ABERRANT SPIKELET AND PANICLE1, encoding a TOPLESS-related transcriptional co-repressor, is involved in the regulation of meristem fate in rice

Authors

  • Akiko Yoshida,

    1. Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-8657, Japan
    Search for more papers by this author
    • Present address: Graduate School of Agriculture and Life Sciences, The University of Tokyo, Yayoi, Bunkyo, Tokyo 113-8657, Japan.

  • Yoshihiro Ohmori,

    1. Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-8657, Japan
    Search for more papers by this author
  • Hidemi Kitano,

    1. Bioscience and Biotechnology Center, Nagoya University, Chikusa, Nagoya 464-8601, Japan
    Search for more papers by this author
  • Fumio Taguchi-Shiobara,

    1. National Institute of Agrobiological Sciences, Tsukuba, Ibaraki 305-8602, Japan
    Search for more papers by this author
  • Hiro-Yuki Hirano

    Corresponding author
    1. Department of Biological Sciences, Graduate School of Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-8657, Japan
      For correspondence (fax +81 3 5841 4056; e-mail hyhirano@biol.s.u-tokyo.ac.jp).
    Search for more papers by this author

For correspondence (fax +81 3 5841 4056; e-mail hyhirano@biol.s.u-tokyo.ac.jp).

Summary

Post-embryonic development depends on the activity of meristems in plants, and thus control of cell fate in the meristem is crucial to plant development and its architecture. In grasses such as rice and maize, the fate of reproductive meristems changes from indeterminate meristems, such as inflorescence and branch meristems, to determinate meristems, such as the spikelet meristem. Here we analyzed a recessive mutant of rice, aberrant spikelet and panicle1 (asp1), that showed pleiotropic phenotypes such as a disorganized branching pattern, aberrant spikelet morphology, and disarrangement of phyllotaxy. Close examination revealed that regulation of meristem fate was compromised in asp1: degeneration of the inflorescence meristem was delayed, transition from the branch meristem to the spikelet meristem was accelerated, and stem cell maintenance in both the branch meristem and the spikelet meristem was compromised. The genetic program was also disturbed in terms of spikelet development. Gene isolation revealed that ASP1 encodes a transcriptional co-repressor that is related to TOPLESS (TPL) in Arabidopsis and RAMOSA ENHANCER LOCUS2 (REL2) in maize. It is likely that the pleiotropic defects are associated with de-repression of multiple genes related to meristem function in the asp1 mutant. The asp1 mutant also showed de-repression of axillary bud growth and disturbed phyllotaxy in the vegetative phase, suggesting that the function of this gene is closely associated with auxin action. Consistent with these observations and the molecular function of Arabidopsis TPL, auxin signaling was also compromised in the rice asp1 mutant. Taken together, these results indicate that ASP1 regulates various aspects of developmental processes and physiological responses as a transcriptional co-repressor in rice.

Introduction

Plant development depends on the function of apical meristems and axillary meristems. Because the axillary meristems (AMs) control the growth of secondary axes, the apparent external morphology of a plant reflects AM activity – namely initiation, maintenance, determinacy and dormancy. Inflorescence architecture relies on the branching pattern, which is ultimately determined by the number and arrangement of the AMs, and timing of transition of the AM from an indeterminate to a determinate state (reviewed by McSteen and Leyser, 2005; Bortiri and Hake, 2007).

There is a wide range of inflorescence architecture in grasses. Rice (Oryza sativa) produces an inflorescence belonging to the raceme class, with long branches, generally called ‘panicles’, whereas wheat (Triticum aestivum) and barley (Hordeum vulgare) form inflorescences without branches, called ‘spikes’ (Itoh et al., 2005). Maize (Zea mays) has two types of inflorescence, the female ear and the male tassel, which differ in morphology and branching pattern (reviewed by Bommert et al., 2005; Bortiri and Hake, 2007; McSteen, 2009). After transition from the vegetative to the reproductive phase in rice, the shoot apical meristem is converted to an inflorescence meristem (IM; rachis meristem), which forms the primary inflorescence axis (rachis) and initiates the primary branch meristem (BM; corresponding to the AM in the inflorescence) (Itoh et al., 2005). Subsequently, the IM degenerates and remains as a trace at the base of the uppermost primary branch in the mature panicle (Figure 1a) (Komatsu et al., 2003b). The primary BM initiates secondary BMs and spikelet meristems (SMs), which develop lateral organs such as rudimentary glumes, sterile lemma, lemma, palea and floral organs (lodicule, stamen and carpel). The secondary BM also initiates SMs. The stem cells in the SMs are consumed by production of carpels (Yamaguchi et al., 2006). Thus, the IM and the primary/secondary BMs have an indeterminate nature, whereas the SM is determinate. At the top of the primary and secondary branches, both primary and secondary BMs are converted to an SM, resulting in the production of terminal spikelets. Thus, meristem fate resulting from transition from the BM to the SM, and degeneration of the IM, greatly affect the size and patterning of the panicle in rice. Elucidation of mechanisms underlying the regulation of meristem fate is an important issue not only for basic developmental studies but also for agronomical studies, as rice is one of the most important crops worldwide.

Figure 1.

 Phenotypes of the panicle in asp1-1 and wild-type.
(a) Schematic representation of panicle architecture in rice. br, bract; dim, degenerated IM; ls, lateral spikelet; pb, primary branch; sb, secondary branch; ts, terminal spikelet.
(b–e) Wild-type panicle (b, d) and asp1-1 panicle (c, e). The arrow and arrowhead indicate a bleached branch and spikelet, respectively.
(f) Close-up view of the primary branch of asp1-1.
(g) Degenerate secondary branches that stopped growing at early developmental stages.
Scale bars = 2 cm (b, c), 5 mm (d, e), 1 mm (f) and 0.5 mm (g).

Many genes involved in the regulation of meristem fate have been identified in grass species such as rice and maize. A reduction in AM initiation results in a decrease in the number of branches and spikelets in mutants of rice and maize (McSteen and Hake, 2001; Komatsu et al., 2003a; Li et al., 2003; Gallavotti et al., 2004; Ikeda et al., 2007; McSteen et al., 2007; Gallavotti et al., 2008a). By contrast, genes associated with high grain yield increase the numbers of branches and spikelets in rice (Ashikari et al., 2005; Ikeda-Kawakatsu et al., 2009; Jiao et al., 2010; Miura et al., 2010; Ookawa et al., 2010). In maize, ramosa mutants (ra1, ra2 and ra3) are associated with an increase in branch and spikelet numbers (Vollbrecht et al., 2005; Bortiri et al., 2006; Satoh-Nagasawa et al., 2006). Among these, ra1 is likely to be a key regulator in the ramosa pathway, because ra2 and ra3 regulate meristem fate by controlling the expression of ra1. Transition from the BM to the SM is regulated by FRIZZY PANICLE1 in rice and branched silkless1 in maize, and transition from the SM to the floret (flower) meristem (FM) is regulated by SUPERNUMERY BRACT (SNB) in rice and indeterminate spikelet1 (ids1) and sister of ids1 (sid1) in maize (Chuck et al., 1998, 2002, 2007, 2008; Komatsu et al., 2003b; Lee et al., 2007). In rice, both branching pattern and spikelet morphology are impaired in panicle phytomer1 (pap1) and pap2 mutants (Takahashi et al., 1998; Kobayashi et al., 2010).

Recent studies have revealed that transcriptional co-repressors play crucial roles in plant development (reviewed by Liu and Karmarkar, 2008, Krogan and Long, 2009). Co-repressors in the Groucho (Gro)/Tup1 family include TOPLESS (TPL)/WUSCHEL-INTERACTING PROTEIN (WSIP) and LEUNIG (LUG) in Arabidopsis (Conner and Liu, 2000; Kieffer et al., 2006; Long et al., 2006). These co-repressors lack DNA-binding activity but are incorporated into transcription complexes by interacting with transcription factors that bind to specific DNA sequences. These co-repressors inhibit expression of the target gene by recruiting histone deacetylases (HDACs) into transcription complexes, and change the chromatin state from active to inactive (Long et al., 2006; Gonzalez et al., 2007; Liu and Karmarkar, 2008; Krogan and Long, 2009). A dominant-negative mutation of TPL (tpl-1) causes severe defects in embryo development: for example, a shoot pole is transformed into a second root pole in the tpl-1 embryo (Long et al., 2002, 2006). The TPL gene and four TOPLESS-RELATED genes (TPR1–TPR4) act redundantly in Arabidopsis development, and a quadruple mutant in combination with RNAi suppression of the 5th gene shows phenotypes similar to those of tpl-1. TPL plays a role in auxin signaling by directly binding to IAA12/BODENLOS (BDL), and by repressing transcription of AUXIN RESPONSE FACTOR5 (ARF5)/MONOPTEROS (MP) (Szemenyei et al., 2008). TPL and TPR4 correspond to WSIP1 and WSIP2, respectively, which were identified as proteins that interact with WUSCHEL (WUS), a promoter of stem cell identity in Arabidopsis (Kieffer et al., 2006). Recently, it was shown that mutation of ramosa1 enhancer2 (rel2) enhances the inflorescence phenotype of ra1 and ra2 in maize, and REL2 encodes a protein similar to Arabidopsis TPL/TPR (Gallavotti et al., 2010). REL2 physically interacts with RA1, suggesting that the REL2/RA1 repressor complex regulates determinacy of the branch meristem in maize inflorescence development.

Here we describe a mutant, aberrant spikelets and panicle1 (asp1), that shows disorganized branching patterns and phyllotaxy, abnormal morphology of spikelets and rachis, premature termination of spikelet development, and reduced numbers of mature spikelets and branches. The asp1 mutant also exhibits de-repression of axillary bud growth and disturbed phyllotaxy in the vegetative phase. Gene isolation revealed that ASP1 encodes a transcriptional co-repressor similar to Arabidopsis TPL and TPR, suggesting that the regulation of multiple genes is compromised in the asp1 mutant. Thus, the molecular evidence and pleiotropic phenotypes in asp1 are consistent with each other. In addition, in agreement with the predicted function of ASP1, auxin signaling and HDAC action are likely to be partially defective in the mutant.

Results

Pleiotropic phenotypes of the panicle in the asp1 mutant

The rice inflorescence (panicle) consists of the rachis (main axis of the inflorescence), primary and secondary branches, and spikelets (Figure 1a,b). The primary branches are arranged in a spiral phyllotaxy, and spikelets are produced on both the primary and secondary branches. We identified a mutant (asp1-1) showing aberrant spikelets and branches forming a small panicle from a mutant collection induced by tissue culture (Figure 1c) (Miyao et al., 2007). We also found three similar mutants (asp1-2, asp1-3 and asp1-4) that showed defects in both spikelets and panicles, and genetic analysis revealed that these mutants were allelic to each other (Figure S1).

The rachis length was reduced to approximately 80% of that of the wild-type in the asp1-1 mutant (Figure 1b,c and Table S1). The primary branches were also shorter and variable in length (Figure 2a). The number of primary and secondary branches fluctuated, and a number of degenerated branches were present (Figure S2). The panicle of asp1-1 showed pleiotropic morphological abnormalities, such as disturbed phyllotaxy and acute curvature of primary branches (Figure 1e,f). Some primary and secondary branches became bleached and stopped growing. Both branch and spikelet development were frequently arrested at earlier stages (Figure 1g).

Figure 2.

 Comparison of branch length and spikelet number between asp1-1 and wild-type.
(a) Length of the primary branch. Bars indicate the standard error.
(b) Number of spikelets. Dark and light green indicate spikelets generated on the primary and secondary branches, respectively. Orange and yellow indicate degenerated spikelets on the primary and secondary branches, respectively.

The number of normal spikelets decreased in both the primary and secondary branches, and varied among plants and panicles in asp1-1 (Figures 1b and 2b). A number of degenerated spikelets were also found (Figure 2b, and see below). The mean number of normal spikelets was reduced in the asp1-1 mutant to <60% of that in wild-type (Figure 2b).

Taken together, these observations suggest that initiation and maintenance of the BM and the SM are compromised, and that subsequent development of lateral organs, such as the branch and spikelets, is partially defective in asp1-1.

Morphological abnormalities of the spikelet in the asp1 mutant

Rice has a unique spikelet, with two tiny vestigial structures (rudimentary glume), two small leaf-like organs (sterile lemma), and one fertile floret (Figure 3a–c) (Bommert et al., 2005; Hirano, 2008; Yoshida et al., 2009). The floret comprises one pistil, six stamens, two lodicules, a palea and a lemma.

Figure 3.

 Phenotypes of spikelets in asp1-1 and wild-type.
(a–c, g–i) Wild-type spikelet (a–c) and asp1-1 spikelet (g–i). The basal region of the spikelets is magnified in (c) and (i). The lemma and palea are removed in (b) and (h).
(d–f) Epidermal surface of the sterile lemma (d), lemma (e) and rudimentary glume (f) in wild-type.
(j–l) Epidermal surface of the sterile lemma (j, k) and rudimentary glume (l) in asp1-1. (m–q) asp1-1 spikelets arrested at various developmental stages. The inset in (m) is a close-up view of the dome-like structure formed at the basal region of the elongated sterile lemma (arrow). Outlines of the rudimentary glume are shown by dotted pink lines in (n) and (o). The apical region of a cone-like structure in (p) is cut off in (q) to show that it has no edge.
(r, s) Dome-like structure, suggesting premature termination of meristem activity.
le, lemma; rg, rudimentary glume; sl, sterile lemma.
Scale bars = 1 mm (a, b, g, h, m–q), 200 μm (c, e, f, k, i, r) and 20 μm (d, j, l, s).

Spikelets in asp1 also showed pleiotropic phenotypes. In spikelets with weak phenotypes, the sterile lemmas and rudimentary glume were elongated and their identities were altered (Figure 3g–i). In some asp1-1 spikelets, the sterile lemma had a chimeric surface comprising the suface of the sterile lemma and the lemma or a similar surface to that of the lemma in wild-type (Figure 3d-f, j, k). In other spikelets, the epidermal surface of the rudimentary glume was similar to that of wild-type sterile lemma (Figure 3l).

Various types of abnormal spikelet were observed in asp1-1: organ initiation and development appeared to be arrested and impaired at various stages. One type of asp1-1 spikelet had only one rudimentary glume, which elongated much more than that of wild-type (almost equivalent to lemma size) (Figure 3m), whereas another type had only one sterile lemma and one rudimentary glume, with abnormal size and identity (Figure 3n). In other cases, the spikelet consisted of one lemma, one elongated sterile lemma and one rudimentary glume (Figure 3o). These arrested spikelets suggest that the function of the spikelet meristem is compromised. Consistent with this, we found a dome-like structure in some spikelets, suggesting that the SM was prematurely terminated (Figure 3m,r,s). Interestingly, we observed some spikelets with a cone-like structure without edges and with identity similar to that of palea/lemma (Figure 3p,q). This type of spikelet developed largely normal floral organs and produced fertile seed.

SEM analysis of early developmental stages

Next, we examined the early stages of inflorescence and spikelet development by scanning electron microscopy. The abnormality in asp1-1 appeared at the stage of primary branch differentiation. The primary BM initiated in a spiral phyllotaxy in wild-type, but this pattern was disturbed and became irregular in asp1-1 (Figure 4a,b). This observation is consistent with the phenotype of mature panicles described above (Figure 1c,e). In wild-type, the IM (rachis meristem) degenerated after production of the primary BM. In asp1-1, by contrast, degeneration of the IM was delayed and the IM was still present at a stage when it had degenerated in wild-type (Figure 4c,d). Spikelet development was roughly synchronized within a primary branch in wild-type, whereas various developmental stages of spikelet and branch were observed within a primary branch in asp1-1 (Figure 4e,f). Thus, the arrangement of primary BM initiation and developmental timing of the IM and SM were compromised in the early inflorescence development of asp1-1. In addition, the bracts at the base of the spikelet, which are usually repressed in wild-type, were extended in asp1-1 (Figure 4e,f).

Figure 4.

 SEM images of inflorescence and spikelet development.
(a, b) Developing inflorescences at the primary branch differentiation stage in wild-type (a) and asp1-1 (b). Arrowheads indicate the IM.
(c, d) Developing inflorescences at the secondary branch differentiation stage in wild-type (c) and asp1-1 (d). The IM was not detected in the wild-type (arrow) due to its degeneration; by contrast, the IM remained in asp1-1 (arrowhead).
(e, f) Developing spikelets on a primary branch in wild-type (e) and asp1-1 (f). All spikelets were at the same developmental stage in wild-type, whereas various developmental stages were observed in asp1-1.
(g) Branching pattern at an early stage in wild-type.
(h) Arrested branches and spikelets in asp1-1. The arrow indicates a putative arrested spikelet meristem, and asterisks indicate arrested branches.
(i) An arrested spikelet, the meristem of which was not maintained (arrow) in asp1-1.
(j) An abnormal spikelet with elongated meristem-like structure (arrow) in asp1-1.
(k) A developing spikelet in which a ring-like primordium (arrow) was initiated in place of the palea/lemma primordia.
(l) Reversion of an SM to a BM. Branches and spikelets developed inside the rudimentary glumes. Arrows indicate developing spikelets, which were morphologically abnormal.
br, bract; brl, branch-like structure; esl, elongated sterile lemma; le, lemma; or, putative arrested spikelet organ; pa, palea; pbm, primary branch meristem; rg, rudimentary glume; sbm, secondary branch meristem; sl, sterile lemma; sm, spikelet meristem.
Scale bars = 50 μm (a–k) and 200 μm in (l).

In wild-type, the secondary BM and the SM initiated sequentially from the primary BM, followed by the development of secondary branches and spikelets with pedicels (Figure 4e,g). In asp1-1, by contrast, we frequently observed a cluster of arrested meristems and branches, which may have formed simultaneously on a branch (Figure 4h).

In later stages, we frequently observed arrested spikelets, in which the meristem terminated after initiation of the palea/lemma primordia (Figure 4i) or which contained a projection similar to a malformed meristem-like dome (Figure 4j), suggesting that meristem maintenance was impaired in asp1-1. In another case, a ring-like primordium that surrounded the meristem was observed (Figure 4k). This ring-like primordium may develop into the cone-like organ described above (Figure 3p,q). Interestingly, we found a cluster of spikelets that developed inside the rudimentary glumes (Figure 4l). These spikelets were supported by branch-like structures and had their own rudimentary glumes. This suggests that the SM, after forming the rudimentary glumes, has reverted to a BM, and that the reverted BM has re-initiated spikelet primordia from the initial stages.

These observations on inflorescence and spikelet development suggest that the ASP1 gene is involved in various aspects of meristem function, such as initiation, maintenance, conversion and degeneration.

ASP1 encodes a TPL-related protein

To determine the function of ASP1 in detail, we isolated the gene by positional cloning. The asp1 locus was confined to within a 12.5 kb region containing two genes (Figure 5a). We found a mutation only in Os08g0162100 in asp1-1: a nucleotide deletion that caused a frame shift and production of a truncated protein (Figure 5b). We also found mutations causing serious defects in Os08g0162100 in three other asp1 mutants (Figure 5b and Figure S1).

Figure 5.

 Gene isolation and molecular characteristics of ASP1.
(a) Positional cloning of ASP1.
(b) Structure of the ASP1 gene and its protein. Deletion of a nucleotide is observed in asp1-1, asp1-3 and asp1-4, and the resulting stop codon is indicated by an asterisk in the protein structure. One nucleotide (G) at an exon–intron junction was changed to A in asp1-2.
(c) Subcellular localization of ASP1. (1) Localization of GFP–ASP1. (2) Bright-field image of onion epidermal cells. (3) DAPI staining of the nucleus. Scale bar = 50 μm.

Next, we searched for knockout lines of the putative ASP1 gene, and found two lines tagged with T-DNA (referred to as asp1-5 and asp1-6 below; Figure 5b). Phenotypic analysis revealed that asp1-5 and asp1-6 were very similar to asp1-1: both asp1-5 and asp1-6 showed small panicles with arrested branches and abnormal spikelets with elongated sterile lemmas and rudimentary glumes (Figure 2, Figure S3 and Table S1). Thus, we conclude that ASP1 is Os08g0162100.

ASP1 is a very large gene of approximately 8.7 kb that comprises 25 exons. It encodes a protein with a lissencephaly type 1-like homology (LisH) domain and a C-terminal to LisH (CTLH) domain in its N-terminal region, and 11 repeats of WD40 domains in its middle and C-terminal regions. The domain organization and protein sequence of ASP1 are very similar to those of TOPLESS (TPL) (BX817862) in Arabidopsis and REL2 (GQ927145 and GQ927146) in maize, which are putative transcriptional co-repressors (Long et al., 2006; Gallavotti et al., 2010).

To characterize ASP1 localization, we fused ASP1 to GFP and introduced the fusion gene into onion epidermal cells (Allium cepa). The results clearly show that the GFP–ASP1 fusion protein is localized in the nucleus, consistent with its putative molecular function (Figure 5c).

ASP1 is expressed in the meristem and lateral organ primordia

Next we examined the spatial expression pattern of ASP1 by in situ hybridization. In the reproductive phase, strong ASP1 expression was initially detected in the IM and the bract primordia, and then in the primary and secondary BMs (Figure 6a–c). In these meristems, the expression level of ASP1 was higher in the epidermal and sub-epidermal cell layers than in the inner regions (Figure 6d). ASP1 transcript was detected throughout the SM, but expression was stronger in the region where the rudimentary glumes and the sterile lemma would initiate (Figure 6e). ASP1 was subsequently expressed in both the SM and the lateral organ primordia of the spikelet, such as the rudimentary glumes, sterile lemma, lemma and palea (Figure 6f,g). These expression patterns of ASP1 are consistent with its putative function deduced from phenotypic analysis as described above.

Figure 6.

 Spatial expression pattern of ASP1.
(a) Inflorescence meristem subtended by the bract.
(b, c) Developing inflorescence at the stage of primary (b) and secondary (c) branch differentiation.
(d) Close-up view of the apical region of a developing primary branch that is initiating a secondary BM.
(e) Developing spikelet just before initiation of the rudimentary glume primordia. Arrowheads indicate the region where the primordia of rudimentary glumes (red) and sterile lemma (blue) will initiate.
(f, g) Developing spikelet that is initiating lateral organs. The arrowhead indicates the region where the palea primordia will initiate.
(h) Shoot apex of the seedling.
(i) Seminal root.
(j) Axillary bud.
(k) Basal region of the shoot, where the crown roots initiate (arrowheads).
(l) Crown root meristem.
am, axillary meristem; br, bract; cr, crown root; im, inflorescence meristem; le, lemma; lp, leaf primordium; pa, palea; pbm, primary branch meristem; rg, rudimentary glume; sam, shoot apical meristem; sbm, secondary branch meristem; sl, sterile lemma; sm, spikelet meristem.
Scale bars = 50 μm.

In the seedling stage, ASP1 transcript was detected in leaf primordia from the P0 stage, but no or very weak expression was detected in the SAM (Figure 6h). ASP1 expression was observed in the seminal root (Figure 6i). In later stages of the vegetative phase, the ASP1 transcript was detected in the AM and the primordia of the crown root (adventitious root) (Figure 6j–l).

Auxin response is compromised in asp1 mutant

In Arabidopsis, TPL has been shown to be involved in auxin signaling (Szemenyei et al., 2008). We therefore examined the phenotypes of asp1 in relation to auxin, such as axillary bud dormancy and phyllotaxy. In wild-type, axillary buds remained dormant in 28-day-old seedlings (Figure 7a). In asp1-2, by contrast, outgrowth of axillary buds was observed at this stage (Figure 7b). At the heading stage, ectopic shoots were formed from the upper nodes in asp1-2, but no such shoots were detected in wild-type (Figure 7c). These observations suggest that dormancy of the AM is de-repressed in asp1-2. In wild-type rice, leaves form in an alternating arrangement (Figure 7d); however, this arrangement was disturbed in asp1-2 (Figure 7e). Thus, both axillary bud dormancy and phyllotaxy are disturbed in asp1, suggesting that ASP1 acts in auxin signaling.

Figure 7.

 Effects of the asp1 mutation related to auxin action.
(a, b) Axillary buds in wild-type (a) and asp1-2 (b) in growing shoot 28 days after germination. Arrowheads indicate axillary buds or axillary shoots.
(c) Shoot phenotypes of the main column at the heading stage in wild-type (left) and asp1-2 (right). In asp1-2, axillary shoots grew from not only the lower nodes (arrows) but also the upper nodes (arrowheads).
(d, e) Shoot branching (tillering) pattern in wild-type (d) and asp1-2 (e). The arrangements of branches (axillary shoots) are indicated schematically in the insets.
(f) Expression levels of OsIAA20 determined by real-time PCR.
(g) Time course of OsIAA20 expression in response to exogenously applied auxin. The scale of the vertical axis is magnified in the inset.
Scale bars = 5 mm (a, b), 5 cm (c) and 1 cm (d, e).

We examined the auxin response of asp1 using the gene OsIAA20 (Os06g0166500), which is closely related to Arabidopsis IAA20 (At IAA20) and is induced by exogenous application of auxin (Arite et al., 2007). The result showed that the level of OsIAA20 in the basal part of the stem, just below the first node, was approximately three times higher in asp1-2 than in wild-type (Figure 7f). Although OsIAA20 expression was induced in response to exogenous auxin in the leaf in both wild-type and asp1-2, the extent of induction was markedly greater in asp1-2 than in wild-type (Figure 7g).

ASP1 function appears to be associated with the regulation of histone deacetylation

TPL functions as a transcriptional co-repressor by recruiting HDACs such as HDAC19, which negatively regulates transcription by chromatin modification (Long et al., 2006; reviewed by Liu and Karmarkar, 2008; Krogan and Long, 2009). In rice, elongation of seedling roots is closely associated with the activity of OsHDAC1: root elongation is inhibited in OsHDAC1 knockout mutants or by application of an HDAC inhibitor (Chung et al., 2009). Thus, we examined the effect of trichostatin A (TSA), an inhibitor of HDACs, on root elongation and axillary bud growth.

In asp1-2, the length of the seminal root decreased and the number of crown roots was reduced compared with wild-type (Figure 8a,b), suggesting that meristem activity of the seminal root and initiation of adventitious roots are compromised in asp1-2. We cultivated rice seedlings in the presence or absence of TSA. The results showed that root elongation was inhibited more efficiently by TSA in asp1-2 than in wild-type, suggesting that this mutant is hypersensitive to TSA (Figure 8c,d).

Figure 8.

 Effects of TSA treatment on asp1 and wild-type.
(a) Length of the seminal root.
(b) Number of crown roots.
(c) Effect of TSA treatment on the length of the seminal root.
(d) Phenotypes of seedlings treated with TSA.
(e) Axillary bud growth in seedlings treated with TSA in liquid culture. Seedlings were germinated and grown on filter paper immersed in sterile water in the presence or absence of TSA. Arrowheads indicate outgrowths of axillary buds.
The seminal root length and crown root number were measured 7 days after germination in (a) and (b). Then, the seedlings were transferred into 20 μm TSA solution (20 ml) or sterile water (20 ml), and incubated at 28°C for an additional 7 days with shaking (e).
Scale bars = 5 cm (d) and 5 mm (e).

Next, we examined the effect of TSA on the dormancy of the AM in liquid culture of seedlings. No axillary bud growth was observed in the absence of TSA in the wild-type, whereas TSA treatment strongly promoted bud growth (Figure 8e). By contrast, the asp1-2 seedling showed bud growth in the absence of TSA (Figure 8e). Thus, TSA treatment mimicked the phenotype of the asp1 mutant, suggesting that HDAC may be associated with ASP1 action.

Discussion

Here, we showed that mutations in the ASP1 gene caused pleiotropic phenotypes in rice development, such as a disorganized branching pattern, aberrant spikelet morphology, disarrangement of phyllotaxy, and release of axillary bud dormancy. As discussed below, these abnormalities are probably closely associated with defects in meristem fate, especially in the AMs. ASP1 encodes a transcriptional co-repressor similar to TPL/TPR in Arabidopsis (Long et al., 2006), suggesting that many genes are controlled by ASP1. This putative molecular function of ASP1 is consistent with the pleiotropic phenotypes observed in asp1. In contrast to single mutants of TPL/TPR, which exhibit no obvious phenotype (Long et al., 2006), the recessive asp1 mutant shows various developmental defects, providing a good opportunity to elucidate the function of TPL-like co-repressors in plants.

Meristem fates are compromised in asp1

In both the inflorescence and spikelet, asp1 mutants showed pleiotropic phenotypes, which are thought to be associated with fate of the reproductive meristems.

First, ASP1 is likely to be involved in the regulation of meristem determinacy. The BM is an indeterminate meristem that sequentially initiates the SM and the next order of BMs, and is finally converted to the determinate SM to form a terminal spikelet at the tip of the branch. In asp1 panicles, the primary branches were short and the number of spikelets decreased. These phenotypes suggest that the BM is precociously converted to the SM, and therefore that BM indeterminacy is partially reduced in asp1. In contrast, the IM (rachis meristem), which starts to degenerate after production of the primary BM in wild-type, was still present in the mutant at the stage of secondary BM differentiation. This observation suggests that the IM has increased indeterminacy in asp1 compared with wild-type. Thus, asp1 showed both increased and reduced indeterminacy depending on the type of meristem. This contrasting phenotype may result from differences in the genes targeted by ASP1 in the different meristem types.

Second, stem cell maintenance in the meristem appears to be partially defective. We observed that the BM and the SM are often prematurely terminated as meristem-like structures in the early stages of inflorescence and spikelet development. Truncated or degenerated branches, in addition to abnormal spikelets arrested at various stages, may also result from a defect in meristem maintenance.

Third, AM initiation also appears to be compromised in the asp1 panicle, as arrangement of the primary BM (phyllotaxy) was disorganized. Taken together, these findings suggest that the meristem fate of the IM, BM and SM appears to be de-regulated in the reproductive development of asp1.

ASP1 is expressed in reproductive meristems, such as the IM, BM and SM. ASP1 may act as a transcriptional co-repressor to repress a number of genes that regulate determinacy and maintenance of these meristems, resulting in pleiotropic phenotypes in asp1 panicles.

Molecular aspects of ASP1 function

ASP1 was found to encode a TPL-like transcriptional co-repressor. Arabidopsis possesses a TPL genes and four TPL-like genes (TPR1TPR4) (Long et al., 2006). These genes are functionally redundant, as a single mutation in one of these genes causes no obvious defect. By contrast, loss of function of all five genes results in homeotic transformation of the shoot into the root (Long et al., 2002, 2006). Rice has two additional genes related to ASP1, ASP1-RELATED1 (ASPR1) and ASPR2. In contrast to Arabidopsis, a single recessive mutation in ASP1 resulted in numerous defects in rice, suggesting that ASP1 and the ASPR genes are not completely redundant, particularly in the reproductive phase. In the vegetative phase, however, except for disorganized phyllotaxy and release of AM dormancy, serious growth defects were not observed in asp1, suggesting that these ASP1-like genes may have overlapping roles. Therefore, ASP1 and the ASPR genes may have distinct and overlapping functions depending on the organ and the developmental stage. Screening of knockout lines of ASPR genes and generating double and triple mutants would provide important information to elucidate the function of transcriptional co-repressors in this family.

WSIP1 and WSIP2 physically interact with WUS, and correspond to TPL and TPR4, respectively, in Arabidopsis (Kieffer et al., 2006; Liu and Karmarkar, 2008). Stem cell maintenance is regulated by a negative feedback loop of WUS and CLAVATA (CLV) genes. In rice, it has been shown that a CLV-like pathway negatively regulates stem cell maintenance, suggesting the existence of a conserved mechanism between Arabidopsis and rice (Suzaki et al., 2004, 2006, 2008; reviewed by Hirano, 2008). Stem cell maintenance is compromised in panicle and spikelet development in asp1, as described above. It is therefore possible that ASP1 is involved in stem cell maintenance together with a WUS-like protein in rice.

TPL appears to be involved in auxin signaling in Arabidopsis (Szemenyei et al., 2008). Consistently, the asp1 mutant showed auxin-related phenotypes in rice: dormancy of the vegetative AMs was released and arrangement of the axillary shoots (phyllotaxy) was disorganized. Expression of OsIAA20 was up-regulated in the asp1 mutant, as in the dwarf10 mutant (Arite et al., 2007), suggesting a close association of OsIAA20 gene expression and axillary bud growth. The OsIAA20 gene has also been shown to be induced by exogenous application of auxin (Arite et al., 2007). Our result indicated that OsIAA20 expression was more highly up-regulated in response to auxin in asp1 than in wild-type. This result suggests that OsIAA20 expression is controlled by feedback regulation to prevent a hyper-response to auxin in wild-type, but this regulation may be compromised in asp1. Thus ASP1 may form a repressor complex with OsIAA20 in this feedback regulation.

The function of transcriptional co-repressors such as TPL and LUG is closely associated with that of histone deacetylases (HDACs), which transform the chromatin state from active to inactive (Long et al., 2006; Gonzalez et al., 2007; Liu and Karmarkar, 2008; Krogan and Long, 2009). Our results indicate that the asp1 mutant is more sensitive to TSA than wild-type with respect to inhibition of root elongation. It is possible that HDAC activity is lowered in asp1 because the ability of the transcriptional co-repressors to recruit HDAC is reduced overall due to loss of function of ASP1. Liquid seedling culture in the presence of TSA promoted axillary bud growth in wild-type, i.e. inhibition of HDAC activity by exogenous TSA mimicked the asp1 phenotype. Taken together, these results support the idea that ASP1 function may be associated with HDAC activity.

Meristem fate and auxin action

A number of studies have reported mutants with sparse inflorescences in maize. Many of these mutants are related to auxin synthesis or its action. The genes sparse infloresence1 and vanishing tassel2 encode enzymes involved in auxin biosynthesis (Gallavotti et al., 2008a; Phillips et al., 2011), whereas barren inflorescence2 (bif2) encodes a protein kinase orthologous to Arabidopsis PINOID (PID), which is required for regulating the localization of PINFORMED1 (PIN1) protein, an efflux carrier of auxin (McSteen and Hake, 2001; McSteen et al., 2007; Skirpan et al., 2009). ba1 encodes a putative transcription factor that physically interacts with BIF2 (Gallavotti et al., 2004; Skirpan et al., 2008). In rice, the LAX gene encodes a protein that is closely related to ba1, and OsPID1, a PINOID ortholog, shows meristem-associated expression in rice inflorescences (Komatsu et al., 2003a; Morita and Kyozuka, 2007). Localization studies for ZmPIN1a and reporter assays using an auxin-responsive promoter indicate that auxin maxima are associated with sites of formation of meristems and organ primordia in maize (Gallavotti et al., 2008b). These finding indicate that auxin action is required for proper initiation and maintenance of the BM and the SM in grasses.

As discussed above, meristem fate is affected by asp1 mutation in rice, but the effects were much milder than the meristem defects observed in sparse inflorescence mutants of maize. It is possible, however, that the meristem defects in reproductive development in asp1 result from disturbed auxin signaling. ASP1 was expressed strongly in the outer layer of the meristem, especially the L1 layer, the presumptive region of organ primordia initiation in the meristem, and the apical region of organ primordia. These spatio-temporal expression patterns of ASP1 are highly similar to the auxin response maxima observed in maize development (Gallavotti et al., 2008b). This suggests a putative association between ASP1 function and auxin action.

In maize, REL2 encodes a transcriptional co-repressor that is similar to Arabidopsis TPL and rice ASP1, and regulates AM fate (Gallavotti et al., 2010). The rel2 mutation enhances branching of the ear in ra1 and ra2 mutants, suggesting that this mutation converts the determinate fate of the spikelet pair meristem into indeterminate BM fate. This observation contrasts with the mutant phenotype of rice asp1, in which transition of the BM to the SM is accelerated. This apparent inconsistency may result from the absence in rice of an ortholog of ra1, a key regulator of the ramosa pathway, and the possibility that the targets of ASP1 are different from those of REL2. REL2 is likely to be recruited by RA1 to the promoter of its target gene (Gallavotti et al., 2010). It would be of great interest to know which proteins physically interact with ASP1, and which genes are targeted by a repressor complex incorporating ASP1. Kernel number appears to be decreased in the rel2 single mutant, and in rel2 ra1 and rel2 ra2 double mutants (Gallavotti et al., 2010). This suggests that REL2 is involved in spikelet development in maize, similar to ASP1 in rice. Further studies on rice ASP1 and maize REL2 should reveal the common and distinct roles of these transcriptional co-repressors in grasses, and provide a deeper understanding of their function in plant development.

Experimental Procedures

Plant materials

The asp1-1 mutant was identified among a collection of mutants that were roughly selected from the TOS17 mutant panel as plants showing abnormal spikelets (Miyao et al., 2007). One strain, CM1272 (asp1-2), which is registered in the Oryzabase database (http://www.shigen.nig.ac.jp/rice/oryzabase/top/top.jsp) as undulate rachis2 (ur2) but has not yet been described in a research paper, was kindly provided by Dr Atsushi Yoshimura (Faculty of Agriculture, Kyushu University, Fukuoka City, Japan). 20KY-1612 (asp1-3) and 99TCM-2827 (asp1-4) were identified among mutants obtained by γ-ray irradiation or N-methyl-N-nitrosourea (MNU) treatment, respectively, of Taichung 65 (T65). Two T-DNA-tagged lines, 3A-04974 (asp1-5) and 2D-20229 (asp1-6), were kindly provided by Dr Gynheung An (Department of Plant Systems Biotech, Kyung Hee University, Yongin, South Korea) (Jeon et al., 2000; Jeong et al., 2006). Nipponbare and Taichung 65 (T65) were used wild-type strains for comparing phenotypes and in situ expression analysis, and for examining the responses to auxin and TSA, respectively.

Morphological analysis

Observations by scanning electron microscopy (SEM) were performed as described previously (Suzaki et al., 2004; Toriba et al., 2010). Young panicles and flowers were fixed in a solution containing 4% w/v paraformaldehyde, 0.25% w/v glutaraldehyde, and 0.1 m sodium phosphate (pH 7.2) at 4°C overnight. After replacement of the fixation solution with 3-methylbutyl acetate, the samples were critical point-dried, sputter-coated with platinum, and observed using a JSM-820S scanning electron microscope (Jeol, http://www.jeol.com/) at an accelerating voltage of 10 kV.

Isolation of ASP1

The primers used for PCR amplification are shown in Table S2. The ASP1 locus was mapped using 2400 F2 plants from a cross between asp1-1 and Kasalath (ssp. indica), and was confined to within a 12.5 kb region containing two genes. A severe mutation was detected in one of the genes, Os08g0162100. The ORF of the putative ASP1 gene was predicted by sequencing the amplified ASP1 cDNA. In allelic tests, asp1-2 and asp1-5 were crossed with asp1-1 and each other. To identify the T-DNA insertion positions in asp1-5 and asp1-6, primers listed in Table S2 were used. For analysis of subcellular localization, constructs to produce a GFP–ASP1 or ASP1–GFP fusion protein were bombarded into onion epidermal cells by a particle-mediated DNA delivery system (PDS1000/He, Bio-Rad, http://www.bio-rad.com/). After overnight incubation at 22°C in the dark, the florescence of GFP was detected by a confocal microscanning laser microscope (Nikon, http://www.nikon.com/). Lastly, cells were stained with DAPI to visualize the nucleus.

In situ hybridization

Samples were fixed in 4% w/v paraformaldehyde and 0.25% w/v glutaraldehyde in 0.1 m sodium phosphate buffer at 4°C overnight, and then dehydrated in a graded ethanol series, replaced 100% ethanol with xylene, and embedded in Paraplast Plus (McCormick Scientific, http://www.mccormickscientific.com/). Microtome sections were applied to silane-coated slides. Digoxigenin (DIG)-labeled RNA, prepared using a DIG RNA labeling kit (Roche Diagnostics, http://www.roche.com) was used as a probe for in situ hybridization.

To prepare the probe for ASP1, a 612 bp fragment in the 3′ region of ASP1 cDNA fragment was amplified using the primer set listed in Table S2. The amplified fragment was cloned into a T-vector (Merck Chemicals, http://www.merck-chemicals.jp/). After linearization of the resulting recombinant plasmid, the RNA probe was transcribed by T7 polymerase (Roche Diagnostics). In situ hybridization and immunological detection of the signals were performed as described by Kouchi and Hata (1993).

Measurement of transcript levels and treatment with auxin and TSA

To determine the transcript levels of OsIAA20, RNA was isolated from the basal part of a 7-day-old seedling using TRIsure reagent (Bioline, http://www.nippongenetics.eu/) and treated with DNase I (Takara, http://www.takara-bio.com/) to remove genomic DNA. To examine the auxin response of OsIAA20, a small section (approximately 1 cm2) was cut from the third leaf blade of a 7-day-old seedling, floated on sterile water for 2 h, and transferred into sterile water containing 50 μm indole-3-acetic acid (IAA). After incubation at 30°C under continuous light for the indicated period, RNA was isolated as described above by the same methods. First-strand cDNA was synthesized from 1 μg total RNA using a PrimScript RT reagent kit (Takara) according to the manufacturer’s instructions. Real-time PCR experiments were performed using Power SYBR Green Master Mix (Applied Biosystems, http://www.appliedbiosystems.com/) and an ABI 7300 real-time PCR system (Applied Biosystems). The ACTIN1 gene was used as a control, and the primers used are listed in Table S2.

To examine the effect of TSA on roots, seeds were germinated on filter paper immersed in 100 μm TSA solution for 7 days. To assess axillary bud outgrowth, 7-day-old seedlings treated by the same method were transferred into 20 ml of 100 μm TSA solution in a 100 ml flask and incubated at 28°C for an additional 7 days with shaking (liquid seedling culture).

Acknowledgements

We thank Drs A. Yoshimura (Kyushu University), A. Miyao, H. Hirochika (National Institute of Agrobiological Sciences) and G. An (Kyung Hee University) for kindly providing asp1 alleles. We thank K. Ohsawa-Yamamoto for technical assistance and technicians at the Institute for Sustainable Agro-Ecosystem Services of the University of Tokyo for cultivation of rice. This research was supported in part by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (numbers 20380005 and 23012011) (to H.-Y.H.), the Global Center-of-Excellence (COE) Program (Integrative Life Science Based on the Study of Biosignaling Mechanisms) of the Ministry of Education, Culture, Sports, Science and Technology, Japan (to A.Y. and Y.O.), and a fellowship from the Ajinomoto Scholarship Foundation (to A.Y.).

Accession numbers: The accession numbers for ASP1, ASPR1 and ASPR2 are Os08g0162100 (AB638269), Os01g0254100 (AB638270) and Os03g0254700 (AB638271), respectively.

Ancillary