Rice LecRK5 phosphorylates a UGPase to regulate callose biosynthesis during pollen development

Plasma membrane-localized OsLecRK5 enhances the activity of UGP1 by phosphorylation for callose biosynthesis during microsporogenesis in rice.

Receptor-like kinases (RLKs) are a large protein family in plants. RLKs localize to the cell surface, where they perceive extracellular cues and transmit them as intracellular signals (Shiu and Bleecker, 2001). Lectin RLKs (LecRLKs), a major subgroup of RLKs, contain an N-terminal carbohydratebinding lectin domain and function in diverse roles from plant development to biotic and abiotic stress responses (Navarro-Gochicoa et al., 2003;Wan et al., 2008). The sgc (small, gluedtogether, and collapsed pollen) mutant is caused by the mutation of a LecRLK gene in Arabidopsis (Wan et al., 2008). LecRLK1 interacts with the N-terminus of AtGSL6, potentially regulating cell plate formation (Dong, 2004). However, these studies did not identify the underlying molecular mechanism.
The rice genome is predicted to contain approximately 173 genes encoding LecRLKs (OsLecRKs) but only a few have been characterized. OsLecRK1, OsLecRK2, OsLecRK3, and OsLecRK4 form a gene cluster, with the first three genes conferring resistance to the rice predator brown planthopper (Liu et al., 2015b). Another rice LecRLK protein, OsLecRK, regulates seed germination and resistance to diseases and insects through actin-depolymerizing factor (Cheng et al., 2013). An important rice blast resistance gene, Pi-d2, encodes a LecRLK that confers resistance to Magnaporthe oryzae (Chen et al., 2006). Although the functions of LecRLKs have been examined in other contexts, how LecRLKs regulate male sterility has not been elucidated.
In this study, we identified a male-sterile mutant, oslecrk5, which is caused by a point mutation in OsLecRK5. The callose wall surrounding each tetrad during male reproductive development is defective in the oslecrk5 mutant. OsLecRK5 is preferentially expressed in anthers and OsLecRK5 localizes to the PM. We established that a conserved lysine residue, K418, controls OsLecRK5 kinase activity and OsLecRK5 phosphorylates the UGPase UGP1 to increase its activity, thus revealing a mechanism by which this LecRLK controls callose biosynthesis during anther development.

Plant materials and growth conditions
Rice plants were grown in South China Agricultural University's paddy field. A mapping population was generated from a cross between oslecrk5 (japonica) and Huanghuazhan (HHZ; indica). For mapping, eight pairs of insertion/deletion molecular markers were designed based on polymorphisms between the japonica and indica genomes (see Supplementary Table S1 at JXB online).

Mutant phenotype characterization
Photographs of whole plants, flowers, anthers, and panicles were taken with a Canon digital camera and a dissecting microscope (Olympus SZx10/DP72). Pollen grains were stained with 1% I 2 -KI solution and observed under a microscope (Olympus CX31). Young spikelets with developing anthers were fixed with 3% (w/v) paraformaldehyde and 0.25% glutaraldehyde in 0.1 M phosphate buffer, pH 7.0, and embedded in Epon 812 resin; semithin sections of 2 μm thickness were cut using a Leica RM 2135 microtome, stained with 0.25% toluidine blue, and photographed using a microscope (Zeiss Axiovert 200).

Callose staining with aniline blue
To observe callose layers, transverse sections of anthers were stained for 10 min at room temperature with 0.01% (w/v) aniline blue in 0.077 M phosphate buffer (pH 8.5; Lu et al., 2014). After being washed with phosphate buffer, sections were visualized under UV light with a confocal laser scanning microscope (Zeiss LSM 7 DUO).

Plasmid construction and rice transformation
For functional complementation, two binary constructs (driven by the native OsLecRK5 promoter) expressing the wild-type OsLecRK5 (pNP::OsLecRK5) or OsLecRK5 fused with a FLAG tag (pNP::OsLecRK5-F) were prepared. For knockout of OsLecRK5, a binary construct (OsLecRK5-KO) was designed and prepared using the CRISPR-GE (Xie et al., 2017) and CRISPR/Cas9 (Ma et al., 2015) systems. Mutations of the target site of T 0 plants were sequenced and analyzed using the DSDecode program (Liu et al., 2015a). To analyze the expression pattern of OsLecRK5, a construct pNP::GUS, in which the β-glucuronidase (GUS) reporter gene was driven by the OsLecRK5 promoter, was prepared for transformation of rice. To test the biological significance of the conserved ATP-binding lysine residue (K418) of OsLecRK5, site-directed mutagenesis using Ω-PCR  was used to create pNP::OsLecRK5 K418E from pNP::OsLecRK5. The pNP::OsLecRK5, pNP::OsLecRK5-F, and pNP::OsLecRK5 K418E constructs were transferred into induced seed calli from the segregant progeny of heterozygous mutant plants (OsLecRK5/oslecrk5); the OsLecRK5-KO and pNP::GUS constructs were transferred into the japonica variety Zhonghua 11 (ZH11) by Agrobacterium-mediated transformation.

Quantitative reverse transcription-PCR analysis
For OsLecRK5 expression analysis, total RNA from rice organs (roots, culms, leaves, and anthers at different stages) was isolated using Trizol reagent (Thermo Fisher Scientific). Total RNA was used to synthesize cDNA from each sample using M-MLV Reverse Transcriptase (Promega) according to the manufacturer's instructions. Quantitative reverse transcription-PCR (qRT-PCR) was conducted using the iQSYBR Green Supermix Detection System (Bio-Rad) with three biological repeats. The rice gene OsActin1 was used as an internal control to normalize target gene expression. Relative expression levels were measured using the 2 (−ΔΔCt) method. Gene-specific primers used for qRT-PCR are listed in Supplementary Table S1.

Subcellular localization and bimolecular fluorescence complementation analysis
To produce the OsLecRK5-GFP construct, the coding region of OsLecRK5 was cloned into the vector pD1-N-GFP (Q. L. Zhu, Y-G. Liu, unpublished data), which carries a P 35S :GFP cassette. The cytomembrane marker RAC3-mCherry was transiently coexpressed in rice leaf sheath protoplasts by polyethylene glycol-mediated transformation (Chen et al., 2010). To prepare the bimolecular fluorescence complementation (BiFC) constructs, OsLecRK5, UGP1, and GSL5 cDNA coding sequences were each cloned into BiFC vectors. Empty vectors and fusion proteins were transiently expressed in Nicotiana benthamiana mesophyll cells, and fluorescence images were obtained using a microscope (Zeiss Axiovert 200).

Pull-down assays
Coding sequences for the OsLecRK5 lectin and kinase domains were cloned into the pMAL-c5X and pET-32a vectors, respectively. Full-length UGP1 was cloned into the pET-32a vector. Proteins were expressed in Escherichia coli BL21 (DE3). TALON® Metal Affinity Resin (TaKaRa Bio) containing 1 μg of maltose binding protein (MBP)-Lectin or MBP was incubated with 1 μg His or His-Lectin in phosphate-buffered saline (PBS). Amylose resin containing 1 μg of MBP-LecRK5-KD or MBP was incubated with 1 μg His or His-UGP1 in PBS. The mixtures were rotated at 4 °C for 8 h. After washing, 20 μl samples were loaded on to a 12% SDS-PAGE gel, and proteins were detected by western blot analysis using anti-MBP or anti-His antibodies (TransGen Biotech) and visualized with Enhanced Chemiluminescence Reagent (Bio-Rad).

Identification and characterization of the rice malesterile mutant oslecrk5
We identified a new male-sterile mutant from our rice mutant library (generated by 60 Co γ-ray irradiation of the japonica cultivar 02428; Ji et al., 2013;Niu et al., 2013;Zhou et al., 2017). We named this mutant oslecrk5 because subsequent analysis showed that its sterility is caused by a mutation in the rice LecRLK gene OsLecRK5 (see below). oslecrk5 mutants had normally developing vegetative tissues and female reproductive organs, but they produced pale yellow anthers and shrunken pollen grains (Fig. 1A-E). Genetic analysis showed that all F 1 plants derived from crossing oslecrk5 and the indica rice variety HHZ displayed a wild-type phenotype. F 2 plants segregated wild-type and mutant plants in a 146:47 ratio, which is equivalent to a 3:1 ratio (χ 2 =0.043, P>0.05), consistent with a single recessive mutation determining the male sterility of oslecrk5.
To identify the gene responsible for the oslecrk5 mutant phenotype, we used 10 193 F 3 individuals derived from the cross between the mutant and HHZ to map the target gene within a 154 kb region on chromosome 2 (Fig. 1F). A bioinformatic analysis predicted 21 candidate genes in this region; we amplified and sequenced the candidates that were expressed in anthers. We found that the 144th nucleotide (C) was deleted from the coding region of LOC_Os02g26160 (http://www. gramene.org/, or Os02g0459600 by http://rapdb.dna.affrc. go.jp/), resulting in a frame shift (Fig. 1F).
We determined the structure of Os02g0459600 by employing Rapid Amplification of cDNA Ends (RACE) to compare the genomic region (http://www.gramene.org) with the full-length cDNA and found that the Os02g0459600 transcript contained a 2088 bp coding sequence. Os02g0459600 is predicted to encode a LecRLK. Based on the number of previously reported rice LecRK genes, we designated Os02g0459600 as OsLecRK5. The predicted OsLecRK5 protein sequence contains 695 amino acids, including an N-terminal lectin domain (amino acids 44-273), one transmembrane region (amino acids 332-354), and a kinase domain (amino acids 359-654; see Supplementary Fig. S1A).
To confirm the function of OsLecRK5 in pollen development, the coding sequences of OsLecRK5 (pNP::OsLecRK5) or OsLecRK5 fused with a FLAG tag (pNP::OsLecRK5-F) were driven by its native promoter and transformed into heterozygous (OsLecRK5/oslecrk5) plants ( Fig. 2A). Both constructs recovered the male-sterile phenotype (Fig. 2B) and the transgene cosegregated with fertility in the T 1 generation: segregants with transgenes were fertile, and those lacking the transgene were sterile (Supplementary Table S2). In contrast, disrupting OsLecRK5 by CRISPR/Cas9 editing in the japonica rice variety ZH11 (Fig. 2C) caused male sterility (Fig. 2D), indicating that the male sterility of oslecrk5 was controlled by a single recessive mutation in Os02g0459600.

OsLecRK5 is mainly expressed in anthers and encodes a plasma membrane-localized protein
To investigate OsLecRK5 expression, we measured its transcript levels in different rice organs using qRT-PCR and found that OsLecRK5 is expressed in leaves, culms, and anthers, but not in roots (Fig. 3A), with the highest expression in anthers. Interestingly, OsLecRK5 expression was highest in stage 8 (meiosis stage) anthers and then gradually declined, becoming nearly undetectable by stage 12 (mature pollen stage). These results imply that OsLecRK5 functions during anther development.
To determine the spatial and temporal patterns of OsLecRK5 expression in planta, we produced plants in which the GUS reporter gene was driven by the OsLecRK5 promoter (pNP::GUS) and visualized the GUS signal in flowers and anthers (Fig. 3B). Consistent with the qRT-PCR results, the GUS signal was limited strictly to anthers at the PMC and meiosis stages (stages 7-8), and was very faint at later stages.
To further verify where OsLecRK5 protein accumulates in planta, we performed immunoblot analysis of the FLAG-tagged protein OsLecRK5-FLAG expressed under the control of the native OsLecRK5 promoter (Fig. 3C). Substantiating our previous results, immunoblot analysis showed that OsLecRK5 accumulated mainly in anthers at stage 7−8 (Fig. 3C). Although qRT-PCR detected OsLecRK5 expression in leaves, our immunoblot analysis barely detected OsLecRK5 in this tissue. Together, these results indicated that although OsLecRK5 is expressed at low levels in some organs, it is highly expressed in anthers during meiosis, implying that it has a crucial role there.
Like other plant LecRLKs, OsLecRK5 was predicted to contain an N-terminal signal and to localize to the PM (Supplementary Fig. S1A; http://www.cbs.dtu.dk/services/SignalP/). To verify its subcellular localization, we introduced CaMV35S-driven OsLecRK5-GFP into rice protoplast cells (Fig. 3D). As expected, the OsLecRK5-GFP signal predominantly colocalized with the PM marker Rac3-mCherry (Chen et al, 2010). These results led us to conclude that OsLecRK5 localizes to the PM.

OsLecRK5 is required for callose deposition during microsporogenesis
To analyze in detail the cellular defects related to the male sterility of oslecrk5, we examined transverse anther sections. Anther development in oslecrk5 was normal prior to PMC meiosis at stage 7 ( Supplementary Fig. S2A). Both wild-type and oslecrk5 anthers consisted of four layers, and PMCs underwent normal meiosis, forming haploid cells by late stage 8 (stage 8b; Supplementary Fig. S2B, C). Between stage 10 and stage 12, the wild-type anther epidermis collapsed and the middle layer, endothecium, and tapetum were mostly degraded. Pollen became vacuolated and underwent normal mitotic divisions to form mature trinucleate pollen grains ( Supplementary Fig.  S2D−F). By contrast, at stage 12, the oslecrk5 anther had swollen epidermis and vacuolated pollen grains ( Supplementary Fig.  S2D−F).
We used scanning electron microscopy imaging to further analyze the cellular defects in oslecrk5. At stage 9, the internal surface of the wild-type tapetum contained abundant Ubisch bodies ( Supplementary Fig. S2G). In oslecrk5, however, the Ubisch bodies were smaller, rounder, and covered by tubular structures (Supplementary Fig. S2G). At stage 12, oslecrk5 pollen grains had abnormally small apertures compared with wild type (Supplementary Fig. S2H).
The cytological defects indicated that OsLecRK5 may affect anther development prior to the uninucleate microspore stage. We therefore performed aniline blue staining to investigate the early events in callose deposition. Our results showed that wildtype PMCs and tetrads were well formed, with a thick callose wall surrounding each cell during microsporogenesis, and callose was not detectable on released microspores in the wildtype anther (Fig. 4). However, a weak callose staining signal was observed during microsporogenesis in the cell wall of oslecrk5 anthers (Fig. 4). These results suggested that callose synthesis and/or deposition were impaired in the oslecrk5 mutant.

OsLecRK5 interacts with UGP1
Given that the PM-localized GSL5 and UGP1 proteins are required for callose synthesis in rice anther development (Kimura et al., 1992;Chen et al., 2007;Shi et al., 2015), we hypothesized that OsLecRK5 may directly interact with GSL5 and/or UGP1. Using BiFC assays in tobacco leaves, we found that OsLecRK5 interacts with UGP1 but not with GSL5 in vivo (Fig. 5A). We further confirmed this interaction with an in vitro pull-down assay using recombinant His-UGP1 and the OsLecRK5 kinase domain tagged with MBP (MBP-LecRK5-KD; Fig. 5B). We found that OsLecRK5 interacted with UGP1, and the OsLecRK5 kinase domain was sufficient to bind UGP1. These results suggested that UGP1 may be a substrate of OsLecRK5 during anther development. Moreover, a BiFC assay showed that OsLecRK5 formed a homodimer and localized on the PM (Fig. 5A). An in vitro pull-down assay confirmed the dimerization of OsLecRK5 (Fig. 5C).

OsLecRK5 activates UGP1 via phosphorylation
Sequence analysis showed that the OsLecRK5 kinase domain contains a conserved ATP-binding lysine residue (K418; Supplementary Fig. S1A). To test the biological importance of K418, we introduced a construct carrying a K to E point mutation at this residue (pNP::OsLecRK5 K418E ) into the oslecrk5 mutant. Unlike the successful complementation test shown in Fig. 2, OsLecRK5 K418E failed to rescue the male sterility of oslecrk5 in any OsLecRK5 K418E transgenic line (Fig. 6A), suggesting that the kinase function of OsLecRK5 requires K418.
To test whether UGP1 is a substrate of OsLecRK5, we conducted an in vitro kinase assay with purified His-UGP1 and MBP-LecRK5-KD. When His-UGP1 and MBP-LecRK5-KD were incubated together with ATP, we detected bands of phosphorylated His-UGP1 and MBP-LecRK5-KD (Fig. 6B), indicating that OsLecRK5 phosphorylates itself and UGP1 in vitro.
To understand the relationship between the phosphorylation status of UGP1 and its UGPase activity in vitro, we examined UGP1 activity by itself and in the presence of either MBP-LecRK5-KD or the kinase-dead MBP-LecRK5-KD K418E . His or MBP-LecRK5-KD alone (negative controls) exhibited no activity, and His-UGP1 alone exhibited weak enzymatic activity (Fig. 6C). However, once MBP-LecRK5-KD was added to His-UGP1, His-UGP1 activity significantly increased. By contrast, adding MBP-LecRK5-KD K418E did not increase His-UGP1 activity. These results clearly indicated that OsLecRK5 phosphorylates UGP1, enhancing its activity and leading to callose biosynthesis.

Discussion
Pollen development is a complicated and finely tuned process in which callose is first deposited on the PMC surface before meiosis, and forms a temporary cell wall during meiosis (Frankel et al., 1969;Stieglitz, 1977). This callose wall separates meiotic cells and protects them from the environment (Heslop-Harrison and Mackenzie, 1967). After meiosis, the callose wall is degraded, releasing the microspores. Several mutants affected in callose wall metabolism exhibit male sterility (Izhar and Frankel, 1971;Warmke and Overman, 1972;Enns et al., 2005;Chen et al., 2007;Toller et al., 2008;Shi et al., 2015). In this work, aniline blue staining demonstrated that initial callose deposition was defective and formation of the callose layer during meiosis was impaired in oslecrk5 anthers (Fig. 4). The defects of tetrads in oslecrk5 were not as clear as those in callose biosynthesis-deficient mutants such as ugp1 and gsl5 (Chen et al., 2007;Shi et al., 2015), which indicates that oslecrk5 may still retain a basic level of callose synthesis for the development of tetrads with minor abnormalities, leading to male sterility ( Supplementary Fig. S3A).
UGPase is a key enzyme in plant carbohydrate metabolism and cell wall biosynthesis (Kleczkowski et al., 2004). A previous study in rice demonstrated that UGP1 knockdown induces abnormal callose deposition during meiosis (Chen et al., 2007), but what regulates UGP1 itself is largely unknown. Our genetic evidence indicated that OsLecRK5 positively regulates callose biosynthesis in rice (Fig. 4). OsLecRK5 interacted with UGP1 in vivo and in vitro (Fig. 5A, B). We further revealed  that OsLecRK5 phosphorylates UGP1 (Fig. 6B), increasing its enzymatic activity (Fig. 6C and Fig. 7). Consistent with our proposed regulatory mechanism, plant UGP1 contains many putative phosphorylation sites (Eimert et al., 1996).
Many LecRLKs are PM-localized receptors (Xin et al., 2009;Huang et al., 2013;Liu et al., 2015a). RLKs dimerize or oligomerize in response to external ligands, activating a kinase cascade that amplifies a signal (Shiu and Bleecker, 2001). We found that OsLecRK5 contains an N-terminal signal peptide and localizes to the PM as a homodimer (Supplementary Fig.  S1A; Fig. 3D and Fig. 5A). OsLecRK5 is a typical Ser/Thr kinase and has a key lysine residue (K418) at a putative ATPbinding site (Supplementary Fig. S1A). Mutating this lysine (K418E) abolished OsLecRK5 function in pollen development (Fig. 6A). An in vitro kinase assay showed that OsLecRK5 can be autophosphorylated (Fig. 6B). These findings imply that the PM-localized LecRLK OsLecRK5 detects a ligand (currently unidentified) and transduces a signal from ligand detection to activate callose biosynthesis during anther development (Fig. 7). Identifying the ligand of OsLecRK5 will be an important goal for future work.
Together, our results reveal that OsLecRK5 phosphorylates UGP1 to activate its activity in callose biosynthesis (Fig. 7). These findings may provide new clues about the regulation of callose biosynthesis related to other stresses and developmental processes.

Supplementary data
Supplementary data are available at JXB online. Fig. S1. OsLecRK5 sequence analysis. Fig. S2. Aberrant anther development in oslecrk5. Fig. S3. Microspore development in wild type and oslecrk5. Table S1. Primers used in this study. Table S2. Genetic analysis of T 1 OsLecRK5 transgenic plants.