LAZY3 interacts with LAZY2 to regulate tiller angle by modulating shoot gravity perception in rice

Summary

Starch biosynthesis in gravity‐sensing tissues of rice shoot determines the magnitude of rice shoot gravitropism and thus tiller angle. However, the molecular mechanism underlying starch biosynthesis in rice gravity‐sensing tissues is still unclear. We characterized a novel tiller angle gene LAZY3 (LA3) in rice through map‐based cloning. Biochemical, molecular and genetic studies further demonstrated the essential roles of LA3 in gravity perception of rice shoot and tiller angle control. The shoot gravitropism and lateral auxin transport were defective in la3 mutant upon gravistimulation. We showed that LA3 encodes a chloroplast‐localized tryptophan‐rich protein associated with starch granules via Tryptophan‐rich region (TRR) domain. Moreover, LA3 could interact with the starch biosynthesis regulator LA2, determining starch granule formation in shoot gravity‐sensing tissues. LA3 and LA2 negatively regulate tiller angle in the same pathway acting upstream of LA1 to mediate asymmetric distribution of auxin. Our study defined LA3 as an indispensable factor of starch biosynthesis in rice gravity‐sensing tissues that greatly broadens current understanding in the molecular mechanisms underlying the starch granule formation in gravity‐sensing tissues, and provides new insights into the regulatory mechanism of shoot gravitropism and rice tiller angle.

Introduction

Rice (Oryza sativa L.) tiller angle is a key agronomic trait determining rice grain yield by affecting planting density. Rice tiller angle is defined as the angle between the vertical line and the side tillers with maximum inclination (Wang et al., 2022). In paddy field, the extreme‐spreading rice plants occupy too much space that reduce grain yield per unit area while the compact plant type influences light capture, light penetration thus photosynthesis efficiency (Gao et al., 2019; Wang and Li, 2008; Xu et al., 1998). Therefore, optimal tiller angle is crucial for high grain yield through achieving ideal plant architecture.

The transition from prostrate growth in wild rice to erect growth of cultivated rice is one of the most important events of rice domestication, which greatly increases rice grain yield (Wang et al., 2022; Xu and Sun, 2021). Some zinc‐finger transcription factors, including PROSTRATEGROWTH1 (PROG1), PROG7 and RICE PLANT ARCHITECTURE DOMESTICATION (RPAD), have been reported to control the tiller angle transition during rice domestication (Hu et al., 2018; Jin et al., 2008; Tan et al., 2008; Wu et al., 2018). Asian cultivated rice is composed of two major subspecies japonica and indica, and their distinct tiller angles underwent strong artificial selection (Wang et al., 2022; Xu and Sun, 2021). Studies in past decades have demonstrated that key genes, such as TILLER ANGLE CONTROL1 (TAC1), TILLER INCLINED GROWTH 1 (TIG1), DWARF2 (D2), TAC3 and TAC4, determine the different tiller angles between japonica and indica rice (Dong et al., 2016; Li et al., 2021; Wang et al., 2022; Yu et al., 2007; Zhang et al., 2019).

Previous studies have found that shoot gravitropism plays a key role in determining rice tiller angle (Gao et al., 2019; Wang et al., 2022). Although plants are sessile organisms, they can sense gravity to redirect their growth direction, thus causing upward growth of shoot (negative gravitropism) and downward growth of root (positive gravitropism) (Morita and Tasaka, 2004). The gravitropic response of higher plants is composed of sequential steps: gravity perception, signal transduction and organ curvature resulting from the asymmetric distribution of auxin (Morita and Tasaka, 2004; Strohm et al., 2012). The classical starch–statolith hypothesis suggested that the sedimented amyloplasts containing starch granules can sense the direction of gravity in gravity‐sensing tissues (Sack, 1991, 1997). In Arabidopsis, endodermal cells were confirmed to be responsible for sensing gravity in hypocotyls and inflorescence stems, which regulate shoot gravitropism and thus branch angle (Fujihira et al., 2000; Fukaki et al., 1998). By contrast, the leaf sheath base regions act as the gravity‐sensing tissues at the seedling stage of Gramineae (Huang et al., 2021; Parker, 1979). However, the regulatory mechanisms underlying starch biosynthesis in rice gravity‐sensing tissues remain to be explored.

The starch‐filled amyloplasts play a key role in regulating shoot gravitropism and thus rice tiller angle (Huang et al., 2021; Okamura et al., 2013, 2015). Phosphoglucomutase (PGM) is a starch biosynthetic enzyme that catalyses the transformation of glucose‐6‐phosphate (G6P) into glucose‐1‐phosphate (G1P) (Santelia et al., 2015; Stitt and Zeeman, 2012). Loss‐of‐function mutant of Oryza sativa plastidic phosphoglucomutase (OspPGM) showed impaired starch biosynthesis and larger tiller angle in rice (Huang et al., 2021; Lee et al., 2016). The YbaB protein family member LA2 could interact with OspPGM in chloroplast to regulate rice tiller angle in the same pathway (Huang et al., 2021). The starch granules in amyloplasts were completely absent in the leaf sheath base of la2 mutant, which causes larger tiller angle and defective shoot gravitropism (Huang et al., 2021). ADP‐glucose pyrophosphorylase (AGP) is a rate‐limiting enzyme that catalyses the first irreversible step in starch biosynthesis (Okamura et al., 2013, 2015). Mutation of the AGPase large subunit gene, OsAGPL1, causes stem‐starch reduction and thus defective shoot gravitropism and large tiller angle (Okamura et al., 2013, 2015). Moreover, compared with agpl1 and agpl3 single mutants, the agpl1 agpl3 double mutant showed lower stem‐starch content, weaker gravitropic response and larger tiller angle, suggesting that starch biosynthesis in amyloplasts affects the magnitude of gravitropic response and rice tiller angle (Okamura et al., 2013, 2015). In addition, previous study also demonstrated that the sedimentation rate of amyloplasts regulates shoot gravitropism and tiller angle in rice (Wu et al., 2013). Loss‐of‐function mutant of the gene Loose Plant Architecture1 (LPA1), the orthologue of Arabidopsis SHOOT GRAVITROPISM5 (SGR5) /INDETERMINATE DOMAIN15 (IDD15) encoding a typical Cys2/His‐2 zinc finger protein, showed reduced sedimentation rate of amyloplasts upon gravistimulation, which causes defective shoot gravitropism and larger rice tiller angle (Kim et al., 2015; Tanimoto et al., 2008; Wu et al., 2013; Yamauchi et al., 1997).

Amyloplast sedimentation may trigger signal transduction to alter auxin transport and induce asymmetric auxin distribution along the direction of gravistimulation, which plays a key role in regulating shoot gravitropism and thus tiller angle in rice (Huang et al., 2021; Wu et al., 2013). In rice, the first identified gene LAZY1 (LA1) regulates rice shoot gravitropism and tiller angle by controlling lateral auxin transport upon gravistimulation (Li et al., 2007; Yoshihara and Iino, 2007). Further studies found that the LA1 interacting protein Brevis Radix Like 4 (BRXL4) could regulate shoot gravitropism and tiller angle by affecting the nuclear localization of LA1 (Li et al., 2019). Moreover, screening for suppressors of la1 revealed the novel function of strigolactones (SLs), a group of carotenoid‐derived branch‐inhibiting hormones that can also attenuate rice shoot gravitropism to modulate rice tiller angle via inhibiting local auxin biosynthesis (Sang et al., 2014). Latest studies showed that the bZIP family transcription factor OsbZIP49 positively regulates rice tiller angle by directly binding to the promoter region to activate the transcription of indole‐3‐acetic acid‐amido synthetase genes OsGH3‐2 and OsGH3‐13, regulating rice tiller angle by controlling local auxin homeostasis (Ding et al., 2021). By analysing the dynamic transcriptome of rice shoots upon gravistimulation, Zhang et al. (2018) identified a core regulatory pathway of rice tiller angle dependent on LA1‐mediated auxin asymmetric distribution and established the direct connection between shoot gravitropism and rice tiller angle. In this pathway, HEAT STRESS TRANSCRIPTION FACTOR 2D (HSFA2D) acts the upstream of LA1 while two redundant transcription factors WUSCHEL RELATED HOMEOBOX6 (WOX6) and WOX11 function downstream of LA1 to regulate shoot gravitropism and rice tiller angle (Zhang et al., 2018). In addition, two transcription factors, OsHOX1 and OsHOX28, could directly bind to the promoter of HSFA2D to regulate rice shoot gravitropism and tiller angle (Hu et al., 2020). A recent study implied that OsmiR167a and its target genes OsARF12, OsARF17 and OsARF25 might participate in HSFA2D‐LA1‐dependent auxin asymmetric distribution pathway to regulate shoot gravitropism and rice tiller angle (Li et al., 2020). These studies established a preliminary genetic regulatory network of rice shoot gravitropism and tiller angle.

Although the significant progress has been achieved in exploring the genetic basis of shoot gravitropism and rice tiller angle, molecular mechanism underlying shoot gravity perception in the regulation of rice tiller angle remains largely unknown. Here, we characterized a chloroplast‐localized tryptophan‐rich protein LAZY3 (LA3) that could interact with LA2 to regulate starch biosynthesis in rice gravity‐sensing tissues. We showed that LA3 is associated with starch granules via TRR domain and functions in the gravity‐sensing tissues. Genetic analysis suggested that LA3 and LA2 function in the same genetic pathway acting upstream of LA1‐dependent auxin asymmetric distribution to control shoot gravitropism and tiller angle in rice.

Results

Phenotypic characterization of the rice tiller angle mutant la3

To elucidate the molecular mechanism underlying rice tiller angle, we focused on the characterization of new components involved in controlling rice tiller angle. A spontaneous rice mutant, designated as la3 with spreading‐out tillers, was isolated from the indica cultivar Guanghui998 (GH998) background. Genetic analysis implied that the la3 is a recessive mutant. Compared with the erect plant architecture of GH998, the la3 mutant showed loose plant architecture with larger tiller angle (Figure 1a–c). Rice tiller angle is strongly associated with shoot gravitropic response (Gao et al., 2019; Wang et al., 2022). To verify the involvement of LA3 in regulating shoot gravitropism, we examined the gravitropic response of the la3 mutant. Compared with GH998, both the seedlings and coleoptiles of la3 displayed reduced gravitropic responses (Figures 1d,e, S1). These results demonstrated that LA3 positively regulates shoot gravitropism, and the defective shoot gravitropism accounts for the larger tiller angle of the la3 mutant.

Figure 1.

Phenotypic characterization and auxin asymmetric distribution analysis of the rice lazy3 (la3) mutant. (a) The gross morphologies of the wild‐type Guanghui 998 (GH998) and la3 mutant at the adult stage. Bars = 20 cm. (b) Comparison of the tiller bases between GH998 and la3 at the adult stage. Bars = 5 cm. (c) Statistical analysis of tiller angle of the wild type and la3. Values are means ± SEM (n = 10). Student's t‐test: **P < 0.01. (d) Shoot curvature of GH998 and la3 seedlings after gravistimulation for 96 h. The arrow indicates the direction of gravity. Bars = 2 cm. (e) Kinetic analysis of shoot curvature of the wild type and la3 seedlings upon gravistimulation. Values are means ± SEM (n = 15). Student's t‐test: **P < 0.01. (f–h) The relative expression levels of OsIAA20 (f), WOX6 (g), and WOX11 (h) at the lower sides and upper sides of the shoot bases in GH998 and la3 seedlings upon gravistimulation. Values are means ± SEM (n = 3). Student's t‐test: **P < 0.01

In addition, we conducted a detailed comparison of the agronomic traits between GH998 and la3 mutant. The phenotypic and statistical analysis showed that there were no significant differences between GH998 and la3 in tiller number and plant height (Figure S2), nor panicle traits including panicle length, primary branch number, secondary branch number, grain number and setting rate (Figure S3a–f), grain length, grain width, 1000‐grain weight and grain yield per plant (Figure S3g–k). These results indicated that LA3 can specifically regulate rice tiller angle and might be used to improve rice tiller angle without affecting other important yield traits.

LA3 could regulate asymmetric distribution of auxin upon gravistimulation

The classical Cholodny‐Went hypothesis proposed that asymmetric auxin distribution mediates the differential growth during gravitropic response (Firn et al., 2000). To examine whether LA3 is involved in auxin redistribution, we detected the expression of the auxin‐responsive marker gene OsIAA20 in the shoot bases of GH998 and la3 seedlings upon gravistimulation. We divided the shoot bases into upper sides and lower sides along the direction of gravistimulation and found that OsIAA20 was asymmetrically expressed in the shoot bases of both GH998 and la3 with preferred expression in their lower sides (Figure 1f). However, the expression of OsIAA20 was significantly reduced in the lower sides of la3 compared with that in the wild type (Figure 1f), implying that the asymmetric distribution of auxin was defective in la3 upon gravistimulation. WOX6 and WOX11 are expressed asymmetrically in response to asymmetric distribution of auxin upon gravistimulation (Zhang et al., 2018). We then detected the expression levels of these two genes in the upper and lower sides and found the preferred expressions of WOX6 (Figure 1g) and WOX11 (Figure 1h) in the lower sides were also reduced significantly in the shoot bases of la3 seedlings. These results revealed that LA3 positively regulates asymmetric distribution of auxin upon gravistimulation and acts upstream of WOX6 and WOX11.

Cloning and functional confirmation of the LA3 gene

To isolate the LA3 gene, we took a map‐based cloning approach and generated an F₂ population by crossing the la3 mutant with ZH11 (a wild‐type japonica variety). LA3 was primarily delimited in the region between the two molecular markers M1 and M2 on the chromosome 3, and LA3 was finally narrowed in a 360‐kb candidate region between molecular markers M4 and M9 by using more recombinant plants and newly developed molecular markers (Figure 2a). Among the 65 candidate genes in the candidate region, we detected 37 expressed genes and sequencing analysis only revealed a single base G insertion at the eighth exon of LOC_Os03g04100 in the la3 mutant, which produces a frameshift mutation (Figures 2a, S4). In order to verify whether the frameshift mutation is responsible for the phenotypes of the la3 mutant, we performed a complementation analysis. The plasmid pLA3C, containing a 7711‐bp genomic DNA fragment consisting of a 2358‐bp upstream sequence, the entire LA3 gene including 11 exons and 10 introns, and an 826‐bp downstream sequence, was introduced into the la3 mutant. We obtained 10 independent transgenic lines, and all of them rescued the large tiller angle (Figure 2b,c) and defective shoot gravitropism (Figure 2d,e) of la3. Therefore, LOC_Os03g04100 is the rice LA3 gene, and its frameshift mutation is responsible for the phenotype of la3.

Figure 2.

Map‐based cloning and confirmation of LAZY3 (LA3) in rice. (a) Map‐based cloning of LA3. LA3 was narrowed down to a 360‐kb candidate region between molecular markers M9 and M4. Numbers under the markers indicate recombinants. (b) The gross morphologies of GH998, la3, the two representative LA3:LA3/la3 complemented lines (pLA3C‐1 and pLA3C‐2) at the adult stage. Bars = 10 cm. (c) Statistical analysis of tiller angle of GH998, la3, and the two representative LA3:LA3/la3 complemented lines (pLA3C‐1 and pLA3C‐2) at the adult stage. Values are means ± SEM (n = 10). Different letters above the column represent statistically significant differences at P < 0.05 (one‐way ANOVA, Tukey's honestly significant difference). (d) Shoot gravitropism of GH998, la3, and the two representative LA3:LA3/la3 complemented lines (pLA3C‐1 and pLA3C‐2) after a 96‐h gravistimulation. Bars = 1 cm. (e) Dynamic changes of shoot curvature of the lines in (d). Values are means ± SEM (n = 15).

To further confirm the function of LA3 and facilitate subsequent studies, we generated its loss‐of‐function mutant lines in japonica rice variety ZH11 by using CRISPR‐Cas9 (CR) gene editing. Using two single guide RNAs (sgRNAs) targeting at different regions of the exons (Figure S5a,b), respectively, we generated multiple independent loss‐of‐function homozygous mutant transgenic lines, and all the CR‐engineered LA3 (CR‐la3) mutant lines showed larger tiller angle (Figure S5c,d) and defective shoot gravitropism compared with that in the wild‐type plants (Figure S5e,f). These results further demonstrated that LOC_Os03g04100 regulates shoot gravitropism and rice tiller angle.

LA3 encodes a chloroplast‐localized tryptophan‐rich protein associated with starch granules

Sequence analysis showed that the putative LA3 protein is composed of 431 amino acids, which shares 84.1% sequence similarity with Arabidopsis EARLY STARVATION 1 (ESV1) (Figure S6), which was previously reported to control the starch metabolism in Arabidopsis thaliana leaves (Feike et al., 2016). The TargetP program (https://services.healthtech.dtu.dk/service.php?TargetP‐2.0) predicted a 38‐amino‐acid chloroplast transit peptide (cTP) at the N‐terminal of LA3, which shows high sequence divergence with the ESV1 (Figure S6). To study its subcellular localization, the LA3‐GFP fusion protein was transiently expressed in protoplasts of the wild‐type indica cultivar 93–11. Results showed that LA3‐GFP was exclusively located in chloroplasts and associated with discrete bodies likely to be starch granules (Figures 3a, S7), which was further confirmed by the completely merged signals between LA3‐GFP and the autofluorescence of amyloplasts (Cy5) (Figure S8). It has been evidenced that the Arabidopsis ESV1 acts as a starch‐binding protein thus showing typical discrete bodies associated with starch granules in the leaf chloroplasts and hypocotyl endodermis plastids in the subcellular localization analysis (Feike et al., 2016; Malinova et al., 2018; Song et al., 2021). To test whether the LA3‐GFP‐associated discrete bodies are starch granules in the chloroplasts, we expressed the fusion protein LA3‐GFP transiently in the la2 mutant, which is defective in starch biosynthesis and the starch granules in amyloplasts are completely absent in the mutant (Huang et al., 2021). Interestingly, LA3‐GFP appeared as diffuse signals in the chloroplast stroma of the mutant la2 rather than being associated with discrete bodies in the wild‐type 93–11 (Figures 3a, S7). These results suggested that the fluorescent discrete bodies in chloroplasts of wild‐type 93–11 are indeed starch granules.

Figure 3.

LAZY3 (LA3) is associated with starch granules via two subdomains in the TRR domain. (a) The close‐ups of the localization of LA3‐GFP in rice protoplast of 93–11 and la2. Bars = 10 μm. (b) Schematic diagram of LA3 and its truncated derivatives. The green, orange and blue boxes indicate the chloroplast transit peptide (cTP) domain, TRR domain and proline‐rich region (PRR) domain, respectively. The folded lines indicate the deleted TRR domains. (c) Subcellular localization analysis of GFP‐tagged LA3 and LA3 truncated derivatives in rice protoplasts. Bars = 10 μm.

We also found that LA3 contains a proline‐rich region (PRR) between P397 and P429 at its C‐terminal and an intermediate tryptophan‐rich region (TRR) between W184 and W313 (Figure S6), hereafter referred to as PRR and TRR domains, respectively. Studies have suggested that conserved TRR was essential for proteins to bind to the starch granules (Christiansen et al., 2009; Lo et al., 2004; Williamson et al., 1997). However, the subdomains or motifs in the TRR domain of ESV1 responsible for the protein‐starch association have not been identified. To further define the subdomains associated with starch granules, we developed a series of LA3‐GFP truncated derivatives, in which different regions of LA3 were deleted (Figure 3b) and transiently expressed them in the rice protoplasts. Subcellular localization analysis showed that both LA3N210 and LA3∆211–313 appeared as diffuse signals in the chloroplast stroma while both LA3N244 and LA3∆211–277 exhibited similar subcellular localization as the LA3‐GFP in punctate starch granules (Figure 3c). These results demonstrated that the two subdomains (G211‐G244 and S278‐W313, referred to as subdomain 1 and subdomain 2, respectively) in the TRR domain are essential for LA3 to be associated with starch granules.

LA3 is involved in starch biosynthesis in shoot gravity‐sensing tissues

The association between LA3 and starch granules prompted us to test whether LA3 is involved in the starch biosynthesis in rice gravity‐sensing tissues. In rice, the starch‐filled amyloplasts in leaf sheath bases and leaf sheath pulvinus are responsible for gravity sensing at the seedling stage and adult stage, respectively (Abe et al., 1994a, 1994b). To this end, we conducted the starch staining to check the starch granules in the rice shoot base of la3 seedlings. As shown in Figure 4, a lot of starch granules in the gravity‐sensing tissues of leaf sheath bases in wild‐type GH998 and the pLA3C complemented transgenic line were stained blue or purple (Figure 4b,d,h,j,k,m), while no starch granules were detected in the la3 mutant (Figure 4c,i,l). Similar to the CR‐la2‐1 mutant, the starch granules were completely lost in the leaf sheath bases of CR‐la3‐6 mutant in ZH11 background (Figure 4e–g). We also conducted the KI/I₂ staining to check the starch granules in the rice leaf sheath pulvinus of CR‐la3‐6 mutant at the adult stage. Results showed that there were no starch granules in the leaf sheath pulvinus of CR‐la3‐6 mutant, while the wild‐type ZH11 was stained blue (Figure S9). However, the grain filling of la3 was not affected (Figure S3g). Taken together, these results indicate that LA3 controls rice tiller angle through regulating the starch biosynthesis in the shoot gravity‐sensing tissues which modulate shoot gravitropism through gravity perception.

Figure 4.

LAZY3 (LA3) is involved in the starch biosynthesis in rice leaf sheath base. (a) Boxed region shows the 0.5 cm of the shoot base used for starch granule staining of the 2‐day‐old seedlings. (b–g) Starch granule staining of the leaf sheath of wild‐type GH998 (b), la3 (c), pLA3C‐1 (d), wild‐type ZH11 (e), CR‐la3‐6 (f), and CR‐la2‐1 (g) at the seedling stage. Bars = 200 μm. (h–j) Starch granules staining of the longitudinal shoot bases of 5‐day‐old GH998 (h), la3 (i), and pLA3C‐1 (j), seedlings. Bars = 100 μm. (k–m) Magnified views of the red boxed in (h), (i), and (j), respectively. Bars = 10 μm.

LA3 can interact with LA2 to regulate rice shoot gravitropism and tiller angle

In our latest study, we have evidenced that the chloroplastic protein LA2 controls shoot gravitropism and tiller angle through interacting with OspPGM to regulate starch biosynthesis in rice gravity‐sensing tissues (Huang et al., 2021). Given the similar roles of LA3 and LA2 in starch biosynthesis in the shoot gravity‐sensing tissues (Figure 4), we then checked whether LA3 can interact with LA2 and OspPGM by using luciferase complementation imaging (LCI) assay. The result showed that LA3 can interact with LA2 but not OspPGM (Figure 5a). The bimolecular fluorescence complementation (BiFC) assay further showed that the interaction between LA3 and LA2 is associated with starch granules (Figure 5b), implying that LA3 might function as a bridge to connect OspPGM‐LA2 complex to the starch granules during starch biosynthesis. To further analyse the genetic relationship between LA2 and LA3, we generated the CR‐la3‐6 CR‐la2‐1 double mutants by crossing the CR‐la3‐6 with CR‐la2‐1. Phenotypic and statistical analysis showed that the tiller angle and shoot gravitropism of the double mutant CR‐la3‐6 CR‐la2‐1 were similar to the single mutants (Figure 5c–f), suggesting that LA3 and LA2 act in the same pathway to regulate rice shoot gravitropism and tiller angle. Considering the result that the subcellular localization of LA3 is dependent on LA2, we proposed that LA3 may act downstream of LA2 to regulate rice shoot gravitropism and tiller angle.

Figure 5.

LAZY3 (LA3) and LAZY2 (LA2) act in the same pathway in regulating rice tiller angle. (a) LA3 interacts with LA2 in tobacco leaves in the luciferase complementation imaging (LCI) assay. (b) LA3 interacts with LA2 in rice chloroplasts in the bimolecular fluorescence complementation (BiFC) assay. Bars = 10 μm. (c) Phenotypes of the wild‐type ZH11, CR‐la3‐6, CR‐la2‐1, and double mutant CR‐la3‐6 CR‐la2‐1 at the adult stage. Bars = 10 cm. (d) Statistical analysis of tiller angle of the ZH11, CR‐la3‐6, CR‐la2‐1, and CR‐la3‐6 CR‐la2‐1. Values are means ± SEM (n = 10). Different letters indicate statistically significant difference determined by one‐way ANOVA, Tukey's honestly significant difference. (e) Shoot gravitropism of ZH11, CR‐la3‐6, CR‐la2‐1, and CR‐la3‐6 CR‐la2‐1 after gravistimulation for 96 h. Bars = 1 cm. (f) Kinetic analysis of shoot curvature of ZH11, CR‐la3‐6, CR‐la2‐1, and CR‐la3‐6 CR‐la2‐1 seedlings grown in light upon gravistimulation. Values are means ± SEM (n = 15).

LA3 acts upstream of LA1 in controlling rice shoot gravitropism and tiller angle

Previous studies have demonstrated that LA1 may function between amyloplast sedimentation and lateral auxin translocation in gravity signal transduction (Yoshihara and Iino, 2007). The fact that LA3 is involved in gravity perception prompted us to test whether LA3 acts the upstream of LA1 to trigger gravity signal transduction. Therefore, we generated CR‐la3‐6 la1 double mutant by crossing the CR‐la3‐6 with la1, and the double mutant showed similar tiller angle to the la1 single mutant but significantly larger than that of CR‐la3‐6 single mutant (Figure 6a,c). In addition, we also examined the gravitropic response of the double mutant CR‐la3‐6 la1 and found that its shoot curvature was similar to that of la1 while significantly reduced compared with CR‐la3‐6 single mutant (Figure 6b,d). Taken together, we concluded that LA3 acts upstream of LA1 in controlling rice shoot gravitropism and tiller angle.

Figure 6.

LAZY3 (LA3) acts upstream of LAZY1 (LA1) in regulating rice tiller angle. (a) Phenotypes of ZH11, CR‐la3‐6, la1 and CR‐la3‐6 la1 at the tillering stage. Bars = 10 cm. (b) Shoot gravitropism of ZH11, CR‐la3‐6, la1, and CR‐la3‐6 la1 after gravistimulation for 96 h. Bars = 1 cm. (c) Statistical analysis of tiller angle of ZH11, CR‐la3‐6, la1, and CR‐la3‐6 la1. Values are means ± SEM (n = 10). Different letters above the column represent statistically significant difference at P < 0.05 (one‐way ANOVA, Tukey's honestly significant difference). (d) Kinetic analysis of shoot curvature of the ZH11, CR‐la3‐6, la1, and CR‐la3‐6 la1 seedlings grown in light upon gravistimulation. Values are means ± SEM (n = 15).

Discussion

Rice tiller angle is an important agronomic trait that contributes greatly to grain yield mainly owing to its large influence on plant density. Shoot gravitropism greatly contributes to rice tiller angle. However, the genetic regulatory networks of shoot gravitropism and rice tiller angle are largely unknown. In this study, we proposed a rice tiller angle regulatory pathway that involves LA3, LA2, OspPGM and LA1 to link the starch‐statolith‐dependent gravity perception, lateral auxin transport, shoot gravitropism and tiller angle (Figure 7). In this pathway, the novel tiller angle regulator LA3 is associated with starch granules in the chloroplast and plays essential roles in the biogenesis of starch‐filled amyloplasts in rice gravity‐sensing tissues. By interacting with LA2, the essential regulator of starch biosynthesis in rice gravity‐sensing tissues, LA3 controls rice shoot gravitropism through defining the prerequisite of gravity perception, possibly forming the LA3‐LA2‐OspPGM complex during starch biosynthesis. The LA3‐LA2‐OspPGM complex‐modulated gravity sensing regulates shoot gravitropism through LA1‐mediated auxin asymmetric distribution, thereby determining rice tiller angle (Figure 7).

Figure 7.

A proposed working model of LAZY3 (LA3) in the control of rice tiller angle. In rice, the starch‐granule‐associating protein LA3 interacts with LA2 at the surface of starch granules in gravity‐sensing tissues to regulates the starch biosynthesis in amyloplasts which modulates gravity sensing of rice shoot. In wild‐type plants, the LA3‐modulated gravity perception controls rice tiller angle by regulating shoot gravitropism through LA1‐mediated asymmetric distribution of auxin. However, loss‐of‐function of LA3 leads to few starch granules, which in turn decreases the magnitude of shoot gravitropism resulting in loose plant architecture. LAT, lateral auxin transport; V, vacuole.

The LA3 protein may directly bind to starch granules via the TRR domain

In this study, we found that the novel tiller angle regulator LA3 was associated with discrete structures likely to be starch granules (Figure S8). By analysing the localization pattern of LA3‐GFP in the la2 mutant whose starch granules are completely lost in the gravity‐sensing tissues, we further confirmed the discrete punctate structures associated with LA3‐GFP are indeed starch granules. In the la2 mutant, loss of starch granules causes the diffused expression of LA3‐GFP in the chloroplast stroma distinct from the discrete punctate structures in wild type (Figures 3a, S7). In Arabidopsis, ESV1, the ortholog of LA3, also directly binds to the starch granules (Feike et al., 2016; Malinova et al., 2018; Song et al., 2021). The esv1 mutant lacks starch granules in their hypocotyl endodermis and root columella cells, and shows reduced hypocotyl and root gravitropism (Song et al., 2021). These studies indicated a conserved role of LA3 proteins in starch biosynthesis and gravitropism across plant species via their role in starch granule formation mediated by the conserved starch‐granule‐binding capacity. However, the detailed starch‐binding motifs of LA3 and its orthologs are still unclear.

In this study, we identified two subdomains, the subdomain 1 (G211‐G244) and subdomain 2 (S278‐W313), responsible for the LA3‐starch granule association in the TRR domain by analysis of the subcellular localization of different LA3‐GFP truncated derivatives (Figure 3b,c). Some well‐known starch‐binding proteins contain carbohydrate‐binding modules (CBMs) were characterized with four highly conserved aromatic amino acids (Tryptophan) that directly involved in the glucan‐binding (Christiansen et al., 2009). Crystal structure analysis showed that glucan‐binding site 1 in CBMs contains two conserved Trp (W) residues, one Glu (E), one Lys (K) and one Asn (N), among which the two Trp residues (WW) form a compact, rigid and surface‐exposed hydrophobic site with an inter‐ring spacing appropriate for binding to α‐1,4‐linked glucoses (Christiansen et al., 2009; Sorimachi et al., 1997; Williamson et al., 1997). The in vitro assays have confirmed that ESV1 preferentially binds to highly ordered, α‐glucans, such as starch and crystalline maltodextrins (Malinova et al., 2018), but the essential starch granule binding site (SBS) was unknown. Although sequence analysis showed that the TRR domains of both LA3 and ESV1 contain four candidate SBSs characterized with WWE(K/R) motifs (Figure S6) which may bind to multiple glucans or interact with glucan chain through multiple contact sites, we found that only the truncated derivatives that neither contain SBS 2 (WWEKW motif in the subdomain 1) nor SBS 4 (WWERW motif in subdomain 2) were dissociated from starch granules in the chloroplast, suggesting that LA3 may directly binds to starch granules through the two WWE(K/R)W motifs in the TRR domain.

LA3‐LA2‐OspPGM complex modulates starch biosynthesis in rice gravity‐sensing tissues

Starch biosynthesis is a complicated metabolic process that involves multiple types of enzymes such as adenosine diphosphate glucose pyrophosphorylase (AGPase), starch synthase (SS), starch branching enzyme (SBE) and starch debranching enzyme (DBE) (Li et al., 2017). Although LA3 is an indispensable component of starch biosynthesis in shoot gravity‐sensing tissues in rice, the grain size, weight as well as the grain yield are not affected in la3 mutant (Figure S3g–k), implicating the distinct mechanism of starch biosynthesis between rice gravity‐sensing tissues and endosperms. Recent studies revealed that rice ESV1 promotes the formation of starch granules in root columellar cells while facilitates the tight packing of the starch granules in grains (Song et al., 2021). In cereal endosperms, ADP‐Glc was produced from Glc‐6‐P by cytosolic PGM and AGPase and then transported into the plastids (Okamura et al., 2013; Slewinski and Braun, 2010; Song et al., 2021). However, in gravity‐sensing tissues and photosynthetic organs, Glc‐6‐P is converted to Glc‐1‐P by PGM and then Glc‐1‐P is metabolized to produce ADP‐Glc, both processes happen in the plastids (Slewinski and Braun, 2010; Song et al., 2021). In addition, sucrose is metabolized to produce Glc‐6‐P in gravity‐sensing tissues while fructose 6‐phosphate (Fru‐6‐P) is used for producing Glc‐6‐P in photosynthesis organs (Slewinski and Braun, 2010; Song et al., 2021; Yu et al., 2001). Although the common enzyme PGM can convert Glc‐6‐P to Glc‐1‐P in plastids in both gravity‐sensing tissues and photosynthetic organs (Kunz et al., 2010; Slewinski and Braun, 2010), our new evidences suggested that it is the specialized factors LA3 and LA2, but not PGM that determine the specificity and compartmentalization of the starch biosynthesis, enabling us to distinguish the starch biosynthesis in gravity‐sensing tissues from that in photosynthetic organs or cereal endosperms in rice.

We found that LA3 could interact with LA2 in rice chloroplasts and the interaction was associated with starch granules (Figure 5b). In our latest study, we confirmed that LA2 could interact with OspPGM to regulate starch biosynthesis and thus rice tiller angle (Huang et al., 2021). Therefore, we postulated that LA3, LA2 and OspPGM may form a LA3‐LA2‐OspPGM protein complex in chloroplasts, and they are all indispensable regulators of starch‐statolith‐mediated gravity perception in rice gravity‐sensing tissues (Figure 7). Among the complex, LA3 and LA2 mainly function in gravity‐sensing tissues while OspPGM acts as a general regulator of starch biosynthesis in different tissues. Although the mechanism of action of the LA3‐LA2‐OspPGM in starch biosynthesis in gravity‐sensing tissues remains unclear, in vitro analysis have suggested that Arabidopsis ortholog of LA3, ESV1 preferentially binds to highly ordered starch‐like glucan (α‐glucan) structures similar to starch (Malinova et al., 2018). Intriguingly, neither the chain elongation nor branching/debranching during amylopectin biosynthesis is strongly affected by the loss of ESV proteins while the amylose in the photosynthetic tissues of esv1 mutant was increased by 60% compared with the wild type (Feike et al., 2016). Therefore, it is very interesting to further explore how LA3 together LA2 and OspPGM synergistically regulates the starch biosynthesis in rice gravity‐sensing tissues.

The starch‐statolith‐dependent pathway acts upstream of LA1‐mediated asymmetric distribution of auxin

The classical Cholodny‐Went hypothesis proposed the asymmetric distribution of auxin results in the differential growth and organ curvature upon gravistimulation (Firn et al., 2000). In this study, we found that the la3 mutant showed defective asymmetric expression of OsIAA20, OsWOX6 and OsWOX11 between the upper and lower sides of shoot bases upon gravistimulation (Figure 1f–h), indicating that asymmetric auxin distribution upon gravistimulation was impaired in la3. Similar asymmetric auxin distribution was also reported in la2 mutants (Huang et al., 2021). Because LA2 acts upstream of LA1‐mediated asymmetric distribution of auxin in controlling shoot gravitropism and rice tiller angle (Huang et al., 2021; Li et al., 2007), we then checked the genetic relationships between LA3 and LA1, and found that LA3 acts upstream of LA1 to regulate shoot gravitropism and rice tiller angle (Figure 6). Previous studies have implied that LA1 may mediate the signal transduction between amyloplast sedimentation and auxin transport (Yoshihara and Iino, 2007). In our study, we demonstrated that LA3 controls rice shoot gravitropism and tiller angle by regulating the biogenesis of amyloplasts in leaf sheath bases (Figure 4), we propose that LA1‐dependent asymmetric distribution of auxin acts downstream of LA3‐LA2‐OspPGM protein complex‐mediated starch‐statolith‐dependent pathway (Figure 7).

Starch‐statolith‐independent pathways in controlling shoot gravitropism and tiller angle in rice

The classical starch–statolith hypothesis suggested higher plants use the sedimentation of starch‐filled amyloplasts to sense the direction of gravity during gravitropism (Sack, 1991, 1997). However, even though the starch granules were absent in the gravity‐sensing tissues of la3 and la2 (Figure 4), these mutants only showed partially defective shoot gravitropism (Figure 5e,f) and moderate increase in tiller angle (Figure 5c,d) (Huang et al., 2021). Consistent with the moderate contribution of the starch‐statolith pathway to rice shoot gravitropism and tiller angle, the starch‐deficient mutants pgm and esv1 in Arabidopsis also showed partially defective shoot gravitropism (Periappuram et al., 2000; Song et al., 2021; Yu et al., 2000). These results suggested that starch is important but not essential for gravity sensing in higher plant.

Researchers proposed that the empty plastids may trigger the residual gravitropic responses because of their mass in the starch‐deficient mutants (Morita, 2010; Nakamura et al., 2019). However, latest study found that the empty plastids in the starch‐deficient mutant esv1 could not sediment in the direction of gravity in endodermal cells (Song et al., 2021), suggesting that the involvement of starch‐statolith‐independent pathways in gravity perception. Besides, the double mutant CR‐la3‐6 la1 exhibits similar phenotypes to la1 but more severe defect in shoot gravitropism and significantly increased tiller angle than CR‐la3‐6 single mutant (Figure 6), suggesting that the starch‐statolith‐dependent and ‐independent pathways at least partially converge at the LA1‐mediated auxin redistribution upon gravistimulation. Further identification of the enhancers of the la3 mutant would provide comprehensive insights into the tiller angle regulatory mechanism independent of starch metabolism.

Taken together, our data revealed that LA3 may form protein complex with LA2 and OspPGM to regulate shoot gravity sensing and thus tiller angle via a starch‐statolith‐dependent shoot gravitropism pathway. This study provides more detailed insights into the genetic regulatory network of shoot gravity perception and thus tiller angle in rice.

Methods

Plant materials

The la3 mutant was in the background of Guanghui 998 (GH998) (Oryza sativa L. subsp. indica), and the CR‐la3 mutants were generated by CRISPR/Cas9 technology. The mutants of CR‐la2‐1 and la1 were reported in our previous study (Huang et al., 2021). The double mutants CR‐la3‐6 CR‐la2‐1 and CR‐la3‐6 la1 were generated by crossing CR‐la3‐6 with CR‐la2‐1 or la1. For field experiments, rice plants were grown either in Beijing or Hainan. For greenhouse experiments, rice plants were grown under a 16‐h light and 8‐h dark photoperiod at 28°C in Beijing.

Analysis of shoot gravitropism

Analysis of shoot gravitropism was performed according to methods described previously (Huang et al., 2021) with minor modifications. After germination at 37°C for 2 days, the rice seeds were grown on 0.4% agar at 28°C for 3 days under a 16‐h light/8‐h dark photoperiod. Then, the seedlings were rotated by 90°C for gravistimulation either in darkness or under the normal 16‐h light/8‐h dark photoperiod. The gravitropic responses were quantified by measuring shoot curvature every 24 h for 4 days, and coleoptile curvature was measured after a 6‐h gravistimulation.

Map‐based cloning of LA3 and sequences analysis of candidate genes

To clone the LA3 gene, the la3 mutant was crossed with ZH11 (compact plant architecture) to generate the F₂ segregation population. The InDel markers were designed based on sequence variations between the genomes of Nipponbare and 93–11. To identify the causal gene, all the coding sequences of the expressed genes in the candidate region were sequenced and compared between GH998 and the la3 mutant. The primers used for mapping are listed in Table S1.

Vector construction and transformation

For complementation test of LA3, the 7711‐bp genomic sequence that contains 2358‐bp promoter, the entire LA3 ORF, and 826‐bp 3′ untranslated region was amplified from the genomic DNA of GH998. PCR products were cloned into the pCAMBIA1300 binary vector to generate the pLA3C construct. The construct was then introduced into Agrobacterium tumefaciens EHA105 and transformed into the la3 calli. To generate the knockout mutants of LA3 (CR‐la3), two single guide RNAs (sgRNAs) targeting at exon 2 or 3 were cloned into the VK005‐01 vector (Viewsolid Biotech, Beijing, China) according to the user's manual. The constructs were introduced into A. tumefaciens EHA105 and transformed into ZH11 calli. PCR products sequencing and hygromycin selection were used to identify the Cas9‐free transgenic plants with homozygous mutations. Agrobacterium tumefaciens‐mediated transformation in rice was performed as described previously (Hiei et al., 1994). Primers used for constructions and genotyping are listed in Tables S2 and S3.

RNA extraction and qRT‐PCR

TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was used to isolate total RNA from rice shoot bases. The total RNA was treated with DNase I (Ambion, Austin, TX, USA) and then reverse transcribed into cDNA using the iScript™ cDNA Synthesis Kit (Bio‐Rad). qRT‐PCR was conducted using the CFX96 Real‐Time System (Bio‐Rad, Hercules, CA, USA) with SsoFast EvaGreen Supermix Kit (Bio‐Rad). Rice UBIQUITIN (LOC_Os03g13170) was used as the endogenous control. Primers used for qRT‐PCR are listed in Table S4.

Subcellular localization analysis of LA3

The full‐length and truncated CDSs of LA3 and full‐length CDS of LA3 without stop codon were amplified from ZH11 cDNA and cloned into pSCYCE‐GFP vector with GFP reporter gene. The constructs 35S _pro :LA3‐GFP, 35S _pro :LA3N244‐GFP, 35S _pro :LA3N210‐GFP, 35S _pro :LA3∆(211–277)‐GFP, 35S _pro :LA3∆(211–313)‐GFP and 35S _pro :LA2‐GFP were transformed into rice protoplasts. After incubation at 28°C for 12 h, the fluorescence of GFP and chlorophyll were observed under a confocal laser scanning microscope at excitation wavelengths of 488 and 647 nm, respectively (FluoView 1000; Olympus, Tokyo, Japan). The 35S _pro :LA3‐YFP plasmid was constructed using a PCR‐based gateway system. The full‐length CDS of LA3 without stop codon was amplified and cloned to pDONR221 vector. The LA3 CDS without stop codon was then inserted into pH7YWG2 vector with YFP reporter gene. The plasmid 35S _pro :LA3‐YFP was introduced into A. tumefaciens EHA105 and infiltrated into Nicotiana benthamiana (N. benthamiana) leaves. The YFP fluorescence was observed at an excitation wavelength of 514 nm 2 days after infiltration. Primers used for subcellular localization analysis are listed in Table S2.

Luciferase complementation imaging (LCI) assay

To generate the constructs of 35S _pro :LA3‐nLUC and 35S _pro :LA2‐cLUC, the full‐length CDSs of LA3 and LA2 without stop codon were amplified from ZH11 cDNA and inserted into 35S _pro :nLUC and 35S _pro :cLUC vectors, respectively. The nLUC and cLUC fusion vectors were transformed into A. tumefaciens EHA105 and then co‐infiltrated into N. benthamiana leaves. The images were taken by NightShade LB 985 In Vivo Plant Imaging System (IndiGO; Berthold Technologies Co., Bad Wildbad, Germany) using beetle luciferin as the substrate (E1603; Promega，Madison, Wisconsin, USA). Primers used for LCI assay are listed in Table S2.

Bimolecular fluorescence complementation (BiFC) assay

The full‐length CDSs of LA3 and LA2 without the stop codon were amplified from ZH11 cDNA and cloned into the vectors pSCYNE (SCN) with nCFP reporter gene and pSCYCE (SCC) with cCFP reporter gene to generate 35S _pro :LA3‐nCFP and 35S _pro :LA2‐cCFP constructs, respectively. The combinations of the constructs were co‐transformed into rice protoplasts. After 12 h, CFP fluorescence and chlorophyll autofluorescence were observed at excitation wavelengths of 405 and 647 nm with confocal microscopy (FluoView 1000; Olympus, Tokyo, Japan), respectively. Primers used for BiFC assay are listed in Table S2.

Starch granules staining assay

Starch granules staining assay was performed as following: starch granules of leaf sheath and leaf sheath pulvinus were stained with 1% KI/I₂ buffer for 1 and 5 min, respectively, and then washed by anhydrous alcohol and water. The leaf sheath bases of 2‐week‐old seedlings and leaf sheath pulvinus at the adult stage were used for the assay, respectively. The sections were observed by using an OLYMPUS CORPORATION microscope (SZX2‐ILLT). The starch granule staining assay of periodic acid Schiff was conducted with a periodic acid Schiff (PAS) Kit (Sigma‐Aldrich, Saint Louis, MO, USA) following the manufacturer's instructions. Eight‐ to 10‐μm paraffin sections were deparaffinized with a microtome (RM2145; Leica, Wetzlar, Germany) and hydrated, immersed in periodic acid solution for 5 min, rinsed with distilled water, and then immersed in Schiff's reagent for 15 min. After dehydrating, clearing and mounting, sections were observed under a Leica DMR microscope and photographed with a Micro Colour Charge‐Coupled Device (CCD) camera (Apogee Instruments, North Logan, UT).

Conflict of interest statement

The authors declare that they have no conflicts of interest.

Author contributions

Y.C., W.W., L.H. and Y.S. designed research, performed experiments and analysed data; W.W., L.H. and Y.C. wrote the manuscript; Y.Y., S.X., X.W., Y.L., J.Z. and G.L. performed some of the experiments; Y.W., and J.L. analysed data; and Y.W. supervised the project, designed research and wrote the manuscript.

Supporting information

Table S1 Primers used for map‐based cloning.

Table S2 Primers used for constructions.

Table S3 Primers used for genotyping.

Table S4 Primers used for qRT‐PCR.

Figure S1 The gravitropism of seedlings and coleoptiles of the rice lazy3 (la3) mutant.

Figure S2 Comparison of the tiller number and plant height between GH988 and lazy3 (la3) mutant.

Figure S3 Characterization of the panicle traits of rice lazy3 (la3) mutant.

Figure S4 DNA sequence information of the mutant position of la3 mutant.

Figure S5 Generation and characterization of CR‐lazy3 (CR‐la3) mutant lines in ZH11 background.

Figure S6 Sequence alignment between LAZY3 (LA3) and ESV1.

Figure S7 LAZY3 (LA3) encodes a chloroplastic protein associated with starch granules in rice.

Figure S8 LAZY3 (LA3)‐GFP is associated with starch granules in rice.

Figure S9 The starch staining of the leaf sheath pulvinus in rice.