OsFIE2 plays an essential role in the regulation of rice vegetative and reproductive development



  • Polycomb group (PcG) proteins are gene repressors that help to maintain cellular identity during development via chromatin remodeling. Fertilization-independent endosperm (FIE), a member of the PcG complex, operates extensively in plant development, but its role in rice has not been fully investigated to date.
  • We report the isolation and characterization of a PcG member in rice, which was designated OsFIE2 for Oryza sativa Fertilization-Independent Endosperm 2. OsFIE2 is a single-copy gene in the rice genome and shows a universal expression pattern.
  • The OsFIE2 RNAi lines displayed pleiotropic phenotypes in vegetative and reproductive organ generation. In unfertilized lines, endosperm formation could be triggered without embryo formation, which indicates that FIE is indeed involved in the suppression of autonomous endosperm development in rice. Furthermore, lateral root generation was promoted early in the roots of OsFIE2 RNAi lines, whereas the primary root was premature and highly differentiated. As the root tip stem cell differentiated, QHB, the gene required for stem cell maintenance in the quiescent center, was down-regulated.
  • Our data suggest that the OsFIE2–PcG complex is vital for rice reproduction and endosperm formation. Its role in stem cell maintenance suggests that the gene is functionally conserved in plants as well as animals.


Polycomb group (PcG) proteins play essential roles in animal and plant life cycles by controlling the expression of important developmental regulators, as well as by regulating cell proliferation (Bracken et al., 2003; Martinez et al., 2006; Köhler & Villar, 2008). PcG proteins were originally discovered in Drosophila, and their essential role in regulating homeotic genes during body formation has been investigated extensively. PcG protein functions in an expanded spectrum, including cell cycle control, cancer, senescence, X-inactivation, cell fate decisions and stem cell maintenance (Morey & Helin, 2010). These structurally heterogeneous proteins form multimeric complexes that exert their functions by modifying chromatin structure and regulating post-translational histone modifications (Lund & van Lohuizen, 2004; Ringrose & Paro, 2004; Bantignies & Cavalli, 2006). There are at least two major PcG complexes in animals: Polycomb repressive complex 1 (PRC1) and Polycomb repressive complex 2 (PRC2) (Kerppola, 2009). PRC2-type complexes contain four core components: enhancer of zeste (E(z) in Drosophila, EZH2 in mammals), extra sex combs (Esc in Drosophila, EED in mammals), suppressor of zeste 12 (Su(z)12 in Drosophila, SUZ12 in mammals) and a nucleosome remodeling factor (Nurf55 (Caf1) in Drosophila, RbAp46/48 (RBBP7/4) in mammals) (Köhler & Villar, 2008; Kerppola, 2009). The global identification of PcG target genes has provided the first insights into the mechanisms that govern stem cell maintenance and imply a dynamic regulation of PcG function during cell differentiation in animals (Valk-Lingbeek et al., 2004; Gil et al., 2005; Akala & Clarke, 2006; Buszczak & Spradling, 2006; Ringrose, 2006). In marine embryonic stem cells (ESCs), PcG target genes are de-repressed in cells deficient in the PRC2 component EED (ESC homolog) and are preferentially activated when differentiation is induced (Boyer et al., 2006; Chamberlain et al., 2008).

Homologs of Drosophila PRC2 components have been cloned in both mammals and plants. However, the duplication of almost all PcG genes allows variations in complex composition, depending on cell or tissue type and developmental phase (Köhler & Grossniklaus, 2002; Hsieh et al., 2003; Lessard & Sauvageau, 2003; Otte & Kwaks, 2003; Kim et al., 2010). Arabidopsis contains several homologs of animal PRC2 components: three E(z) homologs: CURLY LEAF (CLF), MEDEA (MEA), and SWINGER (SWN); three Su(z)12 homologs: EMBRYONIC FLOWER2 (EMF2), FERTILIZATION-INDEPENDENT SEED2 (FIS2) and VERNALIZATION2 (VRN2); and one Esc homolog: FERTILIZATION-INDEPENDENT ENDOSPERM. (FIE). These form three PRC-like complexes that mainly control flower and seed development in Arabidopsis (Kim et al., 2010; Rodrigues et al., 2010).

Fertilization-independent endosperm, a FIS-PRC2-like component of plants with a conserved WD40 domain, is the homolog of ESC in Drosophila and of EED in humans. In Arabidopsis, FIE is the core component of several ESC-E(Z)-like complexes that are mainly involved in early seed development as well as flowering program repression during embryo and subsequent seedling development (Kinoshita et al., 2001; Köhler & Makarevich, 2006; Bouyer et al., 2011). Aberrant phenotypes of fie mutants in Arabidopsis, including fasciated stems, early flowering and premature endosperm development, have been correlated with the de-repression of PcG-targeted differentiation regulators, such as the KNOX and MADS-box genes (Riechmann & Meyerowitz, 1997; Kinoshita et al., 2001; Katz et al., 2004). The universal expression of FIE in wild-type Arabidopsis during the vegetative and reproductive phases implies that it may play a possible role throughout the entire plant life cycle (Köhler & Grossniklaus, 2002).

The essential roles of FIE in endosperm development, which have great economic significance in agriculture, have spurred the identification of homologs of PcG protein FIE-E(z)-like complexes in monocot crops, including maize, barley and rice (Springer et al., 2002; Thakur et al., 2003; Luo et al., 2009; Kapazoglou et al., 2010). Some components of rice PcG proteins have been cloned and analyzed. Among these core components, two Esc-like genes from rice, designated OsFIE1 and OsFIE2, have been identified. OsFIE1 is expressed especially in 4-d post-fertilization endosperm, but no aberrant phenotype was found in its OsFIE1 T-DNA insertion lines (Luo et al., 2009). Recently, it has been reported that OsFIE2 functions in seed development and grain filling via a mechanism distinct from Arabidopsis (Nallamilli et al., 2013). However, the extensive role of OsFIE2 in rice development and its relation to OsFIE1 remain largely unknown.

We report here the isolation and functional characterization of the rice PcG group gene, OsFIE2, which we show to have a broad expression pattern in rice. Significantly reduced levels of OsFIE2 expression resulted in pleiotropic aberrant phenotypes and de-repression of some key developmental regulators, such as the MADS-box (OsMADS3) and KNOX (Oskn3) genes. Our results demonstrate that OsFIE2 plays a critical role, not only in rice reproductive development, but also in vegetative development, especially in stem cell maintenance in the rice root system, where it maintains the suppression of key differentiation regulators, similar to its counterpart in animals.

Materials and Methods

Plant material and growth conditions

A japonica rice, Oryza sativa L. cv Nipponbare, from the National Center for Gene Research, Chinese Academy of Sciences Resources, was used in our experiments. The plants were grown in the field or glasshouse at 30 : 24°C (day : night) under a 12-h photoperiod. The T1 seeds were mainly used for phenotype analyses.

Rapid amplification of cDNA ends (RACE)-based cloning of OsFIE2 cDNA

Total RNA was extracted from fresh seedlings 2 d after germination, using an RNeasy®® Plant Mini Kit (Qiagen, Hilden, Germany). Two micrograms of total RNA were used as starting material for both 3′- and 5′-RACE, employing the SMART™ RACE cDNA Amplification Kit (Clontech Laboratories, Mountain View, CA, USA) according to the manufacturer's instructions. With the 3′-RACE cDNA pool, we performed PCR amplification of the OsFIE2 gene using degenerate oligonucleotide primers (GSP1, 5′-ATHTGGTCNATGAARGAGTTCTGG-3′; GSP2, 5′-ATHTGGTCNATGAARGAGTTTTGG-3′), together with UPM (universal primer A mix). The PCR products were cloned into a T vector, and then sequenced (TaKaRa, Otsu, Japan). Gene-specific antisense primer (GSP3, 5′-GGAAGGTCAGTCCATGTAAAGGATTGCTCAAC-3′) was designed according to this sequence. PCR amplification was performed using the 5′-RACE-Ready cDNA as template. The PCR products were also cloned into the T vector and sequenced (TaKaRa). Next, we amplified the full-length cDNA of OsFIE2 using the primers 5′-AGAGGCGCATGGCGAAGC-3′ and 5′-GAAATTTGAGCTGGGCCAATC-3′ and high-fidelity KOD-Plus-DNA polymerase (TOYOBO, Osaka, Japan). The products were cloned into the T vector and confirmed by sequencing.

Microprojectile bombardment and confocal microscopy

The inner epidermal layers of onion were peeled and placed on Petri dishes containing Murashige–Skoog (MS) basal medium and 2% agar. Particles (10 μl) were coated by precipitating 5 μg DNA purified on Qiagen columns onto 3 mg of water-washed golden particles with 50 μl of 2.5 M CaCl2 and 20 μl of 0.1 M spermidine, followed by washing with 70% ethanol and resuspension in 48 μl of 100% ethanol. The 10-μl particles were then placed on each delivery disk. Onion inner epidermal cells were bombarded at 1100 psi, using a helium biolistic device (Bio-Rad PDS-1000; Du Pont, Hercules, CA, USA). After bombardment, the dishes were sealed and incubated for 18–20 h at 22°C and in the dark for 12–16 h before observation. For microscopy observation, the epidermal cells were mounted in water, viewed with a Leica DMRB inverted microscope (Wetzlar, Germany) and imaged using confocal microscopy (Leica TCS SP2). For the detection of green fluorescent protein (GFP), the selective setting (excitation, 488 nm; emission, 510–550 nm) was used.

Generation of transgenic plants

The 35S-eGFP-OsFIE2 fragments were subcloned into the transient transformation plasmid pUC19, which was bombarded into onion inner epidermal cells for the subcellular localization of OsFIE2. The 1798-bp sequence from the OsFIE2 promoter region from −1912 to −115 bp substitutes for 35s promoter and deletes the coding sequence of OsFIE2 in the above construct including 35S-eGFP-OsFIE2 for OsFIE2 expression pattern analysis. For RNA interference, two inverted fragments (558 bp) of OsFIE2 and an intron from the pKANNIBAL vector (CSIRO Plant Industry, Canberra, ACT, Australia) were subcloned into the pAHC17 vector (Christensen & Quail, 1996) at the unique BamHI site. Subsequently, this construct carrying the ubiquitin (Ubi1) promoter was cut by HindIII and EcoRI and inserted into the pCAMBIA1300 binary vector (Cambia). As a control, a 35S-eGFP fragment was cloned into the pCAMBIA1300 vector as a HindIII–EcoRI fragment. To generate transgenic plants, mature seeds of Nipponbare were used. Biolistic transformation was performed following the protocol of Christou et al. (1991), which was later modified and improved by Rancé et al. (1994). Green fluorescence was detected under a stereomicroscope (SZX12; Olympus, Tokyo, Japan). Only GFP-positive hygromycin-resistant calli were transferred to RNH50 medium for further growth.

Southern analyses

Genomic DNA of wild-type and transgenic seedlings was isolated using cetyltrimethylammoniun bromide (CTAB) extraction buffer, according to Saghai-Maroof et al. (1984). DNA (10 μg) was digested with EcoRI and HindIII, separated on 0.8% agarose gel and transferred to a Hybond™-N+ membrane (Amersham Pharmacia Biotech, Uppsala, Sweden). The first 539 bp of the OsFIE2 coding sequence was used as a probe. Radioactive DNA labeling was prepared using the HexaLabel™ DNA Labeling Kit (Fermentas, Vilnius, Lithuania), according to the manufacturer's recommended procedure. The image was scanned using a Typhoon 9200 Variable Mode Imager (Amersham Biosciences, Piscataway, NJ, USA). sps: 5′-ATGGCGAAGCTAGGGCCGGG-3′; spas: 5′-TCATTCCGGTGACCTCCTGCTCC-3′.

Semi-quantitative (RT-PCR) and quantitative (RT-qPCR) reverse transcription-polymerase chain reaction analyses

RNA extractions were carried out using an RNeasy Mini Kit (Qiagen). The quantity of RNA was detected via spectrophotometer and gel electrophoresis. Reverse transcriptions were performed with 2 μg of total RNA, 5 × first strand buffer, 5 pmol of oligo(dT), 2 μl of 10 mM deoxynucleoside triphosphates (dNTPs) and 100 U ReverTra Ace®® (TOYOBO) in a final volume of 20 μl. Reactions were carried out at 42°C for 1 h in a PTC-200 PCR system (MJ Research; Bio-Rad, Waltham, MA, USA), followed by a 10-min step at 95°C to denature the enzyme, and then cooling to 4°C. rTaq DNA Polymerase (TOYOBO) was used to amplify 2 μl of cDNA product, according to the manufacturer's instructions. The PCR conditions were as follows: 95°C for 1 min, followed by 33 cycles at 94°C for 30 s, 58°C for 30 s, 72°C for 35 s, and, finally, 72°C for 5 min. Specific primers for GAPDH were GS (5′-ACTTTGTTGGTGACAGCAGGTC-3′) and GAS (5′-CAGGTCCATATCATCAGCATCG-3′).

Primers for target genes, such as OsFIE2, OsKN3, MADS1, MADS3, CRL, QHB, LRP1 and OsFIE1, are shown in Supporting Information Table S3. The RNA extracted from rice leaf was used for the detection of Oskn3, OsMads1 and OsMads3 expression via RT-PCR, whereas RNA extracted from rice root was used for the detection of Crl, Qhb and Lrp1 expression. For all reactions, the 5′- and 3′-primers spanned different exons. Thus, the amplification products obtained from the cDNA differed in length from that obtained from any contaminant genomic DNA comprising intron sequences.

RT-qPCRs were performed in a reaction volume of 10 μl containing 0.2 μM of each primer and 2 × SYBR Green Real-time PCR Master Mix (TaKaRa, Tokyo, Japan) for 45 cycles (95°C for 15 s and 60°C for 40 s) with a Light Cycler® 480 (Roche Diagnostics, Mannheim, Germany). The data were analyzed using LinRegPCR (ver. 7.0, bioinfo@amc.uva.nl, Netherlands). Gene expression was normalized to that of Ubiquitin, a house-keeping gene. Specific primers of OsFIE2 in RT-qPCR were IS (5′-GGGGAGGGTAGCATTGAT-3′) and IAS (5′-CCTTCACGGTTGCCTATTG-3′), and specific primers of OsFIE1 in RT-qPCR were IS (5′-ATGGGCCCCACTAGTAGGAA-3′) and IAS (5′-CGGTTGCGCTGTGGATATC-3′). Primers for Ubiquitin were US (5′-AACCAGCTGAGGCCCAAGA-3′) and UAS (5′-ACGATTGATTTAACCAGTCCATGA-3′).

Protein extraction and Western blotting

Young rice seedlings were ground in liquid nitrogen. Protein extraction was performed using Laemmli buffer, according to the protocol of Martínez-García et al. (1999). For each 100 mg of tissue, 300 μl of extraction buffer were used. The protein concentration was determined by Bradford assay, and equal amounts of protein were resolved on 12% sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose membranes (Schleicher & Schuell BioScience, Dassel, Germany). To detect OsFIE2 protein, the membranes were incubated with rabbit anti-OsFIE2 polyclonal antibody in a 1 : 200 dilution for 1 h. As a control, tubulin was labeled using monoclonal anti-α-tubulin (T-9026; Sigma, St. Louis, MO, USA) at a dilution of 1 : 500. Horseradish peroxidase (HRP)-conjugated goat-antirabbit and goat-antimouse IgGs (Pierce, Rockford, IL, USA) were used as secondary antibodies for OsFIE2 and tubulin, respectively, in a 1 : 20 000 dilution. Detection was carried out using SuperSignal West Pico Chemiluminescent Substrate (Pierce), according to the manufacturer's instructions. The first antibodies were prepared according to the protocol of Wang et al. (2008). Fusion proteins containing the 376-amino-acid full protein of OsFIE2 were purified using a glutathione S-transferase (GST) affinity chromatography column (Pierce) according to the supplier's protocol. Purified proteins were collected by SDS-PAGE, which were used for rabbit injection. The injected rabbits were kept at the Center for Disease Control in Hubei Province.

Plant sectioning and microscopy

Plant ovules, leaves and roots were embedded in paraffin. The process of sectioning was performed according to the protocol of Chen et al. (1998). Plant morphology and sections were observed using an inverted microscope (Leica DM IRE2) equipped with a CCD camera (CCD RS image MicroMAX; Princeton Instruments, Inc., Trenton, NJ, USA) or photographed with a Sony-DSC-T9 (Tokyo, Japan) camera. Pollen vitality and sterile ratio were detected by fluorescein diacetate (FDA) staining. Pollen grains were incubated in 10 μg ml−1 FDA solution for 15 min and then excited by ultraviolet light (405 nm) with a Leica DM IRE2 after washing. The root apexes of 7-d-old seedlings were cleared according to the protocol described by Liu & Meinke (1998) before observation.

Plant cultivation and morphological characterization

RNAi and wild-type seeds were surface sterilized with 5% (v/v) NaClO at room temperature for 0.5 h, and then rinsed with sterile water three times for 5 min each. The sterilized seeds were then sown in a Petri dish containing distilled water for germination. The germinated seeds (as judged by radicle emergence) were transferred to lucifugal cups containing 300 ml of distilled water and placed in a plant growth chamber at 30 : 24°C (day : night) under a 12-h photoperiod. The lengths of the embryonic and crown roots, plant height and number of lateral roots were recorded using Image J 1.3 (NIH).

Bromodeoxyuridine (BrdU) incorporation and immunofluorescence staining

To observe the ability of cell differentiation in the root apical meristem (RAM), we incubated germinated seeds of both RNAi and wild-type plants in 10 mM BrdU (Sigma-Aldrich) for 24 h. The immunofluorescence staining procedure was performed according to the protocol of Li et al. (2006).

Statistical analyses

Each experiment was repeated at least three times. One-way ANOVA (SPSS/13; SPSS Inc.) was used to test significant differences between controls and mutants in all replications. P values were analyzed using Student's t-test. Digital pictures were processed using Adobe Photoshop CS2.


OsFIE2 is a conserved PcG gene in rice

Based on the sequence information on the Institute for Genomic Research (TIGR) database (http://www.tigr.org/tdb/e2k1/osa1/) and using SMART RACE strategies, we obtained the full-length (1601-bp) sequence of this gene, named Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2; Number Os08g04270 or Os08g0137100), encoding a deduced protein of 376 amino acids.

The structure of OsFIE2 was analyzed using Spidey, an mRNA-to-genomic alignment program (http://www.ncbi.nlm.nih.gov/IEB/Research/Ostell/Spidey/), which showed that the mRNA sequence comprised 13 exons (Fig. 1a). We blasted the deduced amino acid sequence of OsFIE2 in the National Center for Biotechnology Information (NCBI) megablast database, where similar proteins were found (AAS13489, NP 001060953). SMART software (http://smart.embl-heidelberg.de) predicted the OsFIE2 protein to have six highly conserved WD40 motifs (Fig. S1a). The pattern of basic residues in these motifs shows high homology to WD40-containing PcG proteins in maize (Zmfie1/Zmfie2) and Arabidopsis (FIE), as well as their counterparts in animals, that is, ESC in Drosophila (Simon et al., 1995) and EED in humans (Peytavi et al., 1999; Fig. S1b). This indicates that OsFIE2 is one Polycomb member of the FIE homolog in plants (Fig. S2). Further analyses of the OsFIE2 protein revealed a retinoblastoma protein (Rb) interactive region (amino acid residues 357–361) through SMART/PFAM domains analysis, and a proline-directed kinase (e.g. mitogen-activated protein kinase, MAPK) phosphorylation site (amino acid residues 274–280) through the ELM motif search.

Figure 1.

Structural and expression analysis of Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2). (a) The mRNA sequence comprises 13 exons (E). (b) Reverse transcription-polymerase chain reaction (RT-PCR) analysis of the OsFIE1 and OsFIE2 expression pattern during vegetative and reproductive development of Oryza sativa. RT-PCR was performed on RNA isolated from various rice tissues to determine the expression patterns of the OsFIE2 gene: 1, mature seed; 2, leaf (4 d after germination, DAG); 3, root (4 DAG); 4, spikelets (before flowering); 5, spikelets (after flowering); 6, anther; 7, ovary 3 d after pollination (DAP). As a control, glyceraldehyde-3-phosphate dehydrogenase C (GAPDH) RNA was amplified. (c) Transcript levels of OsFIE2 in different organs and tissues by quantitative reverse transcription-polymerase chain reaction (RT-qPCR): 1, root at 1 wk after germination; 2, stem; 3, leaf; 4, panicle at stage 3; 5, panicle at stage 8; 6, unfertilized ovary; 7, 5-DAP endosperm; 8, 10-DAP endosperm; 9, 14-DAP endosperm; 10, 21-DAP endosperm; 11, 7-DAP embryo.

OsFIE2 expression pattern and OsFIE2 subcellular localization

Unlike OsFIE1, OsFIE2 is universally expressed in both vegetative and reproductive organs in rice, as previous research has reported (Luo et al., 2009; Fig. 1b). The analysis of OsFIE2 promoter activity and the result of RT-qPCR (Figs 1c, S8) were in agreement with each other and further confirmed its universal expression.

To study the subcellular localization of the OsFIE2 protein, we constructed an in-frame fusion of enhanced GFP (eGFP) and OsFIE2 for transient expression assays using a particle gun system. The fusion proteins of OsFIE2:GFP were located in both the nucleus and cytoplasm of the onion epidermal cells (Fig. S3). eGFP protein alone under the control of the CaMV35S promoter was used as a control.

OsFIE2 down-regulation by RNAi

Before creating RNAi transgenic plants, we tested the copy number of OsFIE2. Southern blot analysis was conducted using a radioactively labeled ([α-32P]-dCTP) DNA probe, spanning the first 539 bp of the OsFIE2 coding sequence. As shown in Fig. 2(a), OsFIE2 is a single-copy gene in the rice genome, which confirms the result of bioinformatics analysis in the TIGR database. To analyze the possible function of OsFIE2, we used an RNAi technique to down-regulate OsFIE2 expression. Two inverted fragments of OsFIE2 spaced with an intron were constructed, using the Ubi1 promoter as the driver (Fig. S4). Transformation was carried out using callus culture and a biolistic particle delivery system. We obtained an efficient and stable callus transformation frequency (average of 22.6%) and regeneration rate (average of 61.8%; Table S1). More than 20 independent transgenic lines were obtained and most showed serious abortion. Some produced only a few seeds. These homozygous transgenic lines were used for further analysis.

Figure 2.

The detection of copy number and the down-regulation of Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2). (a) Genomic DNA was extracted from control plants of Oryza sativa as well as full-length OsFIE2-eGFP transformed plants and probed with OsFIE2 cDNA to make a Southern blot. M, λ-HindIII-digested DNA; 1–4, EcoRI-digested DNA; 5–8, HindIII-digested DNA; 1, 5, control; 2–4 and 6–8, OsFIE2-eGFP transgenic lines. (b) Semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR) result of wild-type (1) and different RNAi lines: RNAi-8 (2), RNAi-2-1 (3) and RNAi-1-9 (4). (c) Western blotting with anti-OsFIE2 antibody to confirm the significantly reduced expression of OsFIE2 in RNAi lines: 1, whole protein from control plants as the immunogen; 2–4, whole protein from RNAi lines as the immunogen; anti-α-tubulin serum was used as an endogenous standard.

The transcription levels of RNAi lines were significantly lower than those of control plants (Fig. 2b). OsFIE2 was predicted to encode a 376-amino-acid protein of c. 41 kDa. Western analyses performed with affinity-purified antibodies against OsFIE2 showed that OsFIE2 was barely detected in the seedlings of these abnormal transgenic plants, but did appear in the control (Fig. 2c).

OsFIE2 plays an essential role in reproductive development

More than 20 RNAi lines showed similar phenotypes in our experiments, three of which, RNAi-11, 2-1 and 1-9, were selected for detailed phenotypic analysis. The RNAi plants exhibited multiple phenotypes, as shown in Fig. 3. An apical dominance defect was evident in the stunted RNAi plants (Fig. 3a). At the same time, the lengths of the primary panicles were significantly reduced (Table S1), as were adult plant heights and primary inflorescence lengths (Table S2). On average, RNAi plants flowered a week earlier than wild-type plants, suggesting that OsFIE2 may be required to suppress the differentiation of the vegetative tip meristem during the transition phase. Meanwhile, homeotic conversion of lemmas into palea-like structures was also observed (19.3%, > 200), indicating that OsFIE2 may be involved in maintaining floral organ identity during flower development (Fig. 3e). During male gametophyte development, there was also a reduced amount of pollen and a high proportion of male sterility (Figs 3b–d, S5). Thus, in rice, OsFIE2 also plays a critical role in rice flower organ and pollen development.

Figure 3.

Phenotypes during reproductive development of Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2) RNAi plants. (a) Comparison of control plants (left) and OsFIE2-silenced plants (1-9; right) of Oryza sativa at the same stage. (b–d) Fertile pollen grains from control (b) and sterile pollen of OsFIE2-silenced plants (d); between them is the fluorescent image of a 4′,6-diamidino-2-phenylindole (DAPI)-labeled fertile pollen grain, showing three normal nuclei (c). (e) Normal (left) and abnormally enlarged (right) lemma (arrows) of OsFIE2 RNAi plants. (f) Control (left) and OsFIE2 RNAi seed (right) without fertilization; the asterisk shows early seed development in an unfertilized ovule. Bars: (b) 5 μm; (e) 3 mm.

When RNAi plant ovules were fertilized, we observed both well-developed seeds with mature embryos (Fig. 4a–c,h) and sterile seeds without embryos or with defective embryos arrested at the globular stage (Fig. 4d,g,i). The frequency of the three types of seed showed strong differences among transgenic lines. Fertilization-independent endosperm development was also observed in unfertilized ovules. However, development of the autonomous endosperms was arrested at the cellularization stage (Fig. 4e,f), and thus all seeds with such endosperms were aborted in the early stage. To further confirm fertilization-independent endosperm development, the detailed process was observed in the RNAi line. Among 368 flowers that had been previously emasculated in the RNAi plants, 43 showed expanded ovaries without pollination. These ovaries were sectioned for observation. In the early stage, nucleate endosperm could generate and even occupy all the embryo sac (Fig. 5a,b) and the cellularization process could be triggered (Fig. 5c,d). However, cellularization was never completed (Fig. 5e,f). The endosperm degeneration starts from an uncellularized area and, later, cellularized endosperm cells also collapse in succession (Fig. 5f,h). During this process, embryo was never observed in these ovaries.

Figure 4.

Fertilization-independent endosperm development in Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2) RNAi rice. (a) Differentiated embryo in well-developed seed ovule from RNAi plant of Oryza sativa observed by whole-mount clearing. (b) Fully differentiated embryo in well-developed ovule from RNAi rice. (c) A fully differentiated embryo isolated from the ovule shown in (b). (d) The observation of an enlarged unfertilized ovary from RNAi line 1-9 by whole-mount clearing, showing no embryo formation. (e) Paraffin section showing the area of endosperm in an enlarged unfertilized ovary of an RNAi line. (f) Paraffin section showing autonomous endosperm arrest at the cellular stage in an unfertilized ovary of RNAi line 1-9. (g) Paraffin section showing the embryo at the final stage in the fertilized ovary of an RNAi line. (h) Paraffin section showing the mature embryo in a fertilized ovary of a well-developed seed from RNAi rice. (i) An embryo isolated from a fertilized ovary of an RNAi line. It remains undifferentiated. Bars: (a, b, d–h) 50 μm; (c, i) 100 μm.

Figure 5.

Observation of autonomous development of endosperm in an Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2) RNAi line by paraffin sections stained with toluidine blue O. (a) Nuclear endosperm in embryo sac. (b) Amplification of the boxed area in (a). (c) Cellularized endosperm with cell wall. (d) Amplification of the boxed area in (c). (e) Cellularized endosperm at a later stage. (f) Amplification of the boxed area in (e) showing incomplete cellularization even in a highly cellularized area. (g) Endosperm degenerated finally. (h) Amplification of the boxed area in (g). Bars, 10 μm.

We detected the relative expression levels of OsFIE1 and OsFIE2 by RT-qPCR (Fig. S6) and found both to be down-regulated. This is because the targeted sequence fragment for RNAi has high similarity between OsFIE2 and OsFIE1, suggesting that the aborted endosperm should be attributed to the loss of both OsFIE1 and OsFIE2.

Down-regulation of OsFIE2-enhanced proliferation and differentiation of vascular bundle sheath cells in leaves

Leaf development of OsFIE2 RNAi plants showed severe morphological abnormalities. Externally, leaf blades were smaller, but thicker, curved, and even rolled up like tubes. In some serious transgenic lines, leaf sheaths were distorted (Fig. 6a–d). Transversal sections of leaf blades revealed enlarged and overproliferated vascular bundle sheath cells (Fig. 6e–g,i,k). Some of these sheath cells differentiated into sclerenchyma cells (Fig. 6h,j). Uneven sheath extension among different veins is probably the main reason for the twisted leaves. However, there were no obvious differences in the distribution and morphology of mesophyll, phloem, xylem or epidermal cells between the leaves of wild-type and RNAi plants, indicating the specific cell-type role of OsFIE2 in leaf development.

Figure 6.

Phenotypes in Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2) RNAi leaves. (a) Control. (b–d) RNAi (line 1-9) plant leaf, showing its curled and crinkled morphology. (e) Transversal section of control leaf blade. (f) Transversal section of RNAi leaf blade; notice the over-expanded veins or vascular bundles. (g, i) Magnification of the veins in (e) and (f), respectively. (h, j) Autofluorescence images of (g) and (i). SC, Increased sclerenchyma cells (arrow). (k) The statistical result of the diameter of vascular bundle sheath cells (asterisks in h and j), = 12–18, < 0.05, indicating the much enlarged cell size in the RNAi line. Bars: (a–d) 1 cm; (e, f) 50 μm; (g–j) 10 μm.

Premature primary and adventitious roots as a result of stem cell differentiation in root tips of RNAi lines

The germination of RNAi transgenic seeds was severely retarded compared with that of wild-type seeds (Table S2). After germination, embryonic root development underwent a series of subsequent abnormalities. The embryonic root of RNAi plants elongated significantly more slowly than that of the wild-type (Figs 7a,b,k, S7). As shown in Fig. 7, elongation of both primary and adventitious roots ceased early, but lateral root generation was promoted and longer lateral roots appeared as early as 2 wk after seed germination. Thus, the architecture of the RNAi plant root system was quite different from that of the wild-type (Fig. 7c,d).

Figure 7.

Root system development was affected in Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2) RNAi plants. (a, b) Seed germination of wild-type (a) and RNAi line (b), showing primary roots 3 d after germination (3 DAG). Arrows indicate the primary root emerging from the seeds. (c, d) Seedlings 2-wk after germination, showing promoted lateral root growth (box in d) in wild-type (c) and RNAi plants (d); arrow indicates lateral roots derived from primary root. (e, f) Clearing observation of wild-type (e) and RNAi (f) primary root tips 3 d after germination, showing affected primary root development in OsFIE2 RNAi plants; arrows indicate position of root cap. (g–j) Overview of wild-type and RNAi root tips by clearing technique. (g) Wild-type primary root tip 2 wk after germination. (h) RNAi primary root tip at the same developmental stage has ceased growth. (i) Lateral root primordia on wild-type primary root, 1 wk after germination. (j) Lateral root primordia of RNAi primary root at the same developmental stage. (k) Comparison of primary root length between wild-type (diamonds) and RNAi (squares) plants; = 21. P values calculated for each time point were all < 0.01 except for the third day in (k) (< 0.05). Error bars, ± 1SE. (l) Comparison of lateral root length between wild-type and RNAi plants, < 0.01. Error bars, ± 1SE;= 30. Bars: (a) 2 cm; (b) 1 cm; (c, d) 50 μm; (e, f) 100 μm; (g–j) 50 μm.

Premature primary and adventitious roots in RNAi plants appeared to occur in different root structures that were not yet fully developed. Typically, meristem cells ceased to divide and differentiated at a very early stage, making it difficult to distinguish the typical structures of a root tip, including the meristematic region and even the root cap. Similar phenotypes were observed in 85.2% of roots at the same stage (= 18–20 in three RNAi lines; Fig. 7e,f).

We also observed adventitious root development. Seven days after germination, adventitious roots gradually grew slowly and were shorter and thicker than those of the wild-type, and finally ceased to grow (Fig. 7g,h). However, when most lateral root primordia were still small and located inside root ground tissue in wild-type plants, their counterparts in RNAi plants had already elongated and protruded out of the cortex and epidermis (Fig. 7i,j). This indicates that the premature primary root accelerated lateral root protrusion and elongation.

To further determine why primary root elongation was inhibited and then finally ceased, we carefully observed the root tip at an early growing stage. Compared with the wild-type root tip, the quiescent center (QC) cells of RNAi plants expanded and became larger. Abnormal QC cell division was also observed: 53.1% (= 122) in RNAi line i-11 (Fig. 8a,b). These additional cell divisions most probably contributed to the irregularity of the cell arrangement in the root cortex (Fig. 8c,d).

Figure 8.

Abnormal phenotypes in Oryza sativa Fertilization-Independent Endosperm 2 (OsFIE2) RNAi (line i-11) roots. (a) Wild-type quiescent center (QC) zone 1 wk after germination. (b) RNAi root QC zone at the same time. The insets in (a) and (b) are the magnification of the boxed areas. Asterisks indicate QC cell layer and the arrow indicates abnormal QC cell division. (c) Wild-type root showing its ground tissue, the inset is the enlargement of the boxed area. (d) RNAi root. The inset shows the distorted cell division plane (see marked cells). (e, f) Immunofluorescence image of bromodeoxyuridine (BrdU)-labeled root tips, which reflects cell division activity in wild-type (e) and RNAi (f). (e1, f1) Optical sections of BrdU-labeled RNAi (e1) and wild-type (f1) root tips observed by confocal laser scanning microscopy (CLSM); the arrow indicates the QC zone. (g) RNA level of OsFIE2-related genes detected by semi-quantitative reverse transcription-polymerase chain reaction (RT-PCR). Glyceraldehyde-3-phosphate dehydrogenase C (GAPDH) was used as a control. (h) Relative intensity of gene expression detected by gel electrophoresis: WT, grey bars; RNAi, black bars. 1, OsFIE2; 2, Lrp1; 3, Mads3; 4, Mads1; 5, Oskn3; 6, Crl; 7, Qhb; 8, GAPDH. Bars: (a, b, e, f) 40 μm; (c, d, e1–f1), 20 μm.

To further confirm the developing status of the QC zone and meristem cells, we incorporated BrdU to show their mitotic activity. BrdU is a thymine analog that can be incorporated into proliferating DNA during the S-phase of cell division (Gratzner, 1982). Immunofluorescent experiments can then monitor RAM cell division activities. Li et al. (2006) successfully applied this technique to test the role of glutamate-receptor-like (GLR) 3.1 genes in cell mitotic activity of rice RAM. Following the same technique, we tested the cell division activity of both wild-type and RNAi root tips. The enhanced level of incorporated BrdU was observed in wild-type plants, which showed bright signals, in contrast with the dark QC zone. However, only a very weak signal was observed in RNAi root tips (Fig. 8e,f), indicating that meristem cells were losing their ability to divide and were differentiating instead. Finally, all of the cells ceased to divide and differentiated into different tissues. The QC zone, together with all meristematic cells, ultimately disappeared and later became mature tissue (Fig. 7f,h).

The QC is thought to maintain stem cells by suppressing the differentiation of initial stem cells, and to define the stem cell niche (Aida et al., 2004; Stahl & Simon, 2005). Our result from the RNAi line indicated that root stem cells in RNAi plants were not in proper control. To examine the possible mechanisms that contributed to defective root tip growth and QC quashing in RNAi plants, we tested the expression of LRP1, Crl1 and QHB. LRP1 (Smith & Fedoroff, 1995) and Crl1 (Inukai et al., 2005) were specifically expressed in tissues in which crown and lateral roots were initiated. Our results revealed that the expression levels of both genes were notably enhanced in OsFIE2 RNAi roots. However, QHB, a rice WUS-type gene that is specifically expressed in the QC and is involved in the specification and maintenance of the stem cells in the RAM (Kamiya et al., 2003), showed a lower expression level (Fig. 8g).


OsFIE2 encodes a WD40 Polycomb protein of the PRC2 complex

Polycomb proteins are involved in a wide range of developmental processes in plants. Previous studies have revealed a role for Arabidopsis FIE in both embryonic and postembryonic plant development (Ohad et al., 1999). In the present study, we isolated a homolog of FIE, designated OsFIE2, in rice, a monocot. The deduced amino acid sequence (376 amino acids) of OsFIE2 comprises six WD40 motifs arranged in tandem, showing high homology to Drosophila Extra Sex Comb (ESC) and its counterparts in plants and humans, which are core components of PRC2 (Schubert et al., 2005). Thus, OsFIE2 is a monocot member of the PcG family and belongs to PRC2. We show here that the WD40 motifs containing OsFIE2 have a broad expression pattern. Thus, OsFIE2 may serve as a crucial assembly factor for functional multimeric PcG complexes in distinct developmental processes.

In Arabidopsis, the in vitro interaction between FIE and AtRb (pRb homolog) has been demonstrated by pull-down and surface plasmon resonance measurements (Mosquna et al., 2004). Interestingly, sequence analyses of OsFIE2 (via SMART/PFAM domains analysis) also revealed a possible Rb interactive region (amino acids 357–361) in the last WD40 motif at the C-terminus. Rb, a classic cell cycle regulator, co-localizes with many nuclear PcG components (OsiEZ1, CLF and EMF2 in plants) and may be essential to the efficient association of PcG complexes with chromatin sites in the nucleus (Dahiya et al., 2001). The association between Rb and PcG proteins can form a repressor complex that blocks the entry of cells into mitosis (Dahiya et al., 2001).

A proline-directed kinase (e.g. MAPK) phosphorylation site (amino acids 274–280) was also found in OsFIE2, using the ELM motif search. Currently, PcG-mediated repression maintenance is thought to involve extensive interactions among PcG members and other chromatin factors, which generate a repressive higher order chromatin structure inaccessible to transcriptional activators and chromatin remodeling factors (Calonje & Sung, 2006). In mammalian cells, MAPK signal transduction cascades target PcG protein complex associations of chromatin through phosphorylation, which implicates PcG-mediated repression as a dynamically controlled process (Voncken et al., 2005). Thus, OsFIE2 may also be involved in the chromatin association process regulated by MAPK-mediated phosphorylation.

OsFIE2 is essential for both vegetative and reproductive development in rice

Recent discoveries regarding the control of multiple aspects of plant development by PcG protein complexes suggest that the PcG-mediated cellular memory system is an important mechanism that allows plants to maintain distinct developmental stage transitions via transcriptional control (Guitton & Berger, 2005). In Arabidopsis, FIE participates in the formation of distinct PcG complexes at multiple developmental stages, controlling both seed development and the vegetative–flowering transition program. In Hieracium, general down-regulation of FIE by CaMV35S:hpHFIE confirmed that the gene is essential for multiple developmental processes, including vegetative and reproductive development. However, autonomous endosperm development was not observed (Rodrigues et al., 2008), indicating that the role of FIE may be species dependent. In this study, we found that OsFIE2, the monocot core component of PRC2, is involved in multiple developmental processes in both vegetative and reproductive development.

OsFIE2 RNAi rice plants showed smaller and thicker curly and crinkly leaves than those of wild-type plants. Transversal views of the leaf blade showed increased and enlarged bundle sheath cells (Fig. 6f,h). Despite these changes, the specific distribution of phloem, metaxylem and epidermal bulliform cells still occurred as in wild-type plants. In Arabidopsis, the de-repression of AGAMOUS, a C-class MADS-box gene, resulted in a curled-leaf phenotype of FIE-silenced plants (Katz et al., 2004). The expression of OsMADS3, the rice ortholog of AGAMOUS, was elevated in FIE-silenced rice plants (Fig. 8g,h). However, OsMADS3 is expressed asymmetrically in the floral meristem along the palea–lemma axis and is involved in the repression of lodicule development in wild-type rice plants (Kyozuka & Shimamoto, 2002; Yamaguchi et al., 2006). Thus, different mechanisms appear to be involved in PcG-mediated control of leaf morphology between dicot and monocot plants. Signals from the shoot apical meristem (SAM) are necessary to establish or maintain dorsoventral asymmetry of leaves (Waites et al., 1998). Moreover, genes of Knotted1-like homeobox (KNOX) class 1 encode transcription factors involved in SAM development and maintenance (Scofield & Murray, 2006). Interestingly, ectopic expression of Oskn3, a KNOX-class homeobox gene, also resulted in rumpled leaves with a distorted vein pattern in both rice and tobacco (Postma-Haarsma et al., 1999, 2002). Their resemblance to the malformed leaf phenotypes observed in this work (Fig. 6) and the elevation of the Oskn3 expression level in OsFIE2 RNAi plants shown here (Fig. 8g,h) suggest that Oskn3 is downstream of OsFIE2 in the same regulatory pathway that is vital for normal leaf development.

The reproductive development of RNAi plants shows obvious abnormalities, especially in seed development. In reproductive development, a lower seed ratio was observed in FIE-silenced rice plants. This can be partially accounted for by either homeotic conversion of floral organs or male gamete defects in OsFIE2 RNAi plants. We noticed that, even though mature microspores can be formed in male gametophyte anthers, c. 25.8% of the microspores are sterile, as a result of arrest at an early stage (Fig. 4d). Because OsFIE2 expression was also detected in anthers of wild-type plants, OsFIE2 may also play a role in male gametophyte development. Fertilization-independent endosperm was also observed when anthers were removed before flowering (Figs 4, 5). However, autonomous endosperm development in OsFIE2 RNAi plants never developed to the cellularization completed stage (Figs 4f, 5). Thus, ovary development was arrested at an early stage and never formed a seed-like structure, as observed in Arabidopsis (Ohad et al., 1999). The expression of OsFIE1 was only detected in early ovary development at 3 d after pollination (DAP) (Fig. 1b) and the T-DNA insertion mutant of OsFIE1 showed no abnormal phenotype (Luo et al., 2009). Nallamilli et al. (2013) reported that the reduction in OsFIE2 expression resulted in defects in seed development/grain filling; however, they did not find any evidence for autonomous endosperm development in OsFIE2 RNAi plants. Thus, we carefully checked the endosperm development in our RNAi lines and confirmed again the phenomenon of autonomous endosperm development by repeated experiments. Based on our observations above and the reduction in the OsFIE1 expression level in OsFIE2 RNAi plants (Fig. S6), we conclude that FIE is indeed involved in the regulation of fertilization-independent endosperm development, and these different results might be caused by functional redundancy between OsFIE1 and OsFIE2 during early endosperm development.

OsFIE2 plays an essential role in stem cell maintenance, suggesting its functional conservation in both plants and animals

PcG proteins play critical specification roles in ESCs and in reprogramming differentiated cells in mice (Pereira et al., 2010). In mice, in the absence of EED, which is the homolog of ESC in mammals, ESCs are prone to differentiate (Boyer et al., 2006; Chamberlain et al., 2008), indicating that PcG proteins are required for ESC maintenance in the appropriate area. Unlike animal development, plant embryogenesis generates both SAM and RAM, which remain to continuously elongate roots and shoots. FON2 and FCP1 are two known proteins that functionally diversify to regulate different types of meristem (Suzaki et al., 2006). Unlike FON2, which functions specifically in the SAM, FCP1 works to maintain both the SAM and RAM of rice (Suzaki et al., 2008). PcG protein has also been shown to function in rice SAM, but no evidence indicates functionality in the RAM.

In our study, we found that OsFIE2 was highly expressed in root tips (Fig. 1c) and that the length of primary and adventitious roots of OsFIE2 RNAi plants was significantly reduced. Detailed observations revealed that these root tips ceased to grow as meristem cells lost their ability to divide, but QC cells showed abnormal division, with both finally differentiating. Plant root development is regulated by internal programs and external factors, and proceeds through controlled cell division and cell differentiation (Benfey & Scheres, 2000). Signals derived from QC are the key internal regulatory mechanisms, which maintain the surrounding initial cells in an undifferentiated state, balancing the appropriate elongation of the emerging root with the differentiation of the lateral organs (van den Berg et al., 1998). In OsFIE2 RNAi plants, meristem cell differentiation and QC zone quashing, as well as active lateral root primordia generation, imply that this balance has broken down and that the mechanism for QC maintenance and meristem cell identity is disrupted. Further study revealed that existing QC cells gradually quashed the transcription level of QHB, notably reducing it. QHB, a rice WUS-type homeobox gene that specifically expresses in the QC of the root apex (Kamiya et al., 2003), is thought to be involved in the specification and maintenance of stem cells in the RAM. This further confirmed that OsFIE2 is essential for RAM stem cell maintenance in rice roots. A recent investigation of the moss Physcomitrella patens also confirmed that PpFIE functions to maintain an undifferentiated state of meristematic cells within the apex (Mosquna et al., 2009). These data suggest that regulation of stem cell proliferation and inhibition of their differentiation might be a basic function of the PcG complex in plants as well as in animals.


The work was supported by the ‘973’ Project (2013CB126900 and 2014CB943400) and by the ‘Chen Guang’ project (201154031100).