CURLED LATER1 encoding the largest subunit of the Elongator complex has a unique role in leaf development and meristem function in rice

SUMMARY

The Elongator complex, which is conserved in eukaryotes, has multiple roles in diverse organisms. In Arabidopsis thaliana, Elongator is shown to be involved in development, hormone action and environmental responses. However, except for Arabidopsis, our knowledge of its function is poor in plants. In this study, we initially carried out a genetic analysis to characterize a rice mutant with narrow and curled leaves, termed curled later1 (cur1). The cur1 mutant displayed a heteroblastic change, whereby the mutant leaf phenotype appeared specifically at a later adult phase of vegetative development. The shoot apical meristem (SAM) was small and the leaf initiation rate was low, suggesting that the activity of the SAM seemed to be partially reduced in cur1. We then revealed that CUR1 encodes a yeast ELP1‐like protein, the largest subunit of Elongator. Furthermore, disruption of OsELP3 encoding the catalytic subunit of Elongator resulted in phenotypes similar to those of cur1, including the timing of the appearance of mutant phenotypes. Thus, Elongator activity seems to be specifically required for leaf development at the late vegetative phase. Transcriptome analysis showed that genes involved in protein quality control were highly upregulated in the cur1 shoot apex at the later vegetative phase, suggesting the restoration of impaired proteins probably produced by partial defects in translational control due to the loss of function of Elongator. The differences in the mutant phenotype and gene expression profile between CUR1 and its Arabidopsis ortholog suggest that Elongator has evolved to play a unique role in rice development.

Significance Statement

We revealed that genes encoding the subunit of Elongator complex play an important role in the differentiation of bulliform and sclerenchyma cells in leaf development and in the maintenance of the shoot apical meristem. The mutant leaf phenotypes appeared specifically at the late adult phase of the vegetative development, suggesting a unique role of the Elongator complex in rice.

INTRODUCTION

Plant development depends on the activity of the shoot apical meristem (SAM). In the vegetative phase, the SAM successively generates leaf primordia. After transition to the reproductive phase, the vegetative SAM converts to the inflorescence meristem, which produces flowers. The vegetative phase is further divided into two distinct phases, juvenile and adult, in which different traits are observed in the leaves (for review, see Huijser and Schmid, 2011; Chitwood and Sinha, 2016).

The anatomical and morphological changes that occur in leaves depending on plant growth stage are called heteroblasty. The molecular genetic mechanisms underlying the transition from juvenile to adult phase have been elucidated in Arabidopsis thaliana based on heteroblastic changes in leaves. Micro(mi)RNAs such as miR156 and MIR172, which regulate genes encoding SQUAMOSA PROMOTER BINDING PROTEIN‐LIKE (SPL) and APETALA2 (AP2)‐like transcription factors, are key components in the genetic control of these mechanisms in Arabidopsis (for review, see Huijser and Schmid, 2011; Chitwood and Sinha, 2016). The roles of these miRNAs and transcription factors in the juvenile‐to‐adult transition are conserved in angiosperms. Mutations in Corngrass1 and Glossy15, which encode miR156 and AP2‐like transcription factors, respectively, strongly affect heteroblasty in maize (Zea mays) (Moose and Sisco, 1996; Lauter et al., 2005; Chuck et al., 2007). In addition to these miRNAs, hormones such as jasmonate and gibberellin have been shown to be involved in the juvenile‐to‐adult transition in rice (Oryza sativa) (Tanaka et al., 2011; Hibara et al., 2016).

The rice leaf comprises the leaf blade, leaf sheath and their boundary region, including the ligule and auricle. Rice plants produce 15–20 leaves depending on the strain and growth conditions, and the shape and size of the leaf vary depending on the growth stage. The first to third leaves are formed during embryogenesis (Itoh et al., 2005). The first leaf is very small, and its leaf blade and leaf sheath are indistinguishable. Although the second leaf is also small, the leaf blade is distinguished from the leaf sheath. The first and second leaves are considered juvenile, whereas the third to fifth leaves seem to represent the transition from the juvenile to adult phase (Hibara et al., 2016). Except for the last leaf, the adult leaves are similar to each other in shape and morphological characteristics. The last leaf is called the flag leaf, and is shorter and less slender than the other adult leaves.

Rice has thin and flat leaves, in which large and small vascular bundles run in parallel together with a midrib (Ohmori et al., 2011; Matsumoto et al., 2017; Kubo et al., 2020). Two kinds of cells, bulliform and sclerenchyma cells, keep the leaf blade flat. The bulliform cells are specialized adaxial epidermal cells, which are distinguished from ordinary epidermal cells by their larger size and unique shape, and they keep the leaf blade flat by absorbing water (Itoh et al., 2005). In dry conditions, the leaf blade is curled adaxially due to loss of water from the bulliform cells. The ADAXIALIZED LEAF1 (ADL1) gene is essential for restricting differentiation of the bulliform cells to the adaxial side (Hibara et al., 2009). The sclerenchyma cells, which are formed on both the adaxial and abaxial sides of the vascular bundle, have thick and rigid cell walls, which provide the leaf blade with physical strength. In a loss‐of‐function mutant of the SHALLOR‐LIKE1 (SLL1) gene, which is similar to Arabidopsis KANADI, the leaf blades are curled adaxially due to the failed differentiation of abaxial sclerenchyma cells (Zhang et al., 2009). Leaf development, including shape and size, is regulated by a large number of genes (Fujino et al., 2008; Li et al., 2009; Ohmori et al., 2011; Ishiwata et al., 2013; Yoshikawa et al., 2013; Yasui et al., 2018).

The Elongator complex plays multiple roles in a wide range of eukaryotes, such as plants, animals and fungi (for review, see Versées et al., 2010; Woloszynska et al., 2016; Dalwadi and Yip, 2018). It was initially identified as a factor that directly associates with RNA polymerase II holoenzyme and was shown to be involved in transcriptional elongation (Otero et al., 1999; Wittschieben et al., 1999). In Arabidopsis, the Elongator complex is involved in leaf and root development; in humans, by contrast, it is involved in neuron development and its mutation is associated with familial dysautonomia (for review, see Versées et al., 2010; Woloszynska et al., 2016; Dalwadi and Yip, 2018). Elongator comprises two subcomplexes, i.e., a large subcomplex consisting of the subunits ELP1, ELP2 and ELP3, and a small subcomplex consisting of ELP4, ELP5 and ELP6. ELP1 is the largest subunit and is thought to be a scaffold protein associated with the stability of the whole complex. ELP3 is the main catalytic subunit and contains two distinct domains: a radical S‐adenosyl‐methionine (rSAM) domain and a putative histone acetyltransferase (HAT) domain. The rSAM domain of ELP3 is required to produce the chemically modified wobble uridine of transfer RNA (tRNA), whereas ELP1 and ELP6 provide tRNA‐binding sites (for review, see Dalwadi and Yip, 2018). The HAT domain shows sequence homology to members of the Gcn5‐related N‐acetyltransferase superfamily. From molecular biological and biochemical studies based on these structural observations, Elongator is involved in multiple molecular functions, including transcriptional regulation, translational control and endocytosis (Pandey et al., 2002; for review, see Versées et al., 2010; Woloszynska et al., 2016; Dalwadi and Yip, 2018).

In Arabidopsis, Elongator is involved in diverse biological activities, such as leaf and root development, hormone action and stress/environmental responses (for review, see Woloszynska et al., 2016). The genes encoding the Elongator subunits are called ELONGATA (ELO), except for ELO4 (Nelissen et al., 2005). elo/elp mutants exhibit various developmental defects, such as narrow leaf, disordered venation pattern, short primary root and disturbed hypocotyl elongation (Nelissen et al., 2005; Nelissen et al., 2010; Xu et al., 2012; Woloszynska et al., 2018; Qi et al., 2019). Molecularly, these mutant phenotypes are associated with a partial failure in the regulation of transcription, translation, histone acetylation, tRNA modification and miRNA synthesis (Nelissen et al., 2010; Fang et al., 2015; Leitner et al., 2015). Transcriptome analysis has also shown that the expression of auxin‐related genes is affected by elo/elp mutation: some of the auxin‐related genes have reduced H3K14 acetylation levels, suggesting that Elongator is involved in transcriptional control (Nelissen et al., 2010). Auxin signaling is also affected at the translational level: the abundance of PIN‐FORMED (PIN) proteins such as PIN1 and PIN2, which are responsible for polar auxin transport, is reduced in elo/elp mutants, despite no change at the transcript levels (Leitner et al., 2015). This translational control is associated with a defect in modification of the tRNA wobble uridine, similar to observations in yeast and human (Mehlgarten et al., 2010; Leitner et al., 2015). Thus, much progress has been made toward understanding the function of Elongator in Arabidopsis in the past decade. By contrast, our knowledge about the role of Elongator in other plants is limited, except for few studies (Zhu et al., 2015; Wang et al., 2020).

In this study, we show that Elongator is required for leaf development and meristem function in rice. Initially, we characterized the curled later1 (cur1) mutant, which produced narrow and adaxially curled leaf blades at a late adult stage of vegetative growth. The curled leaf phenotype was related to impaired differentiation of the bulliform and sclerenchyma cells. As compared with wild type (WT), the SAM was reduced in size in cur1 independent of growth stage. Gene isolation revealed that CUR1 encodes a protein similar to yeast ELP1 protein, the largest subunit of the Elongator complex. Disruption of OsELP3, an ortholog of yeast ELP3 and Arabidopsis ELO3, led to defects similar to those observed in cur1, including the appearance of mutant phenotypes at the same growth stage. Thus, Elongator activity seems to be specifically required for leaf development at later stages of the vegetative phase in rice. Transcriptome analysis showed that genes involved in protein quality control, including ubiquitin‐associated protein degradation and molecular chaperoning, were highly upregulated in the shoot apex at later vegetative stages, suggesting that loss of function of Elongator results in a partial defect in translational control.

RESULTS

Leaf phenotype appears at later vegetative stages in cur1

The cur1 mutant was isolated as a morphological mutant that generated slender leaves at a later stage of the vegetative phase (before heading) (Figure 1a). By contrast, leaves at younger growth stages looked normal. Therefore, we observed leaf phenotypes at various stages of plant growth. We characterized the phenotype of every leaf at the point when its whole leaf blade emerged from the sheath of the preceding leaf (Figure S1a). Thus, the leaf characteristics were analyzed at the same developmental stage of the leaf, but at different growth stages of the plant. The leaves were almost mature at this stage and did not grow much more.

Figure 1.

Leaf phenotypes of cur1 and wild type (WT).

(a) Adult rice plants at the growth stage when the flag leaves are emerging, but the panicles are not. WT, 97 days after germination; cur1, 128 days after germination. Note that the WT and cur1 plants are at a similar growth stage, despite the different periods since germination.

(b) Middle part of the leaf blade.

(c) Free‐hand section of the leaf blade.

(d) Width of the leaf blade.

(e) Curling index showing the extent of curling of the leaf blade. Curling index is the distance between the margins of the curling leaf and the width of the leaf blade after flattening (Matsumoto et al., 2017).

(f) Number of small veins in the normal (N‐), intermediate (I‐) and curling (C‐) type leaf. L, long lamina; S, short lamina.

(g) Schematic representation of the timing of the appearance of the cur1 leaf phenotype. gray, N‐type leaf; orange, C‐type leaf; yellow, I‐type leaf; star, flag leaf.

Asterisks (d–f) indicate significant differences (Student’s t‐test: **P < 0.01, *P < 0.05). Error bars represent SE. Scale bars: (b,c) 2 mm.

The leaf width was mostly equivalent between cur1 and WT until approximately the 12th leaf (Figure 1b–d). Thereafter, the leaf width of cur1 became shorter in the 14th leaf, as compared with the WT (Figure 1b–d). We found that the cur1 leaves produced at the later vegetative phase were curled upward (Figure 1c). The curled leaves first appeared after the 14th leaf stage, and the extent of curling became more evident in the 15th leaf onward, hence the mutant name “curled later” (Figure 1e). These observations suggest that the phenotype of the cur1 leaves changes depending on plant growth stage. Hereafter, we refer to the WT‐like leaves produced during the early and middle growth stages of cur1 as “N‐type” (normal) and the curled slender leaves produced at later growth stages as “C‐type” (curled).

The leaf blade consists of the midrib and two laminae formed on either side of the midrib. The large and small veins (vascular bundle) run in parallel in each lamina. We found that the number of small veins was reduced in C‐type leaves in cur1 relative to WT (Figure 1f). We found that the width of the lamina on the two sides often differed from each other in the 13th leaf (Figure 1c), and the number of small veins was reduced in the short lamina of cur1 as compared with WT lamina (Figure 1f and Figure S1b). The phenotype of this leaf seemed to represent an intermediate state of the transition from N‐ to C‐type; therefore, we defined this leaf as “I‐type” (intermediate). The timing of the appearance of the I‐ and C‐type leaves was more or less equal among individual cur1 plants, indicating that loss of CUR1 activity leads to a defect in leaves at an exact stage of plant growth (Figure 1g).

Histological analysis

To identify the cause of the curled leaf phenotype, we performed a histological analysis. Both the bulliform cells and sclerenchyma, which are required to keep leaves flat, were clearly observed in the N‐type leaves of cur1, similar to WT leaves (Figure 2a). By contrast, the bulliform cells were very small or indistinguishable from ordinary epidermal cells in the cur1 C‐type leaves (Figure 2a). In addition, the differentiation of sclerenchyma adjacent to the small vascular bundle was defective in C‐type leaves (Figure 2a,b). These results suggest that a failure in the formation of bulliform cells and sclerenchyma caused the curling of the C‐type leaves in cur1. These developmental defects of the bulliform cells and sclerenchyma were also observed in the short lamina of the I‐type leaf of cur1, but not in the long lamina (Figure 2a). Quantitative analysis indicated that defects in the bulliform cells and sclerenchyma occurred at high frequency both in the C‐type leaves and in the short lamina of the I‐type leaf of cur1 (Figure 2c,d).

Figure 2.

Histological observation of the leaf blade of cur1 and wild type (WT).

(a) Transverse sections of the leaf blade. Red and black arrowheads indicate developmental defects of bulliform cells and the absence of sclerenchyma, respectively.

(b) Close‐up view showing the sclerenchyma abaxial to the small vascular bundle in WT and the absence of it in cur1.

(c) Frequency of the absence of the sclerenchyma in cur1 leaves. I* means the leaf at the same stage as that of the I‐type leaf in cur1. L, long (wide) lamina; S, short (narrow) lamina; n.s., not significant.

(d) Frequency of incomplete differentiation of bulliform cells in cur1 leaves.

(e) Size of the small vascular bundle in the 15th leaf in WT and cur1.

BC, bulliform cell; SC, sclerenchyma; SV, small vascular bundle. (c,d,e) Asterisks indicate significant differences (Student’s t‐test: **P < 0.01, *P < 0.05). Error bars represent SE. Scale bars: (a) 50 μm; (b) 20 μm.

The size of small vascular bundle seemed to be reduced in the C‐type leaves, in addition to showing abnormal cell sizes and arrangement (Figure S1b,c). Measurement of the area clearly indicated a significant reduction in the size of the small vascular bundles as compared with the WT (Figure 2e). Reduced small vascular bundles were also observed in the short lamina of the I‐type leaf (Figure 2a).

Taken together, our observations indicate that defects in tissue differentiation are likely to appear in leaves produced at the later vegetative phase in cur1 via an intermediate state (I‐type leaf).

Meristem activity seems to be related to the appearance of cur1 leaf phenotype

WT and cur1 produced a similar number of leaves (Table 1). By contrast, the period from germination to appearance of the flag leaf (the last vegetative leaf) was longer in cur1 than in WT, consistent with the observation that cur1 displayed a late flowering phenotype. This result indicated that the rate of leaf initiation is slower in cur1 than in the WT.

Table 1.

Total number of leaves and the period from germination to emergence of the flag leaf

	2016		2017
	WT	cur1	WT	cur1
Number of leaves	18.5 ± 0.33	18.7 ± 0.17	17.2 ± 0.19	18.1 ± 0.21
Period (days)	100.8 ± 1.32	122.7 ± 1.11	92.5 ± 1.19	121.5 ± 1.08
n	8	9	13	11

WT, wild type. Plants were grown in pots outdoors in the spring/summer season (day length: 13.5–14.5 h).

Next, we examined meristem activity, which is related to lateral organ differentiation (Tanaka et al., 2012; Suzuki et al., 2019). The size of the SAM was measured after clearing shoot apices obtained at the middle and late stages of the vegetative phase. (The P1 primordium corresponded to 11th and 15th leaf in the middle and late stages, respectively.) The SAM of cur1 was significantly smaller than that of the WT at the middle stage (Figure 3a,b). The SAM became larger in both cur1 and WT as the plant grew, but the SAM remained significantly smaller in the later stage of cur1 as compared with WT. The reduced size of the SAM might be associated with the slower rate of leaf initiation in cur1.

Figure 3.

Size of the shoot apical meristem (SAM) and effect of short daylight (SD) on the leaf phenotype of cur1.

(a) Wild‐type (WT) and cur1 SAM after treatment with a clearing agent. Scale bars: 50 μm.

(b) Height of the SAM. Asterisks indicate significant differences (Student’s t‐test: **P < 0.01). Error bars represent SE.

(c) Schematic representations of the timing of the appearance of the cur1 leaf phenotype in plants grown under the SD condition (winter season). Day length, 9.75–10 h. Gray, N‐type leaf; orange, C‐type leaf; yellow, I‐type leaf; star, flag leaf.

(d) Effect of SD treatment on the appearance of the cur1 leaf phonotype in plants grown outdoors (spring/summer season). For SD treatment, plants were placed in the dark for 14 h/day for 10 days (for treatment details, see Figure S2). C, control (normal growth).

Plants enter the reproductive phase after floral induction: in rice, the SAM, which generates leaf primordia, is converted into the inflorescence meristem, which produces branch meristems. To investigate whether there is a relationship between the formation of C‐type leaves and phase transition, we exposed the cur1 mutant to short day (SD) treatment in two different ways. First, plants were grown in a greenhouse in the winter season (day length, 9.75–10 h). Consequently, the timing of the phenotypic appearance was highly accelerated, occurring from the eighth leaf onward (Figure 3c). Second, two groups of plants were grown outdoors in the spring/summer season (day length: 13.5–14.5 h), one of which was subjected to SD (day length: 10 h) treatment for 10 days (Figure S2). Subsequently, I‐ and C‐type leaves appeared precociously in the group of plants with SD treatment, as compared with control plants without SD treatment (Figure 3d). Collectively, these results suggest that the appearance of the mutant leaf phenotype in cur1 is probably associated with the vegetative‐to‐reproductive phase transition, which is dependent on SD conditions.

CUR1 encodes a protein similar to yeast ELP1, the largest subunit of the Elongator complex

We tried to isolate the gene underlying the cur1 mutation by a combination of rough mapping and the MutMap method (Abe et al., 2012). Consequently, we identified two genes (Os07g0563700 and Os07g0561101) that satisfied both the localization within the mapped region (Figure 4a) and the single‐nucleotide polymorphism index (>0.9) of MutMap analysis. Of these two genes, we focused on Os07g0563700, which had a nonsense mutation in cur1 (Os07g0561101 had a missense mutation). Transformation of cur1 with the WT allele of Os07g0563700 rescued the cur1 leaf phenotype. In the next‐generation plants of transformant no. 2, plants with the transgene showed the WT leaf phenotype, whereas those without it showed the cur1‐like leaf phenotype (Figure S3b). Histological analysis revealed that the rescued cur1 mutants generated both bulliform cells and sclerenchyma in the leaves generated at later developmental stages (Figure 4b and Figure S3c). These observations clearly indicated that the CUR1 gene is Os07g0563700.

Figure 4.

Isolation of the CUR1 gene and phylogenetic tree of CUR1‐related proteins (ELP1 homologs).

(a) Rough mapping of the CUR1 locus.

(b) Rescue of the cur1 leaf phenotype. Transverse sections of the leaf blade showing the formation of normal bulliform cells (BCs) and sclerenchyma (SC) in a T1 plant (#2‐1) of a transformant with a wild‐type CUR1 gene. Scale bars: 50 μm.

(c) Schematic representation of the CUR1 protein. Domains rich in WD40 repeats are indicated in light blue, and the TPR/IKAP domain is indicated in green. TPR, tetratricopeptide repeat.

(d) Phylogenetic tree of ELP1 proteins. The tree was constructed by a neighbor‐joining method using MEGA7 (Kumar et al., 2016). Numbers above the branches indicate the percentage of bootstrap values calculated from 1000 replicates.

CUR1 encodes a protein similar to yeast ELP1 with two WD40 repeat domains in the N‐terminus and a tetratricopeptide repeat in the C‐terminus (Figure 4c). In yeast, ELP1 is the largest subunit of the Elongator complex, which comprises six subunits. Elongator has shown to be involved in multiple functions, including transcription, tRNA modification and histone acetylation (for review, see Svejstrup, 2007; Glatt and Muller, 2013; Ding and Mou, 2015). CUR1 is also orthologous to Arabidopsis ELO2, the mutation of which leads to a narrow leaf phenotype (Nelissen et al., 2005) (Figure 4d).

We examined spatial expression pattern of CUR1 in WT by in situ hybridization. CUR1 was expressed in the SAM and leaf primordia (Figure 5a). Furthermore, reverse transcription–polymerase chain reaction (reverse transcription‐PCR) analysis indicated that CUR1 was expressed in the shoot apices at both middle and late vegetative phase of rice development (Figure 5b).

Figure 5.

Expression of CUR1 transcripts.

(a) In situ localization of CUR1 transcripts visualized by digoxigenin in the shoot apex of a wild‐type (WT) seedling 2 weeks after germination. Scale bars: 50 μm.

(b) Reverse transcription–polymerase chain reaction analysis using RNA isolated from the shoot apex, including the shoot apical meristem (SAM) and leaf primordia (P1–P3).

Knockout lines of OsELP3 show phenotypes similar to cur1

The ELP3 subunit has two distinct domains, rSAM and HAT, which are involved in tRNA maturation and histone acetylation. Mutation of the gene encoding ELP3 causes various biological defects in Arabidopsis, yeast and humans, depending on the organism (Nelissen et al., 2005; Nelissen et al., 2010; Leitner et al., 2015; for review, see Dalwadi and Yip, 2018). Next, therefore, we focused on the function of the ELP3‐like gene (OsELP3) in rice. A BLAST search indicated that the rice genome has a single gene similar to ELP3. Protein alignment revealed that the rSAM and HAT domains of ELP3‐like proteins are highly conserved among distantly related species, suggesting that they have conserved molecular functions (Figure S4).

To explore the developmental role of OsELP3, we disrupted the gene by using CRISPR‐Cas9 technology and obtained three independent knockout lines with biallelic mutations causing a frameshift in the rSAM domain (Figure 6a). All regenerated transformants with biallelic mutation (knockout lines, nos 26, 30, 39) produced curled narrow (C‐type) leaves at the later stages of the vegetative phase, although they produced WT (N‐type) leaves at the early and middle stages (Figure 6b). In addition, we observed the I‐type leaf just before the appearance of the C‐type leaves in these knockout lines. Histological analysis revealed developmental defects in both the bulliform and sclerenchyma cells in these leaves (Figure 6c). These leaf phenotypes were very similar to those observed for the cur1 mutant.

Figure 6.

OsELP3 knockout plants generated by CRISPR‐Cas9 technology.

(a) Target sequence in OsELP3 in the wild‐type (WT) genome and sequences mutated by the CRISPR‐Cas9 method in the regenerated transformants. Both strands of the target sequence of transformants (#26, 30, 39) are shown. All three lines have biallelic loss‐of‐function mutations caused by a frameshift. Blue hyphen and red triangle indicate, respectively, deletion and insertion of the corresponding bases. PAM sequence is underlined. HAT, histone acetyltransferase; gRNA, guide RNA; SAM, shoot apical meristem.

(b) Free hand sections of the leaf blade. Scale bars: 2 mm.

(c) Cross‐sections of the leaf blade tissue. Red and black arrowheads indicate the developmental defects of bulliform cells (BC) and the absence of sclerenchyma (SC) in OsELP3 knockout lines, respectively. Scale bars: 50 μm.

Transcriptome analysis

To examine the effect of the cur1 mutation on gene expression, we performed transcriptome profiling in shoot apices, including the meristem and leaf primordia (P1–P3). The shoot apex was sampled for RNA isolation at two different growth stages corresponding to (i) the middle vegetative stage, where the developing ninth to 11th leaves correspond to the P3–P1 primordia (referred as WT‐M and cur1‐M), and (ii) the late vegetative stage, where the developing 14th–16th leaves correspond to the P3–P1 primordia (referred as WT‐L and cur1‐L) (Figure S5). We expected that the former apices would produce N‐type leaves in the cur1 mutant when matured, whereas the later apices would produce C‐type leaves with the narrow and curled phenotype.

From the RNA sequencing (RNA‐Seq) data, we identified differentially expressed genes (DEGs) between the WT and cur1 shoot apices, and between the middle and late shoot apices (false discovery rate, 0.01; fold‐change, 2). Notably, DEGs were identified in the comparison not only between cur1‐M and cur1‐L but also between WT‐M and WT‐L (Figure 7a). The latter observation suggests the gene expression profile differs in the SAM and leaf primordia (P1–P3) between the middle and late stages of the WT vegetative phase. A large number of genes were upregulated in cur1‐L as compared with WT‐L, whereas fewer genes were downregulated in cur1‐L as compared with WT‐L.

Figure 7.

Transcriptome analysis in wild‐type (WT) and cur1 shoot apices at the middle and late vegetative phase.

(a) Venn diagrams of genes that were upregulated and downregulated in each comparison.

(b) Relative expression levels of genes related to the SCF complex. HSP, heat shock protein; OSK, O. sativa SKP1‐like gene, UbL40, ubiquitin‐fused ribosomal large subunit 40. #N.D., not detected.

(c) Relative expression levels of HSP genes.

(d) Read coverage in the gene body.

We looked for DEGs between cur1‐M and cur1‐L that were also present between WT‐L and cur1‐L (but not WT‐M versus WT‐L), as shown by the mulberry‐colored area of the Venn diagram in Figure 7(a). Among the shared DEGs, which are likely to be associated with the morphological changes in the C‐type leaves in cur1, the number of upregulated genes was much larger than the number of downregulated genes.

We performed gene ontology (GO) term enrichment analysis using PLAZA 4.5, but failed to identify any GO terms helpful to inform developmental insight, although drug‐related GO terms were enriched (Table S1). Therefore, we manually looked for interesting features among the DEGs, which indicated that genes encoding proteins constituting the SCF complex, such as O. sativa SKP1‐like proteins, a cullin and F‐box proteins were highly upregulated (5–50‐fold) in cur1‐L as compared with cur1‐M, WT‐M and WT‐L (Figure 7b, and Tables S2 and S4). The genes encoding ribosome protein‐fused ubiquitin (UbL40) were also upregulated. These proteins are involved in specific‐protein degradation via the 26S proteasome, although the exact function of UbL40 is unclear at present. In addition, genes encoding heat shock proteins (HSPs) and translation‐related proteins were also highly upregulated in the cur1‐L shoot apex, as compared with the other shoot apices (Figure 7c, Figure S5a and Tables S2 and S4). Considering that HSPs are involved in protein refolding, one of the shared features of the genes upregulated in cur1‐L seemed to be an association with protein quality control. We were not able to detect any common features among the shared downregulated genes (Table S3). We also examined the relative expression levels of genes whose mutation is known to cause a narrow leaf phenotype. None of these genes was downregulated in cur1‐L (Figure S5b; Table S4).

In yeast, Elongator interacts with the RNA transcript elongation complex (Otero et al., 1999; Wittschieben et al., 1999) and is implicated in transcription elongation. If CUR1 is also involved in transcriptional elongation in rice, the distribution of mapped reads over the gene body in RNA‐Seq would be expected to differ between cur1 and WT. Unexpectedly, however, the coverage of mapped reads in the gene body was indistinguishable between cur1 and WT (Figure 7d).

DISCUSSION

The cur1 mutation affects both leaf development and meristem activity

In this study, we first characterized the rice mutant cur1, which generates narrow and curled (C‐type) leaves. Histological observation indicated that the C‐type leaves contained undeveloped bulliform cells, no sclerenchyma and small vascular bundles with a disordered cell arrangement. The former two histological defects seem to be related to adaxial curling of the leaf blade, consistent with several previous reports (Zhang et al., 2009; Hu et al., 2010; Zou et al., 2011).

The cur1 mutant showed a heteroblastic change with the prominent feature of timing of the appearance of the mutant leaf phenotypes. In many plants, heteroblastic changes are associated with the juvenile‐to‐adult transition or the vegetative‐to‐reproductive phase transition (Moose and Sisco, 1996, Lauter et al., 2005; Chuck et al., 2007; Usami et al., 2009; Tanaka et al., 2011; Hibara et al., 2016; for review, see Huijser and Schmid, 2011; Chitwood and Sinha, 2016). However, in cur1 the C‐type leaf phenotypes appeared specifically at the late adult phase of vegetative development, suggesting that developmental and/or physiological states in the leaf might differ between the early/middle and late stages of the adult phase. Although the changes in these states were not observed in WT rice, they might be recognized morphologically by the loss of CUR1 activity. The appearance of the intermediate type (I‐type) leaf before the C‐type leaf suggests that this change might be precisely controlled.

There seem to be two possibilities to explain this heteroblastic change, i.e., (i) it may depend on plant age, and (ii) it may be associated with a developmental event such as the vegetative‐to‐reproductive transition. Under the SD treatment that accelerates flowering, the cur1 phenotype appeared at earlier growth stages, suggesting that the heteroblastic change is unrelated to plant age. Furthermore, the flag leaf, the last leaf that is generated in the vegetative phase, emerged soon after the production of C‐type leaves under the SD condition. These results indicate that there may be a strong link between the heteroblastic change and the vegetative‐to‐reproductive phase transition. It is possible that, after perceiving a flowering signal, the vegetative meristem enters a transient state toward its conversion to the inflorescence meristem, and that this “transient meristem” is associated with the heteroblastic change in cur1.

The SAM was significantly smaller in cur1 than in WT. Because the rate of leaf generation was slower in cur1 than in WT, the putative lower activity in the small cur1 SAM might be associated with the slow rate of leaf generation. CUR1 was expressed uniformly in the SAM and leaf primordia independent of plant growth stage. Consistent with this expression, the reduced SAM was observed in both the middle and the late stages of plant growth in cur1. By contrast, despite the continuous expression of CUR1, the leaf phenotype was not observed until the middle growth stage, suggesting that the function of CUR1 is specifically required for normal leaf morphogenesis at the later stages of vegetative growth.

Molecular function of CUR1

Gene isolation revealed that CUR1 encodes a yeast ELP1‐like protein in the Elongator complex. Whereas ELP1 functions as a scaffold protein, ELP3 acts as a catalytic subunit responsible for tRNA maturation and protein acetylation (for review, see Versées et al., 2010; Woloszynska et al., 2016; Dalwadi and Yip, 2018). The knockout lines of OsELP3 showed defects in the differentiation of bulliform and sclerenchyma cells at a specific developmental stage, similar to the cur1 mutant. This result suggests that CUR1 and OsELP3 act together as components of the Elongator complex in rice, and that the developmental alteration in the cur1 mutant is related to a defect in either tRNA maturation or protein acetylation.

The Elongator complex has been shown to be involved in transcriptional elongation, which is associated with histone acetylation, in yeast and Arabidopsis (Otero et al., 1999; Nelissen et al., 2010; for review, see Van Lijsebettens and Grasser, 2014). To address whether CUR1 is related to transcriptional elongation in rice, we performed RNA‐Seq analysis; however, we did not find any differences in the distribution of sequence reads in the 5′ → 3′ direction in the gene body between cur1 and WT, suggesting that overall transcription elongation is not affected in the cur1 mutant.

Although a large number of genes were differentially expressed between the middle and late shoot apices in cur1, the GO analysis did not identify any helpful GO terms providing insight into the developmental function of CUR1. This unexpected result seems to be due to poor GO annotation in rice genomics. Indeed, none of the genes shown in Figure 7 and Figure S6 had GO terms in the PLAZA database.

A prominent feature of transcriptome data revealed by the examination of individual genes revealed that genes encoding components of the SCF complex and HSPs were markedly upregulated specifically in the cur1‐L shoot apex, which is related to the differentiation of C‐type leaves. These proteins function specifically in protein degradation and protein refolding, i.e., in protein quality control. The highly elevated levels of genes encoding these proteins suggest that abnormal proteins may accumulate in cur1‐L shoot apices owing to disturbed translation. Therefore, it seems likely that defects in the leaf development of cur1 might be associated with a disturbance in translational control. This inference is consistent with the above observation that transcriptional elongation was not affected in cur1. In addition, it should be noted that a functional link between Elongator and ubiquitination has been suggested in yeast (Leidel et al., 2009).

Similar to yeast and humans, Elongator in Arabidopsis is involved in producing the wobble uridine modification of tRNA, which is essential for efficient translation (Mehlgarten et al., 2010). In fact, the abundance of PIN proteins is regulated at the translational level via the maturation of tRNAs by Elongator in Arabidopsis (Leitner et al., 2015). Accordingly, it is plausible that CUR1 might play a similar role in translational control in rice. However, we cannot exclude the possibility that the cur1 mutant phenotype resulted from partial defects in the transcriptional control of unknown genes involved in leaf development in rice.

Elongator and plant development

The genes encoding Elongator subunits have been well studied in Arabidopsis, e.g., they have been shown to play important roles in leaf and root development, in addition to environmental responses (Nelissen et al., 2005; Nelissen et al., 2010; Kojima et al., 2011; Xu et al., 2012; Fang et al., 2015; Leitner et al., 2015; Qi et al., 2019; for review, see Ding and Mou, 2015; Woloszynska et al., 2016). Arabidopsis ELO genes such as ELO2/AtELP1 and ELO3/AtELP4 are expressed in the SAM and leaf primordia, similar to rice CUR1 (Nelissen et al., 2010). However, in contrast to these Arabidopsis ELO/AtELP genes, which show localized adaxial expression in the leaf primordia, the rice CUR1 gene was uniformly expressed in the leaf primordia.

The narrow leaf phenotype and altered venation patterns in the rice cur1 mutant are shared by Arabidopsis elo/elp mutants. However, unlike the rice cur1 mutant, elo mutant phenotypes are observed in leaves independent of the growth stage of Arabidopsis (Nelissen et al., 2005; Nelissen et al., 2010). Thus, the involvement of Elongator in heteroblasty seems to be specific to rice. In tomato, silencing of ELP2‐like gene (SlELP2L) promotes leaf senescence, in addition to inhibiting leaf and plant growth (Zhu et al., 2015). Leaf senescence was not observed in Elongator mutants either here in rice or previously in Arabidopsis. Thus, loss of Elongator activity seems to have different effects on development depending on plant species. Indeed, the Elongator complex has been suggested to play a diverse role at the phenotypic level in various organisms (for review, see Mehlgarten et al., 2010; Dalwadi and Yip, 2018). Apart from the functional differences in Elongator genes between rice and tomato, it is interesting to note that a gene encoding E3 ubiquitin ligase is highly upregulated in SlELP2L‐silenced lines (Zhu et al., 2015).

Auxin‐related phenotypes and altered auxin distribution have been observed in Arabidopsis elo mutants (Nelissen et al., 2005; Malenica et al., 2007; Nelissen et al., 2010). Consistent with that observation, transcriptome analysis has shown that the expression of auxin‐related genes is affected by elo mutation (Nelissen et al., 2010). However, in the rice cur1 mutant, the expression of auxin‐related genes was unchanged. Indeed, the cur1 phenotype was distinct from the fishbone mutant, which showed diverse serious phenotypes due to the defective in auxin biosynthesis (Yoshikawa et al., 2014).

The expression of genes responsible for leaf width and bulliform cell growth were not reduced in the cur1‐L shoot apex (Figure S5b) (Fujino et al., 2008; Li et al., 2009; Xiang et al., 2012; Ishiwata et al., 2013; Yoshikawa et al., 2013). In addition, the expression of genes associated with cell cycle and vascular development was not affected markedly in cur1. Thus, genes known to be directly associated with leaf development and growth were not seemingly the cause of the curled leaf phenotype of cur1. However, based on the above hypothesis that translational control is partially compromised in cur1, there remains the possibility that the levels of proteins encoded by these genes are reduced in cur1, similar to PIN1 levels in Arabidopsis elo mutants. In future studies, proteome analysis may provide clues to identifying the factors associated with the defects in leaf development in cur1 and to elucidating the mechanism underlying the heteroblastic change associated with meristem function.

EXPERIMENTAL PROCEDURES

Plant materials and growth conditions

Rice (O. sativa ssp. japonica) variety Koshihikari was used as the WT. The cur1 mutant was identified as a recessive mutant showing narrow leaves among M2 plants of Koshihikari that had been mutagenized with sodium azide. Rice plants were usually grown in pots containing soil (outdoor). Transgenic plants were grown in an NK System Biotron (model LH‐350S, LH‐220S; Nippon Medical & Chemical Instruments, Osaka, Japan).

Histological analysis and in situ hybridization

Leaf segments were dissected, fixed and dehydrated by the methods of Toriba and Hirano (2018). After replacement of the solution with xylene, the leaf tissue samples were embedded in Paraplast Plus (McCormick Scientific, St. Louis, MO, USA). For histological analysis, microtome sections (7 μm) were stained with 1% toluidine blue‐O and observed by a BX50 optical microscope (Olympus, Tokyo, Japan). To measure the size of small vascular bundles, cross‐sections of the 15th leaf blade of WT and cur1 (C‐type) were obtained and the area of the small vascular bundles was measured using ImageJ software (https://imagej.nih.gov/ij/). To measure the size of the SAM, shoot apices were dissected and fixed in acetic alcohol (1:3). After treatment with a clearing agent (16 g of chloral hydrate dissolved in 8 ml of 25% [v/v] glycerol), the samples were observed under differential interference contrast optics. The height of the SAM was measured by using ImageJ software.

To generate the CUR1 probe, partial cDNA fragments were amplified with the primer pair 5′‐CACCCCGTAGCGAACCTCCTCTGT‐3′ and 5′‐CCTCTCCAATCCCCACAATC‐3′ (located in the third exon). The PCR products were then cloned into a pENTR/D‐TOPO vector (Thermo Fisher Scientific, Carlsbad, CA, USA). Labeling of probes and tissue preparation were carried out as described previously (Toriba and Hirano, 2018). In situ hybridization and immunological detection were performed on 10‐μm thick sections by using the method of Toriba and Hirano (2018).

Isolation of the CUR1 gene and complementation test

The CUR1 locus was mapped to a region between molecular markers RM336 (22.0 Mb) and E61310 (23.6 Mb) by using the F₂ population from a cross between cur1 and Nipponbare. To identify CUR1 gene candidates, we determined the whole‐genome sequence of the cur1 mutant by the MutMap method using 38 cur1 homozygotes (Abe et al., 2012). For the complementation analysis, a 7‐kb fragment encompassing the putative CUR1 gene (Os07g0563700) including the promoter region was cloned into a binary vector (Figure S3a). The recombinant plasmid was introduced into Agrobacterium tumefaciens, and transformation of the cur1 mutant was performed as described by Hiei et al. (1994).

Reverse transcription‐PCR analyses

RNA was isolated as described above. After DNase I treatment, first‐strand cDNA was synthesized from 1 μg of total RNA using the SuperScript III First‐Strand Synthesis System (Thermo Fisher Scientific) and the oligo(dT)15 primer. Amplification (30 cycles) was then performed with the primers listed in Table S5.

Generation of knockout mutants of OsELP3 by CRISPR–Cas9 technology

To disrupt OsELP3, 20‐bp sequences corresponding to the SAM domain were selected as the target of guide RNA (gRNA) (Figure 6a). Synthetic oligonucleotides containing the target sequence and adopter sequence were annealed, and subcloned into the gRNA cloning vector, pU6gRNA‐oligo (Mikami et al., 2015). The DNA fragment containing the OsU6 promoter and the gRNA region was then cloned into pZH_OsU3gYSA_MMCas9, as described previously (Tanaka et al., 2017; Yasui et al., 2017). The resulting construct was introduced into scutellum‐derived calli via A. tumefaciens (Hiei et al., 1994). To identify knockout lines of OsELP3, the target genomic region of each regenerating plant was sequenced by using the primers in Table S5 (Figure 6a).

RNA‐Seq experiments

Shoot apices including SAMs and P1–P3 leaf primordia were pooled (>5 apices per sample) and used for RNA isolation. Each sample (WT‐M, WT‐L, cur1‐M, cur1‐L; see Results) was prepared in biological triplicate. Total RNA was extracted by using TRIsure (Bioline, London, UK) and treated with RNase‐free DNase I. mRNA was isolated from total RNA (>2 μg) by using the Ribo‐Zero^™ rRNA removal kit (Illumina, San Diego, CA, USA) to remove rRNA. For RNA‐Seq library preparation, the cDNA was synthesized by using NEB (New England Biolabs, Beverly, MA, USA) Next Random Primers. The resulting libraries were sequenced by using the Illumina HiSeq instrument at GENEWIZ (South Plainfield, NJ, USA), which generated 2 × 150 bp paired‐end reads. The data ranged between 35.8 and 65.8 million reads per library (average 48.3 million). Reads were aligned to the rice reference sequence (Nipponbare) from RAP‐DB (https://rapdb.dna.affrc.go.jp), and an average of 94.9% reads was mapped to the reference genome. A short‐read alignment was performed using hisat2 (v2.0.1) (Kim et al., 2015) with default parameters. To obtain expression data, mapped reads were counted for each gene by featureCounts (Liao et al., 2014) using Oryza_sativa.IRGSP‐1.0.23.gtf from EnsemblPlants (ftp://ftp.ensemblgenomes.org/pub/release‐23) as a gene annotation file. Normalization and differential expression analysis of the expression data were performed by the R package ruvseq (Risso et al., 2014). The uniformity of mapped‐read coverage over the gene body was examined by using the geneBody coverage.py program in the rseqc package (v2.6.6) (Wang et al., 2012). The required reference gene model file (BED12 format) was generated from the gtf (ftp://ftp.ensemblgenomes.org/pub/release‐43) file by using the gtf2bed.pl program in the Genomics Tools package on Github (https://github.com/timothyjlaurent). GO analysis was performed by using the PLAZA database (https://bioinformatics.psb.ugent.be/plaza/versions/plaza_v4_5_monocots/organism/view/Oryza+sativa+ssp.+japonica).

AUTHOR CONTRIBUTIONS

HM, YS and H‐YH conceived research plans; YY, WT, YS and H‐YH supervised the experiments; HM and YY performed molecular and developmental analyses; TI, KS and YS performed genetic analysis and gene isolation; YT and JT performed complementation test; HM, YO and HN performed in silico analysis; HM, YS and H‐YH wrote the article with contributions of all the authors; YS and H‐YH supervised the project and agrees to serve as the author responsible for contact and ensures communication.

CONFLICT OF INTERESTS

The authors declare that they have no competing interests.

Supporting information

Figure S1. Leaf phenotypes of the cur1 mutant.

Figure S2. Time schedule of short daylight treatment.

Figure S3. Complementation of the cur1 mutant by introducing the wild‐type genomic DNA of Os07g0563700.

Figure S4. Alignment of the amino acid sequence of OsELP3.

Figure S5. Plan for harvesting materials for transcriptome analysis.

Figure S6. Transcriptome analysis.

Table S1. List of enriched GO terms.

Table S2. Genes specifically upregulated in cur1‐L in the transcriptome analysis.

Table S3. Genes specifically downregulated in cur1‐L in the transcriptome analysis.

Table S5. Primers used in this study.

Table S4. Gene ID of genes shown in the figures in the main text.

DATA AVAILABILITY STATEMENT

All summary data are included in the article or in Supporting Information online at the journal website. The raw RNA‐Seq reads have been deposited into the DDBJ Sequence Read Archive under accession no. DRA009274.