Rice ABERRANT PANICLE ORGANIZATION 1, encoding an F-box protein, regulates meristem fate

Authors


(fax +81 3 5841 5063; e-mail anagato@mail.ecc.u-tokyo.ac.jp).

Summary

Inflorescence architecture is one of the most important agronomical traits. Characterization of rice aberrant panicle organization 1 (apo1) mutants revealed that APO1 positively controls spikelet number by suppressing the precocious conversion of inflorescence meristems to spikelet meristems. In addition, APO1 is associated with the regulation of the plastchron, floral organ identity, and floral determinacy. Phenotypic analyses of apo1 and floral homeotic double mutants demonstrate that APO1 positively regulates class-C floral homeotic genes, but not class-B genes. Molecular studies revealed that APO1 encodes an F-box protein, an ortholog of Arabidopsis UNUSUAL FLORAL ORGAN (UFO), which is a positive regulator of class-B genes. Overexpression of APO1 caused an increase in inflorescence branches and spikelets. As the mutant inflorescences and flowers differed considerably between apo1 and ufo, the functions of APO1 and UFO appear to have diverged during evolution.

Introduction

Plant architecture is mainly determined by the spatial and temporal arrangement of lateral organs. As all aboveground organs originate from the apical meristem, meristem fate plays a primary role in regulating plant architecture. In the reproductive phase, meristem fate changes drastically in rice (Ikeda et al., 2004). The rachis meristem, a product of the direct conversion of the shoot apical meristem (SAM), forms primary branch meristems and then aborts. The primary branch meristem is finally converted to a spikelet meristem after producing lateral organs (spikelets and secondary branches). The spikelet meristem is converted to floral meristem after formation of two pairs of glumes. The floral meristem forms floral organs, and is finally transformed into the pistil. Thus, the panicle (inflorescence) architecture of rice depends on the timing of rachis meristem abortion, specifying the number of primary branches, and on the conversion of terminal branch meristems to spikelet meristems, specifying the number of spikelets. In other words, the genes temporally regulating meristem fate determine inflorescence architecture. In agriculture, inflorescence architecture is of practical importance, especially for grain crops such as cereals and ornamental crops, because it determines the number of seeds produced and the arrangement of flowers.

In Arabidopsis, several genes have been identified that regulate meristem fate in the reproductive phase. TERMINAL FLOWER 1 (TFL1) inhibits precocious conversion of inflorescence meristems to floral meristems by suppressing the expression of LEAFY (LFY) and APETALA 1 (AP1) in the center of the inflorescence apex; in turn, LFY and AP1 inhibit TFL1 expression in lateral meristems committed to a floral fate (Liljegren et al., 1999; Mandel and Yanofsky, 1995; Ratcliffe et al., 1999). CAULIFLOWER (CAL), FRUITFULL (FUL) and UNUSUAL FLORAL ORGAN (UFO) also enhance the floral fate of lateral meristems (Ferrandiz et al., 2000; Levin and Meyerowitz, 1995; Wilkinson and Haughn, 1995). These studies indicate that complicated regulation of meristem fate occurs during construction of inflorescence architecture in Arabidopsis.

To date, little is known about genes that influence the timing of meristem fate change in rice at the molecular level. The only genes seemingly involved in the control of spikelet number are RCN1 and RCN2 (rice TFL1/CENTRORADIALIS homologs), the overexpression of which causes more inflorescence branches and spikelets caused by delayed conversion of terminal branch meristems to spikelet meristems (Nakagawa et al., 2002). Recently, a major quantitative trait locus (QTL), GN1, affecting spikelet number was isolated (Ashikari et al., 2005). GN1 encodes cytokinin oxidase, the reduced expression of which causes cytokinin accumulation and an increased number of spikelets. Although a detailed developmental analysis of GN1 function has not yet been performed, GN1 is thought to be associated with the duration of inflorescence meristems. In maize, several genes that regulate inflorescence architecture have been identified including fasciated ear2, ramosa2 and ramosa3 (Bortiri et al., 2006; Satoh-Nagasawa et al., 2006; Taguchi-Shiobara et al., 2001). Because rice and maize show distinct inflorescence architectures, the functions of the above genes in rice are not clear.

The ABC model of Arabidopsis can be modified to some extent to apply to floral development in rice. Carpel identity is predominantly determined by the YABBY gene DROOPING LEAF (DL; Nagasawa et al., 2003; Yamaguchi et al., 2004). Class-C gene is duplicated in rice and each copy has a distinct function (Yamaguchi et al., 2006). On the other hand, the function of SUPERWOMAN 1 (SPW1), the ortholog of APETALA3 (AP3), has conserved the function of AP3 (Nagasawa et al., 2003). In Arabidopsis, genes that regulate expression of ABC genes have been studied including LFY and UFO (Busch et al., 1999; Lee et al., 1997; Parcy et al., 1998). In contrast, how organ identity genes are regulated in rice is almost unknown.

We have previously reported on the rice ABERRANT PANICLE ORGANIZATION 1 (APO1) gene that plays a key role in the temporal regulation of meristem fate (Ikeda et al., 2005). The apo1 mutant forms small inflorescences with reduced numbers of branches and spikelets. In addition, apo1 exhibits abnormal floral organ identity and a loss of floral determinacy. In this study, we investigated how APO1 interacts with organ identity genes by analyzing double mutants. We also isolated the APO1 gene using a map-based cloning method, and found that it encodes an F-box protein orthologous to FIMBRIATA (FIM) of Anthirrhinum and UFO of Arabidopsis. Comparative study suggested that the functions of APO1 and UFO have diverged considerably during evolution.

Results

Phenotypes of apo1

As described by Ikeda et al. (2005), apo1 showed a number of abnormal phenotypes in the vegetative and reproductive phases. During vegetative growth, apo1 plants produced leaves rapidly, resulting in 1.5-fold more leaves than the wild type at the mature stage. In the reproductive phase, apo1 exhibited remarkable phenotypes in both panicle (inflorescence) architecture and floral organ identity (Figure 1a). At the early stage of panicle development primary branch primordia were produced by distichous phyllotaxy, in contrast to 2/5 spiral phyllotaxy in the wild type (Figure 1b,c). This phyllotactic change could be the result of a slender rachis meristem (Ikeda et al., 2005). The wild-type rachis meristem usually aborts after forming 10–12 primary branch primordia. However, the apo1 rachis meristem was converted to a spikelet meristem after producing several primary branch primordia. Similar events occurred in the primary branches: precocious conversion of primary branch meristems to spikelet or secondary branch meristems, resulting in a small number of spikelets. Thus, the wild-type APO1 gene should positively regulate primary branch number and spikelet number.

Figure 1.

 Phenotypes of apo1.
(a) Panicle structure. The wild-type plant (left) set 130 seeds per panicle, and the apo1-2 mutant panicle (right) set 30 seeds.
(b) Top view of the wild-type inflorescence.
(c) Side view of the apo1-2 inflorescence.
(d) Wild-type flower comprising two lodicules (arrowhead), six stamens and one pistil.
(e) apo1-2 flower comprising two lodicules (arrowheads), two lodicule-stamen mosaic organs (arrows), three stamens and three pistils.
(f) SEM image of apo1-1 flower.
(g) In situ expression of DL gene in wild-type (left) and apo1-1(right) flowers. DL is expressed in normal and extopic carpel primordia (arrowheads).
(h), (i) Schematic representation of wild-type (f) and apo1 (g) flowers. (j), (k) A model of floral organ identity in wild type (h) and apo1 (i). Abbreviations: ca, carpel; cg, mosaic organ between carpel and glume; le, lemma; lo, lodicule; pa, palea; st, stamen. Scale bars = 10 cm in (a), and 50 μm in (b) and (c).

The apo1 mutation also affected floral organ identities. In apo1 flowers, stamens were frequently replaced with lodicules (petals), and carpels were produced indeterminately (Figure 1d–f). The DL gene, which is specifically expressed in a carpel, was expressed in ectopic carpels (Figure 1g). Ectopic glumes with partial carpelloid characters were occasionally produced outside the carpel whorl. A typical apo1 flower is characterized by an increased number of lodicules, a concomitant decrease in stamens, and indeterminate carpel formation (Figure 1h,i). These phenotypes, together with reduced expression of class-C genes (Ikeda et al., 2005), suggest that APO1 positively regulates class-C genes (Figure 1j,k). Considering the precocious formation of spikelet meristems and prolonged formation of lodicules and carpels, APO1 is likely to be involved in the temporal regulation of meristem identity.

Phenotypes of apo1 dl-sup1 double mutants

To elucidate genic interaction with floral homeotic genes, we produced double mutants with the dl mutant that caused homeotic conversion of carpels to stamens (Figure 2a; Nagasawa et al., 2003). The apo1-1 dl-sup1 plants showed rapid leaf emergence and small inflorescences setting fewer spikelets, similar to apo1-1 (data not shown).

Figure 2.

 Phenotypes of apo1-1 dl-sup1 flowers and expression of floral organ identity genes.
(a)–(c) External appearance of dl-sup1 (a), apo1-1 (b) and apo1-1 and dl-sup1 (c) flowers.
(d) SEM image of apo1-1 dl-sup1 flower.
(e), (f) Cross sections of wild-type (d) and apo1-1 dl-sup1 (e) flowers. Arrowheads in (f) indicate lodicule-like organs.
(g) SEM image of the organ in the central whorl of the apo1-1 dl-sup1 flower.
(h) SEM image of lodicule of the wild-type flower. Bars = 10 mm. (i)–(k) SPW1 expression analysis by in situ hybridization.
(i) Wild-type flower.
(j) apo1-1 flower.
(k) apo1-1 dl-sup1 flower. SPW1 transcripts are detected in numerous lucent organs in the central region. Arrows in (i) and (j) indicate the SPW1 expression in lodicule. Abbreviations: le, lemma; ng, narrow glume; pa, palea. Scale bars = 100 μm in (d), 200 μm in (e) and (f), 10 μm in (g) and (h), and 50 μm in (i)–(k).

The apo1-1 dl-sup1 double mutant exhibited variable floral phenotypes (Figure 2, Table S1). In whorl 3, stamens were decreased, and instead lodicules and lodicule-stamen mosaic organs were formed, as in apo1 (Figure 2b,c). Novel phenotypes were observed in whorl 4. Small, narrow glume-like organs sometimes developed in the outer region of whorl 4 (Figure 2c), which differentiated bulliform cells on the adaxial surface, as in normal palea or lemma. These organs were similar to but thinner and narrower than the glume-like organs formed in whorl 4 of apo1. In the central region where numerous carpels differentiated in apo1-2, numerous (>30) small translucent organs were formed (Figure 2c–f). The surface structure of these organs was similar to that of lodicules (Figure 2g,h). In addition many vascular bundles were observed in these organs, as in wild-type lodicules, in contrast to three vascular bundles in normal carpels (Figure 2e,f). These organs expressed the class-B gene SPW1, which specifies lodicule and stamen identities in whorls 2 and 3, respectively (Figure 2i–k). Judging from these characteristics, these organs were estimated to be lodicules. Lodicule identity is determined by class-A and class-B genes (Kyozuka et al., 2000; Nagasawa et al., 2003). Lodicule formation in whorl 4 indicates that both class A and B genes are expressed in whorl 4 of apo1-1 dl-sup1. Ectopic expression of SPW1 in whorl 4 of apo1-1 dl-sup1 is explained by the fact that DL suppresses the class-B SPW1 gene expression in whorl 4 (Yamaguchi et al., 2004). Class-A gene expression in whorl 4 would be caused by the weakened expression of a class-C gene, as in apo1 single mutants (Ikeda et al., 2005). Therefore, the apo1-1 dl-sup1 phenotype confirms that APO1 positively regulates class-C genes.

Another remarkable abnormality of apo1-1 dl-sup1 was a severe defect in floral determinacy. The number of lodicules in whorl 4 of apo1-1 dl-sup1 was >30 (Figure 2c,d), which was much more than that of the respective single mutants. As the dl mutant shows partially indeterminate stamen formation in whorl 4, the DL gene acts positively in establishing floral determinacy (Nagasawa et al., 2003). In apo1 flowers, floral determinacy is partially lost even in the presence of DL transcripts (Ikeda et al., 2005). Thus, APO1 is also involved in floral determinacy, independently of DL. A severe increase of whorl 4 organs in apo1-1 dl-sup1 indicates that DL and APO1 regulate floral determinacy independently, but redundantly.

Phenotypes of apo1-1 spw1-1 double mutants

Next, we constructed apo1-1 spw1-1 double mutants. In the F2 population of APO1/apo1-1 × SPW1/spw1-1, the segregation frequency of the double mutants (about 2%) was significantly lower than the expected value (6.25%), suggesting that SPW and APO1 are closely linked. The apo1-1 spw1-1 plants showed rapid leaf emergence and small inflorescences with a small number of spikelets, similar to apo1-1 (data not shown).

In rice, SPW1 is known as a class-B floral homeotic gene (Nagasawa et al., 2003). In the spw1 flower, lodicules in whorl 2 and stamens in whorl 3 are converted to narrow glumes and pistils, respectively (Figure 3a). The spw1-1 apo1-1 flowers formed a few narrow glumes in whorls 2 and 3, which were identical to those observed in spw1 flowers (Figure 3b). In whorl 3, wide glumes were also observed occasionally that had trichomes and resembled the glume-like organ formed in whorl 4 of the apo1 (Figure 3c,d). If DL was expressed in whorl 3, carpels would replace the stamens as in spw1 flowers. Thus, APO1 activity should be required for proper function of DL in whorl 3. In whorl 4, a few wide glumes/carpeloid glumes and several carpels that contained no embryo sacs were formed indeterminately (Figure 3b,d, Table S1).

Figure 3.

 Phenotypes of apo1-1 spw1-1 flower.
(a) spw1-1spikelet flower.
(b) apo1-1 spw1-1 flower.
(c) SEM image of narrow glume and wide glume in apo1-1 spw1-1.
(d) Cross sections of an apo1-1 spw1-1 flower.
(e) The apo1-1 spw1-1 spikelet with an extra flower (arrowhead).
(f) Cross section of the apo1-1 spw1-1 spikelet with an extra floral meristem (arrowhead).
(g) Close up view of a protrusion (arrowhead) in the apo1-1 spw1-1 spikelet. Abbreviations: ca, carpel; le, lemma; ng, narrow glume; wg, wide glume. Scale bars = 50 μm in (c), 200 μm in (f) and (d), and 25 μm in (g).

A remarkable characteristic of apo1 spw1 was the formation of an extra flower (Figure 3e,f). This extra flower was always located on the lemma side of whorl 2 (Figure 3e,f). The organization of the extra flower was similar to that of the apo1 flower, having narrow glumes, wide glumes and carpels. Extra flowers were observed in about 10% of spikelets. Another 10% of spikelets had vestiges of extra flowers (Figure 3g). These results showed that APO1 should suppress the formation of ectopic floral meristem redundantly with SPW1.

Cloning of APO1

To elucidate its molecular function, we isolated the APO1 gene using a map-based strategy. Using the F2 population of APO1/apo1-3 × cv. Kasalath (Oryza sativa spp. indica), APO1 was roughly mapped at around 106 cM between two sequence-tagged site (STS) markers, R3819 and C11635, of chromosome 6. This map position supports the close linkage with SPW1 located at around 120 cM. We also found a close homolog of UFO within this region. Occurrence of the mosaic floral organs observed in apo1 resembled the floral phenotypes of Arabidopsis ufo mutants, and prompted us to test the idea that the UFO homolog might be APO1. To test this idea, we examined the nucleotide sequence of this region in the three apo1 alleles. The UFO-like gene comprised two exons and one intron (Figure 4a). Sequencing analysis detected one nucleotide substitution in the exon of each apo1 allele (Figure 4a). Three substitutions caused non-sense mutations at positions 28, 211 and 378 of the deduced amino acid sequence. The mutant phenotype of apo1-3 was rescued when an 8.5-kb genomic fragment containing the UFO-like gene was introduced (Figure S1). Thus, we conclude that APO1 is the UFO-like gene.

Figure 4.

 Sequence analysis of the APO1 gene.
(a) Exon/intron structure of APO1 and mutations in three alleles. Two open boxes indicate exons split by a small intron. Mutations in the three alleles are indicated.
(b) Alignment of the deduced amino acid sequences encoded by APO1, STP, PFM, FIM, UFO, maize fimbriata A and maize fimbriata B. Residues that are >50% identical are shaded in black. Conservative substitutions at an amino acid position are shaded in gray. Dashes denote gaps that were introduced to optimize the alignment. The F-box domain is underlined. Asterisks indicate the completely conserved amino acid and dots indicate the similar residues.
(c) An unrooted phyllogenetic tree generated by ClustalX using the neighbor-joining method from the predicted amino acid sequences of F-box genes from Arabidopsis and rice, and UFO homologs from other plants. Bootstrap values generated with 1000 replicates are indicated before the nodes. The bar represents the branch length equivalent to 0.1 amino acid changes per residue.

APO1 encodes a putative protein of 429 amino acids, with motifs that suggest a role as an F-box protein at the N-terminus (Figure 4b). F-box proteins are characterized as components that bind substrates for ubiquitin-mediated proteolysis (Kipreos and Pagano, 2000). F-box proteins contain a protein–protein interaction motif in the C-terminal region that binds to the target protein (Kipreos and Pagano, 2000). However, we could not identify such a protein–protein interaction motif in APO1 by using the Prosite (http://www.expacy.org) and Pfam (http://pfam.cgb.ki.se) algorithms. Such a protein–protein interaction motif has not been identified in UFO either (Gagne et al., 2002).

In contrast with many studies on UFO homologs in dicots, FIM in Antirrhinum, STANINA PISTILLOIDA (STP) in pea, PROLIFERATING FLORAL ORGANS (PFO) in Lotus japonicus and Imp-FIM (PFM) in Impatiens (Pouteau et al., 1998; Simon et al., 1994; Taylor et al., 2001; Zhang et al., 2003), the function of UFO homologs in monocots is poorly understood. Thus, we searched APO1-like genes in maize gene databases (http://www.maizegdb.org), and found two APO1 homologs with the GSS sequence (ID: ZmGSSTUC11-12-04.37070.1 and ZmGSSTUC11-12-04.31860.1). We designated them ZmFIMA and ZmFIMB, respectively. We aligned the amino acid sequences of APO1 and six UFO homologs (Figure 4b). All seven genes shared high homology along their entire length. For example, APO1 and UFO showed 57.5% identity (90% similarity) in the F-box region and 46% identity (77% similarity) in the C-terminal region (118–416).

To clarify the evolutionary relationships among UFO homologs, we constructed a phylogenetic tree based on the entire amino acid sequences of four rice genes, six Arabidopsis genes and the other genes described above, by the neighbor-joining method (http://clustalw.ddbj.nig.ac.jp/top-j.html). APO1 constituted a small clade with UFO, FIM, STP, PFO, ZmFIMA and ZmFIMB (Figure 4c), whereas three other UFO-like genes of rice were separated from this clade, indicating that APO1 is an ortholog of UFO in rice.

Spatial expression patterns of APO1

The spatial expression pattern of APO1 was analyzed by in situ hybridization. Developmental stages were designated according to Itoh et al. (2005). In the vegetative phase, APO1 transcripts were detected in SAM and P1–P4 leaf primordia (Figure 5a). In the reproductive phase, at stage In2 when the first one or two branch primordia were formed, strong APO1 expression was apparent in the outer several cell layers of the rachis meristem and primary branch meristems (Figure 5b). At stages In3–In5, when primary branches were elongating, APO1 expression persisted in the primary and secondary branch meristems (Figure 5c). Afterwards, APO1 transcripts were detected in the spikelet and floral meristems (Figure 5d,e). Lateral organs of spikelet and floral meristems, such as glumes, lodicules, stamens and carpel, also expressed APO1. Finally, APO1 expression in the floral meristem was retained in the ovule primordium (Figure 5f).

Figure 5.

 Localization of APO1 transctipts in wild type.
(a) Longitudinal section of 3-week-old shoot apex.
(b) Longitudinal section of inflorescence apex when primary branch meristems (arrowheads) are being formed.
(c) Longitudinal section of young panicle in which spikelet or secondary branch meristems (arrowheads) are formed from primary branches.
(d) Longitudinal section of young spikelet when lemma and palea primordia are being formed. Arrowhead indicates spikelet meristem.
(e) Longitudinal section of flower when stamen primordia are being formed. Arrowhead indicates floral meristem.
(f) Longitudinal sections of flower when carpel is being formed. Arrowhead indicates ovule primordium. Abbreviations: ca, carpel; eg, empty glume; le, lemma; P1, P1 leaf primordium; P2, P2 leaf primordium; pa, palea; rg, rudimentary glume; st, stamen. Scale bars = 50 μm.

In summary, APO1 was expressed in apical meristems and the lateral organ primordia throughout development. This expression pattern is consistent with the pleiotropic apo1 phenotypes.

Transcriptional regulation of the APO1 gene

An 8.5-kb genomic fragment covering the APO1 open reading frame (ORF) and 5116 bp of the 5′ flanking sequences (pro5 kb-APO1) (Figure 6a) rescued apo1-3. Prior to this experiment, we introduced an approximately 7-kb genomic fragment containing the APO1 ORF and 2924 bp of the 5′ flanking sequence (pro3 kb-APO1) into apo1-3 (Figure 6a). All of the seven independent transgenic apo1-3 lines formed aberrant plants that were quite different from those of the wild type and apo1-3. The pro3 kb-APO1 plants were larger than the pro5 kb-apo1 plants, and formed larger leaves than the wild type and apo1 (Figure 6b,c). They also produced a larger number of primary branches (>15) than the wild type (10–12) (Figure 6d). In addition, the numbers of secondary branches and spikelet primordia were increased. However, branches and spikelets exhibited remarkable morphological abnormalities in the apical region. Branches were undulated and spikelets were underdeveloped (Figure 6e). Furthermore, unlike the wild type, in which a terminal spikelet developed on each primary branch, the primary branch terminated in a vestige of the meristem or a glume-like organ (Figure 6f). That is, primary branch meristems of pro3 kb-APO1 plants abort after forming a large number of lateral organs (secondary branches and spikelets), in contrast with apo1 in which primary branch meristems are converted precociously to spikelet meristems after producing a small number of lateral organs.

Figure 6.

 Phenotypes of pro 3 kb-APO1 plants.
(a) Structure of APO1 genomic fragments used for transformation.
(b) Mature plants of pro5 kb-APO1 (left) and pro3 kb-APO1 (right).
(c) Flag leaf blades of pro3 kb-APO1 (left), wild-type (middle) and apo1-2 (right).
(d) Panicles of apo1-3 carrying pro3 kb-APO1.
(e) An upper region of primary branch of (b) showing undulated branches and immature spikelets.
(f) Tip of a primary branch. Terminal spikelet is replaced by a vestige of meristem surrounded with hairs (arrowhead).
(g) RT-PCR analysis of APO1 gene in a pro3 kb-APO1 young panicle. –, Wild type; +, pro3 kb-APO1. Scale bar = 5 cm.

These phenotypes of pro3 kb-APO1 plants are opposite to those of apo1-3, suggesting that APO1 is overexpressed in pro3 kb-APO1 plants. RT-PCR analysis revealed a higher level of APO1 transcript accumulation in pro3 kb-APO1 young panicles than in wild types (Figure 6g). As pro3 kb-APO1 is truncated by about 2 kb more than pro5k-APO1 in the 5′ flanking region, and is longer by about 1 kb than pro5k-APO1 in the 3′ flanking region, there may be a repressor motif in one of these regions.

Discussion

Here, we isolated the APO1 gene that encodes an F-box protein and is expressed in apical meristems and lateral organ primordia. As mutant phenotypes of apo1 are observed at various developmental stages, APO1 is thought to be involved in diverse developmental events by regulating the degradation of various proteins. Mutant phenotypes suggest that APO1 functions as a temporal regulator of meristem fate, and as a regulator of floral homeotic genes. The expression of APO1 in all kinds of apical meristems supports this idea.

APO1 is an agronomically important gene affecting leaf number and panicle size

Grain yield in rice is strongly correlated with the number of spikelets. Loss of APO1 activity caused small panicles with a reduced number of inflorescence branches and spikelets. This phenotype indicates that the wild-type APO1 gene suppresses precocious conversion of inflorescence meristems to spikelet meristems, thus ensuring a number of gains. It is expected that overexpression of APO1 causes large panicles with an increased number of spikelets. In fact, the overexpressors transformed with pro3 kb-APO1 produced a larger number of inflorescence branches and spikelets, as a result of retarded conversion of the inflorescence meristems to spikelet meristems (Figure 6d). As the development of branches and spikelets was impaired in these overexpressors, moderate regulation of APO1 expression would be required to increase grain production.

Plant shape is an important trait that is determined by the arrangement and sizes of leaves and branches. The number of leaves is usually determined by heading date. That is, late-heading cultivars produce more leaves than early-heading ones. To produce more leaves without affecting the heading date, it is necessary to shorten the plastochron. Rapid leaf initiation in apo1 suggests that tuning the number of leaves is possible by modulating APO1 expression. A recent study on pla1 and pla2 mutants revealed that precocious leaf maturation results in small leaves and rapid leaf initiation (Kawakatsu et al., 2006; Miyoshi et al., 2004). As PLA1 and PLA2 are expressed in leaf primordia but not in the SAM (Kawakatsu et al., 2006; Miyoshi et al., 2004), APO1 expression in leaf primordia may be responsible for the short plastochron in apo1. However, pla1 SAM is larger than that in the wild type, whereas apo1 SAM is comparable with, or rather smaller than, that in wild type (Ikeda et al., 2005). Thus, APO1 expression in both leaf primordia and SAM might be involved in regulation of the plastochron.

The short plastochron in apo1 as well as in pla1 and pla2 was accompanied by small leaf size (Figure 6c). This indicates that both plastochron and leaf size can be modified simultaneously by regulating APO1 expression.

Interaction between APO1 and floral homeotic genes

In a previous study (Ikeda et al., 2005), mutant phenotypes and in situ hybridization indicated that APO1 positively regulates class-C gene (OsMADS3) expression, but does not affect the class-B gene SPW1. This is also supported by a recent study on the loss-of-function mutant of OsMADS3 that showed apo1-like flowers, i.e. conversion of stamens to lodicules and an increase in carpels (Yamaguchi et al., 2006). However, apo1 spw1 plants generated novel phenotypes. In whorl 3, unlike spw1 flowers in which carpels replaced stamens caused by the ectopic expression of DL (Nagasawa et al., 2003), apo1 spw1 formed narrow organs (Figure 3). This phenotype indicates that DL is not expressed in whorl 3 even when SPW1 activity is lost, and thus suggests that APO1 activity is required for the expression of DL in whorl 3. Accordingly, it is thought that APO1 interacts with class-C genes and DL.

Although APO1 should not affect SPW1 expression as far as floral organ identity is concerned, apo1 spw1 exhibited a novel phenotype: ectopic flower formation from the lemma side of whorl 2 in about 10% of spikelets (Figure 3e). Recently, ectopic florets formed from the rachis outside the normal lemma were reported in the rice supernumerary bract (snb) mutant (Lee et al., 2006). The snb ectopic florets differ from ectopic flowers in apo1 spw1 in that the snb ectopic florets are enclosed by their own lemma and palea, and are formed outside the normal floret (lemma). Ectopic flowers in apo1 spw1 are formed inside the palea after the floral meristem is established. As the ectopic flower is observed only in apo1 spw1, the straightforward interpretation is that APO1 and SPW1 redundantly suppress the formation of extra floral meristems. This interpretation is not easily acceptable, because SPW1 is not expressed in either spikelet meristems or floral meristems (Nagasawa et al., 2003), and a function for meristematic activity or cell proliferation has not been assigned to SPW1. Although ectopic flowers have been reported in dicot mutants of the APO1 orthologs, ufo, fim, stp, and pfo (Levin and Meyerowitz, 1995; Simon et al., 1994; Taylor et al., 2001; Zhang et al., 2003), ectopic flowers were never observed in single apo1 mutants. They are formed only when the expression levels of APO1 and class-B genes are simultaneously reduced. These results suggest that ectopic flowers may be formed when class-B and class-C genes are impaired simultaneously. However, loss of class-B and class-C functions in Arabidopsis (e.g. ap3 ag) does not cause ectopic flower formation (Bowman et al., 1991). Accordingly, reduced activities of APO1, SPW1 and DL may be involved in ectopic flower production, although the mechanism is not clear.

Molecular function of APO1

APO1 encodes an F-box protein that is closely related to Antirrhinum FIM and Arabidopsis UFO. The F-box motif provides substrate specificity for large protein complexes called SCFs (Skp1-Culling-F-box protein complexes), which have E3 ubiquitin ligase activity and target proteins for degradation (Kipreos and Pagano, 2000). As genetic experiments and a binding assay revealed that UFO interacts with the Arabidopsis SKP1 homolog, ASK1 (Durfee et al., 2003; Zhao et al., 2001), UFO is thought to act as the receptor that recruits the substrate. F-box proteins generally contain a protein–protein interaction motif to bind the substrates in the C-terminal portion. However, such a motif and its target protein of FIM/UFO remain unknown. We also could not find a protein–protein interaction motif in the C-terminal portion of APO1.

Our comparative analysis among UFO orthologous proteins from Arabidopsis, Antirrhinum, Pisum sativum, Impatiens, rice and maize showed that high similarity is observed not only in the F-box but also in the C-terminal region (118–416). As the overall structure of UFO may be important for binding with ASK1 (Durfee et al., 2003), similarity in the C-terminal portion does not exclusively represent their function as receptors.

APO1 and UFO show a high similarity of amino acid sequences along their entire length (Figure 4), despite large differences in their mutant phenotypes, and even in their expression patterns (discussed later). The promoter region and/or target proteins for degradation may have diverged between rice and Arabidopsis.

Functional diversification between APO1 and UFO

Although phylogenetic analysis indicates that APO1 is the ortholog of UFO, the loss-of-function phenotypes differ largely between ufo and apo1. Both UFO and APO1 are expressed in inflorescence meristems (Lee et al., 1997), but apo1 forms small panicles with reduced numbers of branches and spikelets (Figure 1a), whereas ufo produces more co-florescences than the wild type (Wilkinson and Haughn, 1995). This difference indicates that APO1 and UFO have opposite effects on inflorescence meristem fate. The conversion of inflorescence meristems to floral (spikelet) meristems is suppressed by APO1, but promoted by UFO. Functional diversification between APO1 and UFO is also suggested by the effect of a loss-of-function mutation on inflorescence meristem size: small inflorescence meristems in apo1 vs. large inflorescence meristems in ufo (Ikeda et al., 2005; Samach et al., 1999).

As for floral organ identity, APO1 should interact with class C (OsMADS3) and DL. In contrast, UFO and the orthologous genes in several eudicot species activate class-B genes (Parcy et al., 1998; Taylor et al., 2001; Wilkinson and Haughn, 1995; Zhang et al., 2003).

The spatial expression pattern of APO1 in flower development is also different from that of UFO. UFO transcripts are localized in the center of an early stage-2 meristem, and then disappear from the floral meristem (Samach et al., 1999). In contrast, APO1 expression in the floral meristem is maintained even after lodicule and stamen formation, thus affecting floral meristem determinacy (Figure 5e,f).

These differences indicate that the rice UFO ortholog, APO1, has acquired novel functions through recruiting different target proteins.

Experimental procedures

Plant materials

The mutants apo1-1, apo1-2 and apo1-3 were described previously by Ikeda et al. (2005). The double mutants apo1-1 dl-sup1 and apo1-1 spw1-1 were generated by crossing heterozygous APO1/apo1-1 with DL/dl-sup1 and SPW1/spw1-1, respectively. Mutant and wild-type plants were grown in pots or in paddy fields under natural conditions. Transgenic plants were grown in a biohazard greenhouse at 30°C during the day and at 25°C during the night.

Histological analysis

For paraffin sectioning, samples were fixed overnight at 4°C in FAA (formalin : glacial acetic acid : 70% ethanol; 1:1:18), and dehydrated in a graded ethanol series. Following substitution with xylene, we embedded the samples in Paraplast Plus (McCormick Scientific, http://www.paraplast.com) and sectioned them at 8-μm thickness using a rotary microtome. Sections were stained with 0.05% toluidine blue and observed with a light microscope.

For scanning electron microscopy (SEM), samples were fixed overnight at 4°C in FAA. After dehydration in a graded ethanol series and substitution with 3-methyl-butyl-acetate, the samples were critical-point dried, sputter coated with platinum and observed under a scanning electron microscope (S-4000; Hitachi Ltd., http://www.hitachi.com) at an accelerating voltage of 10 kV.

Map-based cloning and phylogenetic analysis

For mapping the APO1 locus, a heterozygous APO1/apo1-3 plant (O. sativa L. ssp. japonica) was crossed with cv. Kasalath (ssp. indica), and mutant plants showing the apo1 phenotype in the F2 population were used. The APO1 locus was mapped onto the long arm of chromosome 6 using STS markers obtained from the rice genome database (http://rgp.dna.affrc.go.jp/E/publicdata/caps/index.html). Further mapping delimited the APO1 locus in the 11.4-cM region between two STS markers, R3819 and C11635.

Sequences of UFO homologs were obtained through a search of the plant GDB database (http://www.plantgdb.org). Accession numbers are AB292777 (O. sativa, APO1), AF004843 (P. sativum, STP), S71192 (Antirrhinum majus, FIM); X89224 (Arabidopsis thaliana, UFO), AF047392 (Impatiens balsamina, Imp-FIM) and AY156687 (L. japonicus, PFO). Alignment of amino acid sequences was carried out using ClustalX (Thompson et al., 1997) and Boxshade version 3.1.1 (http://www.ch.embnet.org/software/BOX_form.html).

Complementation test

For complementation tests, the 8.5-kb genomic Nco1 fragment of a bacterial artificial chromosome (BAC) clone containing the candidate gene region was subcloned into the binary vector, and was introduced into apo1-3 plants by the Agrobacterium tumefaciens-mediated transformation method (Hiei et al., 1994).

RNA extraction and semiquantitative RT-PCR analysis

mRNA levels for APO1 were analyzed by semiquantitative RT-PCR. Total RNA was extracted using TRizol Regent (Invitrogen, http://www.invitrogen.com) and cDNA was synthesized from 1 μg of total RNA using Superscript II reverse transcriptase (Invitrogen). cDNA was PCR-amplified using suitable primers, which were localized at the different exons of each gene in order to confirm that the PCR products were derived from the cDNA, rather than from the genomic sequence. The primers used for amplification were: APO1, 5′-TCCGGCAGGTTCTACTGCATG-3′ and 5′-ACGCTGAGACGGCTCTTCTCG-3′; ACTIN1, 5′-CAATCGTGAGAAGATGACCC-3′; and 5′-GTCCATCAGGAAGCTCGTAGC-3′. Following agarose gel electrophoresis, band intensity indicated the relative content of the transcripts.

In situ hybridization

Paraffin sections were prepared as described above and were applied to glass slides coated with APS (3-aminopropyltriethoxysilane) (Matsunami Glasses, Osaka, Japan, http://www.matsunami-glass.co.jp/e-index.html). Digoxigenin-labeled antisense and sense RNA probes for APO1 and DL were generated from full-length cDNAs. The digoxygenin-labeled RNA probes used to detect SPW1 were prepared as described by Nagasawa et al. (2003). In situ hybridization and immunological detection of the hybridization signals were performed as described by Kouchi and Hata (1993). Only the results with antisense probes are shown because the sense probes of these genes showed no specific hybridization signals.

Acknowledgements

We thank K. Miyoshi and N. Kurata for their help in mapping the APO1 locus. We also thank S. Nakata and H. Kimura for their assistance in cultivating rice plants at the Experimental Farm of the University of Tokyo. This work is supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan (14036206 and 16208002 to YN).

Ancillary