• Open Access

Mutations in the F-box gene LARGER PANICLE improve the panicle architecture and enhance the grain yield in rice

Authors

  • Ming Li,

    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Ding Tang,

    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Kejian Wang,

    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Xinru Wu,

    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Lili Lu,

    1. Key Laboratory of Crop Genetics and Physiology of Jiangsu Province/Key Laboratory of Plant Functional Genomics of Ministry of Education, Yangzhou University, Yangzhou, China
    Search for more papers by this author
  • Hengxiu Yu,

    1. Key Laboratory of Crop Genetics and Physiology of Jiangsu Province/Key Laboratory of Plant Functional Genomics of Ministry of Education, Yangzhou University, Yangzhou, China
    Search for more papers by this author
  • Minghong Gu,

    1. Key Laboratory of Crop Genetics and Physiology of Jiangsu Province/Key Laboratory of Plant Functional Genomics of Ministry of Education, Yangzhou University, Yangzhou, China
    Search for more papers by this author
  • Changjie Yan,

    Corresponding author
    1. Key Laboratory of Crop Genetics and Physiology of Jiangsu Province/Key Laboratory of Plant Functional Genomics of Ministry of Education, Yangzhou University, Yangzhou, China
    Search for more papers by this author
  • Zhukuan Cheng

    Corresponding author
    1. State Key Laboratory of Plant Genomics and Center for Plant Gene Research, Institute of Genetics and Developmental Biology, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author

  • Accession numbers: The GenBank accession number for the LP cDNA sequence is HQ018816.

(Tel +86 514 8797 9210; fax +86 514 8799 6817; email cjyan@yzu.edu.cn) or (Tel +86 10 6486 8726; fax + 86 10 6487 3428; email zkcheng@genetics.ac.cn)

Summary

Panicle architecture is one of the most important agronomical traits that directly contribute to grain yield in rice (Oryza sativa L.). We report herein an in-depth characterization of two allelic larger panicle (lp) mutants that show significantly increased panicle size as well as improved plant architecture. Morphological analyses reveal that panicles of two mutants produced more inflorescence branches, especially the primary branches, and contained more grains. Moreover, mutant plants also display more lodging resistance than the wild type. The grain yield per plant in mutants is also increased, suggesting that mutant plants have useful potential for high grain yield in rice breeding. Map-based cloning reveals that LARGER PANICLE (LP) encodes a Kelch repeat-containing F-box protein. RNA in situ hybridization studies display that LP expression was enriched in the branch primordial region. Subcellular localization analyses demonstrate that LP is an endoplasmic reticulum (ER) localized protein, suggesting that LP might be involved in ER-associated protein degradation (ERAD). Using yeast two-hybrid assay and bimolecular fluorescence complementation analysis, we confirm that LP is an F-box protein and could interact with rice SKP1-like protein in an F-box domain-dependent manner. Quantitative real-time PCR results show that OsCKX2, which encodes cytokinin oxidase/dehydrogenase, is down-regulated evidently in mutants, implying that LP might be involved in modulating cytokinin level in plant tissues. These results suggest that LP plays an important role in regulating plant architecture, particularly in regulating panicle architecture, thereby representing promising targets for genetic improvement of grain production plants.

Introduction

Rice (Oryza sativa L.) is one of the most important cereal crops, which provides food for more than half of the world’s population. Given the rapid rise of the world population in this century, food shortage is becoming an even more serious global problem. There is an urgent need for increasing rice yield. Since it is a grain crop, rice panicle architecture, which refers to the number of both primary and secondary rachis branches on a panicle as well as the number of grains setting on these branches, is one of the most important agronomical traits that contribute directly to grain productivity (Sakamoto and Matsuoka, 2008).

A mature rice panicle consists of one rachis (main axis), several primary rachis branches, tens of secondary rachis branches, and more than 100 spikelets (Ikeda et al., 2004). Unlike Arabidopsis, which produces floral meristems directly from the inflorescence meristem (IM) with indeterminate growth habit, the rice inflorescence generates branches and spikelet meristems before producing floral meristems in a determinate pattern (Wang and Li, 2005). In the early stage of rice inflorescence development, after the last foliage leaf emerges, a transition from vegetative to reproductive phase occurs and the shoot apical meristem (SAM) is converted to IM (Ikeda et al., 2004). The IM can be classified into two types of meristems, namely, rachis meristem (RM) and branch meristem (BM), according to their occurrence time and ultimate fates. After producing several BMs (in this case, primary branch primordia), the RM loses its activity and is left as a vestige, called a degenerate point. Subsequently, primary branch primordia begin to elongate and then produce a few secondary branch meristems (primordia) (Ikeda et al., 2004; Itoh et al., 2005). Collectively, in the course of rice inflorescence development, three major factors, namely, timing of IM abortion, conversion of the rachis branch meristem to the terminal spikelet meristem and programme shift of lateral meristem identity along a basal-apical direction on a rachis branch, determine the overall architecture of the rice panicle (Ikeda-Kawakatsu et al., 2009).

In the past decades, several genes that are crucial for meristem identity and ultimately affect rice panicle architecture have been reported. The rice lax panicle mutant is defective in the initiation of inflorescence axillary meristem and has a strongly reduced number of branches and spikelets (Komatsu et al., 2001). LAX encodes a grass-specific bHLH family transcription factor, and its expression is restricted to a special boundary between the SAM and the region where the new meristem is formed (Komatsu et al., 2003a). In rice frizzy panicle (fzp), another classical mutant related to panicle architecture, specification of both terminal and lateral spikelet meristems was blocked and the formation of florets was replaced by sequential rounds of branching (Mackill et al., 1991; Komatsu et al., 2001). FZP encodes an ERF transcription factor and functions to regulate meristem identity at the transition from spikelet meristem to floral meristem (Komatsu et al., 2003b). The rice apo1 mutant shows a precocious conversion of IMs to spikelet meristems and has reduced number of primary branches and spikelets (Ikeda et al., 2005). Molecular studies revealed that APO1 encodes an F-box protein; its overexpression caused an increase in inflorescence branches and spikelets (Ikeda et al., 2007). More recently, gain-of-function alleles of APO1 that result in a higher number of spikelets suggests that the level of APO1 activity regulates the inflorescence form through control of cell proliferation in the meristem (Ikeda-Kawakatsu et al., 2009). Overexpression of RCN1 and RCN2 results in the delay of transition from the vegetative to the reproductive phase and from the branch shoot to the floral meristem, leading to a more branched and denser panicle morphology (Nakagawa et al., 2002). In pap2-1 mutant, the pattern of meristem initiation is disorganized and newly formed meristems show reduced competency to become spikelet meristems, resulting in the transformation of early arising spikelets into rachis branches (Kobayashi et al., 2010).

In addition to genes controlling the establishment of meristem identity, genes regulating meristem activity are also critical for panicle architecture. Gn1a, a quantitative trait locus (QTL), encodes cytokinin oxidase/dehydrogenase (OsCKX2) in rice, regulating the development of IMs by affecting the cytokinin (CK) contents. Reduced expression of OsCKX2 causes CK accumulation in IMs and increases the number of reproductive organs, resulting in enhanced grain yield (Ashikari et al., 2005). The LONELY GUY (LOG) gene, which encodes a novel CK-activating enzyme that works in the final step of bioactive CK synthesis, is required to maintain meristem activity (Kurakawa et al., 2007). The dep1 mutant, derived from a gain-of-function mutation caused by the truncation of phosphatidylethanolamine-binding protein-like domain protein, has enhanced meristematic activity, resulting in a reduced length of the inflorescence internode and an increased number of grains per panicle (Huang et al., 2009).

As mentioned above, many mutants linked to panicle architecture have been characterized and their relevant genes have also been cloned. However, understanding of the molecular basis of rice panicle development remains fragmented. Moreover to date, only a few panicle mutants mentioned above show favourable agronomic traits with high yield potential and can be used or have been used as germplasm resources for improving rice cultivars (Ashikari et al., 2005; Huang et al., 2009). In current rice breeding, shortage of new germplasm resources and/or mutants with important agronomic features is a major barrier to the development of cultivars with higher yield (Wang et al., 2005). In this study, we present two allelic larger panicle (lp-1, lp-2) mutants. Morphological analyses reveal that mutant panicles produce more inflorescence branches, in particular the primary branches, and contain more grains with increased 1000-grain weight. Compared to the wild-type plant, the total grain yield per plant in mutants is also increased, indicating that mutants might be used as an ideal gene resource in rice breeding for higher grain yield. We cloned this gene and found that LARGER PANICLE (LP) encodes a kelch repeat-containing F-box protein. We performed a detailed gene function analysis and found that LP is an ER-localized protein, which is very likely to participate in ER-associated protein degradation, and that LP might be involved in modulating cytokinin homeostasis.

Results

Panicles of the lp-1 and lp-2 show improved architecture, produce more inflorescence branches and contain more grains

To elucidate the molecular mechanisms underlying panicle size and to obtain useful breeding resources for higher grain yield, we identified two rice recessive mutants (lp-1, lp-2) with larger panicle morphology by screening a regenerated plant population derived from the 60Co γ-ray-irradiated japonica cultivar Zhonghua 11. Genetic complement test indicated that the two mutations were located in the same locus. Compared to the wild-type plant, the remarkable characteristic of lp-1 and lp-2 mutants is the larger panicle (Figure 1a,c). First, we examined the branching pattern of lp-1 and lp-2 in more detail. The panicle length in both lp-1 and lp-2 was comparable to that in the wild type (Figure 1d), but the number of primary rachis branches per panicle was significantly higher in lp-1 and lp-2 than in the wild type (Figure 1b,e). On an average, the wild-type panicle consisted of 13.2 primary rachis branches, but the lp-1 and lp-2 panicles consisted of 23.7 and 18.2 primary branches, respectively. The number of secondary branches borne on the primary branches was also determined (Figure 1f). In general, lp-1 and lp-2 panicles contained 60 and 47.3 secondary branches, respectively, while the wild-type panicle consisted of only 39.6 secondary branches. Comparatively, one primary branch in a wild-type panicle produced approximately three secondary branches, while one primary branch in the lp-1 and lp-2 panicles produced 2.5 and 2.6 secondary branches on an average, respectively. This finding indicated that the increase in the total number of secondary branches per panicle should be attributed to the increase in the number of primary branches per panicle. Thus, mutations that happened in lp-1 and lp-2 mutant plants mainly affected the primary branch identity. As a result of increased number of primary and secondary branches, panicles of the two mutants showed larger and denser panicle architecture (Figure 1a–c).

Figure 1.

 Panicle morphological comparison between the wild type and mutants. (a) The panicles phenotype of wild type (WT) (left), lp-1 (middle) and lp-2 (right) in grain filling stage. (b) De-seeded panicles of wild type (left), lp-1 (middle) and lp-2 (right) in grain-filling stage showing the panicle branching. (c) The panicles phenotype of wild type (left), lp-1 (middle) and lp-2 (right) in mature stage. (d–i) Comparison of panicle traits among the WT, lp-1 and lp-2. Values in (d–i) are means ± SE. = 36 panicles in (d–h), and six replicates in (i). Student’s t-test was performed between the WT and the lp-1, the WT and the lp-2, respectively. Single and double asterisks represent significance difference determined by the Student’s t-test at < 0.05 and < 0.01, respectively. Scale bars (a–c) 1 cm.

Next, we determined the total grain number and filled grain number per panicle. Statistical analysis revealed that the total grain number per panicle in lp-1 and lp-2 was significantly increased (Figure 1g). In lp-1 and lp-2 plants, the total grain number per panicle was 348 and 274, respectively, whereas it was only 197 in the wild-type plant. Usually, an increase in the number of grains is always associated with a decrease in the grain filling rate, but no evidence was found for grain filling failure in these two mutants. We found that the grain filling rate in lp-2 was increased by 10% (from 75% to 85%), while the filling rate was slightly decreased from 75% to 73% in lp-1. The filled grain number per panicle was also determined (Figure 1h), and it was found to be higher in the mutants than in the wild type, i.e. 254 filled grains in lp-1, 233 filled grains in lp-2, and 148 filled grains in the wild type. The 1000-grain weight of lp-1 plants was significantly higher than that in the wild-type plants, while that in lp-2 plants was only insignificantly higher (Figure 1i).

The plant architecture is improved and the grain yield is increased in mutants

In addition to the pronounced changes in the panicle features, there were other modest morphological alterations of mutant plants in the vegetative phase. Compared with wild-type plants, both lp-1 and lp-2 plants showed slightly higher statures (Figure 2a,i), erect and upwardly semi-curled (V-type) leaves (Figure 2a,b) and stronger culms (Figure 2c). In the late stage of rice growth, we found that mutant plants showed more lodging resistance than the wild type. Anatomical features showed that mutant plants had more thickened culms (Figure 2d,e,k), more large vascular bundles (Figure 2l) and small vascular bundles (Figure 2m), more layers of sclerenchyma cells (Figure 2f,g), and enlarged large vascular bundles (Figure 2f,h) in mature culms in comparison with wild-type plants. Consistently, the culms mechanical strength of the mutant plants was significantly increased in comparison with wild type (Figure 2o,p). We also found that the number of tillers in the two allelic mutants decreased to different extents (Figure 2j). Generally, the increased grain number per panicle and increased 1000-grain weight are favourable traits for the grain yield, while decreased panicle number (tillers number) per plant would cause reduced productivity. Considering the discrepancy between them in the mutants, we finally tested the effect of the target gene mutations on grain yield by a field plot experiment. Statistical analysis demonstrated that the overall grain yield per plant under field conditions in lp-1 was increased by 11% (from 20.71 g to 23.02 g) and the difference was highly significant (Figure 2n). In lp-2 plants, the grain yield per plant was slightly increased but the difference was not significant (Figure 2n). Thus, the above results suggest that the lp-1 mutant may be used as good breeding materials for higher grain yield.

Figure 2.

 Plant type comparison between the wild-type and mutants. (a–c) Gross morphologies comparison of wild-type (WT) (left), lp-1 (middle) and lp-2 (right) plants at maturity. (a) The whole plant morphology; (b) Leaf blades; (c) Culms. (d–h) Cross-sections of culms of wild-type and lp-1 plants. The cross plane of culm in lp-1 (e) was thicker than that in the wild type (d). (f) Close-up views of the boxed region in (d); (g, h) Close-up views of the boxed regions in (e). Arrow indicates the sclerenchyma cells. SV, small vascular bundles; LV, large vascular bundles. (i–n) Comparison of plant architecture and grain yield between the wild type and mutants. (o–p) Comparison of plant lodging resistance properties. Maximum bending force was measured on the central region (4 cm in length) of the first internodes. Values in (i–p) are means ± SE. = 36 in (i–n) and 20 in (o–p). Student’s t-test was performed between the wild type and lp-1, the wild type and lp-2, respectively. The single and double asterisks represent significance difference at < 0.05 and < 0.01, respectively. Scale bars (a) 10 cm, (b, c) 1 cm and (d–h) 100 μm.

Positional cloning and characterization of LP

To investigate the molecular basis of lp, we performed positional cloning to isolate this gene. From the F2 population generated by crossing lp-1 mutants with the indica variety Dular, 930 homozygous mutant plants were obtained and used for genetic analysis. The LP locus was primarily mapped to the short arm of chromosome 2 between the sequence-tagged site (STS) markers f57 and f35 (Figure 3a). Further mapping using adjacent molecular markers (Table S1) indicated that the LP locus is present in BAC clone AP006160, and the candidate gene region was finally narrowed down to a 38 kb region (Figure 3b). On the basis of the annotations of the rice genome database (http://ricegaas.dna.affrc.go.jp/), this region comprises five putative genes. We identified one of these genes, LOC_Os02g15950, as the most likely candidate. The LOC_Os02g15950 gene comprises two exons and one intron, and encodes a kelch repeat-containing F-box protein (Figure 3c; Figure S1). Sequencing analysis revealed a one nucleotide (T to C) substitution in the second exon of LOC_Os02g15950 in lp-1 mutant plants and a two nucleotide deletion in the second exon of LOC_Os02g15950 in lp-2 mutant plants (Figure 3c,d). The nucleotide substitution in the lp-1 allele resulted in a change from serine to proline at the deduced 472th amino acid residue (Figure 3d; Figure S1). The two nucleotide deletion, which was present just before the stop codon (TAG), caused a translational frameshift and theoretically resulted in a protein with 21 more amino acids in the lp-2 allele than in the wild-type protein (Figure 3d; Figure S1). The mutant phenotype of lp-1 was fully rescued when a 7.9-kb genomic DNA fragment containing the entire LOC_Os02g15950 gene region was introduced (Figure 3e–g). Statistical analysis demonstrated that the rescued line (pCLP) has the comparable values with the wild type on the branches number, grains number and plant height (Figure 3h–l). Thus, we conclude that LP is the LOC_Os02g15950 gene.

Figure 3.

 Map-based cloning of LARGER PANICLE (LP) and the complementation test. (a) LP was mapped primarily to the short arm of chromosome 2 between sequence-tagged site (STS) markers f35 and f57. (b) Fine-mapping of LP. The numerals indicate the number of recombinants. (c) The gene structure and the mutation sites happened in lp-1 and lp-2. (d) Detailed mutation pattern happened in lp-1and lp-2. (e) Constructs for complementary analyses. The Complementation plasmid pCLP contains the entire LP (7930-bp) and the pCLP-CK plasmid used as a control. The 7930-bp genomic DNA fragment in pCLP contains the entire LP coding region (from 3094th to 5520th base-pair), 3093-bp upstream sequence (promoter region, from 1st to 3093th), and 2407-bp downstream sequence (3′ UTR region, from 3095th to 5520th). The dotted line in pCLP-CK indicates the deleted genomic region (from 1225th to 5545th, total 4321 bp in length). (f–g) Plant phenotypic comparison (f) and panicle phenotypic comparison (g) of the transgenic lines harbouring pCLP and pCLP-CK. (h–l) Statistical comparison among the wild type, the rescued lines and rescue control lines, values correspond to means ± SE (= 30). (m) Immunoblot analysis showing that LP encodes a 57 kD protein. Scale bars (f) 10 cm and (g) 1 cm.

KOME cDNA database search (http://cdna01.dna.affrc.go.jp/cDNA/) revealed that three cDNA clones (AK287808, AK064350, and AK073921) with the same length of open reading frame (ORF) (1227-bp) exist for the LP gene locus. Using rapid amplification of cDNA ends (RACE) PCR, we redefined the cDNA sequence and coding sequence (CDS) of LP. Our result showed that the complete cDNA sequence, which differed remarkably from the predicted sequence, is 3202-bp in length with the longest ORF of 1548-bp. To further validate this result, we checked the molecular size of the targeted protein in the wild-type and mutant plants by Western blotting analysis using polyclonal antibody against the LP C-terminal 319-aa polypeptide. In theory, the shorter ORF with a length of 1227-bp would be translated into a protein with 408 amino acids and a molecular weight of 46.3 kDa, whereas the longer ORF with 1548-bp would produce a protein with a molecular weight of 57.4 kDa. Western blotting analysis revealed that a protein approximately 57 kDa in weight was recognized specifically in both tested plants by the antibody against the C-terminal polypeptide of LP (Figure 3m), consistent with the deduced weight (57.4 kDa) of the protein encoded by the longer ORF. Taken together, we concluded that the CDS of LP is 1548-bp in length, encoding a protein with 515 amino acid residues.

Expression patterns of LP

To define the expression pattern of LP, we surveyed its expression in various tissues of rice using quantitative real-time PCR (qRT-PCR). Results revealed that the LP expression level is high in culms, medium in leaf sheaths, SAM and panicles and low in leaves and roots (Figure 4a). With an emphasis on the panicle, analysis of the LP expression at different stages of panicle development was also performed by qRT-PCR. The results showed that LP was more richly transcribed in a young panicle, but its expression decreased gradually with the increasing stages of panicle development (Figure 4b). The transcription of LP in young panicles at early differentiation stages was further examined by mRNA in situ hybridization. Results showed that LP expression was enriched in the primary branch primordia (Figure 4c) and secondary branch primordia (Figure 4d), indicating that LP plays an important role in branch primordial differentiation.

Figure 4.

 Expression patterns analysis of LARGER PANICLE (LP) at the mRNA and protein level. (a) LP expression levels revealed by real-time PCR in various organs. (b) LP expression levels at various stages of panicle development. (c–e) LP expression in young panicle revealed by mRNA in situ hybridization. The longitudinal sections of panicle at middle stage of primary branch primordial differentiation (c) and at early stage of secondary branch primordial differentiation (d) show that the expression of LP was mainly restricted to the rachis-branch primordial zone (asterisks and arrows). Asterisks, the primary rachis-branch primordia; Arrows, the secondary rachis-branch primordia. (e) Background control of in situ hybridization using a sense probe. (f–p) LP-GUS fusion protein expression in various tissues revealed by GUS-staining assay. Expression of LP-GUS in seedling (f), root (g), culm (h), mature leaf (i), leaf sheath (j), panicle (k), spikelet (l), anthers (m), and young panicle at early stage of secondary rachis-branch primordial differentiation (n, o) is analyzed. (p) Hand-cut cross-section of a mature culm showing the expression of LP-GUS mainly localized in vascular bundles. SV, small vascular bundles; LV, large vascular bundles. Values in (a, b) are means ± SE of three independent experiments. Data are displayed as the ratio of expression to rice actin1 RNA. Scale bars (c, d, e, g, o, p) 100 μm, (f, i, l, m, n) 1 mm, (h, j) 2 mm and (k) 5 mm.

To assess the gene expression pattern comprehensively, we further generated transgenic plants expressing LP-GUS driven by its native promoter in the lp-1 background. The wild-type appearance of transgenic plants indicated that the LP-GUS fusion protein functioned as a native protein. β-glucuronidase (GUS) activity was examined histochemically in transgenic plants and results showed that LP-GUS was expressed in coleoptile and embryonic root of germinating seedlings (Figure 4f), vascular tissues of roots (Figure 4g) and culms (Figure 4h,p), leaves (Figure 4i), leaf sheaths (Figure 4j) and panicles (Figure 4k). We took a more detailed observation of the LP-GUS expression in the spikelet and found GUS staining in lemma and palea (Figure 4l), and anthers (Figure 4m). In addition, LP was highly expressed in rachis-branch primordia of a differentiating inflorescence (Figure 4n,o), consistent with the results of mRNA in situ hybridization analysis (Figure 4c,d). Taken together, all the lines of evidence suggest that LP is ubiquitously expressed in plant but with enhanced expression in some tissues, such as young panicles and culms, explaining the observed phenotype of improved panicle and plant architecture in mutant plants.

To check whether the LP expression is regulated by cytokinin or auxin, wild-type seedlings (10 days after germination) were treated with 6-benzylaminopurine (6-BA), naphthalene acetic acid (NAA), or indole-3-acetic acid (IAA), and the LP transcripts were quantified by qRT-PCR. The analysis revealed that no significant changes happened in the LP expression when treated with cytokinin (6-BA) and auxin (NAA and IAA) (Figure S2).

LP is an ER-localized protein

To determine the subcellular localization of the LP, we performed a transient expression experiment of LP in rice leaf protoplasts and onion epidermal cells, respectively. The C-terminus of LP was fused with green fluorescent protein (GFP) under the control of CaMV 35S promoter, and the construct was transferred into rice leaf protoplasts by the polyethylene glycol-mediated method. While expressing GFP alone resulted in ubiquitous signals in the protoplasts (Figure 5a), LP-GFP fused protein was limited to the fibril network within the cytoplasm (Figure 5c), suggesting possible localization in the endoplasmic reticulum (ER). We verified this result by co transformation of LP-GFP with the ER marker protein mCherry-HDEL (Nelson et al., 2007) in rice protoplast cells. The signals of the two fluorescent proteins almost overlapped (Figure 5e–g), suggesting that LP is localized to the ER. The same results were also found when LP-GFP fusion protein was expressed in onion epidermal cells (Figure S3a–h), confirmed that LP is an ER-localized protein.

Figure 5.

 Subcellular localization of LP. (a) Rice protoplast cell expressing GFP alone. (b) Visible light of (a). (c) Protoplast cell expressing LP-GFP. (d) Visible light of (c). (e–g) Rice protoplast cell co-expressing LP-GFP (e) and ER marker mCherry-HDEL (f), with a merged image (g). (h) Visible light of (e). Scale bars 10 μm.

LP is an F-box protein and can interact with rice Skp1-like proteins

Recently, the SCF (Skp1/Cul1/F-box protein/Roc1) protein ubiquitin ligases have been found to be involved in a major pathway for ubiquitylation of proteins for degradation by the 26S proteasome (Kipreos and Pagano, 2000; Petroski and Deshaies, 2005). In the SCF complex, the F-box protein binds directly to SKP1 and functions as a receptor to recruit specific phosphorylated substrates. Previous studies have predicted that the LOC_Os02g15950 gene locus encodes a putative F-box protein (Jain et al., 2007; Piao et al., 2009), as it has features that are characteristic of F-box proteins, such as an F-box domain (149th–194th amino acid residues) at the N-terminal region and a protein-protein interaction domain (kelch motif, 395–441 amino acid residues) at the C terminus (Figure S1). To confirm whether LP functions truly as an F-box protein and can interact with rice SKP1 proteins, we performed a yeast two-hybrid (Y2H) assay. Rice contains at least 32 SKP1 genes (OSKs) (Kong et al., 2007). Among them, five genes (OSK1, 4, 15, 16, and 20), which belong to different gene clades in phylogenetic analysis (Kong et al., 2007), were amplified by PCR and fused to the DNA activation domain of GAL4 in the pGADGH vector; similarly, the LP-coding sequence (1545-bp) was fused to the DNA-binding domain of GAL4 in the pGBKT7 vector. When co transformed into the yeast strain HF7c, all the five tested OSK proteins could interact with LP, as shown by selection for growth on SD/–Leu/–Trp/–His medium and by β-galactosidase activity assays (Figure 6a). The F-box domain was then removed from LP, and interaction assays between LP (ΔF-box) and OSK1, 4, 15, 16 and 20 were also performed; no yeast clones could be observed on the selection medium (Figure 6b). These results suggested that LP is indeed an F-box protein, and that the interactions between LP and OSKs are dependent on the presence of the F-box domain.

Figure 6.

 Yeast two-hybrid assay and BiFC assay between the LP and OSK proteins. (a) Y2H assay of the interaction between LP and OSKs. An empty vector (AD) used as a negative control. (b) Y2H assay of the interaction between LP (ΔF-box) (in which the F-box domain was deleted) and OSKs. (c) BiFC assay between the F-box domain of LP and OSK15. Top panel showing images of rice protoplasts co-transformed with construct expressing LPF-YFPN and OSK15-YFPC. Bottom panel showing images of rice protoplasts co-transformed with construct expressing LPΔF-YFPN and OSK15-YFPC. Scale bars 10 μm.

To further verify whether the F-box domain is indispensable for the interaction between LP and OSKs in living plant cells, bimolecular fluorescence complementation (BiFC) analysis (Walter et al., 2004) was performed using a transient expression system in rice protoplasts. Using the pUC-SPYNE vector, we constructed two kinds of constructs: P35S:LPF-YFPN, which contained the F-box domain (amino acids 149–194) of LP fused with the YFP N-terminal sequence (amino acids 1–155), and P35S:LPΔF-YFPN, which contained the LP other coding region (amino acids 195–515) fused with the YFP N-terminal sequence (amino acids 1–155). Considering that the binding affinity of OSKs with LP is approximately equivalent in yeast, we chose OSK15 as the representative and constructed the P35S:OSK15-YFPC construct, in which the C-terminal sequence (amino acids 156–239) of YFP was fused to the C-terminus of the OSK15. While cells co-transfected with LPΔF-YFPN and OSK15-YFPC produced no or only background fluorescence, a strong fluorescence signal was observed when LPF-YFPN was co expressed with OSK15-YFPC (Figure 6c). These results further validated our previous conclusion that LP can interact with rice SKP1-like proteins in an F-box domain-dependent manner.

Expression of OsCKX2 is down-regulated in young panicles of lp mutants

To date, many genes, such as OsCKX2, LOG, DEP1, RCN1 and LAX, have been found to play essential roles in controlling the panicle architecture (Nakagawa et al., 2002; Komatsu et al., 2003a; Ashikari et al., 2005; Kurakawa et al., 2007; Huang et al., 2009). Using qRT-PCR method, we determined the expression level of these genes in the mutant plants. The expression of OsMADS1, a key gene function in promoting flower meristem determinacy (Jeon et al., 2000; Wang et al., 2010), as well as the LP per se, was also checked in two mutants. Results showed that the transcription level of OsCKX2 at the early stage of panicle development (panicle length, 2–3 cm) was severely down-regulated (Figure 7a). However, no obvious change was detected in the expression of the other six genes in young panicle of two mutants (Figure 7a; Figure S4a–e). This finding suggested that LP is perhaps involved in the maintenance or modulation of dynamic equilibrium of cytokinin through direct or indirect control of OsCKX2 expression.

Figure 7.

 Transcriptional levels of OsCKX2 and LOG in two mutants. Relative expression of OsCKX2 (a) and LOG (b) revealed by qPCR in lp-1, lp-2 and wild-type plants. Data are displayed as the ratio of expression to rice actin1 RNA, data given as means ± SE. All assays were repeated at least three times.

Discussion

LP may act as a positive regulator determining the time of RM abortion in the inflorescence development

At the early stage of rice inflorescence development, transition to the reproductive phase evokes transformation of the SAM to the IM (in this case, RM). Once the RM forms, it starts to generate lateral meristems and further develop into primary rachis branches. After generating a certain amount of lateral meristems, the RM loses its activity and is aborted (Ikeda et al., 2004). Thus, the RM abortion time directly determines the number of primary branches in a panicle (Kobayashi et al., 2010). For a given variety, the overall number of primary branches per panicle is relatively constant, indicating that some genes may exist that control the RM abortion time. Previous studies have shown that APO1, another F-box protein, serves as the negative factor for RM abortion. The RM abortion occurred precociously in an apo1 mutant, leading to a decreased number of primary branches, and RM abortion was delayed on overexpression of APO1 by using apo1-D dominant alleles, leading to the generation of more primary branches (Ikeda et al., 2005, 2007; Ikeda-Kawakatsu et al., 2009). In our study, we found a significantly higher number of primary branches in both lp-1 and lp-2 mutants, indicating that the RM abortion time is delayed in the two mutants. On comparison with APO1, we infer that LP, which is also an F-box protein, may act as a positive regulator determining the time of RM abortion. The secondary branches per panicle in lp-1 and lp-2 mutants were also increased substantially. However, the ratio of the number of secondary branches to that of the primary branches in these mutants is similar to that in the wild type. If the transition from branch meristem to spikelet meristem was delayed, more secondary branches would have been found per primary branch. Therefore, the increase in the number of secondary branches per panicle can be accounted for by the increase in the number of primary branches per panicle. Taking the morphology of the normal branches and spikelet fertility into account, we proposed that the mutations in lp-1 and lp-2 may mainly affect the time of RM abortion.

LP, as an ER-localized F-box protein, might be involved in the ERAD pathway

Many studies have shown that the ER-associated protein degradation (ERAD) pathway plays an important role in directing ubiquitin-mediated degradation of various ER-associated misfolded and normal proteins (Werner et al., 1996; Hampton, 2002). Proteins that fail to undergo correct folding are dislocated from the ER into the cytosol or delivered to vacuoles and then degraded (Vitale and Ceriotti, 2004). In this process, F-box proteins as well as other components of the SCF complex are indispensable for target protein ubiquitination, in which the F-box proteins are responsible for conferring specificity to the complex for the appropriate targets (Margottin et al., 1998; Meusser et al., 2005). Like animal cells, plant cells also have ERAD systems that digest unfolded and aggregated ER proteins (Di Cola et al., 2001; Vitale and Ceriotti, 2004). The subcellular localization result presented herein shows that LP is mainly localized in the ER (Figure 5e–g; Figure S3e–g), which raised the possibility that LP might be involved in the ERAD pathway. Similar results have been reported in human cells (Margottin et al., 1998). h-βTrCP is a WD domain-containing F-box protein. In human HIV-infected cells, h-βTrCP is recruited to ER membranes through a membrane phosphoprotein, Vpu, which combines with the target protein, CD4, for degradation, and then h-βTrCP interacts with cytoplasmic Skp1p and leads to CD4 degradation via the 26S proteasome (Margottin et al., 1998). According to the fluorescence signal, we can only assess that LP is an ER-localized protein, but we cannot determine whether LP is connected to the ER membranes by itself or by another unknown adapter protein just as Vpu in human cells. In rice, F-box proteins form a large gene family comprising 687 potential F-box proteins, and they can be classified into ten subfamilies on the basis of the type of domain(s) present at their C terminus (Jain et al., 2007). The presence of a strikingly large number of F-box proteins in rice implies that functional differentiation and compartmental distribution of F-box proteins are possible. It seems reasonable to assume that special F-box proteins exist that are associated with the ER and their function involves ER protein quality control. To our knowledge, no plant F-box protein that functions in the ERAD pathway has been reported in rice to date. In view of the fact that LP is mainly localized in the ER, we propose that LP might be involved in the ERAD pathway.

Is LP involved in cytokinin homeostasis?

Recent studies have revealed that the functions of many F-box proteins are associated with plant hormones. TIR1, an F-box protein in Arabidopsis, acts as an auxin receptor regulating the stability of Aux/IAA proteins (Gray et al., 2001; Dharmasiri et al., 2005; Kepinski and Leyser, 2005). The role of F-box proteins, such as SLY1 and SNE in gibberellic acid signalling(Dill et al., 2004; Strader et al., 2004) and EBF1/EBF2, ETP1/ETP2 in ethylene signalling (Gagne et al., 2004; Qiao et al., 2009) has also been established. However, thus far, no F-box gene has been reported that is involved in CK metabolic balance or signalling. CK has been implicated in maintaining the apical and axillary meristems in the inflorescence as well as the SAM (Kurakawa et al., 2007; Kyozuka, 2007; Barazesh and McSteen, 2008). Like the other plant hormones, CK activity in plant tissues is thought to be controlled by a balance of synthesis and catabolism (Mok and Mok, 2001; Sakakibara, 2006; Kurakawa et al., 2007). Previous studies have showed that OsCKX2, a cytokinin oxidase/dehydrogenase, plays a critical role in regulating the development of IMs. Reduced expression of OsCKX2 causes CK accumulation in IMs and increases the number of reproductive organs, resulting in enhanced grain yield (Ashikari et al., 2005). Our results show that at the early stage of inflorescence development in the lp-1 mutant, the expression of OsCKX2 gene is significantly decreased; suggesting that OsCKX2 protein accumulation in the young panicle is also reduced. It is reasonable to assume that CK accumulation in mutant inflorescence may increase, resulting in a large panicle and higher grain number, which resembles the phenotype of gn1a mutants. Thus, we hold the opinion that LP might be involved in moderating the CK level in the plant through direct or indirect control of the OsCKX2 expression.

Mutants can be used as ideal genetic resources for higher yield breeding in rice

Improvement of grain yield is the persistent and ultimate goal in rice breeding. Panicle architecture is one of the target characteristics in rice breeding because of its strong association with grain yield. In current rice breeding, the shortage of new germplasm resources with important agronomic features is one of the major barriers to the development of cultivars with higher yield (Wang et al., 2005). Therefore, it is quite important for rice breeding to find new mutants with improved panicle architecture and further isolate genes related to these features. In the present study, we report an in-depth characterization of two allelic larger panicle mutants that show significantly increased panicle size as well as improved plant architecture, suggesting that mutants have useful potential for higher grain yield in rice breeding. More recently, a gene responsible for erect panicle, EP3, was mapped to the same locus (Piao et al., 2009). However, the panicles of ep3 (Hep) were erect and small with less grains, which was cannot be used directly for higher yield in rice breeding. Comparison between mutations in the ep3 and lp (lp-1, lp-2) mutants revealed different mutation patterns. While a single base pair change in ep3 resulted in a truncated protein lacking the Kelch motif, mutations in lp-1 and lp-2 only led to a single amino acid substitution and additional 21 amino acid residues in the C-terminus of LP, respectively, with no change in the Kelch motif. The Kelch motif is well-known as a protein-protein interaction motif, which is responsible for recognizing the specific substrate for degradation. We think that the different mutation sites that happened in the same gene may be the reason for the antagonistic phenotypes of the lp-1, lp-2, and ep3 mutants.

In rice production all over the world, compared to the traditional transplanting method, direct seeding is becoming more popular among rice farmers as it is a labour- and water-saving method for rice planting (Thakur et al., 2004). Varieties suitable to direct seeding are required to have traits such as large panicles, moderate or inferior tillering capacity, and strong resistance to lodging (Dingkuhn et al., 1991; Yamauchi et al., 1993). On the basis of the plant architectural features observed in lp-1 and lp-2 mutants, we hold the opinion that the direct-seeding method is suitable for planting lp-1 and lp-2 mutants. Besides the improved panicles’ architecture, both lp-1 and lp-2 mutants also show improved plant architecture such as erect and upwardly V-type leaves, slightly higher stature and stronger culms with more number of vascular bundles. These traits are favourable for water transport capacity and the mechanical strength of the stem, both of which are important factors for the breeding of high-yielding, lodging-resistant varieties. Consequently, it is reasonable to presume that the two lp mutants, especially lp-1, are useful germplasm resources in molecular breeding for high grain yield.

Experimental procedures

Plant materials and growth conditions

The rice (Oryza sativa L.) larger panicle mutants were identified from the 60Co γ-ray-irradiated japonica cultivar Zhonghua 11. In summer, rice plants were grown in Beijing in a standard paddy field by transplanting one plant per hill at a distance of 15 × 15 cm under conventional cultivation conditions. In winter, plants were grown in Sanya (Hainan Province, China). For evaluation of the yield potential, all data were collected from plants grown in summer in Beijing. The experiments were repeated in 2008 and 2009.

Measurement of lodging resistance-related properties

The maximum bending force was measured using a digital force/length tester according the method described elsewhere (Sunohara et al., 2006). The central region (4 cm) of the first internodes was used for the measurement. We also measured the cLr-value using the main culm in mature plants within 30–40 days after heading according the method described elsewhere (Grafius and Brown, 1954; Sunohara et al., 2006).

Positional cloning

The F2 mapping population was generated from a cross between the lp-1 mutant and the indica rice variety Dular. For fine mapping, STS markers (Table S1) were developed. The corresponding genomic DNA fragments in the narrowed region were amplified from mutants and wild-type plants and sequenced using an Applied Biosystems 3730 sequencer.

Complementation test

A 7.93-kb genomic DNA fragment containing the entire LP coding region, 3093-bp upstream sequence, and 2407-bp downstream sequence was inserted into the pCAMBIA1300 vector (Cambia, http://www.cambia.org/) to generate the construct pCLP. A control construct, pCLP-CK, was generated by digesting pCLP with EcoRI to remove the partial upstream sequences (1873-bp) and the CDS of LP. Constructs were transformed into lp-1 plants by the Agrobacterium-mediated transformation procedure. The T1 transgenic plants and T2 progeny were used for morphological analyses.

Gene expression

Total RNAs were extracted from various tissues of wild-type rice plants using an RNeasy kit (Qiagen, http://www.qiagen.com/). DNase I-treated RNA was used to synthesize cDNA using a reverse transcription kit (Promega, http://www.promega.com/). The amplification of the target genes was analyzed using the CFX96 Real-Time PCR Detection System and software (Bio-Rad Laboratories, Inc., Benicia, California, USA). Primers used in this study are listed in Table S2.

To generate LPpro:LP-GUS transgenic plants, the promoter region (3093-bp) and the entire ORF (no stop codon contained) of LP were PCR amplified and inserted into the vector pCAMBIA 1301 between the BamHI and NcoI sites in-frame with the GUS reporter gene. The construct was transformed into lp-1 plants by the Agrobacterium-mediated transformation procedure. Transgenic plants with wild-type appearance were used for the analysis of GUS activity (Scarpella et al., 2003).

For expression analyses of OsCKX2, LOG, DEP1, RCN1 and LAX in the lp-1 mutant, total RNAs were extracted from lp-1 and wild-type young panicles (2–3 cm in length) by using the same method mentioned above. Quantitative real-time PCR was performed using their corresponding primers (see Table S2).

Exogenous 6-BA, NAA and IAA Treatment

Seeds were sowed and germinated on MS agar medium. After 3 days, the seedlings were transferred to MS liquid media and grown for another 7 days. Then, seedlings were treated with 10−5 m 6-BA or 10−6 m NAA or 10−6 m IAA (Zhao et al., 2009). Total RNA was extracted after 0, 1, 2, 3, 6, 9, 12, 24, 36 and 48 h of treatment and analyzed by qRT-PCR.

5′ RACE and 3′ RACE

Total RNAs were extracted from the young panicle and treated as mentioned above (Gene expression). The 5′ and 3′ RACE of LP was performed according to the protocol of the kit (5′-Full RACE Core Set and 3′-Full RACE Core Set, Takara, http://www.takara-bio.com/). The primers used are listed in Table S3.

The mRNA in situ hybridization

A 410-bp DNA fragment containing the partial 3′ coding region (172-bp) and partial 3′ UTR (238-bp) of LP was amplified by PCR and cloned into the pGEM-T vector. The construct was linearized for use as a template for the synthesis of digoxigenin-labelled sense and anti-sense RNA probes. Tissue fixation and in situ hybridization procedures were performed as described elsewhere (Coen et al., 1990).

Subcellular localization

GFP was fused to the C-terminus of LP in-frame and inserted between the CaMV 35S promoter and the nopaline synthase terminator in PUC19 to generate constructs for transformation of the rice protoplasts and onion epidermal cells. Co transfection of rice protoplasts with GFP-tagged LP and mCherry-HDEL (ER marker) was performed as described elsewhere (Li et al., 2009). For onion cells transformation, those plasmids were bombarded into onion epidermal cells using a PDS-1000/He particle gun (Bio-Rad). The treated samples were observed by confocal laser-scanning microscopy (Leica TCS SP5, Germany).

Yeast two-hybrid assay

The full-length CDS and partial sequence of LP (LPΔF) as well as OSKs were amplified using the primers listed in Table S4. LP and LPΔF were fused to the GAL4 DNA-binding domain in pGBKT7, respectively, while OSKs were fused to GAL4 activation domain in the pGADT7 vector. Constructs were co transformed into yeast strain HF7c.

BiFC assay

Relative DNA fragments (encoding LPF and LPΔF) were PCR amplified and cloned into pUC-SPYNE vector. Similarly, the full-length CDS of OSK15 was inserted into pUC-SPYCE vector. Cotransformation of rice protoplasts and confocal imaging were performed using methods described above. The primers used are listed in Table S4.

Acknowledgements

We thank Dr. Jörg Kudla (Universität Münster) for providing pUC-SPYNE and pUC-SPYCE plasmids used for BIFC. This work was supported by grants from the Ministry of Sciences and Technology of China (grant no. 2009CB118500 and 2008ZX08009) and the National Natural Science Foundation of China (grant no. 30900885).

Ancillary