Recent studies have shown that F-box proteins constitute a large family in eukaryotes, and play pivotal roles in regulating various developmental processes in plants. However, their functions in monocots are still obscure. In this study, we characterized a recessive mutant dwarf and deformed flower 1-1 (ddf1-1) in Oryza sativa (rice). The mutant is abnormal in both vegetative and reproductive development, with significant size reduction in all organs except the spikelet. DDF1 controls organ size by regulating both cell division and cell expansion. In the ddf1-1 spikelet, the specification of floral organs in whorls 2 and 3 is altered, with most lodicules and stamens being transformed into glume-like organs and pistil-like organs, respectively, but the specification of lemma/palea and pistil in whorls 1 and 4 is not affected. DDF1 encodes an F-box protein anchored in the nucleolus, and is expressed in almost all vegetative and reproductive tissues. Consistent with the mutant floral phenotype, DDF1 positively regulates B-class genes OsMADS4 and OsMADS16, and negatively regulates pistil specification gene DL. In addition, DDF1 also negatively regulates the Arabidopsis LFY ortholog APO2, implying a functional connection between DDF1 and APO2. Collectively, these results revealed that DDF1, as a newly identified F-box gene, is a crucial genetic factor with pleiotropic functions for both vegetative growth and floral organ specification in rice. These findings provide additional insights into the molecular mechanism controlling monocot vegetative and reproductive development.
F-box proteins are characterized by a conserved ‘F-box’ motif consisting of approximately 50 amino acids at the N terminus (Bai et al., 1994; Kipreos and Pagano, 2000). Most F-box proteins reported so far have been identified as components of the Skp, Cullin, F-box-containing (SCF) complex that functions in defining substrate specificity in the ubiquitin–proteasome pathway. Several other F-box proteins with function in non-SCF complex have also been found (Clifford et al., 2000; Galan et al., 2001). F-box genes belong to a large family in plants, containing 42 groups, each of which has a distinct domain organization (Xu et al., 2009). Studies have indicated that F-box genes play crucial roles in a number of biological processes in plants, including flower development, self-incompatibility, hormone response, circadian clock and photomorphogenesis, senescence, defense response, embryogenesis and seedling development (Moon et al., 2004; Lechner et al., 2006; Jain et al., 2007; Schumann et al., 2011). Until now, however, our knowledge about the functions of F-box genes in plants has remained very limited. Fewer than 5% in Arabidopsis F-box proteins (Schumann et al., 2011) and only a few proteins in Oryza sativa (rice; Gomi et al., 2004; Ikeda et al., 2007; Jain et al., 2007) have been characterized for their functions.
In addition to the ABC genes, several genes that regulate the expression of ABC genes in Arabidopsis have been identified, including unusual floral organs (UFO; Bowman et al., 1992; Sakai et al., 1995) and Skp1-like 1 (ASK1; Zhao et al., 1999). UFO encodes an F-box protein that mainly regulates B-class genes (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). The loss of UFO function results in homeotic transformation of floral organs in the second and third whorls, without greatly affecting the first and fourth whorls (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). ASK1 shows somewhat similar effects to those of UFO on floral organ specification in terms of their mutant phenotypes. The proteins of these two genes interact to control floral organ identity in whorls 2 and 3. Interestingly, ASK1 is quite similar in sequence to the Saccharomyces cerevisiae (yeast) SKP1 protein, whereas UFO contains an F-box, a motif known to interact with SKP1 in yeast (Zhao et al., 1999). Although these two genes act together in flower development, their functions in vegetative growth are quite different. A mutation in ASK1 affects both vegetative growth and reproductive development, whereas a mutation in UFO does not influence vegetative growth (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). In the ask1-1 mutant, rosette leaves are smaller and internodes in the floral stem are shorter than the wild type (Zhao et al., 1999). Compared with Arabidopsis, much less is known about the regulation of floral organ identity genes in rice (Ikeda et al., 2007; Horigome et al., 2009; Xiao et al., 2009; Duan et al., 2010).
Vegetative growth and reproductive development are both important for high yield in cereal crops because plant architecture is the foundation of reproductive development, and floral development directly influences the quantity and quality of cereal grains. To date, only a few F-box genes with pleiotropic effects on vegetative growth and reproductive development have been identified. In rice, only aberrant panicle organization 1 (APO1), the ortholog of Arabidopsis UFO, has been cloned and functionally characterized for its function in leaf emergence, panicle architecture and floral organ identity (Ikeda et al., 2005, 2007; Ikeda-Kawakatsu et al., 2009).
Here, we report a new pleiotropic F-box gene DDF1 in rice. The ddf1-1 plant is dramatically dwarfed, with much smaller vegetative organs and panicles, and its spikelet exhibits obvious defects in lodicule and stamen specification, resembling the defects found in Arabidopsis ufo mutants. DDF1 is expressed throughout the whole plant, and can regulate the expression of floral homeotic genes involved in lodicule and stamen specification. This work demonstrates that DDF1 is a critical gene for whole-plant development in rice.
ddf1-1 exhibits significant size reduction in the whole plant, except the spikelet
The mutant ddf1-1 was discovered from a breeding population. The ddf1-1 plant is much shorter and thinner than the wild type, and this difference is distinct at the early seedling stage and becomes more apparent as the plants develop. The final plant height of the mutant reaches only about half of that of the wild type (Figure 1a–e and Figure S1). All vegetative organs in ddf1-1, including stems, internodes, leaves and roots (both fibrous and lateral), are significantly shorter and thinner than in the wild type (Figure 1f–j and Figures S2 and S3). In addition, the ddf1-1 panicle is also notably shorter and smaller, consisting of fewer primary and secondary branches and spikelets (Figure 1e,f; Figure S4). These phenotypes suggest that plant growth in ddf1-1 is seriously stunted. However, the size of the spikelet, the numbers of various vegetative organs (e.g. internodes and tillers; Figure S5) and the time from sowing to heading (Figure 1e) remain unaffected in ddf1-1.
Cell division and cell expansion are inhibited in ddf1-1
The size of an organ depends on both cell number and cell size, which are related to cell division and cell expansion, respectively. A decrease in either cell number or cell size, or both, can result in a decrease of organ size. To find out the causes of organ size reduction in ddf1-1, we conducted microscopic observation on the internal structure of vegetative organs of the mutant by paraffin section. The results showed that in the ddf1-1 leaf, parenchyma cells (Figure 2a,b) and epidermal cells (Figure 2c,d) are significantly reduced in size, mesophyll cells and abaxial sclerenchyma cells are notably reduced in number, and the abaxial sclerenchyma cells are even completely absent in some cases (Figure 2a,b). In parallel with the reduction in cell size and cell number, the size of leaf veins in ddf1-1 is reduced and the number of leaf veins is greatly decreased (to 36–44, with an average of approximately 40, compared with 60–76, with an average of approximately 68, in the wild type; Figure 2a,b).
Significant decreases in cell size and cell number are also observed in the ddf1-1 root (Figure 2e–h) and stem (Figure 2i–n). Similar to the case of leaf veins, the number and size of vascular bundles in the ddf1-1 stem are also reduced (to 24–28, with an average of approximately 26, compared with 30–36, with an average of approximately 33, in the wild type; Figure 2k,l). It is noticeable that both the root apical meristem (RAM) region (Figure 2e,g) and the shoot apical meristem (SAM) region (Figure 2i,j) in ddf1-1 are significantly smaller than in the wild type. As all underground organs (roots) and above-ground organs (stems, leaves and flowers) are established from RAM and SAM by cell division, respectively, the decrease of RAM and SAM regions might be the direct cause of the dwarfism and thin stature of ddf1-1.
Based on the above observations, we conclude that the overall dwarfism and thin stature of ddf1-1 is caused by the reduction of both cell number and cell size, suggesting that both cell division and cell expansion are suppressed in ddf1-1.
The expression of cell division/expansion-related genes is downregulated in ddf1-1
The above results of histological analysis suggest that DDF1 probably plays essential roles in plant growth by affecting both cell division and cell expansion in rice. Many genes related to cell division or cell expansion have been known, such as xyloglucan endotransglycosylase-related genes (XTR), α-Expansin (EXPA), Ras-like nuclear GTPase (RAN) and Histone (Tobina et al., 2003; Chen et al., 2011). To better understand the results of histological observation and the role of DDF1 in plant growth, we examined the expression of two genes that are mainly related to cell division (Histone H4 and OsRAN2) and two genes that are mainly related to cell elongation (OsXTR2 and OsEXPA2; Table S1) in ddf1-1 young leaves by quantitative RT-PCR (qRT-PCR). The results showed that the expression levels of all the genes tested are dramatically decreased in ddf1-1 (Figure 3). This is compatible with the morphological phenotype of ddf1-1 and the results of histological analysis, suggesting that DDF1 probably affects cell size and cell number by regulating the expression of cell division/expansion-related genes, so as to direct vegetative organs developing into their normal sizes.
ddf1-1 also shows significant structural abnormalities in florets
The ddf1-1 floret size is normal, but its floral morphology is abnormal (Figure 4a). A wild-type floret comprises four whorls of floral organs: a lemma and a palea (whorl 1), two lodicules (whorl 2), six stamens (whorl 3) and a carpel (whorl 4; Figure 4b,i). The ddf1-1 floret also has four whorls, but its floral organs all exhibit morphological abnormalities to different degrees. The ddf1-1 lemma and palea are both somewhat distorted in shape and cannot close together tightly (Figure 4a). Inside them, the two lodicules in whorl 2 are thoroughly transformed into glume-like organs (GLOs; Figure 4c–f). In some cases there are one or two extra GLOs generated in whorl 2 (Figure 4j,k; Table 1). In whorl 3, the six stamens are transformed into a variable number (usually 4–6) of pistil-like organs (PLOs; Figure 4c–f; Table 1), with occasionally one or two stamens remaining at the site (Figure 4f,k; Table 1). The PLOs are obviously aberrant, and do not have a uniform appearance (Figure 4d,f,j,k). Rarely, a chimera of stamen and pistil tissues can be found in whorl 3 (Figure 4f). Neither the remaining stamens nor the stamen–pistil chimera can produce fertile pollen grains (Figure 4g,h). Therefore, nearly all ddf1-1 florets are male sterile. Nevertheless, the pistil in whorl 4 still has normal function in most ddf1-1 florets, which can produce seeds after being pollinated by wild-type pollens, although the pistil may contain a few (usually one or two) additional styles in some cases (Figure 4f,j).
Table 1. Numbers of floral organs in whorls 2 and 3 in ddf1-1 floretsa
aA total of 236 ddf1-1 florets were observed.
Transverse section analysis of ddf1-1 florets confirmed the results of morphological observation (Figure 4i–k). In particular, it clearly revealed the structural abnormalities in ddf1-1 lemma and palea. The ddf1-1 lemma and palea are thinner than the wild-type ones, but they possess more vascular bundles (there are five and three vascular bundles in the wild-type lemma and palea, respectively, whereas there are seven and five vascular bundles in the ddf1-1 lemma and palea, respectively; Figure 4i–k). These structural abnormalities are likely to cause the distorted shapes of ddf1-1 lemma and palea (Figure 4j,k), which in turn result in an imperfect match between the lemma and palea, leading to a partly opened floret in ddf1-1 (Figure 4a).
To further clarify the morphogenesis of ddf1-1 florets, we examined young spikelets of both the mutant and the wild type at early developmental stages by scanning electron microscopy (Figure 5). In a wild-type floret, floral organ primordia emerge from the floral primordium, and develop into the correct organs from the outermost whorl (lemma and palea) to the innermost whorl (carpel; Figures 5a–d). In the ddf1-1 floret, the primordia of the four-whorl floral organs can also emerge in the correct sequence at the right positions. However, although the primordia in whorls 1 and 4 can correctly develop into lemma/palea (Figure 5e–i) and pistil (Figure 5g–i), respectively, the primordia in whorls 2 and 3 have changed identities, developing into variable numbers of GLOs (Figure 5f–i) and PLOs (Figure 5g–i), respectively.
The above observations indicate that in typical ddf1-1 florets, the identities of lodicule and stamen are seriously altered, and these organs have been transformed into variable numbers of GLOs and PLOs, respectively, whereas the identities of lemma/palea and pistil are hardly affected, suggesting that the DDF1 gene is critical for the specification of lodicule and stamen in rice.
The ddf1-1 phenotype is caused by a single nucleotide substitution
We had previously mapped the DDF1 gene to a 165-kb region on the long arm of chromosome 6 (Li et al., 2011). In this study, using a larger F2 population, and several new SSR and InDel markers (Table S2) that we have developed ourselves, we further narrowed down the position of DDF1 to a 45-kb interval, in which there were eight annotated open reading frames (Figure 6a). Sequence analysis revealed that one of the eight genes had a single base transversion (A→T) inside the third exon, resulting in a conversion of Arg to Tyr at the 251st amino acid in the encoded polypeptide (Figure 6b); no base alteration was found in the other seven genes. By examining this base in six other rice varieties, we found that they all belonged to the wild-type genotype (Figure 6b). In addition, we found that this base happened to be located in a restriction site of enzyme AlwNI, and that its transversion mutation removed the restriction in ddf1-1. Hence, we could conveniently use AlwNI to detect the single nucleotide polymorphism (SNP) at the mutated base (A versus T) in the F2 population. The result showed that this SNP site completely co-segregated with the mutant trait (Figure 6c). The above results strongly suggest that this single base transversion mutation was likely to be the molecular cause of ddf1-1 mutant phenotypes. Therefore, we took this gene as the candidate of DDF1.
To validate the candidate gene, we constructed a plasmid pCAMBIA1300-DDF1 carrying an 8869-bp wild-type genomic DNA, which covered the entire DDF1 candidate gene and its upstream promoter region. By introducing the plasmid into ddf1-1, all the mutant phenotypes (including dwarf stature, smaller stems, leaves, roots and panicles, and aberrant spikelet) were rescued (Figure 6d,e). Thus, the DDF1 gene was confirmed.
DDF1 encodes an LRR-type F-box protein, anchored in nucleolus
DDF1 covers approximately 4.5 kb of the genome, comprising five exons and four introns (Figure 6b). A BLAST search and protein function analysis (http://smart.embl-heidelberg.de) revealed that DDF1 encodes a putative leucine-rich repeat (LRR)-type F-box protein consisting of 519 amino acids (Figure S6). Further analysis identified five other LRR-type F-box proteins with surprisingly high similarities to DDF1 in rice. Among them, Os06g04690 and Os06g04980 have an identity of approximately 96 and 90% to DDF1, respectively, and their genes are both located closely to DDF1 in the genome, implying that there might be tandem or segmental duplication events occurring in that region (Xu et al., 2009). In addition, DDF1 also showed high similarity to HSLF2-S1, an LRR-type F-box protein of Hordeum bulbosum, which is possibly related to self-incompatibility (Kakeda, 2009; Figure S6).
To understand the evolution of DDF1, we constructed a phylogenetic tree based on the full amino acid sequences of DDF1, HSLF2-S1 and another 16 F-box proteins involved in the regulation of flower development, self-incompatibility, hormone response, circadian clock and photomorphogenesis. The results indicated that DDF1 constituted a clade with HSLF2-S1, whereas the other sixteen genes, including APO, UFO and FIM (Fimbriata), which act on flower development, were separate from this clade (Figure 7a), suggesting that DDF1 is possibly an ortholog of Hordeum HSLF2-S1.
Moreover, we constructed a DDF1-GFP fusion protein expression vector to transform young onion epidermal cells with Agrobacterium tumefaciens. The result indicated that the DDF1-GFP fusion protein is expressed specifically in nucleus (Figure 7a), suggesting that DDF1 is a nuclear protein.
The expression pattern of DDF1 is consistent with the phenotypes of ddf1-1
To reveal the expression pattern of DDF1 we constructed a fusion of DDF1 promoter and GUS (β-glucuronidase) reporter gene, and transferred it into rice via A. tumefaciens. A histochemical examination of GUS activity in transgenic plants showed that DDF1 is expressed almost in all tissues throughout the lifetime of the plant (Figure 8a), which is in agreement with the overall plant mutant phenotype of ddf1-1. During early vegetative growth, DDF1 expression is strong in all vegetative organs, including root (Figure 8aA), leaf (Figure 8aB), sheath (Figure 8aC) and young stem (Figure 8aD), but is weak in the stem apical meristem (Figure 8aD). At the jointing stage, DDF1 expression is strong in node regions and medium in elongating internodes, but is very weak or even null in developed internodes (Figures 8aE–G). This is compatible with the inference obtained above that DDF1 plays important roles in both cell proliferation and cell expansion.
The spatial and temporal expression pattern of DDF1 in reproductive organs also appears to be consistent with the phenotypes of ddf1-1. In young panicles, DDF1 expression is strong in both spikelets (especially in lemma and palea) and branches (Figures 8aG,H). As the development of the spikelet proceeds, DDF1 expression in lemma and palea gradually declines, whereas that in stamens gradually increases (Figures 8aI,J). In mature flowers, DDF1 expression is mainly concentrated in stamens, and is also found in lodicules, to a considerable degree, but is weak or even absent in lemma, palea, filaments and pistil (Figures 8aI–L). We also employed the in situ hybridization to analyze the expression pattern of DDF1 in young spikelets at early developmental stages (Figure 8b). DDF1 expression was detected in floral meristems and the primordia of all floral organs, including lemma, palea, lodicule, stamen and carpel, and was especially strong in stamen primordia and developing stamens (Figure 8bA–D). These results were largely in agreement with those of the GUS analysis.
DDF1 regulates the expression of several floral organ specification genes
The floral phenotype of ddf1-1 and the expression pattern of DDF1 in the floret strongly suggest that DDF1 possibly regulates the expression of floral organ identity genes during floral organ differentiation. To address this question, we analyzed the transcription levels of nine floral organ identity genes by qRT-PCR, including three AP1-like genes (OsMADS14, OsMADS15 and OsMADS18), three B-class genes (OsMADS2, OsMADS4 and OsMADS16), two C-class genes (OsMADS3 and OsMADS58) and one carpel specification gene (DL), in young panicles of 2–20 mm in length (i.e. within stage 7 of inflorescence, the period of floral organ differentiation; Ikeda et al., 2004) from ddf1-1 and wild-type plants. The results showed that the two B-class genes OsMADS4 and OsMADS16 are significantly downregulated in ddf1-1, whereas DL is significantly upregulated (Figure 9a). Among the AP1-like genes, OsMADS14 and OsMADS15 appeared to be slightly upregulated and downregulated, respectively, whereas OsMADS18 remained unchanged.
To validate the qRT-PCR results and further discern the relationships between DDF1 and floral organ specification genes, we examined the temporal and spatial expression patterns of two downregulated genes (OsMADS4 and OsMADS16) and one upregulated gene (DL) by in situ RNA hybridization in wild-type and ddf1-1 plants (Figures 9b–s).
The two B-class genes OsMADS4 and OsMADS16 showed similar expression patterns in the wild-type spikelet. They were both expressed in the primordia of lodicule and stamen, as expected, and the expression was detectable throughout the development of the two floral organs (Figures 9b–d,h–j; Nagasawa et al., 2003). In the ddf1-1 spikelet, the two genes were also expressed in the whorls in which they were expected, but their expression levels were significantly reduced. Counter to expectation, no expression was detected in the initiating regions of whorls 2 and 3 where GLOs and PLOs should develop, respectively (Figure 9e,k), and only very weak expression was observed during the development of GLOs and PLOs (Figure 9f,g,l,m).
DL plays a specific role in carpel development (Yamaguchi et al., 2004). In the wild-type floret, DL expression was first detected in the region where carpel primordia should initiate (carpel anlagen). Afterwards, the expression domain of DL was restricted to the carpel anlagen and carpel primordia (Figures 9n–p; Yamaguchi et al., 2004). In the ddf1-1 floret, the DL expression domain was expanded to the primordia in whorl 3, where the ectopic PLOs should develop, and the expression of DL continued during the development of both PLOs and the original carpel (Figure 9s). The expansion of the expression domain might explain the higher DL expression level in ddf1-1 young panicles than in the wild type.
Taken together, the results of qRT-PCR analysis and in situ RNA hybridization analysis were consistent, suggesting that DDF1 positively regulates the expression of OsMADS4 and OsMADS16, and negatively regulates the expression of DL during lodicule and stamen specification.
DDF1 negatively controls APO2 expression
Aberrant panicle organization 2 (APO2) is the ortholog of Arabidopsis LEAFY (LFY) in rice. APO2 shows a similar expression pattern to that of DDF1 in developing spikelets; its loss-of-function mutant also exhibits somewhat similar phenotypes to those of ddf1-1 (Ikeda-Kawakatsu et al., 2012). These similarities between APO2 and DDF1 imply that the two genes might be closely related in function. To understand the relationship between them, we examined APO2 expression in young ddf1-1 panicles by qRT-PCR. The result showed that APO2 expression is enhanced in ddf1-1 compared with that in the wild type (Figure S7), suggesting that DDF1 negatively controls APO2 transcription.
DDF1 is generally required for normal organ size in rice
Cell division and cell expansion are the foundations of plant growth and development, and enable the plant and various organs to develop into the correct sizes. Recent studies have uncovered some key regulatory genes (growth-promoting and growth-restricting factors) and pathways that affect plant organ size by altering either cell number or cell size, or both (Krizek, 2009; Guo et al., 2010; Moyroud et al., 2010). However, the intrinsic mechanisms responsible for organ size variation are yet to be well understood (Guo et al., 2010).
In this study, we have identified an F-box gene DDF1 that appears to play a general role in the control of organ size in rice. In the mutant ddf1-1, all vegetative organs exhibit a significant reduction in size (Figure 1). Inside the organs, both cell number and cell size are apparently reduced (Figure 2), suggesting that both cell division and cell expansion are inhibited in the mutant. Consistent with this, the expression levels of four genes, Histone H4, OsRAN2, OsXTR2 and OsEXPA2, which are essential for cell division and cell expansion in rice (Cho and Kende, 1997; Huang et al., 2000; Uozu et al., 2000; Jan et al., 2004; Yokoyama et al., 2004; Mu et al., 2005; Zang et al., 2010; Chen et al., 2011), are all dramatically downregulated in ddf1-1 (Figure 3). In addition, the expression of DDF1 in culms, as revealed by GUS detection, is concentrated in the cell division and elongation zone of internodes, but is very weak or even undetectable in the mature zone of internodes (Figure 8aD–F). These results strongly suggest that DDF1 positively controls both cell division and cell expansion, and controls organ size by affecting cell division and cell expansion concurrently.
In addition to vegetative organs, DDF1 also affects the size of the inflorescence. Similar to the situation of vegetative organs, the size of panicle in ddf1-1 is also greatly reduced, with significantly fewer primary branches and spikelets (Figure 1e,f and Figure S4). Studies have shown that the number of primary branches in rice is determined by the initial size of the reproductive shoot apex, and that the number of florets per primary branch is determined by cell division activity in the subsequent growth of the apex (Mu et al., 2005). In ddf1-1, the initial size of reproductive shoot apex is obviously smaller than that in the wild type (Figure 2). This is consistent with the phenotype of smaller panicles in ddf1-1.
DDF1 is also required for normal flower development in rice
Floral ABC genes are specifically expressed in the floral meristem and in the primordia of floral organs generated from the floral meristem to elaborate floral organ structures. Proper regulation of floral homeotic gene expression ensures the development of floral organs in the correct number, type and spatial arrangement (Liu and Mara, 2010). To date, several genes that control the development of floral organs via regulating the expression of ABC genes in specific domains have been identified (Bowman et al., 1992; Sakai et al., 1995; Yamaguchi et al., 2004; Liu and Mara, 2010). However, knowledge on the regulation of floral organ identity genes in rice remains very limited.
In this study, we have identified a new F-box gene DDF1 that plays a crucial role in the specification of floral organs in whorls 2 and 3. In most ddf1-1 spikelets, lodicules and stamens in whorls 2 and 3 are transformed into GLOs and PLOs, respectively, whereas the identity specification of lemma/palea in whorl 1 and pistil in whorl 4 is hardly affected (Figure 4). According to the ABC model, the floral phenotype of ddf1-1 implies a reduction of B function and an increase of C function in the mutant florets, suggesting that DDF1 positively regulates B-function genes but negatively regulates C-function genes. This was verified by the results of qRT-PCR and in situ hybridization analyses, in which the expression of B-class genes OsMADS4 and OsMADS16 was significantly reduced, whereas the expression of DL, a specific gene with strong C function, predominantly determining carpel identity in rice (Yamaguchi et al., 2004), was significantly increased, and its expression domain was also expanded from the carpel primordium in whorl 4 to the stamen primordia in whorl 3, where the PLOs subsequently emerged (Figure 9).
Taken together, we have reason to infer that DDF1 affects the specification of floral organs in whorls 2 and 3 by regulating B-function and C-function genes (i.e. enhancing the former, but suppressing the latter). Therefore, DDF1 is also crucial for flower development in rice.
The possible molecular mechanism of DDF1 function
The F-box gene UFO plays a critical role in flower development in Arabidopsis, and controls the specification of floral organs in whorls 2 and 3 by positively regulating the expression of B-class genes (Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). This is quite similar to the function of DDF1 in rice. Studies indicated that UFO activates B-class genes by cooperating with LFY, which encodes a plant-specific transcriptional factor (Weigel et al., 1992; Levin and Meyerowitz, 1995; Honma and Goto, 2000). More recently, it was revealed that UFO functions as a co-factor for LFY-induced transcription of B-class gene AP3, and regulates the activity of LFY in a proteasome-dependent manner (Chae et al., 2008). It has been proposed that the molecular interaction between UFO and LFY in Arabidopsis might represent a general mechanism in plants, as homologs of UFO and LFY have been identified in a wide array of plant species (Chae et al., 2008), and some studies have shown that UFO or its homologs are activators of LFY proteins (Lippman et al., 2008; Souer et al., 2008).
APO1 and APO2 are the orthologs of UFO and LFY in rice, respectively. The mechanism of UFO–LFY interaction has proven to be conserved in rice, although the functions of APO1 and APO2 have partially diverged from those of UFO and LFY (Ikeda et al., 2007; Ikeda-Kawakatsu et al., 2012). APO1 and APO2 both affect vegetative and reproductive development simultaneously. Their loss-of-function mutants display similar phenotypes in plant growth and inflorescence differentiation, namely, shortened plastochron, more leaves and a smaller inflorescence, with fewer spikelets (Ikeda et al., 2005, 2007; Ikeda-Kawakatsu et al., 2012). These similarities between apo1 and apo2 suggest a close relationship between APO1 and APO2. It has been demonstrated that APO2 negatively controls APO1 transcription, and that APO1 and APO2 act cooperatively to control panicle development (Ikeda-Kawakatsu et al., 2012).
Nevertheless, the phenotypes of apo1 and apo2 spikelets are quite different. In apo1, stamens are transformed into lodicules, whereas in apo2, lodicules are transformed into glume-like organs and stamens are transformed into carples and lodicules (Ikeda et al., 2005, 2007; Ikeda-Kawakatsu et al., 2012). In contrast, the apo2 spikelet appears quite similar to that of ddf1-1. Moreover, the APO2 expression pattern in floral meristems and developing floral organ primordia is similar to that of DDF1 as well. These similarities imply that there might be a close relationship between APO2 and DDF1. The functional similarities between UFO and DDF1 mentioned above also provide a clue for the possible connection between APO2 and DDF1. This is preliminarily supported by the result of qRT-PCR analysis (Figure S7), in which DDF1 appears to negatively control APO2 transcription.
In addition, studies have suggested that Arabidopsis LFY and its homologs in other plants have reserved an ancestral role in regulating cell division and arrangement (Moyroud et al., 2010). This has proven true for APO2 in rice, which decides leaf size by regulating the cell number (Ikeda-Kawakatsu et al., 2012). APO1 has also been found to affect cell division, which promotes cell proliferation in the meristem (Ikeda-Kawakatsu et al., 2009). This might result from the interaction between APO1 and APO2, explaining how APO1 and APO2 cooperatively control panicle development in rice (Ikeda-Kawakatsu et al., 2012). Likewise, DDF1 controls organ size via affecting cell division and expansion, and the phenotype of reduced leaf and panicle sizes in apo2 partially resembles that in ddf1-1. The analogue of the relationship between DDF1 and APO2 with that between APO1 and APO2 on the genetic control of cell division and plant development further suggests that DDF1 might also function in connection with APO2 as APO1. In other words, both APO1 and DDF1 might be required to activate APO2 at different sites and times during plant development. Thus, ddf1-1 and apo1 might each reveal a part of the APO2 function.
In conclusion, as a newly identified F-box gene in rice, DDF1 is crucial for vegetative growth and reproductive development. Although its precise mechanism is still unclear, we still have reasons to speculate that DDF1 might act as a co-factor for APO2-induced transcription of floral homeotic genes (e.g. OsMADS4, OsMADS16 and DL), and regulate the activity of APO2 in a proteasome-dependent manner. It will be interesting to further investigate whether and how DDF1 acts as a co-factor to activate APO2 at different sites and times during plant development in rice.
The ddf1-1 mutant was discovered in rice breeding materials (Oryza sativa sub. indica cultivar). As the mutant almost couldn’t set seed, the mutant ddf1-1 was maintained through heterozygotes, which were identified using their progeny lines. The homozygous normal individuals in the progeny lines of ddf1-1 heterozygotes were used as the wild-type control.
Young spikelet, young root, the top-second internode and leaf were observed by paraffin section. The materials were fixed with a formalin: acetic acid: alcohol solution at 4°C overnight, dehydrated in an ethanol series, completely stained in phenodin, rinsed to transparency in xylene, embedded in paraffin, sliced into 8–10-μm sections and attached to microscope slides, and finally observed and photographed using a light microscope (VANOX AH3; Olympus, http://www.olympus-global.com).
Scanning electron microscopy observation
Young panicles (1–10 mm) were observed by scanning electron microscopy (SEM; Duan et al., 2010). Panicles were fixed in 2.5% glutaric dialdehyde and washed with a sodium phosphate buffer (0.1 m, pH 7.0), further fixed in 1% osmic acid for 1–2 h, and again washed with the sodium phosphate buffer (0.1 m, pH 7.0), dehydrated with an ethanol series, incubated in ethanol-isoamyl acetate and then in isoamyl acetate, dried, mounted and coated with gold, and finally observed with an XL30 ESEM scanning electronic microscope (Phillips, http://www.philips.com).
Positional cloning of DDF1
The ddf1 locus was previously mapped onto a 165-kb region between two simple sequence repeats markers, DF2 and DF5 (Li et al., 2011). An F2 population of 2525 mutants was further used to narrow down the ddf1 locus onto a 45-kb region. The candidate gene was identified by sequencing analysis. An 8869-bp DDF1 genomic DNA fragment covering a 3730-bp promoter and a 609-bp 3′ region was subcloned into a binary vector pCAMBIA1300 and introduced into ddf1-1 mutant embryonic calli by Agrobacterium tumefaciens-mediated transformation (Hiei et al., 1994).
Promoter activity analysis
A 3724-bp promoter located upstream of the coding region was fused to the GUS reporter gene with the nopaline synthase terminator and cloned into pCAMBIA1391Z (DDF1-GUS). The DDF1-GUS plasmid was introduced into EHA105 and further into japonica cultivar Zhonghua-15 to generate 46 independent transgenic lines. A histochemical assay for GUS activity in transgenic plants was performed as described by Jefferson et al. (1987).
The DDF1-GFP fusion was made by in-frame fusion of the 1536-bp full-length DDF1 coding sequence with GFP. The fusion gene was inserted into vector pCAMBIA1302T and further into EHA105. Transient expression of the DDF1-GFP fusion and GFP alone (as a control) in onion epidermal cells was performed as described by Eady et al. (2000). The samples were observed with an OLYMPUS IX71 microscope.
Real-time quantitative PCR
The expression levels of genes for cell division and expansion in leaves of a young seedling, the expression traits of DDF1 in roots, leaves, stems, young panices and spikelet, the expression levels of floral identity genes in young panicles (5–10 mm in length) and the expression levels of APO2 in young panicles (5–10 and 10–20 mm in length) were analyzed using real-time quantitative RT-PCR. Isolation of total RNAs was performed using Trizol Reagent kit (Invitrogen, http://www.invitrogen.com). Reverse transcription of total RNA was performed using the PrimescriptTM RT reagent kit (TaKaRa, http://www.takara-bio.com). The cDNA samples were diluted to 8 and 2 ng μl−1. Triplicate quantitative assays were performed using SYBR Premix Ex Taq II (TaKaRa) with a Mastercycler ep realplex4 sequence detection system (Eppendorf, http://www.eppendorf.com). Amplification of Actin was used as an internal control to normalize all data. The primers used for qRT-PCR analysis are listed in Table S1.
mRNA in situ hybridization
mRNA in situ hybridization was essentially performed as previously described by Lai et al. (2002). Young panicles (2–20 mm in length) were fixed with formalin : acetic acid : alcohol solution and embedded in paraffin. The tissues were sliced into 8-mm sections with a rotary microtome and attached to microscope slides. The primers used for mRNA in situ hybridization are listed in Table S3.
We thank Dr Baocai Tan (Chinese University of Hong Kong, Hong Kong) for a critical reading of the article. We are grateful to the anonymous reviewers for offering valuable suggestions to improve the article. We also thank Zhiqin Guan, Rong Zheng, Weiqi Tang, Lihui Lin, Xuzhang Zhang, Damei Mao, Fan Yang, Yichang Liu and Hanjun Wu (Fujian Agriculture & Forestry University) for their help with the experiment. This work was supported by the National Natural Science Foundation of China (30970175 and 30671122), Research Foundation for the Doctoral Program of Higher Education of China (20093515110003), National High Technology Research and Development Program of China (2006AA10Z128), and Natural Science Foundation of Fujian Province of China (B0620001).