Introduction

Lissencephaly 1 (LIS1) was the first gene identified in the pathogenesis of type-1 lissencephaly, a severe brain developmental disease characterized by mislocalization of cortical neurons (Reiner et al. 1993). The amino acid sequence of LIS1 is usually characterized as three distinct regions (Kim et al. 2004). The first region is the LIS1-homology (LisH) motif, which includes residues 1–39 at the N terminus and has recently been recognized as a novel ubiquitous motif that is widely found in 114 eukaryotic proteins (Emes and Ponting 2001). The second region is a coiled-coil region which contains the spanning amino acids 40–85. It was reported that LisH and the coiled-coil fragments were involved in homodimerization of the protein (Tai et al. 2002). Studies on LIS1-interacting proteins, either in human or in mouse models, have suggested that LIS1 may play a role in the pathogenesis of numerous diseases such as male sterility, schizophrenia, neuronal degeneration and viral infections (Reiner et al. 2006). The third distinct region of LIS1 is the WD40 (also known as Trp-Asp) motif, a stretch of approximately 40 amino acid that typically terminates in Trp-Asp (Neer et al. 1994). WD40 motifs usually present as 4–10 tandem repeats to form a functional structure. It was reported that repeated WD40 motifs acted as a site for protein–protein interaction, and proteins containing WD40 repeats serve as platforms for the assembly of protein complexes or mediators of transient interplay among other proteins (Neer et al. 1994; Sondek et al. 1996).

WD40-repeat proteins not only play various key roles in plant biological process, such as signal transduction, cytoskeletal dynamics, protein trafficking, nuclear export, and RNA processing, but are also especially involved in chromatin modification and transcriptional mechanisms (van Nocker and Ludwig 2003). A recent plant study has revealed 85 and 78 WD40 motif-containing proteins in Arabidopsis and rice, respectively (Lee et al. 2008). A large number of genes encoding WD40 motif-containing proteins have been identified in Arabidopsis. For examples, the CONSTITUTIVELY PHOTOMORPHOGENIC 1 (COP1) and SUPPRESSOR OF PHYA-105 (SPA1) are involved in the light signaling/photomorphogenesis and flowering (Hoecker et al. 1999; Holm et al. 2002; von Arnim et al. 1997), respectively. GTP BINDING PROTEIN BETA 1 (AGB1) is required for auxin response and cell division (Ullah et al. 2003). FASCIATA2 (FAS2) is involved in meristem maintenance (Kaya et al. 2001). LEUNIG (LUG) function in floral development (Franks et al. 2002; Liu and Meyerowitz 1995). FERTILIZATION-INDEPENDENT ENDOSPERM (FIE) regulates seed development and flowering (Sorensen et al. 2001). CYCLOPHILIN71 (CYP71), which interacts with histone H3, is essential for chromatin-based gene silencing and organogenesis (Li et al. 2007).

Although extensive studies have been carried out in Arabidopsis, the biological function of WD40 motif-containing genes has seldom been elucidated in depth in the monocot plant rice. One example from rice is Rice Immature Pollen 1 (RIP1), which contained five conservative WD40 domains on its C terminus (Han et al. 2006) and is more highly expressed in anthers than in other tissues. During anther development, a high level of RIP1 transcript accumulates in the late vacuolated and mature pollen stages. RIP1 loss-of-function causes not only abortive male gametophyte development, but also defects in pollen germination (Han et al. 2006). Another documented example of a rice WD40 motif-containing protein is the Salt Responsive WD40 protein (SRWD) subfamily. This gene subfamily can be phylogenetically divided into four groups and consists of five members in rice. Expression profile experiment results revealed that all five members were significantly up-regulated under salt stress, suggesting all of them play important function in salt response (Huang et al. 2008). However, the detailed biological functions of SRWD genes remain to be explored.

Molecular studies in Arabidopsis and rice have revealed that numerous genes are required for gametophyte generation. Because the cytological course of the embryo sac formation is more complicated than microsporogenesis and microgametogenesis, the roles of a majority of the genes related to fertility have been characterized during stamen and pollen development. In flowering plants, male reproductive development requires the differentiation of anther tissues and proper male meiosis to produce microspores (Ma 2005). Male gametophyte development in the anther compartment of the stamen depends on functional cross-talk between gametophytic and sporophytic tissues (Zhang and Wilson 2009). In the last few years, a number of genes controlling anther development, especially for the tapetum degeneration have been characterized in rice. The UNDEVELOPED TAPETUM1 (UDT1) gene, encoding a bHLH transcription factor, is involved in tapetal cell differentiation in rice (Jung et al. 2005). Tapetum Degeneration Retardation (TDR), a putative homolog of AMS in Arabidopsis, is downstream of UDT1 and plays a crucial role in tapetum degeneration and microspore development (Li et al. 2006; Zhang et al. 2008). It was reported that silencing the rice UDP-Glucose Pyrophosphorylase 1 gene inhibits the degeneration of the tapetum (Chen et al. 2007). Carbon Starved Anther (CSA), encoding an R2R3 MYB transcription factor, controls sugar partitioning into the anther and is required for anther development and pollen maturation (Zhang et al. 2010). CYP704B2, a member cytochrome P450 family, is involved in catalyzing v-hydroxylation of C16 and C18 fatty acids and is required for anther cutin biosynthesis and pollen exine formation in rice (Li et al. 2010). PERSISTENT TAPETAL CELL1 (PTC1), encodes a PHD-Finger protein, is a key regulator of tapetal programmed cell death and pollen development in rice (Li et al. 2011a). Rice MADS3 regulates ROS homeostasis during late anther development and is a key transcriptional regulator that functions in late anther development and pollen formation in rice (Hu et al. 2011). Recently, rice APOPTOSIS INHIBITOR5 (API5) was identified as an activator for the tapetum degeneration during pollen development (Li et al. 2011b). Within the anther, microspore mother cells undergo meiosis to produce male gametophytes, which then develop to mature pollen. Defects in meiotic process usually result in abnormal microgametogenesis, which then affects plant fertility. Through generation of numerous rice mutant libraries, genes related to the meiotic process and required for proper fertility in rice have been characterized. PAIR1 and PAIR3 are two novel gene encoding a putative coiled-coil protein that have been determined to be essential for the homologous chromosome pairing in rice (Nonomura et al. 2004a; Yuan et al. 2009), and PAIR2, an ortholog of Arabidopsis ASY1 (Armstrong et al. 2002), has been shown to be required for homologous chromosome synapsis in meiosis I (Nonomura et al. 2004b, 2006). OsRAD21-4 has a crucial role in chromosome condensation, subsequent pairing and synapsis, and separation of chromatids (Zhang et al. 2006b). OsRPA1a is required for DNA repair during meiosis and its knock-down mutants have shown low fertility (Chang et al. 2009). ZEP1, a transverse filament (TF) protein, is the rice homolog of Arabidopsis thaliana ZYP1. ZEP1 is the central element of the synaptonemal complex and regulates the number of crossovers during meiosis in rice (Wang et al. 2010b). Kinesin-1 family member Pollen Semi-Sterility1 (PSS1) is essential for male meiotic chromosomal dynamics, male gametogenesis, and anther dehiscence in rice (Zhou et al. 2011).

In this study, we report the cloning and functional characterization of a novel WD40 motif-containing gene lissencephaly type-1-like 1 (OsLIS-L1) in rice. Two independent mutant alleles, designated oslis-l1-1 and oslis-l1-2, were identified from our T-DNA insertional mutant library (Wu et al. 2003; Zhang et al. 2006a, 2007) and POSTECH, South Korea (Jeon et al. 2000; Jeong et al. 2006), respectively. The two allelic mutants exhibited similar abnormal development phenotypes, such as semi-dwarf, shorter panicle length, and very low fertility. By introducing the OsLIS-L1 genomic fragment into oslis-l1-2 background, we successfully rescued the wild-type (WT) phenotypes, confirming OsLIS-L1 was required for normal fertility in rice. Gene expression profile examination revealed that OsLIS-L1 was relatively highly expressed in stem and panicles, consistent with its function in the cell development in stem and requirement for panicle development. During the panicle developmental process, OsLIS-L1 showed a dynamic expression pattern in consecutive panicle developmental stages. Male meiocyte cytological examination suggested that OsLIS-L1 does not affect the meiosis in pollen mother cells. Histological analysis revealed that the very low fertility was due to the abnormal male gametophyte formation. Histological analysis revealed that OsLIS-L1 regulates the cell proliferation in the first internode of the culms. Our results clearly indicate that OsLIS-L1 is required for the cell elongation in the first internode and pollen development in rice.

Materials and methods

Plant materials and growth conditions

The oslis-l1-1 (03Z11AI23) and oslis-l1-2 (3A-04974) mutants were obtained from the Rice Mutant Database (RMD) (Wu et al. 2003; Zhang et al. 2006a; http://rmd.ncpgr.cn/) and the POSTECH Rice T-DNA Insertion Sequence Database (RISD) (Jeon et al. 2000; Jeong et al. 2006; http://www.postech.ac.kr/life/pfg/risd/), respectively. All plants were planted in the experimental field of Huazhong Agriculture University in Wuhan, China, during the rice-growing season of 2006–2010 with a planting density of 23 plants/m2.

Genotyping of oslis-l1 mutant plants

Polymerase chain reaction (PCR) genotyping of the oslis-l1-segregating population was performed by using the following primers: L1, R1 and NTLB5, L2, R2 and iAPL1, which are listed in Supplementary Table S1. PCR was conducted with an initial step of 94°C incubation for 5 min, followed by 30 cycles of 94°C for 45 s, 55°C for 45 s, and 72°C for 1 min; with a final extension at 72°C for 5 min.

Gene expression analysis

To extract total RNA of various tissues from Zhonghua 11 (Oryza sativa L. ssp. japonica), Trizol reagent (Invitrogen, Carlsbad, CA, USA) was used to isolate total RNA according to the manufacturer’s instructions. Before the RNA reverse transcription step, total RNA was treated with amplification-grade DNase I (Invitrogen) for 15 min to degrade any residual genomic DNA. The synthesis of the cDNA first strand was carried out with Moloney Murine Leukemia Virus (M-MLV) reverse transcriptase (Promega Corp., Madison, WI, USA) according to the manufacturer’s instructions. An aliquot (1 μl) of the reaction mixture was used for PCR. The PCR was performed in an ABI 9700 thermocycler (Applied Biosystem, Foster City, CA, USA) with three replicates. The cycling profile was as follows: 94°C for 5 min; 28 cycles at 94°C for 45 s, 55°C for 45 s, and 72°C for 1 min; and 72°C for 5 min. Next, 20 μl of the PCR product was separated in a 1.0% agarose gel and stained with ethidium bromide for visualization. The primer pairs used for reverse transcription (RT)-PCR were OsLIS-L1RT-L1 and OsLIS-L1RT-R1, OsLIS-L1RT-L2 and OsLIS-L1RT-R2, OsLIS-L1RT-L3 and OsLIS-L1RT-R3, and OsLIS-L1RT-L4 and OsLIS-L1RT-R4. GAPDH was used as the standard in all RT-PCR assays. qRT-PCR was performed in an optical 96-well plate that included 12.5 μl SYBR Premix EX Taq and 0.5 μl Rox Reference Dye II (Takara), 1 μl the cDNA sample, and 0.5 μl of each gene-specific primer, in a final volume of 25 μl on a PRISM 7500 PCR instrument (Applied Biosystems). The reactions were performed at 95°C for 30 s, 40 cycles of 95°C for 5 s, and 60°C for 34 s. Disassociation curve analysis was performed as follows: 95°C for 15 s, 60°C for 1 min, and 95°C for 15 s. Data were collected using the ABI PRISM 7500 sequence detection system following the instruction manual. The rice Ubiquitin5 gene was used as the internal control. All primers for RT-PCR and qRT-PCR are listed in Supplementary Table S1.

Vector construction and rice transformation

An approximately 12.2-kb genomic DNA fragment containing the entire OsLIS-L1 coding region and the 1.4-kb upstream and 3.2-kb downstream sequence was obtained by NheI digestion from the Clemson BAC clone a0023F18. The fragment was inserted into the binary vector pCAMBIA2301 to generate the transformation plasmid pC2301–OsLIS-L1 for the complementation test.

To generate the OsLIS-L1 RNAi vector, a 531-bp fragment was amplified with primers RNAi-L and RNAi-R (restriction sites added for subsequent cloning are underlined in Supplementary Table S1) from FL-cDNA clone AK111830 (http://cdna01.dna.affrc.go.jp/cDNA) and inserted into the pDS1301 vector as described (Chu et al. 2006). The two binary plasmids were introduced into A. tumefaciens EHA105 by electroporation, and the rice transformation was done according to the method of (Wu et al. 2003).

For expression in protoplasts, the full length open reading frame (ORF) of OsLIS-L1 was amplified from FL-cDNA clone AK111830 (http://cdna01.dna.affrc.go.jp/cDNA) by PCR with the following primers: SC-L and SC-R (the XbaI restriction site is shown as the underlined sequence in Supplementary Table 1). The PCR product was cloned into pM999 vector containing the yellow fluorescence protein (YFP) reporter. Plasmids were purified using Qiagen (Valencia, CA, USA) columns according to the manufacturer’s protocol. The plasmids were introduced into Arabidopsis protoplasts (prepared from leaf tissues) by polyethylene glycol-mediated transformation (Yoo et al. 2007). The reported rice transcription factor Ghd7 from our lab (Xue et al. 2008) was used as a reference for nuclear localization. Expression of the fusion construct was monitored at 36 h after transformation, and images were captured with the confocal microscope DMIRE2 (Leica, Germany).

Histochemical assays and microscopic analyses

Spikelets of various anther development stages were fixed in a solution containing 50% ethanol, 5% acetic glacial and 3.7% formaldehyde for 24 h at room temperature, and then replaced with 70% ethanol twice, and fixed anthers were dehydrated through an ethanol series, embedded into Technovit 7100 resin (Heraeus Kulzer), and polymerized at 37°C for 3 days. The sample blocks were sectioned into 2-μm-thick slices with a Leica microtome (Leica, Germany), sections were stained with 0.25% toluidine blue O (Merck), and the microscopic images were captured by a Leica DFC480 Digital Camera system (Leica, Germany).

For TEM, anthers at various developmental stages were prefixed with 2.5% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.4), and were then fixed in 2% OsO4 in phosphate-buffered saline (pH 7.4). Following ethanol dehydration, samples were embedded in epoxy resin (London Resin Company). Ultrathin sections (70 nm) were obtained using a UC6 ultramicrotome (Leica) and were then double stained with 2% (w/v) uranyl acetate and 2.6% (w/v) lead citrate aqueous solution. Observation and capturing of images were performed with an H-7650 transmission electron microscope (Hitachi) at 80 kV and an 832 charge-coupled device camera (Gatan).

Meiotic chromosome observation

The young panicles at meiosis stage were fixed with 5:3:2 of 95% ethanol:glacial acetic acid:chloroform for 24 h at room temperature, placed in 70% ethanol twice, and stored in 70% ethanol at 4°C until observation. The anthers were gently dissected with needles and afterward transferred to a glass slide together with a drop of improved carbol fuchsin. They were then opened gently with a needle to release pollen mother cells, anther wall debris was carefully removed, and a coverslip was added. Finally, the preparation was gently squashed by vertical thumb pressure under a double layer of filter paper and photographed with a Leica DFC480 (Leica, Germany) digital camera system. Images were enhanced with Photoshop CS.

Results

Morphological characterization of the oslis-l1 mutant

We previously generated a rice T-DNA insertional mutant library with over 120,000 available independent lines and with 30,000 known T-DNA insertion sites (Wu et al. 2003; Zhang et al. 2006a, 2007). By screening T1 plants from the tagged population, we identified a mutant oslis-l1-1 that showed retarded vegetative growth and fertility defects. At the harvest stage, the height of the oslis-l1-1 mutant plants was only 48.9 cm, whereas WT plants had a height of 87.3 cm (Fig. 1a; Table 1). The semi-dwarf phenotype of oslis-l1-1 was partially due to its panicle length being shortened to 12.1 cm from 20.5 cm in WT (Fig. 1b; Table 1). The retarded vegetative growth also accompanied by smaller florets (Fig. 1c, d) and very low fertility. In contrast to the WT seed-setting rate (81.4%), the seed-setting rate of oslis-l1-1 mutant was only 11.3% in the same normal field conditions (Table 1). The 10-seed length and width of oslis-l1-1 was only 84.9 and 70.4% of the WT, respectively (Fig. 1e, f; Table 1). The segregation between normal and mutant plants in T1 generation consistent with a Mendelian 3:1 ratio (normal:mutant = 16:4, χ 2 = 0.267 for 3:1), suggesting a recessive mutation in single Mendelian locus controlled the phenotypes.

Fig. 1
figure 1

Morphological characterization of wild-type (WT) and oslis-l1-1 mutant plants. a A WT plant (left) and an oslis-l1-1 mutant plant (right) at heading stage. Bar 10 cm. b A WT panicle (left) and an oslis-l1-1 mutant panicle (right) at heading stage. Bar 2 cm. c A WT spikelet (left) and an oslis-l1-1 (right) mutant spikelet before flowering. d A WT spikelet (left) and an oslis-l1-1 (right) mutant spikelet after removing the lemma before flowering. Bars 1 mm in c and d. e Ten oslis-l1-1 mutant grains length (top) and ten WT grains length (bottom). f Ten oslis-l1-1 mutant grains width (top) and ten WT grains width (bottom). Bars 5 mm in e and f

Full size image
Table 1 Comparison of agronomic traits of oslis-l1 mutants and their WT controls under normal field growth conditions (Wuhan, China, 2009)
Full size table

T-DNA insertion occurs in a lissencephaly type-1-like gene that encodes a protein with WD40-repeat domain

To characterize the T-DNA insertion site of oslis-l1-1, we amplified the sequence flanking the T-DNA insertion using the thermal asymmetric interlaced-PCR technique (Zhang et al. 2007). The sequencing result of the PCR fragment revealed that T-DNA was inserted in the 21st intron of a gene that was located on chromosome 8 (Fig. 2a). The predicted transcript of the tagged gene was 7,617 bp, including a 3,402-bp coding sequence. The locus name of this gene had been registered as LOC_Os08g06480 in The Rice Genome Annotation Project (http://rice.plantbiology.msu.edu/). The gene of LOC_Os08g06480 was annotated encoding a lissencephaly type-1-like homology motif protein, thus we named it OsLIS-L1 (lissencephaly type-1-like 1). OsLIS-L1 comprised 25 exons and 24 introns (Fig. 2a). SMART algorithm analysis (http://smart.embl-heidelberg.de/) indicated that OsLIS-L1 contains four recognizable structural domains: starting from the N terminus, they are the LisH, CTLH, the domain with five WD40 repeats in the middle, and the domain with four WD40 repeats homologous to the β subunit of trimeric G-proteins (Gβ) (Fig. 2e). The predicted protein encoded by the tagged gene contains 1,133 amino acid residues with a molecular mass of 124.8 kDa.

Fig. 2
figure 2

Molecular analysis of the oslis-l1 mutants and RNAi transgenic plants, and the genetic complementation test. a Structure of the OsLIS-L1 gene, position of T-DNA insertion, PCR and RT-PCR primers. Red boxes indicate exons encoding the protein, and blue box and azure box indicate the 5′ and 3′ UTR regions, respectively. Arrows identify primers for genotyping T-DNA insertion and RT-PCR. b Confirmation of co-segregation of T-DNA insertion with oslis-l1-1 mutant phenotype in 20 T1 plants. Lanes 8, 9, 10, and 20 are homozygous (M) which show mutant phenotype; lanes 3, 4, 5, 6, 7, 11, 12, 15, 16, and 19 are heterozygous (H) which show normal phenotype; and lanes 1, 2, 13, 14, 17, and 18 are WT (W) which show normal phenotype. c Confirmation of co-segregation of T-DNA insertion with oslis-l1-2 mutant phenotype in 20 T1 plants. Lanes 2, 7, 13, and 17 are homozygous (M) which show mutant phenotype; lanes 3, 4, 6, 8, 10, 12, 15, 18, and 20 are heterozygous (H) which show normal phenotype; and lanes 1, 5, 9, 11, 14, 16, and 19 are WT (W) which show normal phenotype. d RT-PCR analysis of OsLIS-L1 expression by 4 pairs of RT-PCR primers in leaves from WT, heterozygous (H) and oslis-l1 and oslis-l1-2 at the seeding stage, respectively. e Structure of the whole transcript and 2 different truncated transcripts of OsLIS-L1. f A WT plant (left) and an oslis-l1-2 mutant plant (right) at heading stage. g A WT panicle (left) and an oslis-l1-2 panicle (right) at heading stage. h A positive pDS1301-OsLIS-L1 transgenic plant (left) and a negative pDS1301-OsLIS-L1 transgenic plant (right) at heading stage. i A positive pC2301–OsLIS-L1 transgenic plant (left) and a negative pC2301–OsLIS-L1 transgenic plant (right) at heading stage. Bars 10 cm in f, h, and i. j A positive pC2301–OsLIS-L1 transgenic plant panicle (left) and a negative pC2301–OsLIS-L1 transgenic plant panicle (right) at heading stage. Bars 2 cm in g and j. k Northern blot analysis, with 10 μg of total RNA used in each sample, and a region of OsLIS-L1 serving as a probe. At bottom are the control ribosomal RNAs. A negative transgenic plant (lane 1 in the gel) was used as control, and others are the positive transgenic plants. l RT-PCR analysis of OsLIS-L1 expression in leaves from three positive pC2301–OsLIS-L1 transgenic plants (first three lanes) and a negative pC2301–OsLIS-L1 transgenic plant (last lane) at the seeding stage. Amplification of GAPDH cDNA was used to ensure that approximately equal amounts of cDNA were loaded

Full size image

To test whether the abnormal phenotype could be attributable to T-DNA insertion in the LOC_Os08g06480 locus, a co-segregation test of the T-DNA tag with the mutant phenotype was performed by PCR analysis of T1 and T2 plants. As shown in Fig. 2b, a combination of PCR results of primer set L1 + R1 and L1 + NTLB5 enable us to easily determine the genotype of plants in the T-DNA insertion site. Among the 20 T1 plants we tested, all 4 T-DNA homozygous plants showed semi-dwarf and very low fertility phenotypes, whereas the T-DNA heterozygous and negative plants grew normally. In T2 progeny, we examined more than 1,000 plants and all the T-DNA homozygous plants were semi-dwarf and had low fertility, whereas heterozygous plants and WT plants showed a normal phenotype (data not shown). These results showed that the semi-dwarf and low-fertility mutant phenotype was co-segregated with the T-DNA insertion in OsLIS-L1.

Loss-of-function of OsLIS-L1 deduced the defects in oslis-l1 mutants

To verify that the phenotypes in oslis-l1-1 were due to the mutation of OsLIS-L1, we ordered an allelic mutant (3A-04974), which we designated as oslis-l1-2, from POSTECH, South Korea (Jeon et al. 2000; Jeong et al. 2006). oslis-l1-2 harbored a T-DNA insertion in the 13th exon of OsLIS-L1 (Fig. 2a). Using the same strategy as for oslis-l1-1, we confirmed the genotyping of oslis-l1-2, and the mutant phenotypes co-segregated closely with the T-DNA insertion (Fig. 2c). The mutant oslis-l1-2 was also semi-dwarf with a shorter panicle length and low fertility, similar to that of oslis-l1-1 (Fig. 2f, g; Table 1).

In order to verify OsLIS-L1 function, we also generated OsLIS-L1 RNAi-suppressed lines in WT background. An RNA blot experiment was employed to detect the OsLIS-L1 transcript abundance in the RNAi-suppressed lines. Among the 37 positive transgenic lines, 12 of the plants showed OsLIS-L1 expression level decreased to various extents (Fig. 2k). As expected, all the 12 RNAi-suppressed lines exhibited semi-dwarf height, shortened panicles, and low fertility, which mimicked the same defects with the two allelic T-DNA insertional mutants (Fig. 2h). Additionally, a genetic complementation experiment was performed to further confirm OsLIS-L1 biological function. A construct pC2301–OsLIS-L1, containing the entire OsLIS-L1 ORF, 1.4 kb upstream sequences, and a 3.2 kb downstream region, was introduced into oslis-l1-2 mutant background by Agrobacterium tumefaciens-mediated transformation (Wu et al. 2003). Of the 20 independent transgenic plants generated, 15 restored normal fertility and plant height (Fig. 2i, j), whereas the other negative transgenic plants remained semi-dwarf and low-fertility phenotype similar to that of oslis-l1-1 and oslis-l1-2. RT-PCR analysis showed that the transcripts of OsLIS-L1 in leaves were also restored to the WT level in the positive transgenic plants, while the negative transgenic plants still showed no detectable OsLIS-L1 expression as the mutant background (Fig. 2l). Next, we investigated the detail agronomic traits of the 20 T0 generation complemented oslis-l1 lines with pC2301–OsLIS-L1 (Supplemental Fig. 1a–d). In the progeny of the rescued plants, segregation of pC2301–OsLIS-L1 coincided with normal fertility, panicle length, and plant height (Supplemental Fig. 1e). All these results strongly showed that the oslis-l1 mutant phenotypes were caused by the loss of OsLIS-L1 function.

Different truncated OsLIS-L1 transcripts generated in the two oslis-l1 mutants

To check the effect of T-DNA insertions in OsLIS-L1 at the mRNA level, we carried out RT-PCR with various primer sets (Fig. 2a). In oslis-l1-1, OsLIS-L1 expression was not detectable when primer set RT1, which covered the oslis-l1-1 T-DNA insertion site, was used for RT-PCR (Fig. 2d). The same result was found in primer set RT-2 analysis, which focused on the region downstream of T-DNA insertion site (Fig. 2d). Interestingly, by using the primer sets RT-3 and RT-4, we found that the 5′ region of T-DNA insertion was normally transcribed in oslis-l1-1 suggesting that T-DNA insertion in oslis-l1-1 only partially destroyed the 3′ region of T-DNA insertion site and generated a truncated transcript (Fig. 2d, e). Similarly, we also found the same effect of T-DNA insertion on oslis-l1-2 mRNA in that the 5′ region of T-DNA insertion was normally transcribed (Fig. 2d, e). As shown in Fig. 2a, e, T-DNA was located in the Gβ-protein homology domain in oslis-l1-1, whereas the oslis-l1-2 T-DNA inserted in the space between the five WD40-repeat domain and Gβ-protein homology domain. Therefore, the two different truncated OsLIS-L1 transcripts contained either a disrupted or a defective Gβ-protein homology domain in two allelic mutants, suggesting that the integrity of Gβ-protein homology domain is critical for OsLIS-L1 functions.

OsLIS-L1 does not affect the meiosis in pollen mother cells

Using the meiotic chromosome spreading method, we took oslis-l1-2 as an example for observing chromosomal behavior during the meiosis process of the pollen mother cells to examine whether meiosis was abnormal. We emphasized on examining the homologous chromosome pairing, synapsis, and the integrity as in our previous studies (Chang et al. 2009; Yuan et al. 2009). In meiosis prophase I, from leptotene to diplotene, the chromosomes underwent condensation, synapse and pairing. Compared with WT, no obvious differences were detected in oslis-l1-2 for those stages (Supplemental Fig. 2a–h). At diakinesis, chromosomes condensed to produce 12 bivalents, which were clearly observed both in oslis-l1-2 and WT (Supplemental Fig. 2i, j). Thereafter, the bivalents aligned themselves along the equatorial plane at metaphase I (Supplemental Fig. 2k, l). At anaphase I, the homologous chromosome separated equationally and then generated dyads (Supplemental Fig. 2m, n). Through metaphase II, two groups of chromosomes separated equationally and produced tetrad spores (Supplemental Fig. 2o, p). During those stages, there were no obvious differences between oslis-l1-2 and WT. There are equal division of univalents, normal formation of dyads and tetrads, and no DNA breakage during male meiocyte formation in oslis-l1-2. Our preliminary data indicated that the oslis-l1 mutations did not affect chromosomal behavior during meiosis in pollen mother cells.

Abnormal male gametophyte formation in oslis-l1 mutants

Since oslis-l1-1 and oslis-l1-2 mutant plants both exhibited a very low fertility phenotype, we next checked the mature pollen viability of OsLIS-L1 mutants along with that of the WT by iodium potassium iodide staining assay. We found that 88.4% of the WT plant pollen was viable, with a regular round shape and dark staining under microscope (Fig. 3a). However, only about 27.1 and 8.1% of the oslis-l1-1 and oslis-l1-2 pollen, respectively, was dark stained (Fig. 3b; Table 1). In addition, OsLIS-L1-RNAi suppressed lines showed a faint staining similar to that of oslis-l1 (Fig. 3c), and complementary transgenic plants showed normal dark stained (Fig. 3d). We also examined the oslis-l1-2 male and female gamete fertility by reciprocal crosses between oslis-l1-2 and WT plants. The seed-setting rates was 5.0, 7.0, and 7.9% in three independent panicles, using WT plants as maternal recipient pollinated with oslis-l1-2 pollen. When using oslis-l1-2 plants as maternal recipient pollinated with WT pollen, the seed-setting rates was 45.5, 49.3, and 54.6% in three independent panicles, the seed-setting rates were similar to those of reciprocal crosses between WT and WT plants. These results indicated that the oslis-l1 mutant was almost female fertile and abnormal pollen may be the predominant reason for the low fertility in oslis-l1 mutants.

Fig. 3
figure 3

Mature pollen viability by iodium potassium iodide staining assay and transverse section analysis of the anther development of the WT and the oslis-l1-2 mutant. Mature pollen of WT (a), oslis-l1-2 mutant (b), OsLIS-L1 RNAi transgenic plants (c), and positive pC2301–OsLIS-L1 transgenic plant (d) were stained by iodium potassium iodide solution. Bars 125 μm. Cross sections of WT (e, g, i, k, m, o, q, s) and oslis-l1-2 (f, h, j, l, n, p, r, t) at stage 5 (e, f), stage 6 (g, h), early stage 8 (i, j), late stage 8 (k, l), stage 9 (m, n), stage 10 (o, p), stage 11 (q, r), and stage 12 (s, t). E epidermis, En endothecium, ML middle layer, MMC microspore mother cell, MP mature pollen, Msp microspore, PMC pollen mother cell, Sp sporogenous cell, T tapetum, Tds tetrads. Bars 50 μm. All the plants were grown in the field under normal field growth conditions. Data were obtained from five individuals

Full size image

To characterize the male sterility of the oslis-l1 mutants, we compared the different development stages of the anthers in the WT and oslis-l1-2 mutant plants by histological examination of plastic transverse sections. Recently, rice anther developments were further delineated into 14 stages (Zhang and Wilson 2009). From stages 1–5 (Fig. 3e, f), the microspore and the parietal cell development of oslis-l1-2 was normal in WT as described previously. During these stages, the archesporial cells were divided to form primary parietal cells and primary sporogenous cells. The primary parietal cells then divided to form normal epidermis, endothecium, middle layer, and tapetum. During stages 6–8 (Fig. 3g–l), the microspore mother cells develop into tetrads thorough meiotic division. No obvious differences of anther cell wall and microspore mother cell development were observed from stages 5 to 8 between oslis-l1-2 and WT (Fig. 3e–l).

Subsequently, at stage 9, there was still no obvious difference between WT and the oslis-l1-2 anthers, microspores were released from tetrads, tapetal cells had deeply stained cytoplasm and the middle layer was hardly visible, and the middle layer appeared degenerated and almost invisible (Fig. 3m, n). At stage 10, the WT microspores appeared vacuolated and became round in shape (Fig. 3o), whereas the oslis-l1-2 microspores had undersize and irregular vacuolation (Fig. 3p). At stage 11, the WT and the oslis-l1-2 middle layer and endothecium degenerated, and typical falcate (sickle-like shape) pollen grains were formed in WT anthers (Fig. 3q), whereas oslis-l1-2 produced severely abnormal pollen (Fig. 3r). At stage 12, during mature pollen formation, in the WT anther, mature pollen grains were deeply stained with 0.25% toluidine blue, indicating that the WT microspore is full of starch, lipids, and other storage materials (Fig. 3s) that are required for pollen viability and function. However, in the oslis-l1-2 anther, the pollen grains collapsed and consequently almost no normal pollen grains were released (Fig. 3t). During these stages, the middle layer and the tapetum are normally degenerated, and no obvious differences were observed in oslis-l1-2 anthers compared to WT. Taken together, these findings suggest that the low fertility in oslis-l1 might be due to the failure of normal microspore formation.

Next, we compared the anther cell wall development between the WT and oslis-l1-2 at the level of ultra-structure using transmission electron microscopy (TEM). Consistent with the analysis for semi-thin sections of anthers, no obvious morphological changes were observed between anthers of the WT and oslis-l1-2 from stages 9 to 12 (Supplemental Fig. 3a–h). The middle layer and the tapetum are normally degenerated, and the orbicules also normally formed. Furthermore, pollen wall development was not affected in oslis-l1-2 compared with the WT anthers during stages 9–12 (Supplemental Fig. 3i–p). It seems that the pollen wall development is also not affected in oslis-l1-2 mutants.

OsLIS-L1 controls stem growth by regulating cell proliferation

In order to characterize the semi-dwarf phenotype in oslis-l1 mutant plants, we compared the length of five internodes from the main culms between the WT and oslis-l1-2 mutant plants. Interestingly, we found that the length of the first internode under panicle differed the most between WT (32.1 cm) and oslis-l1-2 mutant plants (22.1 cm), but there were no distinct differences between the lengths of other corresponding internodes (Fig. 4a, b; Table 2). Next, plastic transverse and longitudinal sections were employed to histologically analyze the first internode structure of the WT and oslis-l1-2 mutant. The results showed that the stems of oslis-l1-2 mutant plants consisted of the same layers and cells with the WT (Fig. 4c, d) and had the same stem thickness. Similarly, there was not much difference in cell size between the WT and oslis-l1-2 mutant plants (Fig. 4e–g). Thus, we speculated that the decreased height of the stems in oslis-l1 plants was mostly due to decreased cell proliferation rather than cell size in the first internode.

Fig. 4
figure 4

Morphological characterization of OsLIS-L1 on the stem. a The five internodes under panicle from the main culms at mature stage for the following (from left to right): the first internode of oslis-l1-2, the second internode of oslis-l1-2, the third internode of oslis-l1-2, the fourth internode of oslis-l1-2, the fifth internode of oslis-l1-2, and the first internode of WT, the second internode of WT, the third internode of WT, the fourth internode of WT, the fifth internode of WT. Bar 5 cm. b The length of the five internodes under panicle from the main culms between oslis-l1-2 and WT at mature stage. The data correspond to the mean ± standard error from 20 individuals. c WT and d oslis-l1-2 transverse sections of the first internode under panicle from the main culms at heading stage. e WT and f oslis-l1-2 longitudinal sections of the first internode under panicle from the main culms at heading stage. Bars 50 μm in c–f. g Size of parenchyma cells (PC) in the first internode of WT and oslis-l1-2. The data correspond to the mean ± standard error from 15 individuals. All the plants grown in the field under normal field growth conditions

Full size image
Table 2 Comparison of length of five internodes from the main culms between the wild-type and oslis-l1-2 mutant plants under normal field growth conditions (Wuhan, China, 2009)
Full size table

OsLIS-L1 is relatively highly expressed in stem and panicle and its protein localized in the nucleus

We examined the expression pattern of OsLIS-L1 in various tissues and different developmental stages by quantitative RT (qRT)-PCR. As shown in Fig. 5a, the OsLIS-L1 transcript had a relatively higher abundance in stem and panicles at the heading stage, though other tissues at the same stage such as root, leaf, and sheath, also showed weak expression of OsLIS-L1. This expression pattern of OsLIS-L1 supported its function for internode elongation and panicle development in rice. Furthermore, we examined the dynamic OsLIS-L1 expression pattern in various stages of panicle development by qRT-PCR. qRT-PCR was conducted on total RNA from different young panicle sizes (2.0–20.0 cm). OsLIS-L1 transcripts were detected at a low level at P1 (2.0 cm). With the development of the panicle, OsLIS-L1 transcripts increased and stayed relatively high from P2 to P6 (3.5–11.0 cm). Later, OsLIS-L1 transcripts gradually faded out at P7 (16.5 cm) and P8 (20.0 cm) (Fig. 5a).

Fig. 5
figure 5

OsLIS-L1 expression pattern and its protein subcellular localization. a qRT-PCR analysis of OsLIS-L1 expression levels in root, stem, leaf, and sheath at heading stage. Panicles at various developmental stages according to their length at heading stage (P1–P8). P1 approximately 2.0 cm, P2 approximately 3.5 cm, P3 approximately 4.5 cm, P4 approximately 6.5 cm, P5 approximately 8.0 cm, P6 approximately 11.0 cm, P7 approximately 16.5 cm, P8 approximately 20.0 cm, respectively. Ubiquitin5 was used as a control for normalization. The data shown are means ± SE from three biological repeats and two technical repeats. b Localization of Ghd7–CFP fusion protein. c Localization of OsLIS-L1–YFP fusion protein. d Image of the cell in bright field. e Merged image of b–d. Bars 12 μm in b–e

Full size image

Most WD40-repeat proteins in plants are located in the nucleus (Han et al. 2006; Shi et al. 2005). To determine the subcellular localization of OsLIS-L1, OsLIS-L1 coding sequence was amplified and cloned into vector pM999 in appropriate ORF to fuse with YFP. Rice GHD7 is a known nuclear protein (Xue et al. 2008), and GHD7–CFP fusion protein was used as a marker in this experiment. By using an Arabidopsis mesophyll protoplasts transient gene expression system (Yoo et al. 2007), it was confirmed that the OsLIS-L1–YFP fusion protein was co-localized with the reported rice transcription factor GHD7 in the nucleus (Fig. 5b–e), suggesting that OsLIS-L1 is a nuclear protein.

Identification of OsLIS-L1 ortholog in rice

BLAST searches revealed the existence of two homologues in rice: Os01g15020 and Os03g14980. They possess very similar gene structure with OsLIS-L1. The annotated structure of Os03g14980 and OsLIS-L1 comprises of 25 exons and 24 introns, and Os01g15020 comprises 24 exons and 23 introns (Supplemental Fig. 4a). The cDNA of Os01g15020 and Os03g14980 shared 36 and 39% nucleotide identity with OsLIS-L1, respectively; the Os01g15020 polypeptide shared 62% identity and the Os03g14980 polypeptide shared 67% identity with OsLIS-L1. SMART algorithm (http://smart.embl-heidelberg.de/) analysis of their polypeptide sequences indicated that they also contain four recognizable structural domains similar to OsLIS-L1: starting from the N terminus, they are the LisH, CTLH, the WD40-repeat domain in the middle, and the WD40-repeat domain at the C terminus (Supplemental Fig. 4b, c). The differences are that OsLIS-L1 has a five WD40-repeat domain in the middle, but Os01g15020 and Os03g14980 both have a six WD40-repeat domain in the middle, OsLIS-L1 and Os03g14980 both have a four WD40-repeat domain at the C terminus, whereas Os01g15020 has a five WD40-repeat domain at the C terminus. From the analysis above, the three genes were highly conserved during evolutions.

To understand the expression patterns of the three genes, we extracted the expression data from the CREP database (http://crep.ncpgr.cn/; Wang et al. 2010a). The results from the database for 19 tissues of Minghui 63 (O. sativa L. ssp. indica) with two replications indicated that the three genes are expressed at different levels and exhibit a variety of expression patterns (Supplemental Fig. 4d). In general, Os01g15020 and Os03g14980 showed higher expression in the 19 tissues, but OsLIS-L1 showed lowest expression in the 19 tissues. OsLIS-L1 had a relatively higher expression in stem at the heading stage, 4–5 cm young panicle, and endosperm at 7 days after pollination. This expression pattern supported the fact that disruption of OsLIS-L1 leads to semi-dwarf, low fertility, and smaller seeds phenotypes in rice (Fig. 1; Table 1). Os03g14980 had high expression in the 19 tissues, with the highest expression level in endosperm at 21 days after pollination. Os01g15020 showed a relatively higher expression in endosperm at 7 days after pollination and in flag leaf at 14 days after heading. Interestingly, all the three genes had relatively higher expression in endosperm at various stages, suggesting that they may be involved in the development of endosperm.

Discussion

Gβ-protein homology domain plays an essential role for OsLIS-L1 function

As our study, the Gβ-protein homology domain was either defected or disrupted in the two allelic mutants, while the other three domains, LisH, CTLH, and five WD40-repeat domain, were normally transcribed. This result suggested that Gβ-protein homology domain is indispensable for OsLIS-L1 function in fertility and the first internode elongation. In Arabidopsis, a similar phenomenon was found in protein COP1 (McNellis et al. 1994). COP1 contains three recognizable structural domains: a zinc-binding motif in the N terminus, a putative coiled-coil region, and the domain with multiple WD40 repeats homologous to Gβ in the C terminus. In total 17 available COP1 recessive mutants were classified into three groups according to the severity of the phenotype. All three allelic cop1 mutants with mutations in Gβ-protein domain showed a lethal phenotype, suggesting that the integrity of Gβ-protein domain is critical for COP1 functions. In addition, mutations in the region between the helical domain and the Gβ-protein domain also seriously compromised the COP1 activity (McNellis et al. 1994). Our result in rice and COP1 result in Arabidopsis suggested a conservative function of Gβ-protein domain with WD40 repeats in plants. The Gβ-protein domain may affect OsLIS-L1 function in several possible ways. One possibility is that the loss of or defect in Gβ-protein domain altered the OsLIS-L1 protein configuration, thus decreased the protein stability in plant cell. It is also possible that Gβ-protein domain acts as a functional domain for certain biochemical activities, because several studies have shown that repeated WD40 motifs in Gβ-protein domain may be involved in protein–protein interactions and serve as platforms for the assembly of protein complexes (Smith et al. 1999; Wang et al. 2001). Because of the lack of mutations in the LISH domain and five WD40-repeat domain, we could not study the possible functions of these two domains in OsLIS-L1 at this moment.

OsLIS-L1 profile indicates its function

The spatial and temporal expression pattern of a gene can be indicative of its functional relevance. A wide expression pattern of OsLIS-L1 in different tissues and stages suggests that the gene may play crucial roles at multiple developmental stages. However, a relatively higher transcript level was detected in stem and panicles in mature rice plants by qRT-PCR (Fig. 5a). The plant height of oslis-l1-1 and oslis-l1-2 was reduced to 56.0 and 68.9% of WT, respectively (Table 1). The decreased seed-setting rate in both oslis-l1 mutants also coincided with the high OsLIS-L1 expression level in panicle. Moreover, the transcript of OsLIS-L1 was greatest through the panicle developmental stage P2–P6, which approximately represented stage 6–10. This expression pattern also supports our histological analysis result showing the failure of normal microspores formation in oslis-l1-2. The expression pattern of OsLIS-L1 together with the phenotype observed in the oslis-l1-1 and oslis-l1-2 mutants indicated that the OsLIS-L1 is required for male gametophyte generation.

OsLIS-L1 functions in male gametophyte formation in rice

In flowering plants, the male gametophyte development is very essential for sexual reproduction. The oslis-l1-1 and oslis-l1-2 mutant plants both exhibit very low fertility phenotypes (Table 1), and almost all the pollen lacked starch due to failure of male gametophyte development. From male meiocyte meiosis analysis in the WT and oslis-l1-2 mutant, there were no obvious differences during the meiosis process (Supplemental Fig. 2), suggesting that OsLIS-L1 did not affect the meiosis in pollen mother cells. Based on the histological examination and TEM results of anther development in the WT and the oslis-l1-2 mutant, there were no obvious differences in the four somatic layers and pollen wall during the male gametophyte generation (Fig. 3; Supplemental Fig. 3), indicating that OsLIS-L1 did not function in anther wall and pollen wall development and that the reduced male fertility phenotype was not due to defects in anther wall and pollen wall development. The obvious defects were clearly observed at the later stages of the anther development. The oslis-l1 pollen grains were severely collapsed and consequently no viable pollen grain were released (Fig. 3t). This is a different situation from that frequently reported in the literature in which male sterility is usually associated with defects in tapetum development or the meiotic process (Jung et al. 2006; Li et al. 2006, 2011b; Yuan et al. 2009). Our observation in oslis-l1-2 is different from the other reported male sterile mutants in rice. For example, rice Undeveloped Tapetum1 (Udt1) begins to function at the meiosis stage for tapetum development and pollen mother cell meiosis. Moreover, this gene is needed to degenerate the middle layer (Jung et al. 2005). Mutation of the rice gene PAIR3 exhibits male sterile phenotype mostly due to abnormal male gametophyte formation and lack of bivalent formation in meiosis, but the cells of the parietal layer including the tapetum appeared to be normal (Yuan et al. 2009). The rice (Oryza sativa) APOPTOSIS INHIBITOR5 (API5) regulates the development of male gametophytes through programmed cell death (PCD) during tapetum degeneration (Li et al. 2011b). Combining our present data, we speculate OsLIS-L1 may be involved in the other developmental process for microsporogenesis. Thus, the functional details of OsLIS-L1 on microsporogenesis still needs to be elucidated in future investigation.

The role of OsLIS-L1 in the development of the first internode in rice

Plant height is an important agronomic trait that is related to lodging resistance and fertilizer endurance (Shan et al. 2009). Here, we identified OsLIS-L1 was required for the first internode elongation but not for the other internodes of the culm in rice (Fig. 4a, b; Table 2). The semi-dwarf mutant phenotype of oslis-l1 is quite different with other reporters in rice. For example, mutation of OsH15, OsDHT8, and SPD6 exhibit dwarf phenotype mostly due to the decrease the length of the second, third, and fourth internodes, but the length of first internode was not distinctly different from that of WT plant (Sato et al. 1999; Shan et al. 2009; Wei et al. 2010). In some cases, the length of all the internodes were shortened, led to severe dwarf phenotype in brd1 mutant plants (Mori et al. 2002). The effect on different internodes may be a hint that these genes are involved in different pathways regulating vegetative growth in rice.

Recent studies revealed that defects in cell division and cell expansion are two major reasons for dwarf phenotype in plants (Kumar et al. 2010). Rice height was reported to be affected by DTH8 or OsHAP complexes mainly through the regulation of internode cell elongation (Wei et al. 2010). In OsAPC6 mutant plant, both the cell size and cell number were reduced (Kumar et al. 2010). Given that the oslis-l1-2 first internode cell size was kept the same as that in WT (Fig. 4c–g), we speculated that the reduced plant height in oslis-l1-2 plants was mostly due to a decreased cell proliferation. Sato et al. (1999) also observed reduction in the number of cells in internode in OSH15 mutant in rice due to decreased frequency of cell proliferation in the intercalary meristem. A challenge for future studies will be to elucidate the molecular mechanisms controlling the cell proliferation in the fist internode in rice.