Rice Homeobox Protein KNAT7 Integrates the Pathways Regulating Cell Expansion and Wall Stiffness^,

Rice KNAT7 interacts with a growth master regulator, GRF4, and a secondary wall regulator, NAC31, to coordinate cell expansion and wall thickening in fiber cells.

Abstract

During growth, plant cells must coordinate cell expansion and cell wall reinforcement by integrating distinct regulatory pathways in concert with intrinsic and external cues. However, the mechanism underpinning this integration is unclear, as few of the regulators that orchestrate cell expansion and wall strengthening have been identified. Here, we report a rice (Oryza sativa) Class II KNOX-like homeobox protein, KNOTTED ARABIDOPSIS THALIANA7 (KNAT7), that interacts with different partners to govern cell expansion and wall thickening. A loss-of-function mutation in KNAT7 enhanced wall mechanical strength and cell expansion, resulting in improved lodging resistance and grain size. Overexpression of KNAT7 gave rise to the opposite phenotypes, with plants having weaker cell walls and smaller grains. Biochemical and gene expression analyses revealed that rice KNAT7 interacts with a secondary wall key regulator, NAC31, and a cell growth master regulator, Growth-Regulating Factor 4 (GRF4). The KNAT7-NAC31 and KNAT7-GRF4 modules suppressed regulatory pathways of cell expansion and wall reinforcement, as we show in internode and panicle development. These modules function in sclerenchyma fiber cells and modulate fiber cell length and wall thickness. Hence, our study uncovers a mechanism for the combined control of cell size and wall strengthening, providing a tool to improve lodging resistance and yield in rice production.

Plants have >40 cell types, each with a unique shape and function that depends in part on its wall properties (Chebli and Geitmann, 2017). Plant cell walls are a rigid and plastic network of polysaccharides (cellulose, hemicellulose, and pectins), aromatic compounds (lignin) and glycoproteins that encase plant cells (Bacic et al., 1988; Carpita and Gibeaut, 1993). Cell wall biogenesis and remodeling are closely related to all cell behaviors. For example, pectin demethylesterification affects cell differentiation and organ initiation (Peaucelle et al., 2011), and synthesis and integration of wall products at the division plane is an important step in cytokinesis (Cutler and Ehrhardt, 2002; Mayer and Jürgens, 2004). Cell growth involves cell expansion and wall reinforcement. Turgor pressure-triggered cell expansion requires relaxation of the cell wall, which involves the activities of xyloglucan endotranslycosylases and expansins (McQueen-Mason and Cosgrove, 1994; Whitney et al., 2000; Chanliaud et al., 2004; Che et al., 2015). While the cell walls are expanding, newly synthesized polysaccharides are integrated into the walls to provide rigidity. Upon maturation, secondary wall components are deposited in some types of cells, such as sclerenchyma fiber cells and vessel elements, to confer mechanical strength.

Plants have evolved complex mechanisms to integrate distinct signals to ensure that wall properties are compatible with cell functions (Somerville et al., 2004). Combinatorial controls at different scales are required. Manipulation of enzymatic activities during cell wall biogenesis directly controls cell wall composition and organization; spatiotemporal coregulation of cell wall-related gene expression represents another valid control, as cell wall chemistry is heterogeneous (Brown et al., 2005; Persson et al., 2005). During cell expansion, several kinds of transcription factors (TFs), including basic helix-loop-helix proteins, rice (Oryza sativa) Growth-Regulating Factor 4 (GRF4), APETALA2-type proteins, squamosa promoter binding-like proteins, and the Myeloblastosis (MYB)-like TFs, form a regulatory network (Aya et al., 2014; Che et al., 2015; Duan et al., 2015; Si et al., 2016; Jang et al., 2017; Wu et al., 2017). Wall stiffening is modulated by a regulatory network consisting of NAM, ATAF, and CUC (NAC) and MYB TFs (Zhong et al., 2010, 2011; Zhao, 2016). Moreover, cell wall biogenesis is affected by various developmental signals (Somerville et al., 2004), including plant hormones, such as brassinosteroids, auxin, and gibberellins, as well as light (Wang and He, 2004; Bai et al., 2012; Todaka et al., 2012; Aya et al., 2014; Shan et al., 2014; Huang et al., 2015). However, the mechanism that integrates different regulatory pathways to modulate cell expansion and wall strengthening remains elusive.

Knotted-related homeobox (KNOX) proteins may be involved in integrating these pathways, as their functions are related to diverse physiological processes (Hay and Tsiantis, 2010). KNOX members belong to the plant-specific three-amino acid loop extension superclass of homeodomain proteins (Bürglin, 1997; Hake et al., 2004) and have been clustered into two classes (Kerstetter et al., 1994; Reiser et al., 2000; Magnani and Hake, 2008). The functions of Class I members are implicated in many processes of the plant life cycle; one of their most important roles is to control cell proliferation at the leaf and shoot apical meristems (Vollbrecht et al., 1991; Long et al., 1996; Sato et al., 1999; Belles-Boix et al., 2006). Characterizations of the upstream regulators, interacting cofactors, and downstream effectors have placed Class I KNOX proteins in the central nodes in various physiological processes (Hay and Tsiantis, 2010).

Class II KNOX proteins are less well studied, but their widespread expression pattern indicates that their functions are as diverse as those of Class I (Truernit et al., 2006; Zhong et al., 2008; Chai et al., 2016). The most substantially studied Class II member is KNOTTED ARABIDOPSIS THALIANA7 (KNAT7). It governs secondary wall formation either by interacting with secondary wall TFs, e.g. OVATE FAMILY PROTEIN4 and MYB75, or by being regulated by the secondary wall master TFs, such as SECONDARY WALL-ASSOCIATED NAC DOMAIN PROTEIN/VASCULAR-RELATED NAC DOMAIN6 and MYB46 (Brown et al., 2005; Zhong et al., 2008; Ko et al., 2009; Bhargava et al., 2010; Li et al., 2011, 2012; Gong et al., 2014; Liu et al., 2014). However, the role of KNAT7 in integrating cell growth regulatory pathways is unknown.

Here, we report on the rice class II KNOX-like homeobox member KNAT7, which integrates the regulatory pathways of cell expansion and wall strengthening. A loss-of-function mutation in KNAT7 resulted in enhanced secondary wall biosynthesis and facilitated cell expansion in the mutant; overexpression (OE) of KNAT7 gave rise to the opposite effects. KNAT7 interacted with the secondary wall key regulator NAC31 and the master cell growth factor GRF4 to repress their downstream regulatory pathways. These findings suggest that rice KNAT7 plays an integrative role in coordinating cell size and wall stiffness. Therefore, this study provides insight into the combinatorial control of cell growth and may be instrumental in synergistically improving agronomic traits, especially grain size (and thus yield) and stem strength (and thus resistance of lodging), in crop breeding.

RESULTS

Rice KNAT7 Negatively Regulates Cell Wall Thickening and Mechanical Properties

Cellulose contributes greatly to wall mechanical properties (Cosgrove, 2005). To understand how rice plants build wall rigidity, we performed coexpression analysis with a characterized cellulose synthase gene, CESA4 (Zhang et al., 2009), to screen for key regulators. This analysis identified several NAC and MYB TFs and a Class II homeobox protein, KNAT7 (Supplemental Table S1). Phylogenetic analysis placed rice KNAT7 as an ortholog of Arabidopsis (Arabidopsis thaliana) KNAT7 (Supplemental Fig. S1), but its function in rice is uncharacterized.

To identify the functions of rice KNAT7, we generated a mutant in the rice Zhonghua11 background using the clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR associated protein9 (Cas9) gene editing approach (Supplemental Fig. S2A). Sequence analysis revealed a 19-bp deletion in the second exon of KNAT7, which causes a 48-bp reading-frame shift at the 428-bp site of the coding region and results in a premature translational stop codon (Fig. 1, A and B). Compared to the predicted wild-type KNAT7 protein sequence, the mutated KNAT7 is 137 amino acids in length and lacks the conserved domains of KNOX proteins (Supplemental Fig. S1C). The KNAT7 OE lines were produced by constitutively expressing KNAT7 in the rice variety Nipponbare (Fig. 1C; Supplemental Fig. S2B). In addition to the slightly reduced panicle length and plant height (Supplemental Fig. S2C), the knat7 mutants had an increased mechanical strength, whereas the KNAT7-OE lines had compromised mechanical force (Supplemental Fig. S2D), suggesting that cell wall structure might be affected in these KNAT7-modulated plants. We therefore investigated the epidermal sclerenchyma fiber cells in the internodes of these plants. Scanning electron microscopy (SEM) revealed that the wall thickness of fiber cells was significantly increased in the knat7 mutant but was decreased in the KNAT7-OE lines when compared with the corresponding wild-type plants (Fig. 1, D–F).

Figure 1.

Rice KNAT7 represses wall thickening. A, Schema of KNAT7 gene structure and the mutation site of knat7. The boxes and lines in the diagram indicate exons and introns, respectively. The arrowhead indicates a 19-bp deletion in knat7, which results in a 48-bp reading-frame shift (underlined letters) and a premature translational stop codon (red letters). B, Genotyping the knat7 plants using the primers (F + R) shown in Supplemental Table S4 reveals the 19-bp deletion in the mutant. C, RT-qPCR analysis of KNAT7 expression in the OE plants, showing the relative expression level of KNAT7 to rice HNR. Data represent the mean ± sd of three biological replicates. Lowercase letters indicate significantly different means according to the variance analysis and Tukey’s test (P < 0.01). D and E, SEM graphs of sclerenchyma fiber cells from the internodes of the indicated plants. Bars = 2 μm. F, Measurement of the wall thickness in sclerenchyma fiber cells of the indicated plants. Data represent means ± sd (n = 200 cells from three individual internodes of the indicated plants). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. (G) Cellulose content in internodes of the indicated plants. Data represent the mean ± sd (n = 3 biological replicates). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. ZH11, Zhonghua11; NP, Nipponbare; L, line.

In agreement with the anatomical alterations, cellulose content increased in the knat7 mutants and decreased in the KNAT7-OE lines compared to the corresponding wild-type plants (Fig. 1G). Measurement of the neutral sugar content of cell wall residues showed that the Xyl content, which represents the abundance of xylan, increased in the knat7 mutant but slightly decreased in the KNAT7-OE lines compared to the corresponding wild-type plants (Supplemental Table S2). Moreover, the level of lignin, another component of wall stiffness, was unchanged in the knat7 mutant and slightly increased in the OE plants when compared to the corresponding wild-type plants (Supplemental Table S2). Therefore, rice KNAT7 decreases secondary wall biosynthesis and mechanical strength.

Grain Size Is Altered in the knat7 Mutant and OE Plants

In addition to the changes in wall stiffness, manipulation of rice KNAT7 unexpectedly caused alterations in grain size. The knat7 mutant had larger grains, whereas the KNAT7-OE lines had smaller grains than those of the corresponding wild-type plants (Fig. 2A). Quantitative measurements revealed that the increased grain size of the knat7 mutant resulted from an increase in grain length, grain width, and 1000-grain weight, whereas the opposite phenotypes were observed in the KNAT7-OE lines (Fig. 2, B–D).

Figure 2.

Rice KNAT7 affects grain size. A, Rice grains of the indicated plants. Bar = 5 mm. B to D, Measurement of the grain length, grain width, and grain weight of the indicated plants. Data represent the mean value ± sd (n ≥ 15 grains harvested from at least five plants), except in D, where data represent the mean value ± sd (n = 3 biological replicates). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. E and F, SEM graphs of the glume cells of the indicated plants. Bars = 50 μm. G and H, Statistical analysis of the cell size in the glumes of the indicated plants based on the SEM analyses. Data represent the mean ± sd (n ≥ 205 glume cells in grains from four plants). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. I, Statistical analysis of cell number per glume at the longitudinal direction. Data indicate the mean value ± sd (n ≥ 13 grains harvested from at least five plants). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. ZH11, Zhonghua11; NP, Nipponbare; L, line.

To address the underlying cause of these changes at the cellular level, we investigated the glumes of wild-type, knat7, and KNAT7-OE grains using SEM. Compared to the corresponding wild-type plants, the cell length and cell width in the glumes were significantly higher in the mutant plant and lower in the KNAT7-OE lines (Fig. 2, E–H), with slight changes in the cell number (Fig. 2I). These findings suggest that the altered grain size in these KNAT7-modulated plants is due to the variations in cell size. Hence, rice KNAT7 plays a role in cell-size control in the glumes.

KNAT7 Affects Regulatory Pathways of Cell Wall Thickening and Cell Expansion

We next investigated the biochemical features of rice KNAT7. We transfected GFP-fused KNAT7 into leaf epidermal cells of Nicotiana benthamiana and detected GFP signals in the nucleus (Supplemental Fig. S3A), indicating that it is a putative TF.

To explore whether rice KNAT7 has transactivation activity, we coexpressed the effectors Pro-35S:GAL4BD (BD) or BD fusions to the herpes simplex virus VP16 activation domain (BD-VP16) and to KNAT7 (BD-KNAT7) with the reporter ProGAL4:Luciferase in Arabidopsis protoplasts. In contrast to the strong expression of luciferase promoted by BD-VP16, the reporter transcripts activated by BD-KNAT7 were low and similar to that promoted by BD (Supplemental Fig. S3B). Hence, KNAT7 was unable to activate the expression of the reporter gene. We prepared another reporter construct ProKNOX-ProGAL4:Luciferase by placing four copies of the TGAC motif, the binding element of KNOX proteins, together with the GAL4 binding element. This construct provides binding elements for KNAT7 and BD-VP16. After cotransfecting Arabidopsis protoplasts with the reporter and effector combinations, the luciferase activities promoted by coexpressing BD-VP16 and KNAT7 were significantly decreased compared to those activated by expressing BD-VP16 and GUS (Supplemental Fig. S3C). Hence, rice KNAT7 can function as a transcriptional repressor.

To determine how rice KNAT7 controls both secondary wall formation and cell expansion in planta, we investigated the transcription of genes implicated in the regulatory pathways of secondary wall formation and cell expansion in the KNAT7 modulated plants. NAC31-MYB61-CESAs is one of the identified regulatory pathways for secondary wall cellulose synthesis in rice (Huang et al., 2015); MYB103 is another crucial regulator for rice secondary wall thickening (Ye et al., 2015). Reverse transcription quantitative PCR (RT-qPCR) analyses in the young internodes revealed that the expression of those genes was upregulated in the knat7 mutant but downregulated in the KNAT7-OE lines (Fig. 3, A and C), suggesting that the regulatory pathways for secondary wall formation are affected in these KNAT7-modulated plants.

Figure 3.

Gene expression analysis. A to D, RT-qPCR analysis of the transcription of the examined genes involved in secondary wall formation and cell expansion in young internodes (A and C) and spikelets from 10-cm panicles (B and D) of the KNAT7-modulated plants, showing the expression levels relative to the corresponding wild-type plants. Rice TP1 was used for normalization of the expression of CESA genes, and HNR was used for normalization of the expression of the other examined genes. Data represent the mean ± sd of three biological replicates. *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. ZH11, Zhonghua11; NP, Nipponbare.

Rice GRF4 is a major quantitative trait locus controlling grain size, and expansin genes are its downstream effectors (Che et al., 2015). While GRF4 transcript levels were maintained at a similar level in the young spikelets of the knat7 mutant and KNAT7-OE lines (Fig. 3B), transcription of the examined expansin genes was upregulated in the knat7 mutant but suppressed in the KNAT7-OE lines (Fig. 3D).

Therefore, KNAT7 controls the expression of genes implicated in wall strengthening and cell expansion.

KNAT7 Interacts with NAC31 and GRF4 to Repress the Downstream Regulatory Pathways

The altered transcript levels discussed above prompted us to investigate whether KNAT7 can transcriptionally regulate the examined genes. To this end, we transfected Arabidopsis protoplast cells with the effector construct Pro-35S:KNAT7 and the reporter construct that harbors luciferase driven by the promoters of NAC31, MYB61, MYB103, and the expansin genes. Transactivation activity analysis showed that KNAT7 cannot transcriptionally regulate NAC31, MYB61, MYB103, or the expansin genes (Supplemental Fig. S4). Taken together with the unchanged transcript levels of GRF4 in the transgenic plants (Fig. 3B), KNAT7 might not act alone as a TF to modulate the downstream gene expression in rice.

Transactivation activity analysis further showed that NAC31 transcriptionally activated itself in Arabidopsis protoplasts expressing NAC31 and a reporter construct with the NAC31 promoter driving luciferase (Supplemental Fig. S4A). Cotransfecting KNAT7 in these protoplasts suppressed the self-transactivation activity of NAC31 (Supplemental Fig. S4A), implying that KNAT7 may function through a protein interaction. To test this hypothesis, we performed several experiments that demonstrate protein-protein interactions. Split-luciferase complementation assays showed an interaction between KNAT7 and NAC31 (Fig. 4A), which was confirmed by yeast two hybrid analyses (Fig. 4B). In vivo bimolecular fluorescence complementation (BiFC) and single-molecule fluorescence resonance energy transfer (FRET) analyses revealed that the interactions occur in the nuclei (Fig. 4, C–E; Supplemental Fig. S5). Coimmunoprecipitation (co-IP) experiments in rice protoplasts provided additional biochemical evidence for the interactions (Fig. 4F). To determine the interacting effects, we analyzed the ability of NAC31 to activate MYB61 and MYB103 transcription in the presence of KNAT7 in Arabidopsis protoplasts. Transactivation activity assays demonstrated that the luciferase activities promoted by NAC31 were significantly repressed by coexpression of KNAT7 (Fig. 4G).

Figure 4.

Rice KNAT7 interacts with secondary wall regulator NAC31. A, Split-luciferase complementation assay showing the interaction in N. benthamiana leaves infiltrated with the construct combinations shown at left. Rice GID1 was used as a negative control. Bar = 1 cm. B, Yeast two-hybrid analysis. Yeast cells were grown on SD medium lacking Trp, Leu, His, and Ade. Yeast growth status indicates interactions. C, BiFC analysis of the interaction in N. benthamiana leaves. Infiltrations with the empty vector were used as negative controls (Supplemental Fig. S5). DAPI was used to visualize nuclei. Merge, merged images of enhanced YFP and DAPI. Bar = 20 μm. D, FRET analysis in rice protoplasts verified the interaction. The bottom row shows fluorescence in the cell after photobleaching YFP (AP). Bars = 10 μm. E, Quantification of the FRET efficiency observed in D. FRET efficiency represents the fluorescence change of the donor fluorophore (cyan fluorescent protein [CFP]) after photobleaching YFP. The background indicates the stability of CFP fluorescence before photobleaching. Rice GID1 protein was used as a negative control. Data represent the mean ± sd (n = 10 cells). *P < 0.01 by Welch’s unpaired t test. F, Co-IP analyses in rice protoplasts expressing FLAG-NAC31. GFP-GID1 was used as a negative control. G, Transcription activation was assayed by transfecting Arabidopsis protoplasts with the constructs shown at left. Data represent the mean ± sd of three biological replicates. Lowercase letters indicate the different means according to the variance analysis and Tukey’s test (P < 0.01).

We further determined the relationship between KNAT7 and the cell growth regulator GRF4. Split-luciferase complementation and yeast two hybrid assays revealed the interaction between KNAT7 and GRF4 (Fig. 5, A and B). The KNAT7-GRF4 complex was found to form in the nuclei, based on the results of BiFC and single-molecule FRET analyses (Fig. 5, C–E; Supplemental Fig. S5). Co-IP assays in rice protoplasts biochemically validated the interactions (Fig. 5F). Similarly, we performed transactivation activity analyses to examine the interaction effect. As shown in Figure 5G, KNAT7 suppressed the expression of the expansin genes EXPB3, EXPB17, and EXPA6 promoted by GRF4 in the protoplasts.

Figure 5.

Rice KNAT7 interacts with growth regulator GRF4. A, Split-luciferase complementation assay, showing the interaction in N. benthamiana leaves infiltrated with the construct combinations shown at left. Rice GID1 was used as a negative control. Bar = 1 cm. B, Yeast two-hybrid analysis. Yeast cells were grown on SD medium lacking Trp, Leu, His, and Ade. Yeast growth status indicates interactions. C, BiFC analysis of the interaction between KNAT7 and GRF4 in N. benthamiana leaves. Infiltrations with the empty vector were used as negative controls (Supplemental Fig. S5). DAPI was used to visualize nuclei. Merge, merged images of enhanced YFP and DAPI. Bar = 20 μm. D, FRET analysis in rice protoplasts verifies the interaction. The bottom images show fluorescence in the cell after photobleaching YFP (AP). Bars = 10 μm. E, Quantification of the FRET efficiency observed in D. FRET efficiency represents the fluorescence change of the donor fluorophore (CFP) after photobleaching YFP. The background indicates the stability of CFP fluorescence before photobleaching. Rice GID1 protein was used as a negative control. Data represent the mean ± sd (n = 10 cells). *P < 0.01 by Welch’s unpaired t test. F, Co-IP experiments in rice protoplasts expressing GFP-KNAT7. FLAG-GID1 was used as a negative control. G, Transcription activation was assayed by transfecting Arabidopsis protoplasts with the constructs shown at left. Data represent the mean ± sd of three biological replicates. Letters indicate the different means according to the variance analysis and Tukey’s test (P < 0.01).

Moreover, the mutated KNAT7 protein that contains the first 137 amino acids failed to interact with GRF4 and NAC31 as revealed by split-luciferase complementation analysis (Supplemental Fig. S6). Taken together, these results suggest that KNAT7 interacts with key regulators NAC31 and GRF4 to compromise the expression of downstream genes.

KNAT7 Modules Function during Internode and Panicle Development

Next, we investigated where the KNAT7-NAC31 and KNAT7-GRF4 modules function in planta. Several organs from Nipponbare were used to examine the expression of KNAT7, NAC31, and GRF4. RT-qPCR analyses showed that KNAT7 is ubiquitously expressed at relatively high levels in the internodes and panicles (Fig. 6A), in agreement with the tissues where the phenotypes are displayed in the KNAT7-modulated plants. As developing internodes and panicles essentially provide a time course of cell expansion and wall strengthening (Itoh et al., 2005; Huang et al., 2015; Zhang et al., 2018), we explored the gene expression profile in the developing internodes and panicles of wild-type plants. We prepared total RNA from eight segments of 9-cm wild-type internodes (Huang et al., 2015) and from glumes of developing panicles 0.5 to 15 cm long. RT-qPCR analyses in these tissues revealed that the expression profiles of KNAT7 and NAC31 were similar in the developing internodes (Fig. 6A), indicating that their translated proteins are able to encounter each other during internode development. Specifically, KNAT7 was upregulated at segment 2 (sequentially numbered from the bottom up) and peaked at segments 3 and 4, while NAC31 was upregulated at segment 2 and decreased from segment 3 (Fig. 6A). Segments 2 and 3 of the internodes may be the stages for KNAT7-NAC31 interaction, consistent with the timing of secondary wall initiation, as revealed by anatomical analysis (Fig. 6B; Supplemental Fig. S7A). When these findings are combined with the phenotypes displayed in the internodes of the KNAT7-modulated plants (Fig. 1), our observations validate the KNAT7-NAC31 module in internode development.

Figure 6.

Validation of the KNAT7 modules in developing internodes and panicles. A, Transcription levels of KNAT7, NAC31, and GRF4 in different organs, including the developing internodes and spikelets from the developing panicles of Nipponbare, showing the expression levels relative to rice HNR. Data represent the mean ± sd of three biological replicates. R, root; Sh, leaf sheath; L, leaf. B, Fresh hand-cut cross sections of the developing internodes. S1–S3, three young segments from the bottom up. The red arrows indicate the thickening cell wall. Bar = 50 μm. C, SEM graphs of glumes in 5-, 10-, and 15-cm panicles. Bar = 20 μm. D, Quantification of the length of glume cells examined in C, Data indicate the mean ± sd (n = 50 cells from five spikelets).

GRF4 was mainly expressed in the panicles. Although the transcript abundance of GRF4 gradually declined as the panicles matured, its transcript remained at a relatively high level (Fig. 6A). However, KNAT7 was upregulated in the 10-cm-long panicles and peaked in the 15-cm panicles (Fig. 6A). The in planta interactions between GRF4 and KNAT7 may occur in 10- to 15-cm panicles. This claim was supported by our finding that cell size in the glumes had 1-fold higher expansion in the 10-cm panicles than in the 5-cm panicles, as shown by SEM (Fig. 6, C and D; Supplemental Fig. S7B). Taken together with the alterations in cell size in the grains of KNAT7-modulated plants (Fig. 2), it is likely that the KNAT7-GRF4 module functions during panicle development.

KNAT7 Represses the Regulatory Pathways of Cell Expansion and Wall Strengthening

Given that the KNAT7-NAC31 and KNAT7-GRF4 modules can form in the internodes and panicles, their effects need to be elucidated. MYB61 and MYB103 are the downstream targets of NAC31, and these MYB TFs target secondary wall CESA genes (Huang et al., 2015; Ye et al., 2015). RT-qPCR analyses in the wild-type internode segments showed that the transcription of MYB61, MYB103, and CESAs was sharply upregulated at segment 2, peaked at segment 3, and gradually declined until maturation (Fig. 7A). The expression pattern fits with the timing of the formation of the KNAT7-NAC31 module (Fig. 6A), indicating that the expression of secondary wall regulatory TFs and biosynthesis genes is quickly promoted by the upstream TF genes, e.g. NAC31, in segment 2 and then gradually repressed by the suppressors, e.g. the KNAT7-NAC31 module. Therefore, KNAT7-NAC31 is one of the modules that can slow down wall thickening during internode development.

Figure 7.

Expression profile of genes involved in wall strengthening and cell expansion. A, The transcription level of secondary wall TFs and biosynthesis genes in the developing Nipponbare internode segments. Rice TP1 was used for normalization of the expression of CESA genes, and HNR was used for normalization of the expression of MYBs. The transcription levels at segment 1 (S1) were considered as 1. Data represent the mean ± sd of three biological replicates. B, Transcription level of expansin genes in spikelets from developing Nipponbare panicles. Rice HNR was used for normalization. The transcription levels at 0.5 cm were considered as 1. Data represent the mean ± sd of three biological replicates.

Expansin genes are downstream targets of GRF4 (Che et al., 2015; Li et al., 2018). RT-qPCR analyses revealed that the transcription of the four expansin genes was consistently upregulated in the 0.5- to 10-cm panicles, peaked in the 10-cm panicles, and dropped to a low level in 15-cm panicles (Fig. 7B). The inflection points of these expression profiles match well with the feature of glume cell expansion (Fig. 6, C and D) and the timing to form the KNAT7-GRF4 module (Fig. 6A). Therefore, the KNAT7-GRF4 module can suppress cell expansion during spikelet development.

KNAT7 Coordinates Cell Expansion and Wall Strengthening in Sclerenchyma Fiber Cells

We next addressed whether these KNAT7 modules can act in a specific cell type. Sclerenchyma fiber cells are one of the cell types affected by KNAT7, as we observed the altered wall thickness in these cells and the changed cellulose content in the internodes of KNAT7-modulated plants (Fig. 1). Therefore, we examined the size of sclerenchyma fiber cells in the internodes after maceration treatments. The cell length increased in the knat7 mutant but decreased in the KNAT7-OE lines when compared to the corresponding wild-type plants (Fig. 8A), implying that in addition to the KNAT7-NAC31 module, the KNAT7-GRF4 module also functions in sclerenchyma fiber cells of internodes.

Figure 8.

KNAT7 modules control the growth of sclerenchyma fiber cells. A, Measurement of the fiber cell length in the internodes of the indicated plants. Data indicate the mean ± sd (n = 200 cells from three individual plants). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. B, Measurement of the fiber cell length in the glumes of the indicated plants. Data indicate the mean ± sd (n = 120 cells from three individual plants). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. C and D, SEM graphs of fiber cells from the glumes of the indicated plants. Bars = 2 μm. E, Measurement of the sclerenchyma fiber cell wall thickness in the glumes of the indicated plants. Data represent means ± sd (n = 125 cells from three individual internodes of the indicated plants). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. F, Cellulose content in the glumes of the indicated plants. Data represent the mean ± sd (n = 3 biological replicates). *P < 0.01 by Welch’s unpaired t test represents a significant difference from the corresponding wild-type plants. G and I, A cross section of young internodes (G) and glumes (I). The colored dashed lines indicate the cells harvested by laser microdissection. Bars = 100 μm. FC, fiber cells; PC, parenchyma cells. H and J, RT-qPCR analysis of cells harvested in G and I to show the expression levels relative to rice HNR. The transcription levels detected in PC were considered as 1. Data represent the mean ± sd of three replicates. K Working model of rice KNAT7. KNAT7-GRF4 modulates fiber cell expansion; the interaction represses the transcription of expansin genes activated by GRF4. KNAT7-NAC31 controls wall thickening; the interaction represses the expression of secondary wall regulatory TFs and biosynthetic genes. Therefore, rice KNAT7 integrates regulatory pathways of cell expansion and wall strengthening to coordinate fiber cell growth. ZH11, Zhonghua11; NP, Nipponbare; L, line.

To test the phenotypes in panicles, we performed these examinations in the fiber cells of grain glumes. The fiber cell length in grain glumes increased in the knat7 mutant but decreased in the KNAT7-OE lines when compared to the corresponding wild-type plants (Fig. 8B). Changes in wall thickness of the fiber cells in grain glumes were similar to changes in cell length (Fig. 8, C–E). Alterations in KNAT7 led to increased cellulose content in the knat7 mutant but reduced cellulose levels in the KNAT7-OE lines when compared to the corresponding wild-type plants (Fig. 8F), suggesting that KNAT7 modules regulate fiber cell growth in the panicles.

To obtain molecular support for the above conclusions, we investigated whether KNAT7 and its partners, NAC31 and GRF4, are expressed together in fiber cells. Sclerenchyma fiber cells and parenchyma cells were harvested from the young internodes and glumes by laser microdissection (Fig. 8, G and I). RT-qPCR analyses performed in these cells revealed that the three genes are consistently predominantly expressed in fiber cells of the internodes (Fig. 8H), which was similar to the expression profiles of KNAT7 and NAC31 in glumes (Fig. 8J). Although GRF4 was equally transcribed in both cell types in glumes (Fig. 8J), the three genes were coexpressed in fiber cells in the two examined organs. These results suggest that KNAT7 is an integrative regulator in the control of cell expansion and wall thickening in fiber cells.

DISCUSSION

KNAT7 is a widely studied Class II KNOX protein. Although studies in several plant species have revealed that KNAT7 governs secondary wall thickening, its function differs among species. Arabidopsis KNAT7 can repress lignin biosynthesis in the interfascicular fiber cells (Brown et al., 2005; Zhong et al., 2008; Li et al., 2012) and activate xylan production in xylary tissues (He et al., 2018). The mechanism underlying KNAT7 function remains controversial. Rice has KNOX Class II members similar to those of Arabidopsis, and rice KNAT7, the only rice ortholog of Arabidopsis KNAT7, has an unknown function. In this study, we found that rice KNAT7 interacts with distinct TFs to coordinate cell expansion and wall stiffness.

Rice KNAT7 harbors all of the conserved domains of KNOX proteins, similar to the Arabidopsis ortholog (Ehlting et al., 2005; Persson et al., 2005), and it is coexpressed with secondary wall biosynthetic enzymes and TFs, indicating its role in wall thickening. That putative role in wall thickening has been corroborated in this study, as the rice knat7 mutant had thickened secondary walls and the OE plants had thinner walls. Cell wall composition analyses provided further support. Although rice KNAT7 possesses general TF features, it failed to induce the transcription of downstream TFs and secondary wall biosynthetic genes, based on the results of transactivation activity analyses. Interestingly, we found that rice KNAT7 interacts with NAC31, an upstream master regulator of secondary wall thickening (Huang et al., 2015), thereby repressing the expression of the NAC31 downstream genes MYB61 and MYB103. This finding places rice KNAT7 in the upstream hierarchy of the secondary wall regulatory network, which is distinct from previous reports. Studies in several plant species have found that KNAT7 can regulate secondary wall formation in various ways (Brown et al., 2005; Li et al., 2012; Gong et al., 2014; He et al., 2018), depending upon its target genes and interacting partners (Li et al., 2012; Zhong and Ye, 2012). The divergent regulatory network for secondary wall biosynthesis among different plant species might be one of the reasons why KNAT7 function varies in different plant species and in specific cell types. Rice KNAT7 is hence an upstream regulator in secondary wall thickening. However, the possibilities that this protein can transcriptionally regulate cell wall TFs or be regulated by other hierarchical TFs are not excluded in rice.

More interestingly, we found that the rice knat7 mutant and KNAT7-OE plants differed from each other in grain size, which resulted from altered cell size. Gene expression analyses revealed that several expansin genes proposed to facilitate cell expansion showed different transcript levels among the KNAT7-modulated plants. Rice KNAT7 was further found to interact with GRF4, a major quantitative trait locus controlling grain size by regulating cell expansion (Che et al., 2015; Duan et al., 2015; Li et al., 2018). This interaction suppresses the expression of the expansin genes that are activated by GRF4. Hence, KNAT7 has an unexpected role in controlling cell size. Although previous studies have proposed that the roles of KNAT7 might vary in different tissues and cell types (Li et al., 2012; He et al., 2018), its role in coordinating cell expansion and wall thickening has not yet been found in other plant species.

Wall stiffening generally proceeds along with cell expansion (Cosgrove, 2005; Huang et al., 2015). Correct processing of the two cellular events is largely dependent upon formation of combinatorial modules, such as KNAT7-GRF4 and KNAT7-NAC31, at the right time and the right place. Investigating when and where these modules form not only provided the in planta support for KNAT7 functions, but also offered a better understanding of cell morphogenesis. Examining the expression of KNAT7 and its interacting partners NAC31 and GRF4 in the developing internodes and panicles revealed the spatiotemporal regulation of the KNAT7-GRF4 and KNAT7-NAC31 modules. Based on the expression patterns of the downstream genes during internode and panicle development, the KNAT7 modules were found to repress cell expansion and wall thickening. Wall-thickness and cell-size alterations in fiber cells from internodes and glumes suggest that both regulatory modules can function in one cell type, which was further corroborated by coexpression analyses in the distinct cell types. Therefore, rice KNAT7 can integrate regulatory pathways in fiber cells. KNAT7 interacts with GRF4 to modulate cell expansion by suppressing the transcription of expansin genes, while it binds NAC31 to control wall strength by repressing wall-thickening regulatory pathways (Fig. 8K). This study suggests that KNAT7 plays a role in cell morphogenesis, although further studies on how these two modules cooperatively act in fiber cells are required. As the roles of KNAT7 are diverse, more KNAT7-interacting partners are expected to be found in plants. Hence, spatiotemporal characterization of KNAT7 and its cofactors is of great significance for understanding how KNAT7 functions and how cell growth is programmed.

Coordination of cell expansion and wall strengthening occurs in response to internal and environmental signals. Our data demonstrate that rice KNAT7 plays repressive roles in cell expansion and wall thickening by forming distinct protein complexes. Thus, KNAT7 is a combinatorial regulator of plant cell growth. Serving as an integrative regulator, KNAT7 can respond to various internal and external cues. The presence of various regulatory elements in the promoter region (Supplemental Table S3) solidified the hypothesis that KNAT7 has the potential to integrate multiple developmental and environmental demands to coordinate cell size and wall stiffness. Our study offers a mechanistic view for combinatorial modulation of plant cell growth and may provide a tool for the synergistic improvement of lodging resistance and grain yield in crops.

MATERIALS AND METHODS

Plant Materials and Growth Conditions

All rice (Oryza sativa) plants used in this study, including the wild-type plants, knat7, and the KNAT7-OE plants, were grown in the experimental fields at the Institute of Genetics and Developmental Biology in Beijing and in Lingshui, Hainan Province, China, in different growing seasons. Etiolated rice seedlings for generating protoplast cells were grown in a dark growth chamber at 28°C. Nicotiana benthamiana and Arabidopsis (Arabidopsis thaliana) plants used in this study were grown in a greenhouse under a 16 h light/8 h dark photoperiod at 23°C.

To investigate the agronomic traits, about 24 of the knat7 and KNAT7-OE plants of the T1 generation were planted in the fields. While the plants matured, the representative plants were photographed and subjected to phenotype investigation. Specifically, the grain length and width of 15 grains harvested from five plants were measured by an electronic digital display Vernier caliper. The 200 fully filled seeds from five plants were used to measure the 1,000-grain weight. The plant height and panicle length were obtained by measuring the major tillers from 15 plants.

Generation of the Rice Transgenic Plants

To generate the knat7 mutant, single-guide RNA (sgRNA) sequences targeting KNAT7 (789‒807 bp) were designed using the CRISPR direct database (http://crispr.dbcls.jp/). The targeting oligonucleotides were synthesized and annealed to form sgRNA duplexes and inserted into the guide RNA vector pYLgRNA-OsU3 (Ma et al., 2016). PCR was performed to obtain the expression cassettes of OsU3-sgRNA, which was then inserted into the plant binary vector pYLCRISPR/Cas9-MH to produce the CRISPR/Cas9 construct (Ma et al., 2016). For generation of the KNAT7-OE plants, the full-length coding sequence of KNAT7 was cloned and inserted into the pCAMBIA1300 vector between the rice Ubiquitin promoter and the Nopaline synthase (NOS) terminator. The resulting constructs were transfected into Agrobacterium tumefaciens strain EHA105 and introduced into the wild-type varieties Zhonghua11 and Nipponbare, respectively. The primers used for preparation of the constructs are summarized in Supplemental Table S4.

Bioinformatics Analyses

Coexpression analysis was performed using the RiceFREND database (http://ricefrend.dna.affrc.go.jp/). Rice CESA4 was chosen as a guide gene, and the Pearson correlation threshold value was set above 0.6. The phylogenetic tree of KNAT7 homologs in rice and Arabidopsis was built using MEGA6 software (Tamura et al., 2013) with the neighbor-joining methods and 1,000 bootstrap replicates. Alignment of the sequences of KNAT7 and its homologs in rice and Arabidopsis was conducted using Clustal X (Larkin et al., 2007). The sequences were obtained from rice and Arabidopsis genome databases (http://rice.plantbiology.msu.edu) and the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov/). The cis-acting regulatory elements in the KNAT7 promoter region were analyzed using the PLACE database (http://www.dna.affrc.go.jp/PLACE/).

Microscopy

Fresh hand-cut cross sections of the developing second internodes from Nipponbare were prepared as previously described (Huang et al., 2015). The autofluorescent signals of cell walls were viewed and photographed with 488 nm excitation using a fluorescence microscope (Imager D2, Zeiss). For scanning electron microscopy analysis, the mature second internodes and grains from the wild-type, knat7, and KNAT7-OE plants were harvested and fixed in 4% (w/v) paraformaldehyde (Sigma-Aldrich). The internode and grain samples were sliced with Gillette razor blades. After dehydration through a gradient of ethanol and critical point drying, the samples were sprayed with gold particles and observed with a SEM (S-3000N; Hitachi). To examine the glume cell size of the developing spikelets and mature grains, the glume outer surfaces of 10 spikelets and grains were sprayed with gold particles and observed with a SEM (S-3000N, Hitachi). To measure the cell length of sclerenchyma fibers, the mature internodes and grain glumes were macerated with glacial acetic acid and hydrogen peroxide (v/v, 1:1) at 80°C for 12 h. The tissues were squashed and stained with cellulose binding dye Pontamine Fast Scarlet 4B (Sigma) and photographed with 543 nm excitation using a fluorescence microscope (Imager D2, Zeiss). The data were analyzed and displayed using the ImageJ software.

Cell Wall Composition Analyses

The second internodes and grain glumes from the ∼10 mature wild-type, knat7, and KNAT7-OE plants were collected to prepare cell wall residues. The cell wall residues were treated with pullulanase M1 (Megazyme) and α-amylase (Sigma) in 0.1 m sodium acetate buffer (pH 5.0) for 20 h to remove starch. The destarched alcohol insoluble residues were hydrolyzed by 2 m trifluoroacetic acid. The supernatants were collected and analyzed by gas chromatography mass spectrometry (Agilent) to determine the monosaccharide content as described (Zhang et al., 2009). The remains were further hydrolyzed in Updegraff reagent (acetic acid:nitric acid:water, 8:1:2, v/v). The cooled pellets were thoroughly washed and hydrolyzed with 72% (v/v) sulfuric acid. The cellulose content was examined using the anthrone assay. The lignin content was measured using the acetyl bromide method (Huang et al., 2015).

Subcellular Localization

The full-length coding sequence of KNAT7 was cloned and in-frame fused with GFP and inserted into the binary vector pCAMBIA1300 between the Cauliflower mosaic virus (CaMV) 35S promoter and NOS terminator. The resulting constructs were transfected into A. tumefaciens strain EHA105 and infiltrated into the leaves of 4-week-old N. benthamiana plants. The GFP fluorescent signals were recorded with a confocal laser-scanning microscope (TCS SP5; Leica), and 2 μg/mL of 4′,6-diamidino-2-phenylindole (DAPI, Sigma) was used to visualize the nuclei.

Transactivation Activity Assays

For the TF feature analysis, the full-length coding sequence of KNAT7 was amplified and in-frame fused with the GAL4BD domain and cloned into the vector p2GW7-GAL4BD. VP16 was cloned into p2GW7-GAL4BD as a positive control. The GUS gene was cloned into p2GW7 as a negative control. The effector constructs for function validation were prepared by amplifying the KNAT7, GRF4, and NAC31 genes using the primers shown in Supplemental Table S4 and cloned into the vector p2GW7.

The reporter ProGAL4:Luciferase was prepared by placing five repeats of the Saccharomyces cerevisiae GAL4 binding elements plus the minimal TATA box region of the CaMV 35S promoter upstream of the firefly (Photinus pyralis) luciferase reporter gene. Four copies of the KNOX binding element were synthetized and inserted before the 5×GAL4 motif to generate the reporter construct ProKNOX-ProGAL4:Luciferase. The reporter constructs for function validation were obtained by inserting the promoters (2 kb upstream of the ATG) of EXPB3, EXPB17, EXPA6, MYB61, MYB103, and NAC31 genes (Supplemental Table S4) before the luciferase in the pUC19 vector.

The protoplast cells prepared from 4-week-old Arabidopsis rosette leaves were pairwise cotransfected with the resulting reporter and effector constructs. After a 16- to 20-h incubation, the transfected protoplasts were lysed and luciferase activities were recorded using a dual-luciferase reporter assay system (Promega). The Renilla reniformis luciferase gene driven by the CaMV 35S promoter was included in each assay to monitor transfection efficiency. These experiments were performed three times.

Protein Interaction Analyses

For split-luciferase complementation assays, the coding sequences of KNAT7, GRF4, NAC31, and Gibberellin Insensitive Dwarf1 (GID1) were amplified and in-frame fused with the amino- or carboxyl-terminus of luciferase. The resulting constructs were transfected into A. tumefaciens strain EHA105 and pairwise coinfiltrated into the leaves of 4-week-old N. benthamiana plants. Interactions were visualized by the bioluminescence signal intensity captured using IndiGO software (Berthold). GID1 fused with the N and C termini of the luciferase protein was used to generate the negative controls.

Yeast two-hybrid assays were performed according to the manufacturer’s instructions (Clontech). The coding sequences of KNAT7, GRF4, and NAC31 were cloned and inserted into the pGBKT7 and pGADT7 vectors to generate the bait and prey constructs. The resulting constructs of combinatory tests were cotransformed into the yeast strain Y2HGold. The yeast cells were cultured on synthetic dropout (SD) medium that lacks Trp, Leu, His, and Ade at 30°C for 3–4 d. The interactions were determined based on the yeast growth status.

BiFC analysis was conducted as described (Zhang et al., 2018). In brief, the coding sequences of KNAT7, GRF4, and NAC31 were amplified and inserted into pSPY vectors that contain either amino- or carboxyl-terminal yellow fluorescence protein (YFP) fragments. A. tumefaciens strain EHA105 bacteria containing the constructs were pairwise infiltrated into the leaves of 4-week-old N. benthamiana plants. The interactions were visualized by the fluorescent signal intensity recorded by a confocal laser scanning microscope (Axio imager Z2; Zeiss). The nuclei were stained with 2 μg/mL DAPI. Coinfiltrations of the constructs with the corresponding empty construct were used as negative controls.

For FRET analysis, KNAT7, GRF4, NAC31, and GID1 genes were cloned and inserted into the FRET vector. The resulting constructs were transfected into rice protoplasts that were prepared from 2-week-old etiolated rice seedlings. Acceptor photobleaching FRET experiments were performed as described (Xie et al., 2018). The 405-nm and 514-nm argon ion lasers were used to exciteCFP and YFP fluorescence, respectively. The acceptor YFP fluorescence in the region of interest in the nucleus was bleached by using 50 times 514 nm argon laser line at 100% intensity. After photobleaching, the FRET efficiency was calculated using the following formula: FRET_eff = (I₄ – I₃) × 100/I₄. I₄ represents the CFP intensity after the photobleaching of YFP, and I₃ indicates the CFP intensity before photobleaching. The background FRET efficiency was calculated by measuring the fluctuation of CFP fluorescence before photobleaching. Fluorescence was recorded with a confocal laser-scanning microscope (Axio imager Z2; Zeiss).

For co-IP analysis, the coding sequences of KNAT7 and NAC31 were fused with GFP or FLAG tag and inserted into the pCAMBIA1300 vector between the Ubiquitin promoter and the NOS terminator. The resulting constructs were introduced into the wild-type variety Nipponbare to generate GFP-KNAT7-OE and FLAG-NAC31-OE transgenic plants. The Ubi:FLAG-GRF4 or Ubi:GFP-KNAT7 constructs were transiently transfected into rice protoplasts prepared from seedlings of GFP-KNAT7-OE or FLAG-NAC31-OE plants, respectively. After cultivating overnight, the total proteins were extracted using protein extraction buffer (50 mm Tris-HCl, pH 7.5, 150 mm NaCl, 2 mm EDTA, 0.1% [v/v] Triton X-100, 0.1% [v/v] NP-40, and 1× protease inhibitor cocktail) and incubated with 30 μL of anti-GFP agarose beads (MBL) at 4°C for 2 h. GID1 was used as a negative control. The immunoprecipitations were eluted with 2× SDS loading buffer at 95°C for 5 min. Proteins were separated by SDS–PAGE and detected by immunoblotting analysis using anti-FLAG and anti-GFP primary antibodies.

All the experiments for protein-protein interaction examinations were performed three times. The representative images are shown. The primers used for the construct preparations are included in Supplemental Table S4.

Gene Expression

Different organs, including roots, leaf sheaths, and leaves of 2-week old seedlings and 9-cm developing internodes and young panicles ranging from 0.5 to 15 cm long were collected from Nipponbare plants. The 9-cm developing internodes were cut into nine segments. All segments except for the ninth segments, as well as the spikelets from the developing panicles, were subjected to RNA isolation. Total RNA was extracted using Plant RNA Reagent (Invitrogen). Meanwhile, the 9-cm developing internodes and 10-cm young panicles were harvested from the KNAT7-modulated plants and subjected to RNA isolation. One microgram of total RNA was reverse transcribed to produce cDNA using the PrimeScript RT Reagent Kit (TAKARA) according to the manufacturer’s instructions. RT-qPCR was performed on a cycler apparatus (Bio-Rad CFX96) using the FastStart Universal SYBR Green Master (Roche). The data were analyzed by the 2^-ΔCT method. The cellular-level expression pattern was performed according to the report (Zhang et al., 2018). In brief, the second and third internode segments and spikelets from 10-cm panicles were embedded in paraffin and subjected to laser microdissection. The 15-μm-thick sections were prepared to collect the epidermal sclerenchyma cells and parenchyma cells by the LMD 7000 laser microdissection system (Leica). Total RNA was extracted using the RNeasy micro kit (QIAGEN) and applied for RT-qPCR analysis. Rice Heterogeneous nuclear ribonucleoprotein 27C (HNR) and Triosephosphate isomerise1 (TP1) were used as internal controls for normalization of the expression of TFs and CESA genes, respectively. The primers for RT-qPCR analysis were summarized in Supplemental Table S5. These assays were performed at least three times.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers Os01g54620 (CESA4); Os10g32980 (CESA7); Os03g21820 (EXPA6); Os03g60720 (EXPA7); Os10g40720 (EXPB3); Os04g44780 (EXPB17); Os05g33730 (GID1); Os02g47280 (GRF4); Os03g03164 (KNAT7); Os01g18240 (MYB61); Os08g05520 (MYB103); and Os08g01330 (NAC31).

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Sequence identity of rice KNAT7.

• Supplemental Figure S2. Phenotypes of the KNAT7-modulated plants.

• Supplemental Figure S3. Rice KNAT7 has transcription factor features.

• Supplemental Figure S4. Transactivation activity analysis of rice KNAT7.

• Supplemental Figure S5. KNAT7 interacts with NAC31 and GRF4.

• Supplemental Figure S6. The mutated KNAT7 is unable to interact with NAC31 and GRF4.

• Supplemental Figure S7. Anatomical analysis of the developing internodes and glumes.

• Supplemental Table S1. List of the TFs coexpressed with rice CESA4.

• Supplemental Table S2. Compositional analysis of sugar content of wall residues from the internodes of the KNAT7-modulated plants and the corresponding wild-type plants.

• Supplemental Table S3. Identifying the cis-acting regulatory elements in the 2-kb promoter region of KNAT7.

• Supplemental Table S4. Primers used for construct preparation in this study.

• Supplemental Table S5. Primers used for RT-qPCR analysis in this study.