Volume 231, Issue 3 p. 1073-1087
Full paper
Free Access

LAZY2 controls rice tiller angle through regulating starch biosynthesis in gravity-sensing cells

Linzhou Huang

Linzhou Huang

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

University of Chinese Academy of Sciences, Beijing, 100049 China

These authors contributed equally to this work.

Search for more papers by this author
Wenguang Wang

Wenguang Wang

State Key Laboratory of Crop Biology, College of Life Sciences, Shandong Agricultural University, Tai’an, 271018 China

These authors contributed equally to this work.

Search for more papers by this author
Ning Zhang

Ning Zhang

State Key Laboratory of Rice Biology, Key Laboratory of the Ministry of Agriculture for Nuclear-Agricultural Sciences, Department of Applied Biosciences, Zhejiang University, Hangzhou, 310029 China

These authors contributed equally to this work.

Search for more papers by this author
Yueyue Cai

Yueyue Cai

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

University of Chinese Academy of Sciences, Beijing, 100049 China

Search for more papers by this author
Yan Liang

Yan Liang

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

Search for more papers by this author
Xiangbing Meng

Xiangbing Meng

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

Search for more papers by this author
Yundong Yuan

Yundong Yuan

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

Search for more papers by this author
Jiayang Li

Jiayang Li

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

University of Chinese Academy of Sciences, Beijing, 100049 China

Search for more papers by this author
Dianxing Wu

Corresponding Author

Dianxing Wu

State Key Laboratory of Rice Biology, Key Laboratory of the Ministry of Agriculture for Nuclear-Agricultural Sciences, Department of Applied Biosciences, Zhejiang University, Hangzhou, 310029 China

Authors for correspondence:

YonghongWang

Email:[email protected]

DianxingWu

Email:[email protected]

Search for more papers by this author
Yonghong Wang

Corresponding Author

Yonghong Wang

Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, 100101 China

University of Chinese Academy of Sciences, Beijing, 100049 China

State Key Laboratory of Crop Biology, College of Life Sciences, Shandong Agricultural University, Tai’an, 271018 China

CAS Center for Excellence in Molecular Plant Sciences, Chinese Academy of Sciences, Shanghai, 200032 China

Authors for correspondence:

YonghongWang

Email:[email protected]

DianxingWu

Email:[email protected]

Search for more papers by this author
First published: 27 May 2021
Citations: 14

Summary

  • Rice (Oryza sativa) tiller angle is a key component for achieving ideal plant architecture and higher grain yield. However, the molecular mechanism underlying rice tiller angle remains elusive.
  • We characterized a novel rice tiller angle mutant lazy2 (la2) and isolated the causative gene LA2 through map-based cloning. Biochemical, molecular and genetic studies were conducted to elucidate the LA2-involved tiller angle regulatory mechanism.
  • The la2 mutant shows large tiller angle with impaired shoot gravitropism and defective asymmetric distribution of auxin. We found that starch granules in amyloplasts are completely lost in the gravity-sensing leaf sheath base cells of la2, whereas the seed development is not affected. LA2 encodes a novel chloroplastic protein that can interact with the starch biosynthetic enzyme Oryza sativa plastidic phosphoglucomutase (OspPGM) to regulate starch biosynthesis in rice shoot gravity-sensing cells. Genetic analysis showed that LA2 regulates shoot gravitropism and tiller angle by acting upstream of LA1 to mediate lateral auxin transport.
  • Our studies revealed that LA2 acts as a novel regulator of rice tiller angle by specifically regulating starch biosynthesis in gravity-sensing cells, and established the framework of the starch-statolith-dependent rice tiller angle regulatory pathway, providing new insights into the rice tiller angle regulatory network.

Introduction

Rice (Oryza sativa L.) tiller angle, a key component of rice plant architecture, is defined as the angle between the main culm and its tillers (Xu et al., 1998). In cultivation, tiller angle acts as the primary limiting factor for plant density of a rice variety, thus greatly determining rice grain yield (Xu et al., 1998; Wang & Li, 2008; Ikeda et al., 2013; Gao et al., 2019). Elucidating the genetic control of rice tiller angle would contribute substantially to high-yield breeding goals.

The genetic basis of rice tiller angle is complex and is associated with many loci, each with small contributions to phenotypic variance (Li et al., 1999; Shen et al., 2005; Li et al., 2006; Wang et al., 2011; Dong et al., 2016). Several quantitative trait loci (QTL) have been identified as major contributors including TILLER ANGLE CONTROL 1 (TAC1), TAC3, Dwarf2 (D2) and qTAC8 (Li et al., 1999; Qian et al., 2000; Shen et al., 2005; Yu et al., 2005, 2007; Dong et al., 2016; He et al., 2017). In addition, the transition from prostrate growth of wild rice to erect growth of cultivated rice is controlled mainly by several domestication-related genes such as PROSTRATE GROWTH 1 (PROG1), PROG7, RICE PLANT ARCHITECTURE DOMESTICATION (RPAD) and TILLER INCLINED GROWTH 1 (TIG1) (Jin et al., 2008; Tan et al., 2008; Hu et al., 2018; Wu et al., 2018; Zhang et al., 2019).

Studies in past decades have demonstrated that shoot gravitropism has important roles in determining rice tiller angle (Abe & Suge, 1993; Abe et al., 1996; Li et al., 2007, 2019; Yoshihara & Iino, 2007; Okamura et al., 2013, 2015; Wu et al., 2013; Zhang et al., 2018). The gravitropic response of higher plants involves sequential steps, including gravity perception, signal transduction and organ curvature resulting from asymmetric distribution of auxin (Tasaka et al., 1999; Strohm et al., 2012). Higher plants use the sedimentation of starch-filled amyloplasts as the statolith to sense the direction of gravity during gravitropism (Sack, 1991, 1997). The columella cells in the root cap act as the statocytes for gravity perception in root while the endodermal cells in the inflorescence stem and hypocotyl are responsible for sensing gravity in Arabidopsis shoot (Tasaka et al., 1999; Hashiguchi et al., 2013). Distinct from Arabidopsis, the basal region of the leaf sheath, also known as the leaf sheath pulvinus is the gravity-perceptive tissue in Gramineae (Dayanandan et al., 1977; Parker, 1979; Kaufman et al., 1987). In rice, the starch-filled amyloplasts in leaf sheath bases of seedlings and leaf sheath pulvinus of culms are responsible for gravity sensing (Abe et al., 1994a,b).

Genes involved in starch metabolism such as Arabidopsis phosphoglucomutase encoding gene PGM (Periappuram et al., 2000; Yu et al., 2000) and STARCH EXCESS 1 (SEX1) could regulate plant gravitropism via modulating gravity perception (Yu et al., 2001; Ritte et al., 2002, 2006; Baunsgaard et al., 2005; Kotting et al., 2005). With the absence of starch granules in the hypocotyl endodermal cells and root columella cells, both the root and hypocotyl of pgm1 mutant plants exhibit defective gravitropic responses (Caspar et al., 1985; Caspar & Pickard, 1989; Kofler et al., 2000; Yu et al., 2000; Streb et al., 2009). The loss-of-function of SEX1 results in excess starch and enlarged starch granules in both inflorescence stem and hypocotyl, leading to enhanced sensitivity to gravistimulation in Arabidopsis (Vitha et al., 2007). ADP-glucose pyrophosphorylase (AGP) is a rate-limiting enzyme that catalyzes the first irreversible step in starch biosynthesis (Santelia et al., 2015). Loss-of-function of OsAGPL1, one of the large subunits of AGP, can increase tiller angle by suppressing starch synthesis in rice stems which in turn inhibits the gravitropic response of rice shoots (Okamura et al., 2013, 2015). In addition, overexpression of CO2-RESPONSIVE CONSTANS, CONSTANS-LIKE and TIME of CHLOROPHYLL a/b BINDING PROTEIN1 (CCT) PROTEIN (CRCT) showed increased starch content and tiller angle (Morita et al., 2015), suggesting complicated roles of starch metabolism in controlling rice tiller angle. Besides, it has been demonstrated that the sedimentation of amyloplasts has vital roles in determining rice tiller angle (Wu et al., 2013).

In addition to gravity perception, genes involved in auxin transport and redistribution upon gravistimulation also are required for the regulation of rice tiller angle. LAZY1 (LA1), the first identified gene in controlling rice tiller angle, regulates rice shoot gravitropism by affecting auxin transport which determines the asymmetric distribution of auxin upon gravistimulation (Li et al., 2007; Yoshihara & Iino, 2007). Screening for suppressors of la1 revealed that strigolactones (SLs) can regulate tiller angle by attenuating rice shoot gravitropism via inhibiting local auxin biosynthesis (Sang et al., 2014). Using dynamic transcriptome analysis of rice shoots, Zhang et al., (2018) deciphered a core regulatory pathway of rice tiller angle mediated by the LA1-dependent asymmetric distribution of auxin. In this pathway, the gravistimulation-responsive HEAT STRESS TRANSCRIPTION FACTOR 2D (HSFA2D) positively regulates the expression of LA1 and LA1-mediated auxin asymmetric distribution, which consequently induces asymmetric expression of two transcription factors, WUSCHEL RELATED HOMEOBOX6 (WOX6) and WOX11, thus determining tiller angle (Zhang et al., 2018). A recent study showed that the transcription factors OsHOMEOBOX GENE 1 (OsHOX1) and OsHOX28 could bind to the promoter of HSFA2D and regulate rice tiller angle by suppressing the HSFA2D-LA1 module (Hu et al., 2020). Moreover, the LA1 interacting protein Brevis Radix Like 4 (BRXL4) has been confirmed to regulate shoot gravitropism and tiller angle by affecting nuclear localization of LA1 (Li et al., 2019). Other studies found that α-1,3-FUCOSYLTRANSFERASE (FUCT) and OsmiR167a also can regulate rice tiller angle through auxin transport and distribution (Harmoko et al., 2016; Li et al., 2020).

Despite the significant progress in dissecting the genetic basis of rice tiller angle, the regulatory network of tiller angle is still largely unknown. Here, we report the isolation and characterization of a novel rice tiller angle gene LAZY2 (LA2), which encodes a chloroplast-localized protein crucial for the starch biosynthesis in amyloplasts of the gravity-sensing leaf sheath bases. We show that LA2 acts in the same pathway with OspPGM to regulate tiller angle and shoot gravitropism through LA1-mediated asymmetric distribution of auxin, revealing the framework of starch-statolith-dependent shoot gravitropism regulatory pathway in controlling rice tiller angle.

Materials and Methods

Plant materials

The LAZY2 mutant la2 and Ubipro:LA2-3×Flag transgenic lines were in the 93-11 (Oryza sativa L. subsp. indica cv 93-11) background, and the la1, CRISPR-Cas9 (CR) engineered LAZY2 mutant CR-la2, rice plastidic phosphoglucomutase (OspPGM) encoding gene mutant CR-osppgm, CR-la2 CR-osppgm mutant, Ubipro:LA2-GFP and Ubipro:LA2Δ30-GFP were in a ZH11 (Oryza sativa L. subsp. japonica cv Zhonghua 11) background. The la1 CR-la2-1 double mutant was generated by crossing the la1 with CR-la2-1, and homozygous lines were confirmed by sequencing analysis. Rice plants were grown either in paddy fields or glasshouses in Beijing and Hainan. Tobacco (Nicotiana benthamiana) plants were grown in soil in glasshouses at 22°C under a 16 h : 8 h, light : dark photoperiod for 40 d before infiltration.

Analysis of shoot gravitropism

After germination at 37°C for 2 d, the rice seeds were grown on 0.4% agar at 28°C for 3 d under a 16 h : 8 h, light : dark photoperiod. Then, the seedlings were transferred to darkness and rotated by 90°C for gravistimulation. The gravitropic responses were quantified by measuring shoot curvature every 12 h for 3 d for the light-grown seedling and every 2 h for 6 h for the etiolated rice shoot. For the gravitropism analysis of adult rice plants, individual plants were placed horizontally at the heading stage for 3 wk and the culm curvatures were determined by measuring the gravitropic bending of the 3rd and 4th internodes.

Lateral auxin transport (LAT) assay

A lateral auxin transport assay was implemented according to a method described previously (Zhang et al., 2018) with minor modifications. In brief, 3-d-old dark-grown coleoptile segments (1 cm) were deprived of endogenous IAA. The segments were laid horizontally in 0.6-ml tubes with their apical ends inserted into 10-μl agar blocks containing 500 nM 3H-IAA or 500 nM benzyl adenine (BA) and subjected to a 2.5-h transport in darkness at 28°C. Then, 0.5-cm sections from the nonsubmerged ends of segments were split evenly into upper and lower halves. After 24 -h incubation in 400 μl scintillation liquid, the radioactivity of each half (n = 8–13) was counted using a liquid scintillation counter (1450 MicroBeta TriLux; Perkin-Elmer, Waltham, MA, USA).

Map-based cloning of LA2

In order to isolate LA2, the la2 mutant was crossed with the japonica variety Xiushui 110 (compact plant architecture) to generate the F2 segregation population. InDel markers were designed based on sequence variations between the genomes of Nipponbare and 93-11. The primers used for mapping are listed in Supporting Information Table S1.

Generation of transgenic plants

For complementation test of LA2, the genomic sequence of LA2 was amplified from 93-11 with specific primers (Table S2), and cloned into the pCAMBIA1300 (CAMBIA) vector to generate LA2pro:LA2 plasmid. To generate Ubipro:LA2-3×Flag plasmid, the coding sequence (CDS) of LA2 was amplified by specific primers (Table S2) and ligated to SC-3×Flag plasmid (Xu et al., 2012), forming the 35Spro:LA2-3×Flag intermediate plasmid. Then, the LA2-3×Flag sequence was amplified by specific primers (Table S2), and cloned into binary vector 1460 (Wang et al., 2012). To generate CR-la2 mutants and CR-Osppgm mutants in a ZH11 background, two single guide RNAs (sgRNAs) targeting different exons were designed for each gene and cloned separately into pYLCRISPR/Cas9Pubi-H vector as reported previously (Ma & Liu, 2016). To generate the CR-la2 CR-osppgm double mutant line, the LA2-sgRNA1 together with OspPGM-gRNA1 were constructed into pYLCRISPR/Cas9Pubi-H. The sgRNAs sequences of each gene are listed in Table S2 and primers used for genotyping are listed in Table S3. All generated plasmids were introduced into Agrobacterium tumefaciens stain EHA105 and transformed into indicated rice calli according to a method described previously (Hiei et al., 1994).

Quantitative RT-PCR

Total RNAs were isolated from shoot bases using a TRIzol kit (Invitrogen) according to the user’s manual. Total RNA (2 µg) was treated with DNase I (Ambion) and used to synthesize cDNA with the AMV Reverse Transcription System (Promega). Quantitative reverse transcription (qRT)-PCR was implemented with SsoFast EvaGreen Supermix Kit (Bio-Rad) on the CFX96 Real-time system (Bio-Rad) following the manufacturer’s instructions. Rice Ubiquitin gene was used as an internal control and gene specific primers are listed in Table S4. The 2−ΔΔCT method (Livak & Schmittgen, 2001) was used to calibrate relative expression levels of target genes with the reference gene.

Subcellular localization analysis

For subcellular localization analysis of LA2, DNA fragments encoding full-length CDS of LA2 and LA2Δ30 (deletion of amino acid residues 1 to 30 at N-terminal end) were amplified from 93-11 cDNA by specific primers (Table S2), respectively, and cloned into CaMV35Spro:GFP vector (Lin et al., 2009) with In-fusion HD Cloning Kit (Clontech, TaKaRa, Shiga, Japan), generating 35Spro:LA2-GFP and 35Spro:LA2Δ30-GFP transient expression plasmids. For subcellular localization analysis of OspPGM, we first constructed an intermediate pSCYCE-GFP vector driven by cauliflower mosaic virus (CaMV) 35S promoter via replacing the SCFP3AC155 sequence of pSCYCE vector (Waadt et al., 2008) with green fluorescent protein (GFP). The 720-bp GFP tag was amplified from pH7FWG2 vector (Karimi et al., 2002) with specific primers (Table S2) and ligated to pSCYCE. Then the CDS of OspPGM was amplified from ZH11 cDNA and ligated to pSCYCE-GFP with In-fusion HD Cloning Kit (Clontech), forming 35Spro:OspPGM-GFP plasmid. The Ubipro:GFP plasmid was generated by cloning the 720-bp GFP into binary vector 1460 (Wang et al., 2012). To construct the Ubipro:OspPGM-GFP plasmid, specific primers (Table S2) were used to amplify the CDS of OspPGM from ZH11 cDNA and the fragment was cloned into Ubipro:GFP plasmid with In-fusion HD Cloning Kit (Clontech). The generated plasmids CaMV35Spro:GFP, 35Spro:LA2-GFP, 35Spro:LA2Δ30-GFP, pSCYCE-GFP and 35Spro:OspPGM-GFP were transformed into rice protoplasts via polyethylene glycol (PEG)-mediated transformation (Bart et al., 2006). After 12-h incubation in the dark, the GFP signals and autofluorescence of chloroplasts were examined under confocal microscope at excitation wavelengths of 488 and 647 nm, respectively (FluoView 1000; Olympus, Tokyo, Japan). The Ubipro:GFP and Ubipro:OspPGM-GFP plasmids were transformed intro A. tumefaciens strain EHA105 and infiltrated into abaxial epidermis of N. benthamiana leaves as reported previously (Liu et al., 2010). For confocal microscope analysis, mesophyll cells in the abaxial leaf discs 3 d post-infiltration were imaged at the aforementioned excitation wavelengths.

Firefly luciferase complementation imaging (LCI) assay

In order to generate 35Spro:LA2-nLUC and 35Spro:LA2-cLUC plasmids for LCI assay in N. benthamiana leaves, the CDS of LA2 was amplified from ZH11 cDNA and ligated to the modified 35Spro:nLUC and 35Spro:cLUC vector, respectively. The 35Spro:OspPGM-nLUC and 35Spro:OspPGM-cLUC plasmids were constructed through cloning the PCR products amplified by specific primers to the 35Spro:nLUC vector and 35Spro:cLUC vector, respectively. The cDNA prepared from ZH11 was used as template for PCR amplification and primer sequences are listed in Table S2. All plasmids were transformed into A. tumefaciens strain EHA105 and abaxial epidermis of N. benthamiana leaves were infiltrated with Agrobacterium suspensions as reported previously (Chen et al., 2008). After infiltration for 2–3 d, the abaxial epidermis was smeared with 1 mM beetle luciferin (Promega; E1603) and imaged under NightShade LB 985 In Vivo Plant Imaging System (IndiGO; Berthold Technologies Co., Bad Wildbad, Germany).

Bimolecular florescence complementation (BiFC) assays in rice protoplasts

In order to construct the plasmids for BiFC analysis, the CDSs of LA2 and OspPGM were amplified from ZH11 cDNA and ligated to pSCYCE (SCC) and pSCYNE (SCN) vectors (Waadt et al., 2008) to generate pSCN-LA2 and pSCC-OspPGM, respectively. The plasmids were transformed into rice protoplast as described previously (Bart et al., 2006). Confocal microscope analyses for cyan fluorescent protein (CFP) signals and chloroplast autofluorescence were conducted at the excitation wavelengths of 405 and 647 nm, respectively (FluoView 1000; Olympus). Primers used for BiFC assay are listed in Table S2.

Preparation of recombinant proteins

Codon-optimized LA2 (named LA2O) was synthesized and cloned into the pGEX-6P-1 (GE Healthcare, Chicago, IL, USA). The MBP-LA2 expression plasmid was generated by cloning LA2O into EcoRI/SalI digested pMAL-c2X (New England Biolabs, Ipswich, MA, USA) with specific primers. Site-directed mutagenesis primers were used to generated the MBP-LA2L121F through overlapping PCR. The generated plasmids were transformed into E. coli strain Transetta (DE3) and bacteria cells were cultured in lysogeny broth (LB) medium at 37°C to reach a concentration of OD600 = 0.6. After induction with 0.6 mM isopropyl-β-d-1-thiogalactopyranoside (IPTG) at 28°C for 6 h, cells were harvested and lysed using a JNBIO high pressure homogenizer (Shunxinwangchang Co., Beijing, China). The lysates were centrifuged at 15 294 g for 30 min at 4°C and the supernatants were collected for affinity chromatography or further analysis. Primers used for recombinant protein preparation are listed in Table S2.

Antibody preparation

Approximately 5 mg of GST-LA2O fusion protein purified with Glutathione Sepharose 4 Fast Flow (GE Healthcare) was used to raise polyclonal antibodies against LA2 in rabbit. The specificity of LA2 antiserum was determined by immunoblotting analysis with Ubipro:LA2-3×Flag transgenic lines (Fig. S1). For detecting endogenous LA2 protein, plant Actin was used as an internal control and the dilutions for immunoblotting were 1 : 5000–1 : 10 000 for anti-LA2 polyclonal antibody and 1 : 10 000 for anti-Actin monoclonal antibody (M20009L; Abmart Inc., Shanghai, China).

Starch granule staining assay

A starch granule staining assay was conducted with a periodic acid Schiff (PAS) Kit (Sigma-Aldrich) following the manufacturer’s instructions. Briefly, the shoot bases of 5-d-old rice seedlings were fixed with FAA (formalin/acetic acid/alcohol) at 4°C for 1 d, followed by serial dehydration and infiltration, and embedded in paraffin (Paraplast Plus; Sigma). Tissues were sliced into 10-μm sections with a microtome (RM2145; Leica, Wetzlar, Germany), and then deparaffinized and hydrated, followed by staining in periodic acid solution for 5 min. After rinsing with distilled water, sections were immersed in Schiff’s reagent for 15 min and then subjected to dehydrating, clearing, and mounting sequentially. Finally, slides were observed under a Leica DMR microscope and photographed with a Micro Color Charge-Coupled Device (CCD) camera (Apogee Instruments, North Logan, UT, USA).

Results

Identification of rice tiller angle mutant la2

We isolated a 60Co γ-ray radiation-induced rice mutant la2 from the radiation mutagenesis library in the indica cultivar 93-11(wild-type, WT) background (Fig. 1a). Distinct from the compact architecture of the WT plants at maturity, the la2 mutant exhibited significantly increased tiller angle (Fig. 1a–c). Comparison of the agronomic traits between 93-11 and la2 showed that there were no significant differences in plant height, tiller number, leaf angle or culm diameter, nor in panicle traits including number of primary or secondary branches, grain number per panicle, setting rate, 1000-grain weight or grain yield per plant (Fig. S2). These results suggested that the causative LA2 gene is specifically involved in the control of rice tiller angle.

Details are in the caption following the image
Phenotypic characterization of the rice LAZY2 mutant la2. (a) The gross morphologies of 93-11 and la2 at maturity. Bars, 20 cm. (b, c) Comparison of the tiller bases (b) and tiller angle (c) between 93-11 and la2 at maturity. Bars, 10 cm. Values in (c) are means ± SD (n = 10). Student’s t-test: **, P < 0.01. (d) Shoot curvature of 93-11 and la2 seedlings after gravistimulation for 72 h. The arrow indicates the direction of gravity. Bars, 2 cm. (e) Dynamic changes of shoot curvature of 93-11 and la2 upon gravistimulation. Values are means ± SD (n = 15). Student’s t-test: **, P < 0.01. (f) Shoot gravitropism of 93-11 and la2 adult plants after gravistimulation for 3 wk. The top panel, shoot curvature of 93-11 (left) and la2 (right). The bottom panel, culm curvature of 93-11 (left) and la2 (right). The yellow arrows, red arrows and blue arrows indicate the 2nd, 3rd and 4th nodes respectively, while the black arrow indicates the direction of gravity. Bars, 20 cm. (g) Statistical analysis of the culm curvature of 93-11 and la2 in (f). Values means ± SD (n = 10). Student’s t-test: **, P < 0.01.

Because shoot gravitropism has a predominant role in the formation of rice tiller angle (Gao et al., 2019), we examined whether the enlarged tiller angle of la2 is caused by defective shoot gravitropism. We measured the shoot curvature of 3-d-old seedlings upon gravistimulation, and found that the gravitropic bending of la2 was severely impaired compared with 93-11 (Fig. 1d,e). After reorientation by 90° for 3 wk at the heading stage, the culm curvatures of the third and fourth internodes of la2 were significantly decreased compared with that of 93-11 (Fig. 1f,g), suggesting defective shoot gravitropism of la2 at the adult stage, which accounts for the increased tiller angle of la2 (Fig. 1a–c).

The la2 mutant is defective in auxin redistribution upon gravistimulation

It is well known that an asymmetric distribution of auxin mediated by auxin carriers has a key role in organ curvature in plant gravitropism (Firn et al., 2000; Morita & Tasaka, 2004; Su et al., 2017). To characterize the involvement of auxin redistribution in LA2-mediated rice shoot gravitropism, we first examined the expression of the auxin-responsive marker gene OsIAA20 in the shoot bases of 93-11 and la2 seedlings upon gravistimulation. After gravistimulation, we divided the shoot bases into upper and lower portions along the direction of gravistimulation, and found that OsIAA20 was asymmetrically expressed in the shoot bases of 93-11 with preferred expression in the lower sides in 93-11 while the expression of OsIAA20 was significant decreased in the lower sides in la2 compared with that in the WT (Fig. 2a), implying impaired asymmetric distribution of auxin in la2 upon gravistimulation. Further measurement of LAT in coleoptiles by monitoring 3H-IAA movement showed that LAT was significantly reduced in la2 (Fig. 2b). Our previous study demonstrated that the asymmetric distribution of auxin can induce the asymmetric expression of WUSCHEL RELATED HOMEOBOX6 (WOX6) and WOX11 to control rice shoot gravitropism and tiller angle (Zhang et al., 2018). Thus, we examined the expression of these two genes and found that the preferred expressions of WOX6 (Fig. 2c) and WOX11 (Fig. 2d) in the lower sides also were reduced significantly in the shoot bases of la2 seedlings. Consistently, the etiolated la2 shoots grown in darkness displayed severe defects in shoot curvature after they were rotated by 90° (Fig. 2e,f). Taken together, these results suggested that LA2 regulates rice shoot gravitropism via LAT-mediated asymmetric distribution of auxin.

Details are in the caption following the image
LAZY2 (LA2) regulates rice shoot gravitropism through modulating lateral auxin transport. (a) Expression level of OsIAA20 at the upper sides and lower sides of the shoot bases in 93-11 and la2 upon gravistimulation (Os, Oryzasativa). Values are means ± SE (n = 3). Student’s t-test: **, P < 0.01. (b) Comparison of lateral auxin transport (LAT) in 93-11 and la2 coleoptiles. The Cpm ratio indicates the radioactivity of the lower side to that of the upper side of coleoptiles upon gravistimulation for 2.5 h. Values are means ± SE (n = 10–13). Student’s t-test: **, P < 0.01. (c, d) Relative expression of WOX6 (c) and WOX11 (d) at the upper sides and lower sides of the shoot bases in 93-11 and la2 upon gravistimulation. Values are means ± SE (n = 3). Student’s t-test: **, P < 0.01. (e) Shoot gravitropic responses of 93-11 and la2 in darkness upon gravistimulation for 6 h. The black arrow indicates the direction of gravity. Bars, 1 cm. (f) Dynamic changes of the shoot curvature of 93-11 and la2 in the dark. Values are means ± SD (n = 15). Student’s t-test: **, P < 0.01.

Map-based cloning of LA2

In order to isolate the LA2 gene, we took a map-based cloning approach. LA2 was narrowed down to a 62-kb region between molecular markers M9 and M6 (Fig. 3a). Within this region, there are eight open reading frames (ORFs). Sequencing of these eight ORFs in la2 showed a point mutation only at the fifth exon of LOC_Os02g08380, and this nucleotide substitution from C in 93-11 to T in la2 results in an L121F amino acid substitution (Figs 3a, S3). A genetic complementation test was performed to verify whether LOC_Os02g08380 is the LA2 locus. The plasmid LA2pro:LA2 harboring a 5622-bp genomic sequence (consisting of a 1777-bp upstream sequence, the entire ORF including seven exons and six introns, and an 822-bp downstream region) was introduced into the la2 mutant. We obtained 15 independent transgenic lines and all of them rescued the phenotypes of la2, including tiller angle (Fig. 3b,c) and shoot gravitropism (Fig. 3d,e). Therefore, LOC_Os02g08380 is the rice LA2 gene, and its L121F amino acid substitution is responsible for the phenotype of la2.

Details are in the caption following the image
Map-based cloning and conformation of LAZY2 (LA2) in rice. (a) Map-based cloning of LA2. LA2 was narrowed down to a 62-kb candidate region between molecular markers M9 and M6. Numbers under the markers indicate recombinants. The red triangles indicate LA2 and the blue arrow indicates the 1-bp substitution from C (in 93-11) to T (in la2) at the fifth exon of LA2 which results in an L121F amino acid substitution in la2 (red arrow). The black boxes represent the coding regions and the brown boxes indicate the untranslated regions (UTRs). The black lines between boxes represent introns. (b) The gross morphologies of 93-11, la2 and the representative LA2pro:LA2/la2 complemented lines (−1 and −2) at the tillering stage. Bars, 20 cm. (c) Statistical analysis of the tiller angle of 93-11, la2 and LA2pro:LA2/la2 (−1 and −2). Values are means ± SD (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 93-11, la2 and LA2pro:LA2/la2 (−1 and −2) after gravistimulation for 72 h. The arrow indicates the direction of gravity vector. Bars, 2 cm. (e) Dynamic changes of shoot curvature of the lines in (d). Values are given as means ± SD (n = 15).

In order to further confirm the function of LA2 in controlling rice tiller angle and facilitate subsequent studies, we generated loss-of-function mutations of LA2 in the japonica rice variety ZH11 through CRISPR-Cas9 (CR) gene editing. Using two sgRNAs separately targeting the first and third exons, we generated multiple independent transgenic lines for each target site and identified four kinds of homozygous mutants harboring insertion/deletion (Indel) alleles through PCR screening and sequencing analysis (Fig. S4a). Compared with the WT plants, all the CR-engineered la2 mutant lines displayed significantly increased tiller angle (Fig. S4b,c). Immunoblotting analysis showed that the LA2 protein was absent in these mutants (Fig. S4d), suggesting that all of them were null mutants of LA2. Among them, CR-la2-1 was chosen for subsequent analysis in this study.

LA2 encodes a chloroplastic protein

Sequence analysis showed that the putative LA2 protein is composed of 185 amino acids (Figs 3a, S3), with a predicted mass of c. 20 kDa. The TargetP program (http://www.cbs.dtu.dk/services/TargetP/) predicted a 25-amino acid chloroplast transit peptide (cTP) at the N-terminal end of LA2 (Figs 3a, S3). To confirm its subcellular localization, the GFP protein was fused to the C-terminal end of LA2, and the LA2-GFP fusion protein was transient expressed in rice protoplasts. Confocal microscopy showed definitive localization of LA2-GFP in chloroplasts supported by the completely merged signals with the chlorophyll autofluorescence (Fig. 4a). The same subcellular localization also was observed in the mesophyll cells of N. benthamiana leaves in which the LA2-GFP fusion protein, driven by its native promoter, was transiently expressed (Fig. S5). When the first 30 amino acid residues at the N-terminus of LA2 were deleted, the LA2Δ30-GFP fusion protein signals could not be observed in chloroplasts (Fig. 4a). To verify the biological significance of the subcellular localization of LA2, we generated stable expression constructs of LA2-GFP and LA2Δ30-GFP driven by the maize ubiquitin promoter and introduced them into the CR-la2-1 mutant. Ten and 12 independent transgenic lines were obtained for LA2-GFP and LA2Δ30-GFP, respectively. Phenotypic analysis showed that only those lines carrying intact LA2-GFP could rescue the tiller angle phenotype of CR-la2-1 (Fig. 4b,c), suggesting that the cTP at the N-terminus of LA2 is indispensable for its subcellular localization and function in controlling rice tiller angle.

Details are in the caption following the image
LAZY2 (LA2) encodes a rice chloroplastic protein and L121 is crucial for the stability of the LA2 protein. (a) Confocal microscope analysis of 35S:GFP (GFP), 35S:LA2-GFP (LA2-GFP), and 35S:LA2Δ30-GFP (LA2Δ30-GFP ) in the rice chloroplast (GFP, green fluorescent protein). Bars, 10 µm. (b) Statistical analysis of tiller angle of ZH11, CR-la2-1, and representative Ubipro:LA2-GFP/CR-la2-1 (LA2-GFP-1 and-2) and Ubipro:LA2Δ30-GFP/CR-la2-1 (LA2Δ30-GFP-1 and -2) at tillering stage. Values are means ± SD (n = 8). Different letters above the column represent statistically significant difference at P < 0.05 (one-way ANOVA, Tukey’s honestly significant difference). (c) Phenotypes of ZH11, CR-la2-1, Ubipro:LA2-GFP/CR-la2-1 (LA2-GFP) and Ubipro:LA2Δ30-GFP/CR-la2-1 (LA2Δ30-GFP) at tillering stage. Bars, 20 cm. (d) Transcription level of LA2 in the shoot bases of 93-11 and la2 seedlings. qLA2-1 and qLA2-2 are primers located upstream and downstream of the L121F mutation site in la2 respectively. Data are means ± SD (n = 3). (e) Detecting LA2 at protein level in 93-11 and la2 by immunoblotting analysis with the LA2 antibody. Plant actin was used as an endogenous control.

Smart (http://smart.embl-heidelberg.de/) analysis showed that LA2 has an annotated YbaB DNA-binding domain at the C-terminus (amino acids 85 to 177) in which the L121F mutation is localized (Fig. 3a). Multiple sequence alignment showed that LA2 shares the same conserved domain with Arabidopsis STIC2, STIC2-Like (STCL) and Haemophilus influenzae (Hi) and Escherichia coli (Ec) YbaB proteins (Fig. S6a), falling into the YbaB protein family. Phylogenetic analysis showed that homologs of LA2 were ubiquitously distributed in the Viridiplantae, including land plants, moss and algae (Fig. S6b). To uncover how the L121 residue regulates the expression of LA2, we designed primer pairs located upstream and downstream of the mutation site, and examined the transcription level of LA2 in both la2 and 93-11 by qRT-PCR. The results showed that the transcription levels of LA2 in la2 seedlings were comparable to that in 93-11 (Fig. 4d). Immunoblotting analysis revealed the absence of LA2 protein in mutant plants (Fig. 4e), suggesting that L121 is involved in maintaining LA2 stability. To rule out the possibility that L121F may disrupt the recognition between LA2 antibody and la2, we expressed the MBP-LA2 and MBP-LA2L121F recombinant proteins in vitro (Fig. S7a). Immunoblotting analysis showed that comparable signal intensity was detected between MBP-LA2 and MBP-LA2L121F when equal amounts of recombinant proteins were loaded (Fig. S7b). In addition, qRT-PCR analysis showed that neither the LA2 transcripts nor the protein accumulation of LA2 was changed significantly in response to gravistimulation (Fig. S8).

LA2 functions in gravity perception by regulating starch metabolism in amyloplasts

It is well-known that higher plants use the sedimentation of starch-filled amyloplasts to sense the direction of the gravity vector (Sack, 1991, 1997). To test whether LA2 is involved in gravity perception, we first examined starch granules, the major inclusions of amyloplasts, in the gravity-sensing leaf sheath base in la2 and 93-11 (Fig. 5a). The result showed that the starch granules in amyloplasts were completely absent in the gravity-sensing cells of la2 (Fig. 5c,f). By contrast, large numbers of starch granules were stained purple and aggregated in the leaf sheath base cells in 93-11 (Fig. 5b,e) and the complemented transgenic line (Fig. 5d,g). However, the seed development of la2 was not affected, suggesting that LA2 is specifically involved in the regulation of starch metabolism in gravity-sensing cells in rice. In higher plants, the starch-statolith hypothesis postulated the sedimentation of starch-filled amyloplasts in the direction of the gravity vector to trigger gravity sensing (Sack, 1991, 1997). Therefore, the above results suggested that LA2 is involved in the starch metabolism in amyloplasts, which regulates shoot gravitropism by modulating gravity perception, thus controlling rice tiller angle. Interestingly, although the starch granules in amyloplasts were completely absent in the gravity-sensing cells of la2, the la2 seedling still retained a proportion of gravity response capacity (Fig. 1d–g), implying the existence of another regulatory pathway independent of starch in controlling rice shoot gravitropism and tiller angle.

Details are in the caption following the image
LAZY2 (LA2) is involved in the biogenesis of amyloplasts of rice leaf sheath base. (a) Boxed region shows the 0.5 cm of the shoot base used for starch granule staining of the 5-d-old 93-11, la2 and LA2pro:LA2/la2-1 seedlings. (b–d) Starch granule staining of leaf sheath bases of 93-11 (b), la2 (c) and LA2pro:LA2/la2-1 (d). Bars, 100 µm. (e–g) Close-ups of the red boxed regions in (b), (c) and (d), respectively. Bars, 100 µm.

LA2 acts in the same pathway with OspPGM in regulating rice tiller angle

In Arabidopsis, the plastid-localized PGM has been well-characterized in regulating starch biosynthesis in chloroplasts and shoot gravitropism, and its ortholog in rice OspPGM also was involved in the regulation of starch synthesis (Caspar et al., 1985; Caspar & Pickard, 1989; Kofler et al., 2000; Periappuram et al., 2000; Yu et al., 2000; Streb et al., 2009; Lee et al., 2016). The similar function in starch biosynthesis prompted us to speculate that LA2 and OspPGM may function in the same pathway. To test this hypothesis, we first analyzed the subcellular localization of OspPGM. The transient expression of OspPGM-GFP in both rice protoplast and N. benthamiana showed chloroplast-localized signals (Fig. S9). Moreover, a cTP localized in the first 50 amino acid residues was identified by the subcellular localization analysis of the OspPGMΔ50-GFP fusion protein, which showed incapability of chloroplast-localization (Fig. S9a). Next, we checked whether LA2 interacted with OspPGM. A luciferase complementation assay in N. benthamiana leaves showed the interaction between LA2 and OspPGM (Fig. 6b). A BiFC test further demonstrated that the interaction between LA2 and OspPGM occurred in rice chloroplast (Fig. 6c), in accordance with their subcellular localization.

Details are in the caption following the image
LAZY2 (LA2) and Oryzasativaplastidicphosphoglucomutaseencodinggene (OspPGM) act in the same pathway in the control of rice tiller angle. (a) Schematic illustration and sequence alignment of the two gRNA target sites of OspPGM in the CRISPR-Cas9(CR) engineered CR-osppgm mutant lines. The sgRNA target sequences are highlighted with red font and the red boxes indicate protospacer-adjacent motif (PAM) sequences. The blue capitals represent the 1-bp insertion in the target sites of the CR-osppgm mutants. The black boxes represent coding region of OspPGM while the brown ones indicate untranslated regions. (b) LA2 interacts with OspPGM in LUC assay. The pseudocolor bar showed the luminescence intensity in the image. (c) LA2 interacts with OspPGM in the rice chloroplasts. Bars, 10 µm. (d) Phenotypes of CR-osppgm-1 and CR-osppgm-2 at the tillering stage. Bars, 20 cm. (e) Statistical analysis of tiller angle of the CR-osppgm-1 and CR-osppgm-2 lines. Values are means ± SD (n = 10). Different letters above the column represent statistically significant difference at P < 0.05 (one-way ANOVA, Tukey’s honestly significant difference). (f) Statistical analysis of tiller angle of the CR-la2-1CR-osppgm-1 double mutant line. Values are means ± SD (n = 10). Different letters above the column represent statistically significant difference at P < 0.05 (one-way ANOVA, Tukey’s honestly significant difference). (g) Gross morphologies of the ZH11, CR-la2-1, CR-osppgm-1 and CR-la2-1CR-osppgm-1 at tillering stage. Bars, 20 cm.

In order to confirm the involvement of OspPGM in controlling rice tiller angle, two sgRNAs separately targeting the second and ninth exons were designed, and used to generate CRISPR-Cas9 engineered osppgm mutants in a ZH11 background. Homozygous osppgm allelic mutant lines from different target sites, designated as CR-osppgm-1 and CR-osppgm-2 respectively, were obtained through PCR screening and sequencing analysis (Fig. 6a,d,e). Both mutants exhibited a 1-bp T insertion in the target sites which resulted in a premature stop codon (CR-osppgm-1) or frame shift (CR-osppgm-2) in the OspPGM coding region (Fig. 6a). In comparison with WT plants, the CR-osppgm mutants displayed significantly increased tiller angle similar to that of CR-la2-1 mutant plants (Fig. 6d,e), demonstrating that OspPGM acted as a negative regulator in controlling rice tiller angle. Again, we generated CR-la2-1 CR-osppgm-1 double mutant using CR technology and compared it to the corresponding single mutants. Phenotypic characterization and statistical analysis showed that the double mutant exhibited a larger tiller angle than the WT plants, but this was indistinguishable from the CR-la2-1 or CR-osppgm-1 single mutant in plant architecture (Fig. 6f,g). Collectively, these results suggested that LA2 and OspPGM act in the same pathway in regulating rice tiller angle.

LA2 acts upstream of LA1 in controlling rice shoot gravitropism and tiller angle

LA1-mediated asymmetric distribution of the auxin pathway is essential in controlling rice tiller angle (Li et al., 2007; Yoshihara & Iino, 2007; Zhang et al., 2018). To establish the link between LA2-involved gravity perception and LA1-mediated auxin distribution, we crossed the CR-la2-1 mutant with la1 and identified the la1 CR-la2-1 double mutant in the F2 population by sequencing. The la1 CR-la2-1 double mutant showed similar plant architecture to the la1 single mutant with a spread-out growth habit, while its tiller angle was significantly increased compared with CR-la2-1 single mutant (Fig. 7a,b). Gravitropic response analysis showed that the shoot curvature of la1 CR-la2-1 was comparable to that of la1 mutant, but significantly reduced when compared with CR-la2-1 single mutant (Fig. 7c,d). Previous studies have identified the key role of LA1 in determining auxin redistribution upon gravitropism, during a stage between amyloplasts sedimentation and gravitropic bending (Abe et al., 1994a,b; Godbole et al., 1999; Li et al., 2007; Yoshihara & Iino, 2007). Together with the above results, we concluded that LA2 acts upstream of LA1 in controlling rice shoot gravitropism and tiller angle.

Details are in the caption following the image
LAZY2 (LA2) acts upstream of LA1 in the control of rice tiller angle. (a) The plant architecture of Oryzasativa L. subsp. japonica cv Zhonghua 11 (ZH11), CRISPR-Cas9 (CR) engineered CR-la2-1, la1 and la1CR-la2-1 at the tillering stage. Bars, 20 cm. (b) Statistical analysis of tiller angle of ZH11, CR-la2-1, la1 and la1CR-la2-1 at the tillering stage. Values are means ± SD (n = 10). Different letters above the column represent statistically significant difference at P < 0.05 (one-way ANOVA, Tukey’s honestly significant difference). (c) Shoot gravitropic responses of ZH11, CR-la2-1, la1 and la1CR-la2-1 in darkness upon gravistimulation for 6 h. The arrow indicates the direction of gravity. Bars, 1 cm. (d) Statistical analysis of shoot curvature of the lines in (c). Values are means ± SD (n = 6). Different letters above the column represent statistically significant difference at P < 0.05 (one-way ANOVA, Tukey’s honestly significant difference).

Discussion

As a determinant of rice plant architecture, tiller angle is a major factor influencing plant density in rice production. However, the regulatory pathways and networks of rice tiller angle are largely unknown because the identified tiller angle genes are limited. In this study, we identified a novel rice tiller angle control gene LAZY2 (LA2) and established a framework of a rice tiller angle regulatory pathway that links tiller angle, starch-statolith-dependent gravity perception, asymmetric auxin distribution and shoot gravitropism. The previously unknown LA2 protein does not contain any functionally characterized domain that suggests enzymatic activity or other biochemical function in plant, except for a transit peptide indicative of plastid localization. We proposed a pathway that involves LA2, Oryza sativa plastidic phosphoglucomutase (OspPGM) and LA1 to explain how and to what extent the tiller angle is regulated by the starch-statolith-mediated gravity perception in rice (Fig. 8). In this pathway, the chloroplast-localized LA2 and OspPGM regulate the accumulation of starch granules, thus modulating gravity perception by controlling the biogenesis of starch-filled amyloplasts in gravity-sensing cells in rice shoots. The LA2-modulated gravity perception then regulates rice shoot gravitropism through LA1-mediated asymmetric distribution of auxin, thereby determining rice tiller angle (Fig. 8).

Details are in the caption following the image
A proposed working model of LAZY2 (LA2) in the control of rice tiller angle. In rice, LA2 acts in the starch-statolith-dependent pathway with Oryzasativa plastidic phosphoglucomutase (OspPGM) in gravity perception cells to regulate the starch biosynthesis in amyloplasts which modulates gravity sensing. In wild-type plants, the LA2-modulated gravity perception controls rice tiller angle by regulating shoot gravitropism through LA1-mediated asymmetric distribution of auxin. However, loss-of-function of LA2 leads to few starch granules, which in turn decreases the magnitude of shoot gravitropism resulting in loose plant architecture. The absence of starch granules in gravity-sensing cells but moderate changes of shoot gravitropism and tiller angle in la2 imply a starch-statolith-independent pathway in controlling rice shoot gravitropism and tiller angle. The black arrow marked with 'g' indicates the direction of gravity vector.

LA2 regulates tiller angle by modulating starch biosynthesis in gravity perception cells

Gravitropism consists of sequential steps including gravity perception, signal transduction and auxin redistribution-mediated organ bending (Tasaka et al., 1999; Strohm et al., 2012). With respect to gravity perception, the mainstream view is the starch-statolith hypothesis which considers the high-density, starch-filled amyloplasts to be the key factor in sensing gravistimulation and initiating the signaling through its sedimentation (Sack, 1991, 1997). A variety of studies have shown that the magnitude of gravitropism is determined largely by the starch content in amyloplasts (Caspar & Pickard, 1989; Kiss et al., 1989, 1996, 1997; Weise & Kiss, 1999). In this study, we found that LA2 encodes a chloroplastic protein (Figs 4a, S5). The chloroplast is a kind of plastid characterized by an intact thylakoid membrane system and a high concentration of chlorophyll that conducts starch biosynthesis in plant photosynthetic tissues (Barkan, 1998; Wise, 2007). A previous study showed that the translocon of the outer membrane of chloroplasts (TOC) complex is involved in gravity signal transduction within the statocytes in root gravitropism (Stanga et al., 2009). Moreover, the characterization of phosphoglucomutase encoding gene (PGM) in Arabidopsis also demonstrated the important role of chloroplastic protein in determining shoot gravitropism or branching angle by regulating starch-statolith-dependent gravity perception (Caspar et al., 1985; Caspar & Pickard, 1989; Kiss et al., 1989; Streb et al., 2009). We thus postulated that LA2 acts as a novel regulator of starch-statolith-mediated gravity perception in gravity-sensing cells.

We showed that starch granules, the major inclusions of amyloplasts acting as the key factor for gravity perception and signal transduction, were completely lost in the leaf sheath base of la2 seedlings (Fig. 5c,f), suggesting that LA2 controls rice tiller angle by regulating starch metabolism which modulates gravity perception by determining the turnover of starch in amyloplasts. In addition, we demonstrated that LA2 could interact with OspPGM in rice chloroplast (Fig. 6c), the protein homolog of Arabidopsis plastid-localized PGM that reversibly catalyzes the inter-conversion between Glc-6-P and Glc-1-P in starch biosynthesis (Periappuram et al., 2000; Yu et al., 2000). Given the role of OspPGM in rice starch biosynthesis (Lee et al., 2016), the generation and characterization of CR-osppgm single mutant lines and CR-la2-1 CR-osppgm-1 double mutant (Fig. 6a,d–g) not only confirmed that OspPGM was involved in controlling rice tiller angle, but also demonstrated that LA2 is indeed involved in the starch metabolism pathway with OspPGM in controlling rice tiller angle, and is potentially specialized in starch biosynthesis in rice chloroplasts. The involvement of LA2 in starch metabolism, an earlier biochemical process that defines the prerequisite determining the magnitude of gravitropism via modulating gravity perception by amyloplast sedimentation, provides new insight into the regulatory mechanism of rice tiller angle.

LA2 acts upstream of LA1-mediated asymmetric distribution of auxin in controlling shoot gravitropism and rice tiller angle

In higher plants, the Cholodny–Went hypothesis proposed that the asymmetric distribution of auxin between two sides of the gravitropic organs induced by gravistimulation resulted in differential growth and organ curvature (Firn et al., 2000), defining the core role of auxin redistribution in gravitropic bending. In this study, we showed that the asymmetric expression of OsIAA20 between the upper and lower sides of shoot bases upon gravistimulation was severely impaired in la2 (Fig. 2a), indicating that asymmetric auxin distribution upon gravistimulation was abnormal in la2. This was further supported by the significantly reduced lateral auxin transport (LAT) in la2 shoots (Fig. 2b), suggesting that LA2 regulates shoot gravitropism and rice tiller angle in an auxin redistribution-dependent manner.

Given the essential role of LA1 in determining asymmetric auxin distribution (Godbole et al., 1999; Li et al., 2007; Yoshihara & Iino, 2007) and the impaired redistribution of OsIAA20 expression upon gravistimulation and LAT in the la2 mutant (Fig. 2a,b), we checked the genetic relationships between LA2 and LA1. The results suggested that LA2 acts upstream of LA1 in the control of shoot gravitropism and rice tiller angle (Fig. 7a–d). Different from the la2 mutant, the number and distribution of starch granules and the sedimentation of amyloplasts in gravity-sensing cells are normal in la1 (Abe et al., 1994b; Godbole et al., 1999), suggesting that LA1 is likely to function at a step between amyloplast sedimentation and auxin transport during gravity signal transduction (Godbole et al., 1999; Li et al., 2007; Yoshihara & Iino, 2007). Moreover, the asymmetric expression of WUSCHEL RELATED HOMEOBOX6 (WOX6) and WOX11, two transcription factor genes that act downstream of LA1-mediated asymmetric auxin distribution and connect shoot gravitropism with rice tiller angle (Zhang et al., 2018), were significantly suppressed in la2. Because we have demonstrated that LA2 controls tiller angle and shoot gravitropism by regulating the biogenesis of amyloplasts in leaf sheath base (Fig. 5), we propose that LA1-dependent asymmetric distribution of auxin acts downstream of LA2-mediated starch metabolism in the control of shoot gravitropism and hence rice tiller angle.

LA2 is homologous to bacteria YbaB proteins but may have a novel role in rice

In the study, we showed that LA2 encodes a previously unknown protein that is homologous to bacteria YbaB proteins (Fig. 3a), which has been implicated in DNA-binding activity and global regulation of gene expression in bacteria, and has positive roles in membrane protein biogenesis in Escherichia coli (Cooley et al., 2009; Skretas & Georgiou, 2010; Jutras et al., 2012). Two homologs of LA2, suppressor of Tic40 (STIC)2 and STIC-like (STCL) participate in chloroplast protein targeting in Arabidopsis (Bedard et al., 2017). Interestingly, neither single mutant stic2 or stcl, nor the stic2 stcl double mutant were characterized by defects in shoot gravitropism and branching angle (Bedard et al., 2017), suggesting a novel function of LA2 distinct from its homologs in Arabidopsis.

As a chloroplast protein, LA2 also is distinct from Arabidopsis LZY proteins (AtLAZY1-AtLAZY4) that share five conserved domains, which seem to be decisive for the proper function of AtLZY proteins (Yoshihara et al., 2013; Ge & Chen, 2016; Yoshihara & Spalding, 2017, 2020; Furutani et al., 2020; Jiao et al., 2021).

In our study, we found that LA2 is involved in the biogenesis of amyloplasts (Fig. 5) and acts in the common pathway with OspPGM to control rice tiller angle (Fig. 6). As a general regulator of starch biosynthesis, OspPGM is essential for pollen fertility and the osppgm mutation could result in male sterility due to impaired starch biosynthesis in pollen (Lee et al., 2016). Intriguingly, there was no significant decrease in yield traits in the la2 mutant (Fig. S2), suggesting a specific role of LA2 in regulating starch metabolism in gravity-sensing tissue. Possibly, the specificity of LA2 in controlling biosynthesis of starch in amyloplasts in the leaf sheath base could split the general function of OspPGM into starch biosynthesis in the sites of gravity perception and that of nongravity perception sites. Because the chloroplastic localization of OspPGM protein is unchanged in the la2 mutant (Fig. S10), and considering the interaction between LA2 and OspPGM (Fig. 6b,c), we propose that LA2 may act as a cofactor that is indispensable for OspPGM to normally synthesize starch in rice gravity perception sites, which in turn regulates tiller angle via the starch-statolith-dependent shoot gravitropism regulatory pathway.

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

In this study, we identified a rice tiller angle regulatory pathway that links LA2, OspPGM, starch-statolith-mediated gravity perception, and LA1-dependent asymmetric distribution of auxin and shoot gravitropism (Fig. 8). In this pathway, the chloroplast-localized LA2 and OspPGM regulate starch metabolism and define the prerequisite of gravity perception by determining the starch biosynthesis in amyloplasts. However, the limited defects of shoot gravitropism in seedlings and moderate increase in tiller angle in la2 (Fig. 1), imply that alternative pathways independent of the starch-statolith hypothesis also may be involved in the control of rice shoot gravitropism and tiller angle. In addition, as the la1 CR-la2-1 double mutant displays a significant increase in tiller angle (Fig. 7a,b) and decreased shoot gravitropism (Fig. 7c,d) compared with the CR-la2-1 single mutant but similar tiller angle and shoot gravitropism to la1 single mutant, we proposed the involvement of a starch-statolith-independent pathway that regulates rice tiller angle in an LA1-dependent manner (Fig. 8).

Taken together, our data demonstrate that LA2 specifically controls rice tiller angle via a starch-statolith-dependent shoot gravitropism pathway. As the starch granules were completely absent in leaf sheath base in la2 together with the moderate contribution to rice tiller angle, further identification of suppressors and enhancers of LA2 will provide more detailed insights into the synergistic regulation of tiller angle by starch-statolith-dependent and -independent pathways.

Acknowledgements

We thank Yaoguang Liu (South China Agricultural University) for providing pYLCRISPR/Cas9Pubi-H vector, and Jian-Min Zhou (Institute of Genetics and Developmental Biology, Chinese Academy of Sciences) for providing vectors for firefly luciferase complementation imaging assay. This work was supported by grants from the National Natural Science Foundation of China (91935301), the National Key Research and Development Program of China (2016YFD0101801), the Strategic Priority Research Program ‘Molecular Mechanism of Plant Growth and Development’ of CAS (XDB27010100), and the Top Talents Program ‘One Case One Discussion (Yishiyiyi)’ from Shandong Province.

    Author contributions

    LH and WW designed research, analyzed data and wrote the paper; LH, WW and NZ performed experiments; YC, YL, XM and YY performed some of the experiments; YW, DW and JL analyzed data; DW generated the rice mutant material; and YW supervised the project, designed research and wrote the manuscript. LH, WW and NZ contributed equally to this work.

    Data availability

    Sequence data from this study can be found in the Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/) under the following accession numbers: LA2 (LOC_Os02g08380), OspPGM (LOC_Os10g11140), LA1 (LOC_Os11g29840), OsIAA20 (LOC_Os06g07040), WOX6 (LOC_Os03g20910), WOX11 (LOC_Os07g48560) and Ubiquitin (LOC_Os03g13170).