The OsMOB1A–OsSTK38 kinase complex phosphorylates CYCLIN C, controlling grain size and weight in rice

Abstract

Grain size and weight are crucial yield-related traits in rice (Oryza sativa). Although certain key genes associated with rice grain size and weight have been successfully cloned, the molecular mechanisms underlying grain size and weight regulation remain elusive. Here, we identified a molecular pathway regulating grain size and weight in rice involving the MPS ONE BINDER KINASE ACTIVATOR-LIKE 1A–SERINE/THREONINE-PROTEIN KINASE 38–CYCLIN C (OsMOB1A–OsSTK38–OsCycC) module. OsSTK38 is a nuclear Dbf2-related kinase that positively regulates grain size and weight by coordinating cell proliferation and expansion in the spikelet hull. OsMOB1A interacts with and enhances the autophosphorylation of OsSTK38. Specifically, the critical role of the OsSTK38 S322 site in its kinase activity is highlighted. Furthermore, OsCycC, a component of the Mediator complex, was identified as a substrate of OsSTK38, with enhancement by OsMOB1A. Notably, OsSTK38 phosphorylates the T33 site of OsCycC. The phosphorylation of OsCycC by OsSTK38 influenced its interaction with the transcription factor KNOTTED-LIKE HOMEOBOX OF ARABIDOPSIS THALIANA 7 (OsKNAT7). Genetic analysis confirmed that OsMOB1A, OsSTK38, and OsCycC function in a common pathway to regulate grain size and weight. Taken together, our findings revealed a connection between the Hippo signaling pathway and the cyclin-dependent kinase module in eukaryotes. Moreover, they provide insights into the molecular mechanisms linked to yield-related traits and propose innovative breeding strategies for high-yielding varieties.

Phosphorylation of the cell cycle regulator CYCLIN C by the ONE BINDER KINASE ACTIVATOR-LIKE 1A–SERINE/THREONINE-PROTEIN KINASE 38 kinase complex positively regulates grain size and weight by coordinating cell proliferation and expansion in the spikelet hull.

IN A NUTSHELL.

Background: Grain size and weight are crucial determinants of rice yield, yet the molecular mechanisms governing their regulation remain incompletely understood. The Hippo signaling pathway and the cyclin-dependent kinase (CDK) module (CKM) are conserved pathways involved in regulating various cellular processes across eukaryotes, but their roles in grain size regulation were unclear.

Question: What is the role of the Hippo signaling pathway and its interaction with the CKM in regulating grain size and weight in rice?

Findings: This study identified TINY1/OsSTK38, a nuclear Dbf2-related kinase, as a positive regulator of grain size and weight in rice. Notably, OsSTK38 interacts with OsMOB1A, which enhances the autophosphorylation of OsSTK38, a conserved mechanism in eukaryotes. Furthermore, OsSTK38 phosphorylates and interacts with OsCycC, a component of the CKM. Remarkably, the phosphorylation of OsCycC by OsSTK38 influences its interaction with the transcription factor KNAT7, which regulates cell expansion genes like EXPANSINs. Through genetic analyses, it was demonstrated that OsMOB1A, OsSTK38, and OsCycC function in a common genetic pathway to coordinate cell proliferation and expansion, thereby regulating grain size and weight in rice.

Next steps: Further investigations are needed to elucidate the molecular mechanisms by which the OsMOB1A–OsSTK38–OsCycC module regulates KNAT7 and other downstream targets. Exploring the potential involvement of this pathway in regulating other developmental processes and its crosstalk with other signaling pathways would be valuable. Conducting functional studies to understand the conservation and divergence of this pathway across different plant species is also warranted. Additionally, investigating the potential applications of this pathway in breeding programs for improving grain yield and quality could have significant agricultural implications.

Introduction

Rice (Oryza sativa) is a major crop grown worldwide, and the enhancement of rice yield is of utmost importance in ensuring global food security in the context of a growing population. The yield of rice is primarily influenced by 3 key factors: the effective tiller number, grain number per panicle, and grain weight. Grain size indicators, such as grain length, grain width, grain thickness, and the grain length-to-width ratio, are also intricately linked to grain weight. Grain size plays a pivotal role in determining crop yield, emphasizing the significance of identifying and harnessing genes associated with grain weight as a pivotal strategy for augmenting food production in the future (Xing and Zhang 2010; Huang et al. 2013).

Spikelet development plays a critical role in determining grain size and weight in rice. The growth of the spikelet hull is governed by cell proliferation and cell expansion (Li and Li 2016). Previous research has revealed numerous genes and quantitative trait loci encoding proteins that influence grain size and weight through their regulation of cell proliferation and/or expansion in the spikelet hull (Huang et al. 2009; Che et al. 2015; Duan et al. 2015; Hu et al. 2015; Liu et al. 2015; Duan et al. 2017; Liu et al. 2017, 2018; Hu et al. 2018; Sun et al. 2018; Qiao et al. 2021). These proteins are involved in different signaling pathways, including the G protein signaling pathway, ubiquitin–proteasome pathway, mitogen-activated protein kinase signaling pathway, plant hormone signaling pathways, and transcriptional regulatory pathways (Li and Li 2016; Ren et al. 2023). However, the fine molecular genetic network governing grain size and weight remains largely unexplored.

The Hippo signaling pathway is a conserved pathway that plays a crucial role in regulating various cellular processes, including organ size, cell proliferation, apoptosis, and stem cell self-renewal (Hergovich et al. 2006). It is present in diverse organisms ranging from yeast and fruit flies to humans. At the core of this pathway is the nuclear Dbf2-related kinase (NDR) protein, which exhibits a conserved structure. It consists of an N-terminal regulatory domain that interacts with the Mps One Binder (MOB) protein, a serine autophosphorylation site within the kinase domain, and a downstream threonine that can be phosphorylated by upstream kinases belonging to the GCK/Ste20 family (Hergovich et al. 2006). NDR kinase, found in various species, can further phosphorylate substrates, such as Cdc14 and members of the YAP/Yki transcription factor family, thereby regulating various cellular activities (Gógl et al. 2015). Disruptions in the Hippo signaling pathway have been associated with several diseases in humans, including cancer (Zheng and Pan 2019). Additionally, the involvement of the NDR protein in the Hippo signaling pathway has been demonstrated in fission yeast (Schizosaccharomyces pombe) and nematodes (Caenorhabditis elegans; Hergovich et al. 2006; Lee et al. 2019). Among plants, the NDR-type AGC kinase TaAGC1 of wheat (Triticum aestivum) positively regulates host resistance against Rhizoctonia cerealis by modulating the expression of genes associated with reactive oxygen species and defense (Zhu et al. 2015). In Arabidopsis (Arabidopsis thaliana), AtNDR2/4/5 interacts with AtMOB1A/1B and is involved in pollen development and pollen germination (Zhou et al. 2021). However, the precise role of the Hippo signaling pathway in controlling grain size and weight has not yet been confirmed.

The Mediator complex is a highly conserved protein complex found in eukaryotes that serves as a crucial link between transcription factors and RNA polymerase II (RNAP II) during transcriptional regulation (Asturias et al. 1999; Poss et al. 2013; Harper and Taatjes 2018). It comprises 4 primary modules: the head, middle, tail, and kinase modules. The head, middle, and tail modules collectively form the core of the Mediator complex, establishing physical interactions between RNAP II and sequence-specific transcription factors (Jeronimo and Robert 2017). On the other hand, the cyclin-dependent kinase (CDK) module (CKM), an independent component of the Mediator complex, consists of CDK8, cyclin C (CycC), mediator subunit 12 (MED12), and mediator subunit 13 (MED13). When associated with the core Mediator complex, CKM exerts a substantial influence on gene transcription. While the Mediator complex has been implicated in various aspects of plant growth, development, and responses to environmental stimuli (Agrawal et al. 2021), its involvement in the regulation of rice grain size and weight remains uncertain.

In this study, we present findings regarding the contribution of rice NDR kinase [O. sativa SERINE/THREONINE-PROTEIN KINASE 38 (OsSTK38)] to the coordinated regulation of cell proliferation and cell expansion in the rice lemma as well as its role in governing grain size and weight. Specifically, we demonstrate that ONE BINDER KINASE ACTIVATOR-LIKE 1A (OsMOB1A) plays a crucial role in facilitating the phosphorylation of OsCycC by OsSTK38. The phosphorylation of OsCycC by OsSTK38 influenced its interaction with the transcription factor O. sativa KNOTTED-LIKE HOMEOBOX OF ARABIDOPSIS THALIANA 7 (OsKNAT7). Our genetic analysis further supports the involvement of the OsMOB1A–OsSTK38–OsCycC module in the common pathway responsible for controlling grain size and weight in rice. These findings not only establish a connection between the evolutionarily conserved Hippo signaling pathway and the CDK module in eukaryotes, but also elucidate the underlying mechanisms by which they regulate grain size and weight in crops.

Results

Tiny1 forms small grains through coordinated alterations of cell size and cell number

To investigate the mechanisms determining rice grain size and weight, we acquired a small-grain mutant, tiny1, from a rice T-DNA mutation library (Wan et al. 2009). The tiny1 mutant exhibited a semi-dwarf phenotype and a reduced grain size (Fig. 1, A to C). In comparison with the wild type (WT; Nipponbare), the grain length of tiny1 was reduced by 12.89%, its grain width decreased by 5.46%, and its 1,000-grain weight decreased by 29.48% (Fig. 1, E to G). These results suggest that TINY1 plays a positive role in regulating grain size and weight.

Figure 1.

Tiny1 forms small grains through coordinated alterations of cell size and cell number. A) Mature paddy rice grains of Nipponbare (Nip) and tiny1. Bar, 5 mm. B) Brown rice grains of Nip and tiny1. Bar, 5 mm. C) Plant architecture of Nip and mutant tiny1 plants at the reproductive stage. Bar, 10 cm. D) Panicles of Nip and tiny1. Bar, 10 cm. E) and F) Grain length (n = 15) and width (n = 15) of Nip and tiny1. G) The 1,000-grain weight (n = 3) of Nip and tiny1.H) Plant height (n = 12) of Nip and tiny1. I) Panicle length (n = 12) of Nip and tiny1. J) Grain number per panicle (n = 12) of Nip and tiny1. K) and L) SEM analysis of the outer surface of Nip and tiny1 lemmas. Bar, 50 μm. M) Average length (n = 10) and (N) width (n = 10) of outer epidermal cells in Nip and tiny1 lemmas. O) Outer epidermal cell number (n = 10) in the longitudinal direction in Nip and tiny1 lemmas. P) Outer epidermal cell number (n = 10) in the transverse direction in Nip and tiny1 lemmas. Bars E) to J) and M) to P) represent mean ± Sd. Student's t-tests were used to generate the P-values.

Mature tiny1 plants exhibited significantly reduced height compared with Nipponbare plants (Fig. 1, C and H). Similarly, tiny1 panicles were shorter than Nipponbare panicles but displayed a significantly greater grain number (Fig. 1, D, I, and J). These findings indicate that TINY1 plays a role in regulating the balance between grain number and size in rice.

To elucidate the cellular mechanism by which TINY1 influences grain size, we conducted scanning electron microscopy (SEM) observations. Compared with Nipponbare, the number of epidermal cells in the mature grain lemma of tiny1 increased by 10.28% in the longitudinal direction and 23.55% in the transverse direction. However, the length and width of the epidermal cells in the tiny1 lemma decreased by 24.54% and 30.82%, respectively (Fig. 1, K to P). These results indicated that the per-unit area in tiny1 contained more cells, but the sizes of these cells were smaller, which led to smaller spikelets in tiny1. Based on these results, it can be inferred that TINY1 likely influences grain size by coordinating the balance between spikelet cell proliferation and cell expansion.

To investigate the regulatory role of TINY1 in cell proliferation and expansion, we examined the expression levels of several cell-cycle-related genes. Reverse transcription-quantitative PCR (RT-qPCR) analysis revealed significant upregulation of OsCDKB;2, OsCDC20, OsE2F2, OsCycB;2, OsCycH1;1, OsCycD4;1, OsCycU4;1, and OsCycU4;3 in the tiny1 mutant compared with in Nipponbare (Supplementary Fig. S1). Furthermore, we utilized flow cytometry to analyze the cell division rate in young panicles (2 cm in length) of both Nipponbare and tiny1, which involved determining the DNA content at different stages of the cell cycle. The results showed an increased proportion of cells in S-phase and G2/M-phase with a higher DNA content in tiny1, while the percentage of G1-phase cells with a 2C DNA content decreased after the completion of the new cell cycle (Supplementary Fig. S2). The RT-qPCR analysis of cell-cycle-related genes was consistent with the DNA content. However, despite the increased number of epidermal cells in the tiny1 spikelet hulls, the overall grain size in tiny1 was smaller than that in Nipponbare due to the smaller size of the individual epidermal cells. Notably, the downregulation of certain genes encoding proteins known to positively regulate cell expansion, such as SPL13/GWL7, PGL1, and GS2, was observed in tiny1 (Supplementary Fig. S3). The above results indicate that TINY1 mutation accelerates cell cycle progression and promotes cell division, yet it restricts cell expansion, leading to the formation of smaller organs.

TINY1 encodes an NDR, OsSTK38

Genetic analyses of tiny1/Nipponbare plants showed that their F₁ progeny had a WT-like phenotype; moreover, the F₂ plants segregated at a 3:1 ratio (x² = 0.31 < x² = 0.05, 1) of WT-like (221 plants) to mutant-like (76 plants), indicating that the tiny1 phenotype was caused by a single recessive nuclear mutation.

To clone TINY1, we created an F₂ population from a cross of the tiny1 mutant and the indica variety Dular. A mapping-based cloning method was used to isolate TINY1. Initially, the TINY1 locus was mapped between molecular markers rm1-2.3 and rm1-4.9 on the Chromosome 1 short arm. Subsequently, we narrowed the locus to a 40 kb region between the markers rm1-4.61 and rm1-4.65 in 300 homozygous F₂ plants. Within this region, we identified 5 open reading frames: LOC_Os01g09180, LOC_Os01g09190, LOC_Os01g09200, LOC_Os01g09206, and LOC_Os01g0212 (Fig. 2A). We sequenced and analyzed these 5 open reading frames and discovered a T-DNA insertion occurring 118 bp upstream of the start codon of LOC_Os01g09200 (Fig. 2B). These results indicated that LOC_Os01g09200 was a potential candidate gene.

Figure 2.

TINY1 encodes a nuclear Dbf2-related kinase, OsSTK38. A) The TINY1 locus was initially mapped to the short arm of Chromosome 1 between markers rm1-2.3 and rm1-4.9 and then delimited to a 40 kb region between markers rm1-4.61 and rm1-4.65 that contained 5 ORFs. The numbers beneath the marker positions indicate the number of recombinants. B)TINY1/OsSTK38 gene structure. The coding sequence is shown as a black box, and introns are indicated using black lines. ATG and TGA represent the start codon and the stop codon, respectively. The position of the T-DNA insertion in tiny1 is shown. C) Mature paddy and D) brown rice grains of WT, tiny1, gTINY1;tiny1 #1, and gTINY1;tiny1 #2. Bar, 5 mm. E) Grain length (n = 15) and F) width (n = 15) of WT, tiny1, gTINY1;tiny1 #1, and gTINY1;tiny1 #2. G) The 1,000-grain weight (n = 3) of WT, tiny1, gTINY1;tiny1 #1, and gTINY1;tiny1 #2. Bars E) to G) represent mean ± Sd. Significant differences (P < 0.05) are indicated by letters using 1-way ANOVA with Tukey's multiple comparisons test.

To further investigate the relationship between the small-grain phenotype and the T-DNA insertion, we conducted cosegregation analysis of F₂ populations resulting from a cross between tiny1 and Nipponbare. The results showed that the small-grain phenotype cosegregated with the T-DNA insertion (Supplementary Fig. S4). Subsequently, to confirm that LOC_Os01g09200 was responsible for the tiny1 phenotype, we performed a genetic complementation experiment. A genomic fragment of LOC_Os01g09200 (referred to as gTINY1) containing the upstream 2 kb and downstream 1 kb sequences was introduced into the tiny1 callus, resulting in the generation of 12 independent transgenic lines. All 12 independent transgenic lines in T₀ progeny completely rescued the mutant phenotype according to a comparison of plant height and panicle shape between transgenic plants and WT or tiny1 mutant plants. In the T₁ progeny, the grain length, grain width, 1,000-grain weight, plant height, and panicle length of all 12 lines were restored to WT levels (Fig. 2, C to G and Supplementary Fig. S5). In addition, LOC_Os01g09200 was knocked out using CRISPR editing tool. Unfortunately, we did not obtain regenerated plants in 2 attempts, and we speculated that loss of function of the LOC_Os01g09200 product results in embryonic death. In Arabidopsis, LOC_Os01g09200 homologous genes NDR2/4/5 affected embryonic development (Yoon et al. 2021). We determined the expression level of LOC_Os01g09200 in tiny1 mutant and found that it reached 10% of the WT level (Supplementary Fig. S6). Considering the plant phenotype of tiny1, we speculated that tiny1 was a knock-down mutant of LOC_Os01g09200. Based on the above experimental results, we concluded that TINY1 corresponds to LOC_Os01g09200.

Since TINY1 corresponds to OsSTK38 in the TIGR database (http://rice.uga.edu/), we replaced TINY1 with OsSTK38 in subsequent analyses. TINY1/OsSTK38 encodes a protein kinase belonging to the AGC protein kinase family and is classified within a subfamily of NDR/large tumor suppressor (LATS) kinases. The structural characteristics of NDR kinases are highly conserved across different species. NDR kinases are widely distributed in plants, animals, and fungi and play important roles in various cellular processes, including cell division, cell morphology, and cell death. In mammals, NDR/LATS kinases have been associated with immune responses and cancer development (Hergovich et al. 2006; Zheng and Pan 2019). In plants, NDR kinases participate in disease resistance, pollen development, germination, and embryogenesis (Zhu et al. 2015; Yoon et al. 2021; Zhou et al. 2021). However, the precise functions of OsSTK38 and its homologous genes in rice remain largely unclear.

Expression pattern and subcellular localization of OsSTK38

Previous studies have reported that NDR kinases in Arabidopsis are predominantly localized in the nucleus and cytoplasm (Zhou et al. 2021). Given the conservation of NDR kinases, it is reasonable to speculate that OsSTK38 may exhibit a similar cell localization. To confirm this, we conducted cotransfection experiments using the fusion construct proE35S:OsSTK38-GFP and a well-established nuclear-cytoplasmic localization marker (proUbi:OsGSK5-tagRFP) in rice protoplasts (Ying et al. 2018). The results showed complete colocalization between proE35S:OsSTK38-GFP and proUbi:OsGSK5-tagRFP, providing strong evidence of the expression of OsSTK38 in both the cytoplasm and nucleus (Supplementary Fig. S7).

Furthermore, we employed RT-qPCR to investigate the expression pattern of OsSTK38. Our results revealed that OsSTK38 displayed higher expression levels in root, stem, and panicle tissues, particularly in organs characterized by active proliferation. Conversely, mature leaves exhibited relatively low expression levels of OsSTK38 (Supplementary Fig. S8A). The expression pattern observed for GUS driven by the OsSTK38 promoter in transgenic plants was generally consistent with the RT-qPCR analysis results (Supplementary Fig. S8B).

Overexpression of OsSTK38 results in large grains

To further investigate the impact of OsSTK38, we generated transgenic lines by introducing a proActin:OsSTK38 construct into Nipponbare, resulting in the generation of 18 independent transgenic lines. At least 10 overexpression (OE) lines in T₁ progeny showing a large grain phenotype were obtained. Two independent lines (OsSTK38-OE #2 and OsSTK38-OE #6) with high LOC_Os01g09200 expression were further analyzed (Fig. 3C). The OE lines (OE #2, OE #6) were taller and exhibited larger grains, with significant increases in grain length, grain width, and 1,000-grain weight (Fig. 3, A and B and Fig. 3, D to F). However, there was no significant difference in panicle length and a slight decrease in the grain number per panicle (Supplementary Fig. S9). These results suggest that OsSTK38 positively regulates grain size.

Figure 3.

Overexpression of OsSTK38 results in large grains. A) Plant architecture of WT, OsSTK38-OE #2, and OsSTK38-OE #6 plants at the reproductive stage. Bar, 10 cm. B) Mature paddy rice grains of WT, OsSTK38-OE #2, and OsSTK38-OE #6. Bar, 5 mm. C) Relative expression levels of OsSTK38 in WT and OsSTK38-OE young panicles (n = 3). D) Grain length (n = 15) and E) width (n = 15) of WT and OsSTK38-OE transgenic lines. F) The 1,000-grain weight (n = 3) of WT and OsSTK38-OE transgenic lines. G) SEM analysis of the outer surface of WT and OsSTK38-OE #2 lemmas. Bar, 50 μm. H) Average length (n = 15) and I) width (n = 15) of outer epidermal cells in WT and tiny1 lemmas. J) Outer epidermal cell number in the longitudinal direction in WT and OsSTK38-OE #2 lemmas. K) Outer epidermal cell number in the transverse direction in WT and OsSTK38-OE #2 lemmas. Bars C) to F) and H) to K) represent mean ± Sd. Student's t-tests were used to generate the P-values (comparing each line to the WT).

Subsequently, we examined the effects of OsSTK38-OE on cell number and cell size in the mature grain lemma. The OsSTK38-OE #2 line displayed a decrease in cell numbers (reduced by 16.91% in longitudinal sections and 18.55% in transverse sections) and an increase in cell size (increases in cell length by 32.91% and cell width by 22.40%) within the lemma compared with the WT (Fig. 3, G to K). These results further verified the role of OsSTK38 in regulating grain size through the synergistic modulation of cell proliferation and expansion within spikelet hulls.

OsMOB1A interacts with OsSTK38 and enhances its autophosphorylation

Multiple studies have demonstrated that NDR kinases form complexes with MOB kinase coactivators, and these complexes constitute essential and evolutionarily conserved components of the Hippo signaling pathway regulating cell proliferation and morphogenesis in eukaryotes (Hergovich et al. 2006; Meng et al. 2016; Parker et al. 2020; Zhou et al. 2021). Thus, we identified a homologue of MOB in rice, OsMOB1A (Supplementary Figs. S10 and S11B). We examined whether OsSTK38 interacts with OsMOB1A. Initially, we confirmed the interaction between OsMOB1A and OsSTK38 through yeast 2-hybrid assays (Fig. 4A).

Figure 4.

OsMOB1A interacts with OsSTK38 and enhances its autophosphorylation. A) OsSTK38 interacts with OsMOB1A in yeast cells. Yeast cells were cultured on SD/-Trp-Leu and SD/-Trp-Leu-His-Ade media. AD, GAL4 activation domain; BD, GAL4 DNA-binding domain; SD, synthetic defined. B) BiFC assays showing that OsSTK38 interacts with OsMOB1A in rice protoplasts. OsSTK38-nYFP was cotransformed with OsMOB1A-cYFP in rice protoplasts. 34.47% cells exhibiting YFP fluorescence. The OsGSK5-nYFP/OsMOB1A-cYFP and OsSTK38-nYFP/OsARF4-cYFP were used as negative controls. Bar, 10 µm. C) OsSTK38 binds OsMOB1A in vitro. MBP-OsMOB1A was incubated with GST-OsSTK38, pulled down by GST-OsSTK38 and detected by immunoblotting with an anti-GST antibody. IB, immunoblot. D) Interaction between OsSTK38 and OsMOB1A in the Co-IP assays. Anti-GFP beads were used to immunoprecipitate OsSTK38-MYC proteins. Gel blots were probed with anti-MYC or anti-GFP antibodies. E) OsSTK38 protein sequence, with potential activity sites. F) OsSTK38 Ser-322 is solely responsible for OsSTK38 autophosphorylation. Autophosphorylated GST-OsSTK38, GST-OsSTK38^K145M, GST-OsSTK38^S322A, GST-OsSTK38^T101A, and GST-OsSTK38^T485A were detected by anti-α-phospho-(Ser/Thr) antibody. G) OsMOB1A enhances OsSTK38 autophosphorylation in vitro. The autophosphorylated GST-OsSTK38 and GST-OsSTK38^S322A were detected by anti-α-phospho-(Ser/Thr) antibody.

Next, we performed bimolecular fluorescence complementation (BiFC) analysis to test the interaction between OsSTK38 and OsMOB1A in plant cells. OsSTK38 was fused to the N-terminus of yellow fluorescent protein (OsSTK38-nYFP), and OsMOB1A was fused to the C-terminus of yellow fluorescent protein (OsMOB1A-cYFP). The interaction between OsGSK5 and OsARF4 has been demonstrated by BiFC assay (Hu et al. 2018). To serve as negative controls, we constructed 2 vectors: OsGSK5-nYFP and OsARF4-cYFP. When OsSTK38-nYFP and OsMOB1A-cYFP were coexpressed in rice protoplasts, confocal laser scanning microscopy observations revealed strong YFP fluorescence in both the cytoplasm and nucleus. However, when OsGSK5-nYFP/OsMOB1A-cYFP and OsSTK38-nYFP/OsARF4-cYFP were cotransformed, no fluorescence was observed (Fig. 4B). The results indicate that OsSTK38 binds to OsMOB1A in plant cells.

Furthermore, pull-down assays were used to judge whether OsSTK38 directly interacted with OsMOB1A. A glutathione S-transferase (GST)-tagged OsSTK38 fusion protein (GST-OsSTK38) and a maltose-binding protein (MBP)-tagged OsMOB1A fusion protein (MBP-OsMOB1A) were expressed in Escherichia coli cells. As shown in Fig. 4C, GST-OsSTK38 physically interacted with MBP-OsMOB1A but showed no interaction with the negative control (GST) in vitro. Coimmunoprecipitation analysis was also used to examine the association of OsSTK38 with OsMOB1A in Nicotiana benthamiana. OsSTK38-MYC and OsMOB1A-GFP were coexpressed in N. benthamiana leaves. Total protein was isolated and incubated with GFP beads for the immunoprecipitation of OsMOB1A-GFP. Anti-MYC and anti-GFP antibodies were used to detect the immunoprecipitated proteins. The presence of the OsSTK38-MYC protein was detected in the immunoprecipitated OsMOB1A-GFP complex, indicating the association of OsSTK38 with OsMOB1A in vivo (Fig. 4D). The above results demonstrate that OsSTK38 directly interacts with OsMOB1A both in vitro and in vivo.

The activation of NDR kinases requires 3 conditions: the binding of the MOB protein to the N-terminal regulatory (NTR) domain of NDR kinase, the autophosphorylation of serine residues in the kinase domain, and the phosphorylation of a threonine residue in the hydrophobic C-terminal domain by upstream GCK/Ste20 family kinases (Hergovich et al. 2006). It has been demonstrated that MOB is essential for NDR kinase phosphorylation and activation in humans and yeast. MOB can stimulate the autophosphorylation of NDR kinases (Devroe et al. 2004). To verify whether this mechanism is conserved in plants, we first identified potential phosphorylation sites in OsSTK38 through phylogenetic analysis. In human NDR1, there are 3 conserved phosphorylation sites, T74, S281, and T444 as well as an ATP-binding site, K118 (Tamaskovic et al. 2003; Devroe et al. 2004). The corresponding sites in OsSTK38 were T101, S322, T485, and K145 (Fig. 4E and Supplementary Fig. S12). We purified WT GST-OsSTK38 and different mutant forms of GST fusion proteins (GST-OsOsSTK38^T101A, GST-OsOsSTK38^S322A, GST-OsSTK38^T485A, and GST-OsSTK38^K145M) and incubated them in an in vitro kinase assay buffer. As anticipated, immunoblotting with anti-phospho-(Ser/Thr), which recognizes phosphorylated serine and threonine residues, detected an unambiguous band of phosphorylated GST-OsSTK38 (P-GST-OsSTK38). In contrast, the phosphorylation levels of GST-OsSTK38^T101A and GST-OsOsSTK38^T485A were slightly lower than that of GST-OsSTK38, while GST-OsSTK38^S322A completely lost autophosphorylation activity, similar to the kinase-dead mutant GST-OsSTK38^K145M (Fig. 4F). To investigate whether OsMOB1A can enhance the autophosphorylation of OsSTK38, we coincubated MBP-OsMOB1A with GST-OsSTK38 or GST-OsSTK38^S322A in in vitro kinase assay buffer. In the presence of MBP-OsMOB1A, the phosphorylation level of GST-OsSTK38 substantially increased, while GST-OsSTK38^S322A completely lost autophosphorylation activity^S322A (Fig. 4G).

Furthermore, we examined the OsSTK38 phosphorylation sites and incubated GST-OsSTK38 and MBP-OsMOB1A in an in vitro kinase assay followed by liquid chromatography-tandem mass spectrometry. We thus identified 3 viable phosphorylation sites of OsSTK38 (T101, S322, and T485; Supplementary Fig. S13). The mass spectrometry results further support the in vitro phosphorylation kinase assay. Totally, these results suggest that S322 of OsSTK38 is the major site of its autophosphorylation and that OsMOB1A enhances the autophosphorylation of OsSTK38.

OsMOB1A acts in a common pathway with OsSTK38 to control grain size and weight

Given the interaction between OsMOB1A and OsSTK38, along with the ability of OsMOB1A to stimulate the autophosphorylation of OsSTK38, we hypothesized that OsMOB1A may influence grain size. To test this hypothesis, we employed CRISPR‒Cas9 technology to knock out the OsMOB1A gene and obtained Osmob1a knockout lines (Supplementary Fig. S14F). Furthermore, we generated 15 lines overexpressing OsMOB1A-OE by transforming the proActin:OsMOB1A construct into Nipponbare callus (Supplementary Fig. S14, C and D). The Osmob1a knockout lines showed a dwarf phenotype and had significantly smaller grains with a reduced grain length, width, and 1,000-grain weight, resembling the tiny1 phenotype (Supplementary Figs. S14, A, B, E and S14, H to J). In contrast, 2 independent lines (OsMOB1A-OE #1 and #2) with high LOC_Os03g38020 expression were further analyzed (Supplementary Fig. S14G). OsMOB1A-OE lines (#1, #2) exhibited an increased plant height, grain yield per plant, and a larger grain size, with an increased grain length, width, and 1,000-grain weight compared with WT (Supplementary Figs. S14, C to E and S14, H to J).

Next, we examined mature grain lemma epidermal cells in the WT, Osmob1a, and OsMOB1A-OE transgenic lines (Supplementary Fig. S15, A to C). Although the Osmob1a lemma displayed an increased number of epidermal cells, the cells in the Osmob1a lemma were smaller than those in the WT lemma. Conversely, OsMOB1A-OE lines displayed the opposite phenotype compared with Osmob1a, indicating that OsMOB1A plays a positive role in grain growth by coregulating cell proliferation and expansion within the spikelet hull (Supplementary Fig. S15, D to G).

Because OsMOB1A stimulated the autophosphorylation of OsSTK38 and similar grain phenotypes were observed in the tiny1 and Osmob1a knockout lines, we hypothesized that OsMOB1A acts upstream of OsSTK38 in regulating grain size. To verify this hypothesis, 3 experiments were conducted as follows. First, we introduced the proActin:OsSTK38 construct into the Osmob1a mutant. The result showed that OsSTK38-OE/Osmob1a lines restored the small-grain phenotype of Osmob1a (Fig. 5, A to D). Secondly, we crossed Osmob1a-1 with tiny1 to obtain Osmob1a-1/tiny1. We observed that the grains of Osmob1a-1/tiny1 were similar to those of Osmob1a-1 and tiny1 and did not produce smaller grains (Fig. 5, E to H). In addition, we also introduced the proActin:OsMOB1A construct into the tiny1 mutant. Unexpectedly, we found that the OsMOB1A-OE/tiny1 line not only showed restoration of the small-grain phenotype of tiny1 but also exhibited even longer grains than the OsMOB1A-OE line (Fig. 5, I to L). This result did not seem to support our hypothesis that OsMOB1A and OsSTK38 act in different pathways. On the contrary, since tiny1 is a knock-down mutant of LOC_Os01g09200 (Supplementary Fig. S6), OE of OsMOB1A perhaps enhances the phosphorylation level of OsSTK38, leading to larger grains. Based on the above 3 genetic transformation experiments, we speculated that OsMOB1A and OsSTK38 act in a common pathway and that OsMOB1A is genetically epistatic to OsSTK38.

Figure 5.

OsMOB1A acts in a common genetic pathway with OsSTK38 to control grain size and weight. A) Mature paddy rice grains of WT, Osmob1a-1, OsSTK38-OE #2, and OsSTK38-OE/Osmob1a-1. Bar, 5 mm. B) Grain length (n = 15) and C) width (n = 15) of WT, Osmob1a-1, OsSTK38-OE #2, and OsSTK38-OE/Osmob1a-1.D) The 1,000-grain weight (n = 3) of WT, Osmob1a-1, OsSTK38-OE #2, and OsSTK38-OE/Osmob1a-1.E) Mature paddy rice grains of WT, tiny1, Osmob1a-1, and Osmob1a-1/tiny1. Bar, 5 mm. F) Grain length (n = 15) and G) width (n = 15) of WT, tiny1, Osmob1a-1, and Osmob1a-1/tiny1. H) The 1,000-grain weight (n = 3) of WT, tiny1, Osmob1a-1, and Osmob1a-1/tiny1.I) Mature paddy rice grains of WT, tiny1, OsMOB1A-OE #1, and OsMOB1A-OE/tiny1. Bar, 5 mm. J) Grain length (n = 15) and K) width (n = 15) of WT, tiny1, OsMOB1A-OE #1, and OsMOB1A-OE/tiny1. L) The 1,000-grain weight (n = 3) of WT, tiny1, OsMOB1A-OE #1, and OsMOB1A-OE/tiny1. Bars B) to D), F) to H) and J) to L) represent mean ± Sd. Significant differences (P < 0.05) are indicated by letters using 1-way ANOVA with Tukey's multiple comparisons test.

OsSTK38 interacts with and phosphorylates OsCycC

To obtain downstream substrates for OsSTK38, we employed the full-length OsSTK38 protein as bait to identify downstream substrates, screening for interacting proteins in a yeast library. Notably, OsCycC was identified in 5 independent repeats. The interaction between OsSTK38 and OsCycC was further confirmed using yeast 2-hybrid analysis. In contrast, a direct interaction between OsMOB1A and OsCycC was not observed (Fig. 6A).

Figure 6.

OsSTK38 interacts with and phosphorylates OsCycC. A) OsCycC interacts with OsSTK38 rather than OsMOB1A in yeast cells. Yeast cells were cultured on SD/-Trp-Leu and SD/-Trp-Leu-His-Ade media. AD, GAL4 activation domain; BD, GAL4 DNA-binding domain; SD, synthetic defined. B) BiFC assays showing that OsSTK38 interacts with OsCycC in rice protoplasts. OsSTK38-nYFP was cotransformed with OsCycC-cYFP in rice protoplasts. About 42.68% cells exhibiting YFP fluorescence. The OsGSK5-nYFP/OsCycC-cYFP and OsSTK38-nYFP/OsARF4-cYFP were used as negative controls. Bar, 10 µm. C) OsSTK38 binds OsCycC in vitro. MBP-OsCycC was incubated with GST-OsSTK38, pulled down by GST-OsSTK38 and detected by immunoblotting with an anti-GST antibody. IB, immunoblot. D) Interaction between OsSTK38 and OsCycC in Co-IP assays. Anti-GFP beads were used to immunoprecipitate OsSTK38-MYC proteins. Gel blots were probed with anti-MYC or anti-GFP antibodies. E) OsSTK38 phosphorylates OsCycC in vitro. Phosphorylated MBP-OsCycC was separated by phos-tag SDS‒PAGE. F) OsCycC protein sequence and potential phosphorylation sites. G) Thr-33 strongly influences the phosphorylation of OsCycC. Phosphorylated MBP-OsCycC^T33A, MBP-OsCycC^S45A, MBP-OsCycC^T74A, and MBP-OsCycC^T89A were separated by phos-tag SDS‒PAGE.

Subsequently, a BiFC assay was conducted to elucidate the interaction between OsSTK38 and OsCycC. Constructs were generated for OsSTK38-nYFP (N-terminal fusion of OsSTK38 with yellow fluorescent protein) and OsCycC-cYFP (C-terminal fusion of OsCycC with yellow fluorescent protein). The coexpression of OsSTK38-nYFP and OsCycC-cYFP in rice protoplasts resulted in robust YFP fluorescence in the cytoplasm and nucleus observed via confocal laser scanning microscopy. However, similar to the BiFC assay between OsSTK38 and OsMOB1A, no fluorescence was observed when OsGSK5-nYFP/OsCycC-cYFP and OsSTK38-nYFP/OsARF4-cYFP were cotransformed (Fig. 6B).

Furthermore, a pull-down assay was performed to confirm the direct interaction between OsSTK38 and OsCycC. We expressed the OsSTK38 protein fused with a GST tag (GST-OsSTK38) and the OsCycC protein fused with an MBP tag (MBP-OsCycC) in E. coli cells. Figure 6C demonstrates the physical interaction between GST-OsSTK38 and MBP-OsCycC in vitro, while no interaction was observed with the negative control (GST). Coimmunoprecipitation analysis was employed to investigate the interaction between OsSTK38 and OsCycC in N. benthamiana. OsSTK38-MYC and OsCycC-GFP were coexpressed in the leaves of N. benthamiana. Total protein was then extracted and incubated with GFP beads for the immunoprecipitation of OsCycC-GFP. Immunoprecipitated proteins were detected using anti-MYC and anti-GFP antibodies. The presence of OsSTK38-MYC protein in the immunoprecipitated OsCycC-GFP complex confirmed the interaction between OsSTK38 and OsCycC in vivo (Fig. 6D).

NDR kinase can phosphorylate substrates, thereby regulating various physiological and developmental processes (Hergovich et al. 2006; Hergovich 2013). In animals, p21 and YAP/TAZ have been identified as downstream substrates of NDR1 kinase (Dong et al. 2007; Zhao et al. 2007; Hao et al. 2008; Varelas et al. 2008; Cornils et al. 2011). Given the interaction between OsSTK38 and OsCycC, we investigated the potential phosphorylation of OsCycC by OsSTK38. First, we assessed kinase activity in vitro by incubating MBP-OsCycC with GST-OsSTK38 and GST-OsSTK38^S322A in kinase detection buffer. Phosphorylated MBP-OsCycC was detected only when GST-OsSTK38 was present, while no phosphorylation of MBP-OsCycC was observed in the presence of both GST and GST-OsSTK38^S322A. These results indicated that OsSTK38 could phosphorylate OsCycC in vitro, with the S322 site playing a crucial role in its kinase activity (Fig. 6E).

To determine the phosphorylation sites of OsCycC targeted by OsSTK38, we employed GPS 6.0 software to predict potential phosphorylation sites of OsCycC, and T33, S45, T74, and T89 were identified as potential sites of OsCycC phosphorylation by OsSTK38 (Fig. 6F). Site-directed mutagenesis was performed at each of these 4 sites to replace the residues with alanine to abolish phosphorylation, and phosphorylation levels were subsequently examined using GST-OsSTK38 and GST-OsSTK38^S322A. Our findings revealed that MBP-OsCycC^T33A was the only mutated form that did not exhibit phosphorylation bands, indicating that T33 serves as the critical site for the OsSTK38-mediated phosphorylation of OsCycC (Fig. 6G).

Furthermore, we examined the OsCycC phosphorylation sites and incubated GST-OsSTK38 and MBP-OsCycC in an in vitro kinase assay followed by liquid chromatography-tandem mass spectrometry. We thus identified 1 viable phosphorylation site of OsCycC (T33) (Supplementary Fig. S16). The mass spectrometry results further support the in vitro phosphorylation kinase assay. Taken together, these results further verified that T33 is a major site of OsCycC for phosphorylation by OsSTK38 in vitro.

The Thr-33 phosphorylation status of OsCycC affects interactions with KNAT7

Protein phosphorylation mediated by kinases often influences substrate stability. To examine the impact of the phosphorylation of OsCycC on its stability, we conducted a cell-free degradation assay by incubating in vitro-expressed MBP-OsCycC with protein extracts from both WT and tiny1 seedlings. Negligible alterations in the protein abundance of MBP-OsCycC were observed in both the WT and tiny1 protein extracts (Supplementary Fig. S17). Thus, the phosphorylation of OsCycC by OsSTK38 is unlikely to impact its protein stability.

Subsequently, we investigated the impact of OsCycC phosphorylation on its subcellular localization. The transient transformation of rice protoplasts with pE35S:OsCycC-GFP revealed the localization of OsCycC-GFP in both the nucleus and cytoplasm. No notable difference in subcellular localization patterns was observed between OsCycC-GFP, OsCycC^T33A-GFP, and OsCycC^T33D-GFP (Supplementary Fig. S18). In conclusion, the phosphorylation of OsCycC by OsSTK38 was unlikely to substantially affect the protein stability or subcellular localization of OsCycC.

Recent studies in Arabidopsis have demonstrated that CycC interacts with WRKY75 and ABI5 transcription factors, inhibiting their transcriptional activation and consequently impacting downstream gene expression (Guo et al. 2022; Lu et al. 2023). Thus, we speculated that the phosphorylation of OsCycC might modulate its interactions with other proteins. To test this hypothesis, we utilized the full-length OsCycC protein as bait to identify potential downstream substrates by screening for interacting proteins in a yeast library. We identified KNAT7 as a candidate protein interacting with OsCycC. Initially, we confirmed the interaction between OsCycC and KNAT7 using Y2H (Fig. 7A). Subsequently, we introduced T33 mutations (to alanine, Ala, or aspartate, Asp) in OsCycC to mimic its dephosphorylated or phosphorylated states, respectively, and assessed the interactions of the mutant proteins with KNAT7. The Y2H experiments revealed that OsCycC^T33D retained its ability to interact with KNAT7, akin to KNAT7 itself. Interestingly, OsCycC^T33A exhibited a loss of interaction with KNAT7 (Fig. 7A). This observation was corroborated by BiFC assays, which demonstrated that OsCycC^T33D, but not OsCycC^T33A, retained its ability to interact with KNAT7 in rice protoplasts (Fig. 7, B and C). KNAT7 encodes a rice Class II KNOX-like homeobox transcription factor that exerts a negative regulatory influence on rice grain size (Wang et al. 2019). Thus, these data indicate that OsMOB1A/OsSTK38 may affect the interaction between OsCycC and KNAT7 through phosphorylation modification, thereby regulating grain size.

Figure 7.

The Thr-33 phosphorylation status of OsCycC affects interactions with KNAT7. A) OsCycC and OsCycC^T33D rather than OsCycC^T33A interact with KNAT7 in yeast cells. Yeast cells were cultured on SD/-Trp-Leu and SD/-Trp-Leu-His-Ade media. AD, GAL4 activation domain; BD, GAL4 DNA-binding domain; SD, synthetic defined. B) BiFC assays showing that OsCycC and OsCycC^T33D rather than OsCycC^T33A interact with KNAT7 in rice protoplasts. KNAT7-nYFP was cotransformed separately with OsCycC-cYFP (39.49% cells exhibiting YFP fluorescence), OsCycC^T33A-cYFP and OsCycC^T33D-cYFP (40.12% cells exhibiting YFP fluorescence) in rice protoplasts. Bar, 10 µm. C) Negative controls for the BiFC analysis for the interaction between OsCycC and KNAT7. In rice protoplasts, the negative controls had no fluorescent signal. Bar, 10 µm.

OsMOB1A, OsSTK38, and OsCycC function in a common genetic pathway to regulate grain size and weight

As OsCycC served as a substrate for OsSTK38, we investigated its role in grain size and weight regulation. We obtained OsCycC-OE lines driven by the OsActin1 promoter and OsCycC knockout lines generated using CRISPR‒Cas9 technology. Ten knockout lines were verified by PCR sequencing in T₀. All the lines showed a reduced plant height in T₀ progeny. Two Oscycc knockout lines (Oscycc-1, Oscycc-2) in T₁ progeny reduced plant height, exhibited smaller grains, and decreased yield per plant compared with the WT (Supplementary Fig. S19). We obtained 16 OsCycC-OE lines in T₀ and selected 2 independent lines (#5, #14) with high LOC_Os09g32680 expression to further analyze (Supplementary Fig. S19E). The OsCycC-OE lines (#5, #14) in T₁ displayed slightly increased plant height, larger grains, and incremented yield per plant, providing evidence that OsCycC positively regulates grain size and weight (Supplementary Fig. S19). SEM analysis revealed that the reduced grain size and weight of Oscycc-1 were primarily attributed to significantly smaller cells, despite an increased cell number, similar to the phenotypes observed in tiny1 and Osmob1a-1 (Supplementary Fig. S20).

Since the interaction between the human MOB protein and NDR kinase can enhance NDR kinase activity (Devroe et al. 2004), we investigated whether OsMOB1A regulated the phosphorylation of OsCycC by OsSTK38. Figure 8A demonstrates a substantial increase in the abundance of phosphorylated MBP-OsCycC bands in the presence of GST-OsMOB1A, whereas the mutant kinase GST-OsSTK38^S322A did not produce any MBP-OsCycC phosphorylation bands, regardless of the presence of GST-OsMOB1A.

Figure 8.

OsMOB1A, OsSTK38, and OsCycC function in a common pathway to regulate grain size. A) OsMOB1A promotes the phosphorylation of OsCycC by OsSTK38. Phosphorylated MBP-OsCycC was separated by phos-tag SDS‒PAGE. IB, immunoblot. B) Mature paddy rice grains of WT, tiny1, OsCycC-OE #1 and OsCycC-OE/tiny1. Bar, 5 mm. C) Grain length (n = 20) and D) width (n = 20) of WT, tiny1, OsCycC-OE #1, and OsCycC-OE/tiny1. E) The 1,000-grain weight (n = 5) of WT, tiny1, OsCycC-OE #1, and OsCycC-OE/tiny1. F) Mature paddy rice grains of WT, tiny1, Oscycc-1, and Oscycc/tiny1. Bar, 5 mm. G) Grain length (n = 20) and H) width (n = 20) of WT, tiny1, Oscycc-1, and Oscycc/tiny1. I) The 1,000-grain weight (n = 3) of WT, tiny1, Oscycc-1, and Oscycc/tiny1. Expression levels of J)EXPA6, K)EXPA7, L)EXPB3, and M)EXPB17 in WT, tiny1, and Oscycc panicles. ACTIN1 was used to normalize expression. Values are mean ± Sd relative to the WT value, set at 1. Three biological replicates were used (n = 3). *P < 0.05; **P < 0.01 compared with the WT (Student's t-test). Bars C) to E) and G) to I) represent mean ± Sd. Significant differences (P < 0.05) are indicated by letters using a 1-way ANOVA with Tukey’s multiple comparisons test.

Subsequently, we examined the relationship between OsCycC and OsSTK38 regarding grain size and weight. First, we introduced the proActin:OsCycC construct into the tiny1 mutant. The grain length and weight of the OsCycC-OE/tiny1 lines were significantly greater than those of tiny1 but were reduced compared with those of OsCycC-OE. However, no significant difference in grain width was observed between tiny1 and OsCycC-OE, both of which showed notably shorter grain widths than OsCycC-OE (Fig. 8, B to E). Owing to the fact that tiny1 was a knock-down mutant of LOC_Os01g09200, OE of OsCycC can complement the phenotype of grain length and weight of tiny1 mutant. Secondly, we crossed Oscycc-1 with tiny1 to obtain Oscycc/tiny1. We observed that the grains of Oscycc-1/tiny1 were similar to those of Oscycc-1 and tiny1 and did not produce smaller grains (Fig. 8, F to I). Overall, our genetic experiments suggest that OsCycC plays a role in grain size and weight regulation, at least partly downstream of OsSTK38.

Notably, 4 EXPANSIN genes, EXPA6, EXPA7, EXPB3, and EXPB17, are downstream regulation genes of KNAT7-GRF4 (Wang et al. 2019). The expression of 4 EXPANSIN genes was upregulated in the knat7 mutant, whereas it was repressed in the KNAT7-OE line (Wang et al. 2019). Both OsSTK38 and OsCycC positively regulate grain size, as evidenced by their mutant spikelet epidermal cells being smaller in size when compared with those of the WT. Consequently, we analyzed the expression levels of the above 4 EXPANSIN genes in young spikelets of tiny1 and Oscycc. In line with the results in the KNAT7-OE lines, the expression levels of EXPA6, EXPA7, EXPB3, and EXPB17 were diminished in both tiny1 and Oscycc spikelets when compared with levels in the WT (Fig. 8, J to M). These results further support the notion that OsMOB1A, OsSTK38, OsCycC, and KNAT7 function in a common pathway to regulate grain size and weight.

Haplotype analysis of OsMOB1A, OsSTK38, and OsCycC

The natural genetic variation of OsMOB1A, OsSTK38, and OsCycC was investigated by collecting single-nucleotide polymorphisms (SNPs) in their coding regions from the 3K Rice Genome Database (3K-RG). Two synonymous SNPs were identified in both OsMOB1A and OsSTK38 (Supplementary Fig. S21). Sequence analysis revealed that OsMOB1A and OsSTK38 homologues were widely distributed in eukaryotes, particularly in the grass family. Homologues of OsMOB1A and OsSTK38 from sorghum, maize, dogwood, and dicotyledonous short-stalked grass were found to share ∼90% and 98% sequence similarity with OsMOB1A (Supplementary Figs. S10 and S11B) and OsSTK38 (Supplementary Figs. S11A and S12), respectively. This high sequence similarity in different species indicates that OsSTK38 and its homologues have been subjected to strong purifying selection, enabling less variation and a more conserved function. Additionally, 3 SNPs were identified in the coding region of OsCycC, including vg0919499255 and vg0919499304, which represent nonsynonymous mutations (Supplementary Fig. S22A). Haplotype analysis showed that 4 main haplotype populations exist in OsCycC (Hap3 was the reference type, Nipponbare). The analysis of 3K-RG grain length phenotype data indicated a significant decrease in grain length in Hap3 compared with Hap1, Hap2, and Hap4, which can likely be attributed to the vg0919499304 mutation. Notably, Hap1 and Hap2 are the predominant haplotypes found in indica rice, whereas Hap3 is the major haplotype in japonica rice (Supplementary Fig. S22, B and C). Further investigation is warranted to determine how vg0919499304 affects grain size, particularly in terms of differentiating between japonica and indica rice.

Discussion

Grain size and weight are crucial factors determining rice yield. Numerous genes associated with grain size regulation have been identified, including well-known genes, such as GS3 and GW2, that have been extensively targeted for breeding purposes (Fan et al. 2006; Song et al. 2007). However, the fine molecular genetic network governing rice grain size and weight remains poorly understood. This study elucidates the molecular mechanism by which TINY1/OsSTK38 regulates grain size and weight. Through map-based cloning, cosegregation, complementation experiments, and OE assays, we identify OsSTK38 as a positive regulator of grain size and weight. OsSTK38 is a NDR belonging to the AGC protein kinase family. NDR kinases are widely distributed in plants, animals, and yeast and play crucial roles in various cellular processes, including cell division, cell morphology, and cell death. NDR kinases are involved in diverse processes in plants, including disease resistance, pollen development and germination, and embryogenesis (Zhu et al. 2015; Yoon et al. 2021; Zhou et al. 2021). This study expands our understanding of the role of NDR kinases in regulating organ size in plants.

Furthermore, we elucidate the molecular mechanism underlying the regulation of grain size and weight through the OsMOB1A–OsSTK38–OsCycC regulatory module. Our findings demonstrate that the OsMOB1A–OsSTK38 kinase complex controls grain size and weight by modulating the phosphorylation of OsCycC, a key component of the downstream substrate-mediated kinase module, through multiple mechanisms. The Osmob1a, tiny1, and Oscycc mutants exhibit significant reductions in grain length, width and 1,000-grain weight compared with WT. Conversely, the OE of OsMOB1A, OsSTK38, and OsCycC in transgenic plants results in significant increases in grain length, width, and 1,000-grain weight. SEM analysis confirmed that OsMOB1A, OsSTK38, and OsCycC exert specific effects on both cell size and cell number, thereby influencing the regulation of grain size through coordinated modulation of cell proliferation and division. OsMOB1A, OsSTK38, and OsCycC were more highly expressed in panicles (Supplementary Figs. S8A and S23). The spikelets of Osmob1a, tiny1, and Oscycc exhibited substantial increases in the number of longitudinal and transverse cells, while the length and width of their lemma cells were markedly reduced compared with those of WT. Notably, in tiny1, the length and width of mature seed lemma cells were reduced by 24.54% and 30.82%, respectively, in comparison with WT, despite a 10.28% increase in the number of longitudinal cells and a 23.55% increase in the number of transverse cells. These findings indicate that the primary factor contributing to the smaller grains of tiny1 is restricted cell expansion. Additionally, the OE of OsSTK38 leads to the development of larger grains, characterized by a 32.91% increase in the length and a 22.40% increase in the width of mature grain lemma cells compared with those of WT, despite decreases in the number of longitudinal and lateral cells by 16.71% and 18.55%, respectively, compared with those of WT. These results are consistent with findings obtained for OsGSK5/TGW3 and KNAT7, which also regulate grain size and weight through the coordinated modulation of cell expansion and cell proliferation (Ying et al. 2018). Despite a significant reduction in average grain length, width, and 1,000-grain weight, the tiny1 mutant exhibited a remarkable increase in the grain number per panicle compared with WT. These results suggest that OsSTK38 plays a role in regulating the trade-off between grain size and grain number per panicle. This concept finds support in previous studies that have demonstrated the involvement of GSN1 and OsMKKK10 in influencing the balance between grain size and the grain number per panicle (Guo et al. 2018). Further investigation is warranted to elucidate the regulatory mechanisms through which OsSTK38 controls the trade-off between grain size and the grain number per panicle. Moreover, OsSTK38 influences plant height. Thus, these findings indicate that OsSTK38 plays a role in regulating the development of both nutritional and reproductive organs in rice.

Eukaryotes rely on Hippo signaling pathways to regulate various developmental processes, including cell proliferation and morphogenesis. The core components of these networks have been conserved throughout billions of years of evolution and are also found in animals and fungi (Huang et al. 2005; Hergovich et al. 2006; Lee et al. 2019; Zheng and Pan 2019). Within these pathways, MST/hippo kinases activate NDR or LATS kinases (Harvey et al. 2003; Chan et al. 2005; Hergovich 2013), both of which belong to the AGC family of protein kinases. NDR kinases form a regulatory complex by binding to highly conserved MOB kinase activators, thereby controlling a diverse range of effector proteins (Tamaskovic et al. 2003; Devroe et al. 2004; He et al. 2005; Lai et al. 2005; Kohler et al. 2010; Meng et al. 2016; Parker et al. 2020). In Arabidopsis, the interaction between AtNDR2/4/5 and AtMOB1A/B plays a role in regulating pollen formation and germination (Zhou et al. 2021). Here, we identified OsMOB1A, which interacts with OsSTK38. Our findings demonstrate that OsMOB1A promotes the autophosphorylation of OsSTK38, a conserved mechanism in eukaryotes. Genetic analysis indicated that OsMOB1A regulates grain size through a common genetic pathway with OsSTK38. Previous studies have demonstrated the involvement of AtMOB1A in regulating embryogenesis, root development, and fertility through auxin- and JA-related signaling (Cui et al. 2016; Xiong et al. 2016; Guo et al. 2020; Zhou et al. 2021). Further investigation should be performed to determine whether OsMOB1A and OsSTK38 influence organ development through auxin or other hormones.

Although several substrates of NDR kinases, such as p21 and YAP/TAZ, have been identified in animals (Hao et al. 2008; Varelas et al. 2008; Cornils et al. 2011), their counterparts in plants have remained unclear. This study identifies OsCycC as a substrate of OsSTK38. Cyclin C interacts with CDK8, leading to the phosphorylation of the carboxy-terminal structural domain of the RNAP II large subunit. Notably, the kinase module of the mediator complex consists of CDK8, Cyclin C, MED12, and MED13, and it plays a crucial role in transcriptional regulation. Cyclin C serves as a subunit of the plant mediator protein complex, acting as a bridging molecule between transcription factors and RNAP II, thereby influencing transcription (Poss et al. 2013; Jeronimo and Robert 2017; Agrawal et al. 2021). In Arabidopsis, Cyclin C has been implicated in plant immunity, seed germination, and the salt stress response (Zhu et al. 2014; Guo et al. 2022; Lu et al. 2023). In pea (Pisum sativum), Cyclin C is also known to promote flowering and support normal reproductive development (Hasan et al. 2020). Our findings indicate that OsCycC acts as a downstream substrate of the OsMOB1A–OsSTK38 kinase complex, positively regulating rice grain size. The phosphorylation of a protein typically affects its stability, subcellular localization, and interactions with other proteins. Our investigation aimed to examine the impact of OsSTK38-mediated phosphorylation on OsCycC and revealed that phosphorylation does not influence its stability, and subcellular localization. Recent studies in Arabidopsis have demonstrated that CycC interacts with WRKY75 and ABI5 transcription factors, inhibiting their transcriptional activation and consequently impacting downstream gene expression (Guo et al. 2022; Lu et al. 2023). Biochemical experiments revealed the interaction between OsCycC and KNAT7. Additionally, we observed that OsSTK38-mediated phosphorylation of OsCycC influences its interactions with KNAT7. Previous reports have indicated that KNAT7 negatively regulates grain size through the modulation of cell expansion in rice. Furthermore, it was discovered that KNAT7 interacts with GRF4/GS2 (Wang et al. 2019). Drawing upon this molecular and genetic evidence, we present a proposed model elucidating the control of rice grain size through the OsMOB1A–OsSTK38–OsCycC-mediated mechanism (Fig. 9). Phosphorylated OsCycC, catalyzed by the upstream OsMOB1A–OsSTK38 kinase complex, interacts with KNAT7, thereby triggering the activation of EXPANSIN gene expression. Nevertheless, in cases of OsSTK38 mutation that reduces OsCycC phosphorylation modification, the interaction with KNAT7 is hindered, resulting in the inhibition of EXPANSIN gene expression. Consequently, this leads to the production of smaller seeds in tiny1. In conclusion, our results have unveiled a molecular mechanism by which the OsMOB1A–OsSTK38–OsCycC regulatory module governs the size and weight of rice grains. Additional research is required to fully elucidate the intricate molecular mechanisms underlying this pathway.

Figure 9.

A proposed model illustrating the mechanism of grain size control through the OsMOB1A–OsSTK38–OsCycC regulatory module. The OsMOB1A–OsSTK38 kinase complex phosphorylates OsCycC, and the resulting phosphorylated OsCycC interacts with KNAT7, triggering the activation of downstream EXPANSIN genes expression. In cases of OsSTK38 mutation, OsCycC becomes incapable of phosphorylation, leading to its inability to interact with KNAT7. Consequently, this inhibition results in the suppression of downstream EXPANSIN genes expression, ultimately leading to the development of a small-grain phenotype in tiny1. Likewise, the loss of OsCycC function resulted in the inhibition of downstream EXPANSIN genes expression, causing a small-grain phenotype in Oscycc.

These results establish Cyclin C as a substrate of NDR kinase in rice, shedding light on the connection between the Hippo signaling pathway and the CDK8 module (CKM). This study has implications for research on animals. First, the Hippo signaling pathway governs cell proliferation, cell expansion, tissue development, and overall development in animals. The discovery of Cyclin C as a substrate of NDR kinase in plants offers valuable insights into the functionality and regulatory mechanisms of the Hippo signaling pathway across diverse species, possibly revealing signaling pathways and regulatory factors in animals. Secondly, investigating the cross-regulatory mechanism between these 2 signaling pathways can elucidate the interactions and regulatory networks that govern the coordination of diverse signaling pathways in regulating cellular functions and physiological processes.

Materials and methods

Plant materials and growth conditions

Previously, we generated a rice (O. sativa) T-DNA insertion mutant library (Wan et al. 2009). In this T-DNA insertion mutant library, there were 1,734 mutant materials with phenotypic variations, from which the tiny1 mutant was derived.

The rice plants were planted under natural conditions in the experimental fields of Chinese Academy of Agricultural Sciences in Langfang, Hebei, from May to October in 2020 and 2023. Each rice plant was transplanted in 8 replicates, arranged in 5 rows with a plant density of 25 × 25 cm.

Phenotyping and cellular analysis

Mature grains were scanned using an EPSON V700 PHOTO scanner (Epson, Japan). Grain length and width were measured with an SC-G seed counting and weighing device (Wseen, China), while 1,000-grain weight was measured using an electronic analytical balance. The lemma cell sizes and cell numbers in mature rice grains were observed using a SEM (S-4800, Hitachi). ImageJ software was used to measure cell length and width. Cell numbers were determined based on the average grain length and cell length.

Identification of the TINY1 gene

The positional cloning strategy followed that described by Zhang et al. (1994). The tiny1 mutant was crossed with the rice variety Dular (ssp. indica), and the resulting F₁ plants were self-pollinated to generate an F₂ mapping population. First, 20 tiny1-like samples in F₂ were collected for preliminary mapping. Initially, the TINY1 locus was mapped between molecular markers rm1-2.3 and rm1-4.9 on the Chromosome 1 short arm. To precisely locate the TINY1 locus, molecular markers were developed based on the polymorphism differences between 2 varieties (Nipponbare and Dular). Subsequently, we narrowed the locus to a 40 kb region between the markers rm1-4.61 and rm1-4.65 in 300 homozygous F₂ plants. Within this region, the candidate genes were predicted and sequenced according to the Rice Genome Annotation Project database. Additionally, the tiny1 mutant was crossed with Nipponbare, and the resulting F₂ population was used for genetic cosegregation analysis. The molecular markers used are listed in Supplementary Table S1.

RNA extraction and RT-qPCR analysis

Total RNA was extracted using the RNAprep Pure Plant Kit (DP432, Tiangen), and first-strand cDNA synthesis was carried out using the FastKing RT Kit (KR116, Tiangen). RT-qPCR was conducted using the Taq Pro Universal SYBR qPCR Master Mix (Q712-02, Vazyme), with OsACTIN (LOC_Os03g50885) employed as the internal control. The Bio-Rad iQ5 real-time system was utilized, and the relative expression levels were calculated using the 2^−ΔΔCt method. All RT-qPCR primers used are listed in Supplementary Table S2.

Flow cytometric analysis

Fresh young panicles (0.2 g, 2 cm in length) from both the WT and tiny1 plants were immersed in 500 μL of nuclear extraction buffer. The panicles were finely chopped using a sharp blade and then filtered through a 50 μm filter after a 60 s interval. Subsequently, 2,000 μL of DAPI staining buffer was added, and the samples were incubated in darkness for 2 min. The resulting nuclear suspension was subjected to analysis using the CyFlow Space Flow Cytometer (Sysmex Partec, Muenster, Germany) and the accompanying FloMax software. For each assay, the ploidy of a minimum of 10,000 nuclei was recorded. The counts of isoploid and tetraploid nuclei were tallied, and the percentage of cells with a 4C DNA content was determined by dividing the number of cells with a 4C DNA content by the total numbers of isoploid cells and cells with a 4C DNA content.

Plasmid construction and plant transformation

All constructs used in this study were prepared using the In-Fusion HD Cloning Kit (639648, Clontech). The 12 kb genome sequence of TINY1/OsSTK38 was amplified from Nipponbare using the primers gTINY1-F and gTINY1-R. Subsequently, it was cloned into the KpnI and HindIII sites of the pCAMBIA2300 vector, resulting in the generation of the gTINY1 plasmid. Similarly, a 2.1 kb sequence was amplified from Nipponbare using the primers proOsSTK38-GUS-F and proOsSTK38-GUS-R and was then cloned into the BamHI and EcoRI sites of the pCAMBIA1391z vector, generating the proOsSTK38:GUS plasmid. The construction of the second series of recombinant vectors involved the same kit and methods. The complete coding sequences of OsSTK38, OsMOB1A, and OsCycC were cloned into the SmaI and XbaI sites of the dual vector pCAMBIA2300-proActin to generate corresponding OE materials. To generate knockout material, the CRISPR/Cas9 method was employed (Xie et al. 2017). All constructs were confirmed by sequencing. The primers used for the construct are found in Supplementary Table S3. The gTINY1 construct was transferred to tiny1 for genetic complementation. For genetic analysis, the proACT:OsMOB1A and proACT:OsCycC constructs were introduced into both tiny1 and Nipponbare. The proACT:OsSTK38 constructs were introduced into both Osmob1a and Nipponbare. Other plasmids were transferred to Nipponbare. All transgenic rice materials were obtained by Agrobacterium-mediated transformation.

Yeast 2-hybrid assay

For the Y2H library screening assay, the full-length coding sequence of OsSTK38 was amplified using gene-specific primers and inserted into the EcoRI and BamHI restriction sites of the pGBKT7 vector to generate the BD-OsSTK38 construct. Screening was performed according to the yeast protocol manual (Clontech). Similarly, the full-length coding sequences of OsMOB1A, OsCycC, and KNAT7 were amplified using gene-specific primers and cloned into the EcoRI and BamHI restriction sites of the pGADT7 vector, resulting in the AD-OsMOB1A, AD-OsCycC, and AD-KNAT7 constructs. The construction of the BD-OsCycC, BD-OsCycC^T33A, and BD-OsCycC^T33D construct was performed using the same method employed for the BD-OsSTK38 construct. The bait and prey constructs were cotransformed into yeast cells (strain: Y2H GOLD). The cotransformed yeast clones were subsequently serially diluted (1:10) and spotted onto the indicated deficient medium to assess growth at 30 °C.

BiFC analysis

The full-length coding sequences of OsSTK38, KNAT7, and OsGSK5 were inserted into the BamHI site of the pVYNE vector, resulting in the generation of OsSTK38-nYFP, KNAT7-nYFP, and OsGSK5-nYFP, respectively. Similarly, the full-length coding sequences of OsMOB1A, OsCycC, OsCycC^T33A, OsCycC^T33D, and OsARF4 were inserted into the BamHI site of the pSCYCE vector, generating OsMOB1A-cYFP, OsCycC-cYFP, OsCycC^T33A-cYFP and OsCycC^T33D-cYFP, and OsARF4-cYFP, respectively. OsGSK5-nYFP and OsARF4-cYFP served as negative controls. These constructs were then transformed into rice protoplasts for transient coexpression. After 12 h of incubation, the fluorescence signals were examined using a confocal laser scanning microscope (Leica TCS SP2). For fluorescence microscopy using a Leica TCS SP2 confocal microscope, the setup is as follows: YFP and Chlorophyll b channels are excited with 488 and 514 nm argon lasers, respectively. Laser intensities are kept low, between 1% to 3% for YFP and 1% to 5% for Chlorophyll a, to avoid photobleaching and phototoxicity. Collection is performed using bandpass filters with emission ranges of 500–550 nm for YFP and 650 to 700 nm for Chlorophyll b, with center wavelengths at approximately 514 and 680 nm, respectively. Gain settings are adjusted to 700 to 800 for YFP and 600 to 700 for Chlorophyll b to optimize signal-to-noise ratio without saturation.

Pull-down assay

The full-length coding sequence of OsSTK38 was cloned into the NdeI and SalI sites of the pPGH vector, resulting in the generation of GST-OsSTK38. Similarly, the full-length coding sequences of OsMOB1A and OsCycC were cloned into the KpnI and HindIII sites of the pML303 vector, generating MBP-OsMOB1A and MBP-OsCycC, respectively. The constructs were transformed into E. coli BL21 (DE3) cells, and the resulting recombinant proteins were induced with 0.8 mM IPTG (isopropyl-β-D-thiogalactopyranoside) at 18 °C.

For the purification of proteins with GST tags, BeyoGold GST-tag Purification Resin (P2250, Beyotime) was utilized, while proteins with MBP tags were purified using Amylose Resin (E8021S, NEB).

Pull-down assays were performed by incubating the GST fusion protein (10 μg) with BeyoGold GST-tag Purification Resin (P2250, Beyotime) at 4 °C for 1 h, followed by the addition of the indicated recombinant protein (10 μg). Incubation was then continued for an additional 2 h. Subsequently, the beads were thoroughly washed, subjected to SDS‒PAGE, and probed with either an anti-GST antibody (dilution 1:1,000, lot #AE001, Abclonal) or an anti-MBP antibody. The bait protein was specifically probed with an anti-MBP antibody (dilution 1:1,000, lot #AE016, Abclonal).

Coimmunoprecipitation assay

The coding sequences of OsMOB1A and OsCycC were individually cloned into the pCAMBIA1300-eGFP vector, resulting in the fusion proteins OsMOB1A-GFP and OsCycC-GFP, respectively. Similarly, the coding sequence of OsSTK38 was inserted into the pCAMBIA1300-MYC vector to generate the fusion protein OsSTK38-MYC. Each construct was transformed into Agrobacterium GV3101 cells and coinfiltrated into N. benthamiana leaves in the indicated combinations. Coimmunoprecipitation (Co-IP) was conducted following a previously described protocol (Gao et al. 2021). The immunoprecipitates were separated in 10% SDS–PAGE and examined by immunoblotting with anti-GFP (dilution 1:1,000, lot #AE012, Abclonal) and anti-MYC antibodies (dilution 1:1,000, lot #AE010, Abclonal).

Subcellular localization

The full-length coding sequence of OsSTK38 without its termination codon was amplified using PCR and cloned into the XbaI and BamHI sites of the pAN580 vector, resulting in the pro35S:OsSTK38-GFP construct. The full-length coding sequence of cytoplasmic nucleus-localized OsGSK5 was inserted into the KpnI site of the pRTVcRFP vector, generating the proUbi:OsGSK5-tagRFP construct. Both constructs were transiently cotransformed into rice protoplasts extracted from 14-d-old seedlings by the PEG-mediated transformation method (Cui et al. 2019). After 12 h at 28 °C in the dark, the transformed protoplasts were observed using confocal laser scanning microscopy (Leica TCS SP2). For the subcellular localization experiments involving OsCycC-GFP, OsCycC^T33A-GFP, and OsCycC^T33D-GFP, the same construct method employed for OsSTK38-GFP was followed. OsCycC-GFP, OsCycC^T33A-GFP, and OsCycC^T33D-GFP were then cotransformed with the nuclear localization marker NSL-mCherry in rice protoplasts. Fluorescence signals were examined using a confocal laser scanning microscope (Leica TCS SP2) after 12 h of incubation. For fluorescence microscopy using a Leica TCS SP2 confocal microscope, the experimental setup is as follows: GFP, mCherry/tagRFP, and Chlorophyll b channels are excited with an argon laser at 488 nm for GFP, 514 nm for Chlorophyll b, and a helium–neon laser at 543 nm for mCherry/tagRFP. Laser intensities are kept low, between 1% to 3% of maximum power for GFP and mCherry/tagRFP and 1% to 5% for Chlorophyll b, to prevent photobleaching and phototoxicity. Bandpass filters are set for GFP (500 to 550 nm), mCherry/tagRFP (570 to 620 nm), and Chlorophyll b (650 to 700 nm) emissions with center wavelengths at ∼510, 590, and 680 nm, respectively. The gain settings are adjusted to 700 to 800 for GFP and mCherry/tagRFP and 600 to 700 for Chlorophyll b to optimize the signal-to-noise ratio without saturation.

GUS staining

GUS staining was performed as previously described (Cui et al. 2019). GUS transgenic plants were briefly stained in a buffer solution at 37 °C for 10 h. Subsequently, chlorophyll was extracted using 70% (v/v) ethanol.

Phylogenetic analysis

The phylogenetic tree was constructed by comparing the STK38 and MOB1A proteins using ClustalW. The neighbor-joining method in MEGAX was used to construct the phylogenetic tree based on the paired output. The construction of the phylogenetic tree was performed with the following parameters: complete deletion and bootstrap procedure (1,000 replicates). The alignments and phylogenetic trees in Newick format are provided as Supplementary Files 1 to 4.

Cell-free degradation assays

Cell-free degradation assays were conducted with slight modifications following a previously described method (Gao et al. 2021). The total proteins of WT and tiny1 were quickly extracted and suspended in 1 mL of cell-free buffer containing 25 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 10 mM NaCl, 1 mM Phenylmethanesulfonyl fluoride, 5 mM ATP, and 1 mM DTT. The above extracts were incubated with equal amounts of MBP-OsCycC recombinant protein, which was produced by E. coli, for the specified duration at room temperature. Subsequently, the resulting samples were analyzed using an anti-MBP antibody (dilution 1:1,000, lot #AE016, Abclonal) and anti-actin antibody (dilution 1:1,000, lot #AC009, Abclonal).

In vitro phosphorylation assays

GST-OsSTK38 and GST-OsMOB1A were generated by cloning the full-length coding sequences of OsSTK38 and OsMOB1A into the NdeI and SalI sites of the pPGH vector, respectively. MBP-OsMOB1A and MBP-OsCycC were generated by cloning the full-length coding sequences of OsMOB1A and OsCycC into the KpnI and HindIII sites of the pML303 vector, respectively, for expression in E. coli. To generate a kinase-dead variant of OsSTK38 (OsSTK38^K145M), the 5′ fragment of OsSTK38^K145M was amplified using the primers GST-OsSTK38-F and GST-OsSTK38^K145M-R, while the 3′ fragment was amplified using the primers GST-OsSTK38^K145M-F and GST-OsSTK38-R. Subsequently, the 5′ and 3′ fragments of OsSTK38^K145M were fused and cloned into the NdeI and SalI sites of the pPGH vector via a triple fragment fusion reaction, resulting in the plasmid GST-OsSTK38^K145M. GST-OsSTK38^T101A, GST-OsSTK38^S322A, and GST-OsSTK38^T485A were generated using a similar method. Additionally, MBP-OsCycC^T33A, MBP-OsCycC^S45A, MBP-OsCycC^T74A, and MBP-OsCycC^T89A were amplified using specific primers and cloned into the KpnI and HindIII sites of the pML303 vector. The protein purification method was the same as that employed for the pull-down assay. The purified proteins were incubated with kinase reaction buffer [25 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 1 mM DTT, and 100 μM ATP] at 37 °C for 1 h. For the separation of phosphorylated proteins, 50 μM phos-tag (AAL-107, Wako) and 100 μM MnCl₂ were added to 6% SDS–PAGE gels. After electrophoresis, the gels were subjected to 3 washes of 10 min each with transfer buffer containing 10 mM EDTA, followed by an additional 10 min wash with transfer buffer (without EDTA). The phosphorylated products were analyzed using phos-tag SDS‒PAGE. For OsSTK38 autophosphorylation, the samples were detected through immunoblotting using anti-phospho-(Ser/Thr) antibody (dilution 1:1,000, lot #9631s, Cell Signaling Technology). The relative protein levels were quantitatively analyzed using ImageJ. To examine the phosphorylation sites of OsCycC, the SDS–PAGE gel piece containing phosphorylated MBP-OsCycC was harvested and digested by trypsin (Promega) at 30 °C overnight. The samples were then analyzed using liquid chromatography-tandem mass spectrometry, as previously described (Wang et al. 2013).

Haplotype network analysis

SNPs were obtained from the RFGB database (https://www.rmbreeding.cn/), which encompasses 3,000 rice accessions (3K rice genomes). Haplotype analysis of OsCycC was conducted by selecting 3 SNPs within the mRNA region. Haplotype network analysis was conducted using the Rice Variation Map v2.0 online tool (http://ricevarmap.ncpgr.cn).

Statistical analysis

Statistical analyses were performed as described in each figure legend. Statistical data are provided in Supplementary Data Set 1.

Accession numbers

Sequence data from this article can be found in the RAP-DB libraries under the following accession numbers: OsSTK38 (Os01g0186700), OsMOB1A (Os03g0577200), OsCycC (Os09g0504400), KNAT7 (Os03g0123500), OsActin (Os03g0718100), CDKA;1 (Os03g0118400), CDKA;2 (Os02g0123100), CDKB;2 (Os08g0512600), CDKD;1 (Os05g0392300), CDC20 (Os02g0700100), E2F (Os12g0158800), Rb1 (Os11g0533500), CycB2;1 (Os04g0563700), CycB2;2 (Os06g0726800), CycH1;1 (Os03g0737600), CycD4;1 (Os09g0466100), CycU4;1 (Os10g0563900), CycU4;3 (Os10g0563900), SPL13 (Os07g0505200), PGL1 (Os03g0171300), GW8 (Os08g0531600), GW2 (Os02g0244100), SRS5 (Os11g0247300), GIF1 (Os11g0615200), GS5 (Os05g0158500), GS2 (Os02g0701300), GS3 (Os03g0407400), EXPA6 (Os03g0336400), EXPA7 (Os03g0822000), EXPB3 (Os10g0555900), and EXPB17 (Os04g0530100).

Supplementary data

The following materials are available in the online version of this article.

Supplementary Figure S1. Expression levels of cell cycle-related genes in Nipponbare (Nip) and tiny1 panicles (supports Fig. 1).

Supplementary Figure S2. Flow cytometry analysis of 4C DNA in Nipponbare (Nip) and tiny1 glumes (supports Fig. 1).

Supplementary Figure S3. Expression levels of the indicated genes involved in regulating grain size in Nipponbare (Nip) and tiny1 panicles (supports Fig. 1).

Supplementary Figure S4. Cosegregation analysis of T-DNA insertion in the segregating tiny1 segregation population (supports Fig. 2).

Supplementary Figure S5. A genomic fragment of LOC_Os01g09200 complements the short panicle and semi-dwarf phenotypes of tiny1 (supports Fig. 2).

Supplementary Figure S6. Expression levels of LOC_Os01g09200 in Nipponbare (Nip) and tiny1 panicles (supports Fig. 1).

Supplementary Figure S7. Subcellular localization of OsSTK38-GFP in rice protoplasts (supports Fig. 2).

Supplementary Figure S8. Expression pattern of OsSTK38 (supports Fig. 2).

Supplementary Figure S9. Overexpression of OsSTK38 causes tall plants but reduces the grain number per panicle (supports Fig. 3).

Supplementary Figure S10. Amino acid sequence alignments were performed for OsMOB1A and its homologues across different species (supports Fig. 4).

Supplementary Figure S11. Phylogenetic tree of NDR1 kinase (A) and MOB1A kinase activator (B) in different species (supports Fig. 4).

Supplementary Figure S12. Amino acid sequence alignments were performed for OsSTK38 and its homologues across different species (supports Fig. 4).

Supplementary Figure S13. Identification of 3 OsSTK38 phosphorylation sites by LC/MS (supports Fig. 4).

Supplementary Figure S14. OsMOB1A positively regulates grain size and weight (supports Fig. 5).

Supplementary Figure S15. Cytological analysis of transgenic lines revealed that OsMOB1A regulates grain size by coordinating changes in cell size and cell number within the spikelet hull (supports Fig. 5).

Supplementary Figure S16. Identification of the OsCycC T33 phosphorylation site by LC/MS (supports Fig. 6).

Supplementary Figure S17. Cell-free degradation assay of OsCycC (supports Fig. 7).

Supplementary Figure S18. Subcellular localization of OsCycC-GFP, OsCycC^T33A-GFP, and OsCycC^T33D-GFP (supports Fig. 7).

Supplementary Figure S19. OsCycC positively regulates grain size and weight (supports Fig. 8).

Supplementary Figure S20. Cytological analysis of transgenic lines revealed that OsCycC regulates grain size by coordinating changes in cell size and cell number within the spikelet hull (supports Fig. 8).

Supplementary Figure S21. Two synonymous SNPs were identified in both the OsMOB1A (A) and OsSTK38 (B) mRNA regions based on 3K rice germplasms (supports Fig. 8).

Supplementary Figure S22. Haplotype analysis of OsCycC across the open reading frame region in 3,000 rice accessions (3K rice genome; supports Fig. 8).

Supplementary Figure S23. GUS staining of proOsMOB1A:GUS lines and proOsCycC:GUS lines in panicles (supports Fig. 8).

Supplementary Table S1. Primers used for mapping and cosegregation.

Supplementary Table S2. Primers used for RT-qPCR.

Supplementary Table S3. Primers used for constructs.

Supplementary Table S4. Agronomic traits comparison among WT, Osmob1a-1 line, OsMOB1A-OE #1 line, Oscycc-1 line, OsCycC-OE #14 line, and OsSTK38-OE #2 line in 2023 in Langfang.

Supplementary Data Set 1. Summary of statistical analyses.

Supplementary File 1. Alignment used for phylogenetic analysis of MOB1A.

Supplementary File 2. MOB1A phylogenetic tree in Newick format.

Supplementary File 3. Alignment used for phylogenetic analysis of STK38.

Supplementary File 4. STK38 phylogenetic tree in Newick format.

Author contributions

G.C. conceived the main experiments. Z.Z. and G.C. cloned the OsSTK38 gene. J.W. and S.W. went on the transformation experiments. J.S. went on with the expression analysis. G.C. and J.G. went on with the phosphorylation analysis. L.Z. went on the subcellular localization. Z.C. and Y.C. went on with the field management. Z.Z., G.C., and X.C. contributed to the total data analysis. X.S. provided valuable suggestions for the manuscript. W.P.Q. provided valuable suggestions for writing the manuscript. Z.Z. and T.L. designed and guided the manuscript writing.

Funding

This research was supported by the National Key Research and Development Program of China (2022YFF1001700) and National Natural Science Foundation of China (32372152).

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Dive Curated Terms

The following phenotypic, genotypic, and functional terms are of significance to the work described in this paper:

• TINY1 Gramene: Os01g0186700

• TINY1 Araport: Os01g0186700

• grain size Planteome: TO:0000397

• WRKY75 Gramene: AT5G13080

• WRKY75 Araport: AT5G13080

• ABI5 Gramene: AT2G36270

• ABI5 Araport: AT2G36270

• KNAT7 Gramene: Os03g0123500

• KNAT7 Araport: Os03g0123500

• spikelet AmiGo: PO:0009051 AmiGO2

• OsSTK38 Gramene: Os01g0186700

• OsSTK38 Araport: Os01g0186700

• OsMOB1A Gramene: Os03g0577200

• OsMOB1A Araport: Os03g0577200

• OsCycC Gramene: Os09g0504400

• OsCycC Araport: Os09g0504400