW2 was markedly increased in OE‐1, whereas slightly reduced or unchanged in osacl5‐2 (Figure 4v–x). These three genes are major QTLs that play negative roles in grain size (Fan et al., 2006; Huang et al., 2009). Whereas, the expression levels of other two detected size‐related genes, OsGSN1 and OsGW8, were not significantly changed (Figure S9). It suggested that OsACL5 may affect grain size via mediating the expression of a set of size‐related genes.

Expression pattern and subcellular localisation of OsACL5

To examine the expression pattern of OsACL5, we performed quantitative RT‐PCR (qRT‐PCR) and promoter‐GUS reporter gene assay in various tissues at different stages. The qRT‐PCR assay indicated that OsACL5 was expressed in all detected tissues, especially in roots, stems and leaves (Figure 5a). Besides, to further verify the quantitative data, we assayed the GUS (β‐glucuronidase) activity in several OsACL5 _pro :GUS transgenic lines. OsACL5 _pro :GUS was expressed in the roots of 3‐day‐old seedlings (Figure 5b1), embryos of both germinating seeds (Figure 5b2) or fresh mature seeds (Figure 5b3), leaf sheath (Figure 5b4), coleoptile and radicle (Figure 5b5, 6). OsACL5 was also expressed in seeds during grouting period (Figure S10). Especially, GUS activities were detected in the stele of the main roots and lateral roots (Figure 5b, 7–9), as well as in the root hairs (Figure 5b10). Besides, OsACL5 expression was also observed in floral meristem and floral organ primordial of different stages (Figure 5b11, 12), the vascular systems of young flower (Figure 5b13) and the vascular bundles of mature stems (Figure 5b14) in the GUS transgenic plants. These observations indicated that OsACL5 revealed more intense expression in vasculature tissue, suggesting that OsACL5 possibly play important roles in vascular development in rice.

Figure 5.

Expression pattern of OsACL5. (a) Relative expression level of OsACL5 in different tissues. (b) GUS staining of tissues from OsACL5 _pro :GUS transgenic plants. 3‐day‐old seedling (b1), germinating seed (b2), fresh mature seed (b3), leaf sheath (b4), coleoptile (b5), radicle (b6), stele of the main root (b7), lateral root (b8), stele of the lateral root (b9), root hairs (b10), floral meristem and floral organ primordia (b11, 12), young floret (b13), and cross section of mature stem (b14). (c) Subcellular localisation of OsACL5‐eGFP in rice protoplast cells. Strong green fluorescent signals were visualized in cytoplasm, nuclei and plasma membrane. (d) Empty vector 35S:eGFP was used as control. (e) OsACL5‐eGFP and IAA17‐RFP co‐localised in nuclei. (f) OsACL5‐eGFP and mCherry‐OsRAC3 colocalised in plasma membrane. Scale bars: (b1–b4, b14) 1 mm; (b5–b13) 20 μm; (c–f) 10 μm.

To investigate the subcellular localisation of OsACL5, we generated the 35S:OsACL5‐eGFP plasmid and transiently expressed it in the protoplasts of rice. Observation of GFP fluorescence showed that OsACL5‐eGFP localised at nucleus, cytoplasm and plasma membrane (PM; Figure 5c), in comparison with the 35S:eGFP control which predominantly localised at nucleus and cytoplasm (Figure 5d). Furthermore, the 35S:OsACL5‐eGFP and 35S:IAA17‐RFP (a nuclear localisation marker) co‐localised at nucleus (Figure 5e); meanwhile, the 35S:OsACL5‐eGFP and 35S:mCherry:OsRAC3 (a PM localisation marker) co‐localised at PM (Figure 5f), respectively. Similarly, OsSPMS1‐GFP, a Spm synthase gene fused with a GFP tag, subcellular localised in the nucleus, PM and cytoplasm (Tao et al., 2018). Taken together, these results indicated that OsACL5 was ubiquitously expressed in various plant tissues and widely localised in the nucleus, cytoplasm and PM, suggesting it possibly play multiple roles in rice.

Natural variations in OsACL5 among different rice germplasm resources

We investigated the natural variation in OsACL5 DNA sequence across over 4726 different rice accessions, based on the online database of RiceVarMap v2.0 (Zhao et al., 2021). We performed the haplotype analysis by using these germplasm resources. There were seven haplotypes in the coding region and untranslated region (UTR) of OsACL5 (Figure 6a). Hap1 and Hap4 were mainly found in Japonica and Aus, respectively. While Hap2 was mainly found in Indica. Hap5 and Hap6 were Japonica‐ and Indica‐specific haplotypes (Figure 6b). Hap1 and Hap2 were predominantly found in Japonica and Indica, and other five haplotypes only with low proportion in rice (Figure 6b), suggesting that OsACL5 was selected during the domestication of rice. To further investigate whether the haplotypes differences of OsACL5 in Japonica and Indica affect the PAs levels, we examined the PAs levels among four Japonica and three Indica species. The results indicated that, in general, the levels of Put and Spd were higher in Japonica than in Indica, and the Spm levels did not change among these detected species. Especially, the T‐Spm levels almost keep at the same very low level in all of these seven species (Figure 6c,d). Taken together, these observations suggested that OsACL5 plays a conserved function on regulating T‐Spm levels during the domestication of rice.

Figure 6.

Haplotype analysis of OsACL5, and the PA measurement among Japonica and Indica species. (a) Haplotypes of OsACL5 and haplotype frequency in natural population based on genomic variations. (b) Haplotype network of OsACL5. Size of circles indicates proportional for a given haplotype. The frequencies for different species were displayed by colour. The contents of PAs (c) and T‐Spm (d) among four Japonica and three Indica rice species. The bars in Figure 6d were magnified by 30 times in ordinates.

OsACL5 regulates PA biosynthesis and affects ethylene biosynthesis

In plants, Put is converted to Spd by Spd synthases, and then further converted to Spm or T‐Spm catalysed by Spm or T‐Spm synthase, respectively (Michael, 2016; Takano et al., 2012). To verify whether OsACL5 regulates the PA biosynthesis, we detected the PA contents using the shoots and leaves of 2‐week‐old seedlings. Overexpression of OsACL5 resulted in obvious increase in T‐Spm contents (raised by 13 times in OE‐1 and nine times in OE‐2; Figure 7a), indicating that the OsACL5 indeed serves as a T‐Spm synthase in rice. Whereas, it was confusing that there was no significant reduction in T‐Spm contents in osacl5 mutants compared to those of WT (Figure 7a). Besides, the Spd contents reduced by 55.18% in OE‐1 and by 36.06% in OE‐2, respectively. Meanwhile, the Spm contents reduced by 61.26% in OE‐1 and by 59.44% in OE‐2, respectively (Figure 7a). Besides, the T‐Spm contents in OE leaves significantly increased under high temperature, high light intensity and low humidity (Figure S11c), suggesting that T‐Spm may play a role in response to environmental stress. Aminopropyl group serves as the common precursor for PA and ethylene, implying a competitive relationship between the PA and ethylene biosynthesis (Bitrián et al., 2012; Tao et al., 2018; Yu et al., 2019). To investigate the link between PA and ethylene biosynthesis in OE and osacl5 plants, first we examined the transcriptional levels of two key ethylene biosynthesis genes, OsACS2 and OsACS6. The results indicated that both of OsACS2 and OsACS6 were significantly downregulated in OE plants while upregulated in osacl5 mutants (Figure 7b). Next, we measured the ethylene evolution in fresh mature seeds. The observation indicated that the ethylene evolution was markedly reduced in OE plants, whereas no significant increment can be seen in osacl5 mutants (Figure 7c). Taken together, OsACL5 regulates PA biosynthesis and affects ethylene biosynthesis by a competing way, which therefore regulates the development of leaf and veil system, as well as seed size and yield as shown in the proposed working model (Figure 7d).

Figure 7.

OsACL5 regulates leaf development, plant height and grain size in rice by mediating PA biosynthesis and affecting ethylene biosynthesis. (a) Contents of Put, Spd, T‐Spm and Spm in WT, OE and osacl5 plants. (b) Expression level of ethylene biosynthesis genes. (c) Ethylene evolution. FW, Fresh weight. Data are presented as mean values of three biological replicates with SD. Significance of data is tested by Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). (d) Proposed working model of OsACL5 in regulating leaf development, plant height and grain size in rice. The green arrows indicate the biosynthesis pathway of PA and ethylene, and the red arrows indicate the back‐conversion pathway of PA. The blue arrows indicate PA and ethylene biosynthesis competing for the common precursor of dcSAM. Temperature/light/humidity and NO signals are involved in those physiological processes. 500 indicates light intensity (μmol m⁻² s⁻¹), and 65% indicates humidity. ADC, arginine decarboxylase; AIH, agmatine iminohydrolase; CPA, N‐carbamoylputrescine amidohydrolase; SPDS, Spd synthase; SPMS, Spm synthase; PAO, polyamine oxidase; SAM, S‐adenosylmethionine; SAMS, SAM synthase; SAMDC, SAM decarboxylase; dcSAM, decarboxylated S‐adenosylmethionine; ACC, 1‐aminocyclopropane‐1‐carboxylic acid; ACS, ACC synthase; ACO, ACC oxidase.

Discussion

T‐Spm is generated by the action of T‐Spm synthase encoded by ACAULISS (ACL5) in plants (Takahashi, 2018). In this study, we used a reverse genetic strategy to elucidate the role of OsACL5. Interestingly, the OE plants showed significantly adaxially rolled leaves in an environmentally‐dependent manner. In detail, from the cool morning to the hot midday in the rice growing season, the leaves became more and more rolling with the temperature and light intensity increasing and humidity decreasing, especially the leaves rolled like shallot shape at midday since around 2‐month‐old stage after germination (Figure 1c–f), and the rolled leaves will recover to almost flat in the next morning. The significantly higher LRI quantified the extent of leaf rolling (Figure 1j). The photo‐sensitive leaf rolling 1 (psl1) mutant exhibited rolling leaves in response to high light intensity and low humidity during midday (Zhang et al., 2021). Further investigation indicated that the higher temperature and light intensity, and lower humidity cooperatively regulate leaf rolling in OE plants, and the high temperature possibly is the main reason for this biological phenomenon (Figures 2 and 7d).

Several studies found that the bulliform cells (BCs) play crucial roles on leaf rolling. The srl1‐1 and srl1‐2, loss‐of‐function of SEMI‐ROLLED LEAF1 (SRL1), showed adaxially rolled leaves due to the increased numbers of BCs at the adaxial cell layers (Xiang et al., 2012). Considering the similar phenotype of adaxially rolled leaves in OE lines, we performed cross section analysis of the second leaf from the top. Unexpectedly, we could not find significant difference in the number and size of BCs between OE and WT plants (Figure S3), suggesting that the phenotype of leaf rolling in OE plants was not caused by the morphological changes in BCs. Similarly, the BCs number and size in cld1, a leaf rolling mutant, showed no significant difference compared to those of WT (Li et al., 2017). However, we found that the morphogenesis of the 6‐day‐old young shoots was altered in OE plants, and they showed more underdeveloped vascular bundles compared to those of WT (Figure S4). The cross sections of psl1 young leaves showed underdeveloped vascular bundles that are related to leaf rolling (Zhang et al., 2021). Especially the markedly reduced size in metaxylem of midribs and large veins in OE leaves (Figure 1k–o), suggesting that the dysfunctional defects in veins in OE plants is the possible reason for leaf rolling. On the contrary, thickvein (tkv), a ACL5 allele mutant, developed thicker veins in leaves and in inflorescence stems in Arabidopsis. The defect in auxin transport led to the abnormal vascular phenotypes (Clay and Nelson, 2005). What's more, our observation showed that the size and number of the metaxylem vessels in OE roots were significantly reduced compared to those of WT (Figure 3m–t). Miyamoto et al., reported that the expansion of metaxylem vessel in rice root was repressed under exogenous T‐Spm treatment (Miyamoto et al., 2019). Their findings were in accordance with our results that OE lines with higher T‐Spm level exhibit smaller metaxylem vessels.

Leaf rolling is a concomitant response of increased stomatal resistance to reduce leaf water potential. The observations exhibited that the stomatal density was increased in OE plants may result in markedly reduced water potential in leaves at midday (Figure S7a,c,d). The cld1 defected in leaf epidermis which reduced the water‐retaining capacity and caused water deficits in leaves, which contributed to the main reason of leaf rolling (Li et al., 2017). However, the stomatal conductance and transpiration rate of OE plants did not show significant difference from WT (Figure S7e,f). This might be due to there were no changes in stomatal size in OE leaves (Figure S7b). These observations suggested that the smaller vein system in OE plants reduced the ability of water transport resulting in leaf rolling. Besides, RNA‐seq analysis showed that 775 DEGs were upregulated under midday conditions in OE lines and those DEGs were mainly related to metabolic pathway and biosynthesis of secondary metabolites (Figure S5a–c). OsRL14 and OsZFP7 were significantly downregulated in OE‐1 at midday (Figure S5d,e). rl14‐1 showed incurved leaves on the adaxial side and OsZFP7 overexpression lines showed curling leaves (Fang et al., 2012; Liu et al., 2018b). It suggested that downregulation of OsRL14 and OsZFP7 might be another reason for the phenotype of rolling leaf in OE plants.

In OE plants, the T‐Spm contents were markedly increased over nine times, meanwhile, the Spd contents were decreased possibly due to the consumption for compounding higher T‐Spm contents (Figure 7a). Simultaneously, the Spm contents decreased possibly due to the weaker ability of competition of Spm synthase than that of T‐Spm synthase for the common precursor of Spd in OE plants (Figure 7a). In addition, the T‐Spm contents markedly increased under high temperature, high light intensity and low humidity in the OE leaves (Figure S11), suggesting that T‐Spm may be involved in response to stress. In consideration of the biosynthesis of PAs (Spd, Spm and T‐Spm), it requires for the common precursor of SAM with ethylene biosynthesis (Bitrián et al., 2012; Tao et al., 2018). We detected the ethylene evolution as expected, the OE plants had lower ethylene evolution compared to the WT and osacl5 mutants (Figure 7c), that is in accordance with the downregulated transcriptional levels of two key ethylene synthase genes OsACS2 and OsACS6 (Figure 7b). We also noticed that the relative expression of OsACS2 and OsACS6 were increased but the ethylene evolution did not significantly upregulate in osacl5 mutants, the possible explanation is that the relative expression levels of six detected ethylene oxidase OsACO (OsACO1, 2, 3, 4, 5, 7) were downregulated in osacl5 (Figure S12). Above all, our data indicated that the OE plants with high contents of T‐Spm is due to the markedly increased contents of OsACL5 enzyme, thus enhancing the ability of competition for SAM, that results in the decrement in ethylene evolution. It is suggested that a fine cross‐network effect existed in rice to regulate the balance of PA and ethylene pathways, subsequently co‐regulated the development of leaf and veil system, as well as seed size and yield in rice as shown in the proposed working model (Figure 7d).

In addition, we found osacl5 did not significantly decrease in the contents of T‐Spm (Figure 7a). Tao et al., reported that overexpression of OsSPMS1 did not promote the accumulation of Spm (Tao et al., 2018). T‐Spm can be back‐converted to Spd catalysed by OsPAO1 in vitro (Liu et al., 2014a), suggesting that OsPAO1 may be involved in the dynamic balance of T‐Spm in plants. Thus, the intracellular PA contents including T‐Spm might not only be determined by the activity of a certain enzyme but also affected by metabolism or other unknown mechanisms.

PAs can induce NO release, and PAs directly or indirectly act through interaction with signalling molecules (H₂O₂, NO) and phytohormones (Kamiab et al., 2020; Tun et al., 2006). Our results indicated that all PAs (Put, Spd, T‐Spm and Spm) can induce NO release (Figure S13), similar to Yang's findings that PAs can sufficiently induce NO release in soybean cotyledon node callus (Yang et al., 2014). We also observed that the OE lines with higher NO production compared to WT and osacl5 plants (Figure S14), implying that higher T‐Spm levels may result in more NO release. Taken together, PAs and NO may co‐regulate environmentally‐dependent leaf rolling and grain size (Figure 7d), while more studies are required to reveal the underlying mechanism.

The reports on how PA biosynthesis affects plant architecture and yield in crops is still limited. OsSPMS1 negatively regulated plant height, grain number and size, and yield in rice (Tao et al., 2018). In Arabidopsis, acl5 showed overproliferation of xylem vessels with severe dwarf phenotype (Kakehi et al., 2008; Takano et al., 2012). The loss‐of‐function mutant pao5 contains twofold higher T‐Spm levels and exhibits growth defects (Kim et al., 2014). In this study, we found that overexpression of OsACL5 reduces yield while knockout of OsACL5 increases yield (Figure 4k,l). Though the photosynthesis rate and the chlorophyll contents were unchanged, the area of blade leaf of OE‐1 was markedly smaller than that of WT and osacl5‐2 (Figure S15), suggesting that the total of photosynthetic products in OE‐1 maybe much less than that in WT and osacl5‐2. Besides, the results of SEM indicated that the cell size of the glume outer surfaces of the mature grains was reduced in OE‐1 and increased in osacl5‐2 (Figure 4s,t). In addition, the expression levels of OsDEP1, OsGS3 and OsGW2 were upregulated (Figure 4v–x). These three genes are major QTLs that play negative roles in grain size (Fan et al., 2006; Huang et al., 2009; Song et al., 2007). It suggested that OsACL5 may affect grain size via mediating the expression of a set of size‐related genes. Furthermore, the haplotypes analysis indicated that OsACL5 is selected during Indica‐Japonica differentiation, and OsACL5 plays a conserved function on regulating T‐Spm levels during the domestication of rice (Figure 6).

In conclusion, we found that overexpression of OsACL5 with over nine‐time higher T‐Spm contents exhibits smaller seeds, smaller metaxylem vessels, and the higher T‐Spm triggers environmentally‐dependent leaf rolling; meanwhile, the knockout mutant osacl5 showed increment in the grain length, 1000‐grain weight and yield, suggesting that OsACL5 may be an important target gene for rice molecular breeding. Manipulation operates the T‐Spm contents at a certain level by genetic engineering, which might be a new strategy to enhance grain yield.

Materials and methods

Plant materials and growth conditions

Rice plants (Oryza sativa cv. Zhonghua 11 (ZH11), and the transgenic plants were grown in a greenhouse under 14‐h‐light long‐day conditions at 28 °C day/light cycles, or grown under normal field conditions in the rice growing season at South China Botanical Garden, Chinese Academy of Sciences, Guangzhou.

Plasmids construction and plant transformation

To generate the OsACL5pro:GUS construct, a 2695 bp OsACL5 promoter sequence was inserted into Pcambia1305.1 at EcoR I and Nco I sites. To construct the overexpression vector, the full length coding sequence of OsACL5 was cloned into a POX vector with an ubiquitin promoter cloned from maize (Cornejo et al., 1993) at Kpn I and BamH I sites. To generate the gene editing constructs, two appropriate targeting sites were genome edited by a CRISPR/Cas9 vector system provided by Dr. Yao‐Guang Liu (Ma et al., 2016). The generated binary constructs were introduced into Agrobacterium tumefacines strain EHA105 and were used for transformation as described protocol (Li et al., 2011). The primers for generating the constructs were listed in Table S1.

qRT‐PCR analysis

To analyse gene expression, total RNA was extracted from 2‐week‐old seedlings using the RNeasy plant min kit (Qiagen). Genomic DNA contamination was removed by RNase‐free DNase I treatment. Next, the first‐strand cDNA was synthesised with Primescript cDNA synthesis kit (TAKARA) and oligo (dT). Finally, the relative transcriptional level was assayed using synthesised cDNAs and the primers.

Cytological observation and microscopy

The first internodes of stem from the top at elongation stage and the 5‐day‐old roots were fixed, respectively, in a fixative solution, followed by a dehydration series with ethanol and infiltration with xylene. Next, the tissues were embedded in paraffin and then were sliced into about 8 μm sections as described previously (Huang et al., 2018). For vibrating section, middle parts of the second leaf from the top were immediately embedded in 4% (m/v) agarose. Samples were sliced into 25 μm thick with a vibratome (Leica VT1000S) and stained with 0.05% toluidine blue. Samples were observed with an Olympus BX51 microscope. For SEM analysis, the mature grains were critical‐point dried and coated with gold as described previously (Liu et al., 2022a), and SEM images were taken by a scanning electron microscope (EVO MA15, Zeiss).

Stomatal observation

The analysis of leaves' stomatal density and size were performed using 2‐week‐old seedlings as previously described (Schuler et al., 2018). Samples were collected from the middle parts of the second leaf from the top, and then samples were cleared in a Hoyer's solution and HCG solution as described (Chen et al., 2011). Finally, stomatal density was counted between two neighbouring small veins in an area of 0.06 mm². Stomatal sizes were defined as stomatal height multiplied by stomatal width (μm²), and were measured using ImageJ software.

Measurement of stomatal conductance, transpiration rate and chlorophyll content

Three 2‐month‐old leaves (the second leaf from the top) were used for stomatal conductance and transpiration rate analysis, the measurement was conducted as described previously (He et al., 2021). The light intensity settings for the system were 800 μmol m⁻² s⁻¹ (am) and 1400 μmol m⁻² s⁻¹ (pm). The contents of chlorophyll were measured as described previously (Chazaux et al., 2022).

Measurement of leaf water potential and LRI

For measuring the leaf water potential, the second leaf from the top of 2‐month‐old plants were used by a WP4C dewpoint potentiometer (Gene, Hong Kong, China). For measuring the LRI, 2‐month‐old plants were used. The measurements were as follows: expanded the second leaf from the top and determined the greatest width of the leaf (Lw); at the same site, measured the natural distance of the leaf margins (Ln); the LRI (%) = (Lw − Ln)/Lw × 100 as described previously (Shi et al., 2007).

Subcellular localisation assay in rice protoplasts

The full‐length of OsACL5 (Os02g0237100) coding sequence was amplified by specific primers containing Xba I site, and inserted into the Xba I site of 35S:eGFP vector through Multis One Step Cloning Kit (#cat: C113‐02, VazymE). Rice protoplasts isolation and transfection were performed as described previously (Liu et al., 2022b), and then observed by microscope.

Measurement of PA by HPLC and ethylene by GC

To analyse the PA contents, the plant samples of WT, OE, osacl5 were collected. 0.3–0.5 g samples per plant were pulverised and the PAs were extracted following a previous method (Liu et al., 2014a). The benozylated PAs were assayed with a programmable Agilent HPLC using a reverse‐phase column (ZORBAX SB‐C18 4.6 × 250 mm, PACKING LOT #: B17225, US), and detected at 254 nm. The ethylene evolution was measured as described previously (Tao et al., 2018). The gas chromatograph (Model GC‐17A; Shimadzu Co., Kyoto, Japan) fitted with a flame ionisation detector and an activated alumina column (200 cm × 0.3 cm) with an injector temperature of 120 °C, column temperature of 60 °C, and detector temperature of 60 °C as described previously (Shan et al., 2012).

RNA‐Seq and haplotype analysis

For RNA‐Seq analysis, leaves were sampled under midday and morning conditions, respectively. The cDNA library was constructed and sequenced using Illumina Novaseq6000 by Gene Denovo Biotechnology Co. (Guangzhou, China). Data processing and analysis were performed by using the cloud service (http://www.genedenovo.com/). The haplotype analysis was performed by using 4726 different rice accessions based on the online database of RiceVarMap v2.0 (http://ricevarmap.ncpgr.cn/; Zhao et al., 2021).

Analysed the enzymatic activity of OsACL5 in vitro

The OsACL5 coding sequence was cloned into the pColdI vector, and then the generated pColdI‐OsACL5 plasmid was transformed into Escherichia coli for protein expression and purification following a previous method (Liu et al., 2014b). The obtained recombinant OsACL5 proteins were incubated with the substrates (Spd and dcSAM) as previously described (Yoon et al., 2000). Finally, the generated PAs were detected by HPLC.

Detection of NO signals in roots

Nitric oxide (NO) was detected using DAF‐FM DA (diaminofluorescein FM diacetate) Kit (#cat: S0019, Beyotime). First, the roots of 7‐day‐old seedlings were loaded with 10 μm DAF‐FM DA in 20 mm HEPES/NaOH buffer (pH 7.4) for 30 min, then washed three times in fresh HEPES/NaOH buffer. Finally, images were taken by an Olympus BX51 microscope and the signal intensities of green fluorescence were quantified as described previously (Jin et al., 2011).

Statistics

All the data represented the average with the standard deviation (SD) of the average from three biological experiments. Significant difference was determined by paired two‐tailed Student's t‐tests. P < 0.05 was considered significant.

Conflict of interest

The authors declare that they have no conflict of interests.

Author contributions

T.L., H.M., and T.Q. designed the research; T.L., H.M., T.Q., and H.L. analysed the data; H.M., T.Q., H.W., Z.Y., G.P., Z.L., J.N., H.Y., H.T., and L.X. performed the experiments; T.L. wrote the article.

Supporting information

Figure S1 Phylogenetic analysis of OsACL5 and homologues from Arabidopsis and tomato.

Figure S2 OsACL5 serves as a T‐Spm synthase in vitro analysis.

Figure S3 The analysis of the size of bulliform cells.

Figure S4 Overexpression of OsACL5 affected the development of vascular bundles at 6‐day‐old shoots (folded leaves).

Figure S5 Differentially expressed genes (DEGs) analysis by KEGG enrichment, and the expression of two known leaf rolling related genes.

Figure S6 Phenotype analysis among WT, OE and osacl5 plants at 7‐day‐old stage.

Figure S7 Overexpression of OsACL5 increased stomatal density and affected leaf water potential.

Figure S8 Seed morphology of WT, OE, and osacl5 plants after removing the hulls.

Figure S9 Expression of grain size‐related genes, OsGSN1 and OsGW8, in seeds.

Figure S10 OsACL5 was expressed in seeds during grouting period.

Figure S11 T‐Spm contents increased under high temperature, high light intensity and low humidity in OE leaves.

Figure S12 The relative expression of ACC oxidase (ACO) determined by qRT‐PCR.

Figure S13 PAs enhanced NO signals in ZH11.

Figure S14 NO signals were strongly enhanced in OE lines.

Figure S15 The analyses of photosynthesis, chlorophyll contents and leaf area.

Table S1 List of primers used in this study.