BRASSINOSTEROID UPREGULATED1, encoding a helix-loop-helix protein, is a novel gene involved in brassinosteroid signaling and controls bending of the lamina joint in rice.

Brassinosteroids (BRs) are involved in many developmental processes and regulate many subsets of downstream genes throughout the plant kingdom. However, little is known about the BR signal transduction and response network in monocots. To identify novel BR-related genes in rice (Oryza sativa), we monitored the transcriptomic response of the brassinosteroid deficient1 (brd1) mutant, with a defective BR biosynthetic gene, to brassinolide treatment. Here, we describe a novel BR-induced rice gene BRASSINOSTEROID UPREGULATED1 (BU1), encoding a helix-loop-helix protein. Rice plants overexpressing BU1 (BU1:OX) showed enhanced bending of the lamina joint, increased grain size, and resistance to brassinazole, an inhibitor of BR biosynthesis. In contrast to BU1:OX, RNAi plants designed to repress both BU1 and its homologs displayed erect leaves. In addition, compared to the wild type, the induction of BU1 by exogenous brassinolide did not require de novo protein synthesis and it was weaker in a BR receptor mutant OsbriI (Oryza sativa brassinosteroid insensitive1, d61) and a rice G protein alpha subunit (RGA1) mutant d1. These results indicate that BU1 protein is a positive regulator of BR response: it controls bending of the lamina joint in rice and it is a novel primary response gene that participates in two BR signaling pathways through OsBRI1 and RGA1. Furthermore, expression analyses showed that BU1 is expressed in several organs including lamina joint, phloem, and epithelial cells in embryos. These results indicate that BU1 may participate in some other unknown processes modulated by BR in rice.

We have previously isolated the brd1 mutant that had a defective BR biosynthetic gene, BRD1 (OsBR6ox/OsDWARF) in rice (Mori et al., 2002). Analyses of brd1 mutants clarified many BR effects in rice. The brd1 mutant showed severe dwarfism, sterility, small seeds, abnormal vascular bundles, shortened internodes and inhibition of elongation of the coleoptile and mesocotyl in the dark (Mori et al., 2002;Hong et al., 2002). The lamina joint, which is a border region between the leaf blade and sheath, is an especially sensitive organ to BR and it is severely bent after exposure to active BRs (Wada et al., 1981). Loss of function mutants of BR biosynthesis, d2 and d11, have erect leaves because of BR deficiency (Hong et al., 2003;Tanabe et al., 2005). In addition, BR-deficient rice mutants have much shorter grains than wild-type (WT ) (Mori et al., 2002;Hong et al., 2003;Tanabe et al., 2005). On the other hand, hyperproduction of BR by overexpression of DWARF4, which encodes a BR biosynthetic gene, increased seed size and yield in Arabidopsis (Choe et al., 2001). In rice, overexpression of DWARF4 increased grain filling, but there was no mention of its effect on grain size (Wu et al., 2008). Some components of BR signaling have been also identified in rice. OsBRI1 and OsBZR1 have been respectively identified as counterparts of BRI1 and BZR1 in Arabidopsis and their loss of function mutants display BR deficient mutant -like phenotypes such as erect leaves (Yamamuro et al., 2000;Bai et al., 2007). OsBZR1 directly binds to the promoter of DWARF AND LOW TILLERING (DLT) gene, whose 8

Isolation of BR-induced Genes
To identify novel BR-regulated genes, RNA were obtained from young shoots of 50-day-old brd1 mutants, with or without 40 nM of BL treatment for 24 hrs and analyzed using a 22K rice microarray. 96 of 21,938 genes were up-regulated by BL application, as defined by a more than 2-fold difference in the fold change values that represent ratios of hybridization signals between BL-treated and control brd1 plants.
Several Arabidopsis genes reportedly involved in BR signaling, such as BEE1-3 and BIM1-3, contain basic/helix-loop-helix (bHLH) motifs. Therefore, we searched the bHLH motif among the 96 upregulated genes and identified three genes. In this paper, we report one of the three genes with the bHLH motif that we have designated as BU1 (Os06g0226500) (Fig.1A), whose induction by exogenous BL has been confirmed by real-time PCR analysis both in brd1 and WT plants (Supplemental Fig. S1 and Fig. 4A).
We will provide details of our microarray results and other up-regulated genes in future publications.

In silico Analysis of BU1 Protein
Transcription factors with the bHLH motif regulate a large array of target genes and play a pivotal role in cellular signaling in living organisms (Ledent and Vervoort, 2001;Toledo-Ortiz et al., 2003;Li et al., 2006). A bHLH protein has two functional domains, the basic region and the HLH region. The former is required for binding to DNA to regulate the expression of genes, while the latter is required for interaction with other bHLH proteins for hetero-or homodimerization (Murre et al., 1989). In rice, more than 160 bHLH genes have been identified (Li et al., 2006). Although the BU1 protein is not described by Li et al. (2006), it has an HLH motif and is included in the bHLH protein family (Fig. 1A). Since BU1 lacks the basic region needed for binding to DNA, it is classified into Group D of bHLH proteins, the non-DNA binding protein family (Li et al., 2006). In rice, there are three proteins which have high identity with BU1. However, Os02g0747900 may not be a functional protein, since Os02g0747900 lacks one helix motif of the HLH domain (Fig. 1A). BU1 and these proteins have a high similarity with PACLOBUTRAZOL RESISTANCE 1 (PRE1) of Arabidopsis, an HLH protein which also lacks the basic region. Since PRE1 functions in gibberellin (GA) signaling (Lee et al., 2006), we expect that BU1 may also play an important role in plant hormone signaling, especially BR. Therefore, we used phylogenetic analysis to determine the similarity of the HLH domain of BU1 to those of BEE1 and BIM1, which are bHLH proteins related to BR signaling in Arabidopsis (Fig. 1B). Since the NJ tree indicates that the similarity between BU1 and BEE1 or BIM1 is very low, BU1 and its homologs may be a novel protein family involved in BR signaling. In contrast to BU1, because the induction of the three homologs by exogenous BL was not detected in the WT shoot (data not shown), we preferentially analyzed BU1 in this paper.

BR Action Phenocopies Rice Plants Overexpressing BU1
To characterize BU1, we made transgenic rice plants overexpressing BU1 (BU1:OX) under the control of a maize Ubiquitin promoter. Many T0 plants showed enhanced bending of the lamina joint, but most of them were sterile. As a result, we obtained seeds from only two overexpressed lines, BU1:OX-3 and 4 ( Fig. 2A). The expression level of BU1 is higher in OX-3 and much higher in OX-4 than in WT plants  Fig. 2D and E). Since, it is known that bending of the lamina joint and grain size are regulated by BR (Yamamuro et al., 2000;Mori et al., 2002;Hong et al., 2003), these results suggest that BU1 may be involved in BR signaling or biosynthesis. In addition, spikelet numbers were drastically reduced in OX-4, although they were similar in WT and OX-3 (data not shown). This observation may be the result of poor growth due to high transcript levels of BU1 in OX-4.

BU1
:OX also has some characteristic abnormalities in the pattern of internode elongation ( Fig. 2F), progression of tillers in nodes of aerial part (Fig. 2G) , and lignification and development of crown roots in these nodes (Supplemental Fig. S2).
The first internode was severely shortened, but the 5th and 6th internodes were slightly elongated. Panicles did not emerge perfectly from the flag leaf sheath as a result of the shortened first internode in OX-4.

BU1 Is Involved in BR Signaling, Not BR Biosynthesis
In Arabidopsis, a BR signal component, BZR1, was isolated by genetic screens using elongation of hypocotyls with or without BRZ (Wang et al., 2002). Similar to hypocotyls in Arabidopsis, elongation of the mesocotyls and 2nd internodes in rice grown in complete darkness was severely dependent on BR (Yamamuro et al., 2000).
To ascertain whether BU1 functions in BR signaling or biosynthesis, we used the inhibition of elongation of the mesocotyl and 2nd internode in dark with or without BRZ, similar to the test performed in Arabidopsis hypocotyls. Mesocotyls and 2nd internodes in WT were elongated in half-strength MS medium without BRZ, but shortened in half-strength MS medium containing BRZ ( Fig. 3A and B). On the other 1 1 hand, in BU1:OX-4 , the length of the 2nd internodes was not affected by BRZ.
Mesocotyls of OX-4 were greatly elongated compared to WT and the elongation was not much affected by BRZ. These results indicate that BU1:OX-4 is less sensitive to BRZ than wild-type.
Bending of the lamina joint is very sensitive to BR. BU1:OX shows enhanced bending of the lamina joint as described above. In contrast, the loss of function mutant of OsBRI1, d61, has erect leaves and it is less sensitive to BL. We generated d61 plants overexpressing BU1 and observed the resulting phenotypes, especially their lamina joints. BU1 overexpressor in the d61 background showed increased bending of the lamina joint (Fig. 3C) and a pattern of internode elongation similar to BU1:OX-4 ( Fig.   3D), indicating that overexpression of BU1 complemented the BR-deficient phenotype of d61.
In addition, the WT and OX-4 did not differ markedly in their level of expression of BRD1, a gene that encodes a P450 that catalyzes the conversion of 6-deoxocastasterone (6-Deoxo CS) to castasterone (CS) (Supplemental Fig. S3A). We also measured endogenous BR contents in BU1:OX-4 and WT. There were no obvious differences between OX-4 and WT in the amounts of both CS (presumably the most active endogenous BR in rice) and 6-Deoxo CS, the main precursor of CS (Supplemental To test the response to other plant hormones, we performed real-time PCR analysis using RNA samples of seedlings subjected to a variety of hormones and chemical treatments. Previous research showed that genes induced by auxin overlapped with those induced by BR, and both hormones synergistically modulate plant architecture (Yin et al., 2002;Nemhauser et al., 2004). However, expression of BU1 was not induced by IAA or by GA 3 . On the other hand, ABA, a known antagonist to BR response, repressed expression of BU1 (Fig.4A). The mechanism whereby ABA represses the primary signaling outputs of BR in Arabidopsis (Zhang et al., 2009) may be conserved in rice as well.
BU1 was greatly upregulated by exogenous BL in WT plants. In addition, BU1:OX showed many BR-related phenotypes. These results suggest that BU1 may act as an early response gene in BR signaling. We tested the response of BU1 expression to exogenous BL treatment with cycloheximide (CHX), an inhibitor of de novo protein synthesis. A previous report in Arabidopsis indicated that induction of primary response genes by BR signaling, e.g., BEE1-3, did not need de novo protein synthesis (Friedrichsen et al., 2002). Induction of BU1 by exogenous BL was not affected by CHX (Fig. 4B). This result indicates that induction of BU1 does not need de novo protein synthesis and BU1 is a primary response gene to BR signaling.
To determine whether the induction of BU1 by endogenous BRs is through OsBRI1, a transmembrane BR receptor, we compared the response of BU1 to exogenous BL between the d61-2 (OsbriI) mutant and its WT, T65 (Fig. 4C). In T65, induction of BU1 by exogenous BL was 8.7 times more than the control, and in the d61-2 mutant, 6.6 times greater than the control. The expression level of BU1 in d61-2 after BL treatment was about one-half compared to that in T65. These results indicate that BU1 is less sensitive to exogenous BL in d61-2 compared to T65 and it is induced through OsBRI1 by BRs. On the other hand, the sensitivity of BU1 to exogenous BL still remained in d61-2. This result suggests that the induction of BU1 by BRs may be also through another pathway distinct from the one associated with OsBRI1. Recent research showed that RGA1, encoding the rice heterotrimeric G protein alpha subunit, was also involved in BR signaling in rice (Wang et al., 2006;Oki et al., 2008). To determine whether the induction of BU1 by BRs is not only through OsBRI1 but also through RGA1, we compared the response of BU1 to exogenous BL between T65 and the d1 mutant, a loss of function mutant of RGA1 in the T65 background ( Fig. 4D). In the WT (T65), induction of BU1 by BL was 5.9-fold to control but only 2.7-fold in the d1 mutant. The expression level of BU1 in d1 after BL treatment was about one-third compared to that in T65. These results indicate that BU1 is less sensitive to exogenous BL in d1 and induced through both OsBRI1 and RGA1 pathway by BRs.

Expression Analysis of BU1
We compared the expression level of BU1 in various organs by real-time PCR analysis (Fig. 5A). The expression level of BU1 is higher in the lamina joint before bending than the lamina joint after bending. Aside from the lamina joint, BU1 is highly expressed in the leaf blade, leaf sheath and the panicle at heading stage. To analyze the expression pattern of BU1 in detail, we inserted a 2 kb segment of the promoter region of BU1 to drive the GUS gene, then transformed rice plants with this gene cassette. We performed the histochemical GUS assay using these transgenic plants (Fig. 5B). GUS staining was observed in various organs. We examined the lamina joint region of the second and 8th leaf (Fig. 5B (1 and 2) 9-day-old seedlings ( Fig. 5B (3)). The veins in the leaf blade and sheath were well stained ( Fig. 5B (4 and 5)). Using a light microscope, we observed staining in phloem in transverse sections of leaf blades (Fig. 5B (6)). In the reproductive stage, staining was observed in the ovule and filaments before anthesis (Fig. 5B (7)). In particular, staining was observed in the vascular bundles of the ovule, lemma and palea ( Fig. 5B (8, 9, 10)).
In the seed, epithelial cells were stained in the embryo 3 days after imbibition ( Fig. 5B (11, 12)). The expression of BU1 in the lamina joint, verified by two independent experiments ( Fig. 5A and B (2)), strongly suggests that BU1 is involved in bending of the lamina joint in WT plants. In addition, the results of GUS staining suggest that expression of BU1 that was detected by real-time PCR in the leaf blade and panicle may be localized in the vascular bundles (phloem) in these organs.
Expression of BU1 is higher in the embryo than in the endosperm; however, BU1 expression is not influenced by imbibition (Fig. 5C). Although elongation of the 2nd internode and mesocotyl is very sensitive to BR and these organs were elongated in BU1:OX compared to WT, in the presence of BRZ in dark ( Fig.3A and B), GUS staining was not observed in 10-day-old seedlings grown in the dark (data not shown).
This result suggests that BU1 may not be involved in BR response in the absence of light in WT plants.
To determine the intracellular localization of BU1 protein, we constructed a plasmid DNA for fusion protein in which BU1 was fused to the C-terminus of an enhanced green fluorescent protein (eGFP), that we subsequently introduced into rice coleoptiles via particle gun bombardment. Dispersed fluorescent signal of eGFP-BU1 fusion protein throughout the coleoptile cells was observed (Supplemental Fig. S4). Transcription factors are basically localized in nucleus. Therefore, BR signal transduction by BU1 may be mediated by a different mechanism.

Rice Plants Suppressing Both BU1 and Its Homologues Show Erect Leaves
We generated transgenic plants suppressing BU1 using the 3'-UTR region (340bp) as a trigger region (BU1:RNAi), and investigated the effect of silencing by morphological evaluation. The angle of the lamina joint and other morphological characters in rice suppressing BU1 did not markedly change as compared to WT plants ( Fig. 6A to C). As described in Fig. 1A, there are three proteins which have a high similarity in amino acid sequence to BU1. We supposed that these three proteins may have functions that overlapped with those of BU1, therefore we generated transgenic plants suppressing both BU1 and these homologs using the whole ORF region (270bp) of BU1 as a trigger region (BU1F:RNAi). BU1F:RNAi plants showed erect leaves in contrast to BU1:OX (Fig. 6D to F). In addition, we compared the sensitivity between BU1F:RNAi and WT to BL by using the lamina joint bending assay. In BU1F:RNAi, the angles of the lamina joint were narrower than those of the WT at all concentrations of BL (Fig. 6G), indicating that the sensitivity of BU1F:RNAi to BL is lower than that of the WT. Moreover, the expression levels of BU1 homologs were reduced in the lamina joint of BRZ-treated WT seedlings as BU1 (Supplemental Fig. S5)  BU1:OX shows abnormal phenotypes that includes previously known and also possibly new BR effects. The length of the first internode of culm is shortened but that of the 5th internode is elongated, and the 6th internode protrudes in BU1:OX ( Fig. 2F and G). d61 has a relatively longer first internode and shorter lower internodes (Fig. 3D) in comparison to BU1:OX. This result suggests that alternation of length of internodes in BU1:OX may be a characteristic BR effect on growth and development of culm.
Progression of tillers in lower aerial nodes (Fig. 2F)   growth-related genes but homeostatic-related genes. Therefore, genes regulated by BU1 in the epithelium may be different from those in the phloem.
In this study, we showed that BU1 is a novel primary response gene to BR signaling through both OsBRI1 and RGA1. Previous reports describe that OsBZR1 functions as a transcriptional repressor for downregulating DLT and BR biosynthesis genes such as AtBZR1 (He et al., 2005;Bai et al., 2008;Tong et al., 2009). On the other hand, real-time PCR clearly showed that BU1 is a primary response gene upregulated by exogenous BL, suggesting that BU1 may be upregulated by a novel transcription factor (maybe an activator) distinct from OsBZR1.
In addition, we compared the expression of known BR-marker genes between WT and two independent BU1:OX lines. However, BR-marker genes, OsXTR1, OsXTR3, OsBLE2 and OsBLE3 (Uozu et al., 2000;Yang et al., 2003Yang et al., , 2006, were not upregulated in BU1:OX (data not shown). Therefore, downstream genes of BU1 may be different from known BR-related genes regulated by the OsBZR1 pathway in rice. BU1 protein is categorized as a putative non-DNA binding bHLH protein, since it lacks a basic region (Fig. 1A). Therefore, downstream genes of BU1 may be regulated by more complex mechanisms, different from the general mechanism of transcriptional regulation of target genes by a transcription factor. Molecular mechanism of non-DNA binding HLH proteins is understood well in human Inhibitor of DNA binding (Id) proteins. Id proteins heterodimerize with other DNA binding bHLH protein partners via HLH motif and abolish their functions as transcription factors, binding to the cis-element of their target genes (Benezra et al., 1990;Sun et al., 1991). Consequently, expression of their target genes by partners of Id proteins is abolished by Id proteins.
Likewise, BU1 may also interact or form a complex with putative BR-negative regulators (maybe bHLH proteins), and inhibit their functions as transcription factors.
Taking into account these studies and perspectives, we propose a model of BR signaling in rice (Fig. 7). BU1 has high similarity with PRE1, encoding a HLH protein involved in GA signaling in Arabidopsis (Lee et al., 2006). In addition, elongation of internodes in darkness is affected not only by BRs but also by GAs, suggesting that BU1 may also be involved in GA signaling. To test whether BU1 is also a GA signal component, we performed two kinds of experiments, the 2nd leaf sheath elongation assay and the α -amylase induction assay. First, we compared the effect of paclobutrazol, an inhibitor of GA biosynthesis, against 2nd leaf sheath elongation in BU1:OX and its WT. BU1:OX showed shortened leaf sheaths and displayed sensitivity to paclobutrazol similar to the WT (Supplemental Fig. S6A and B). Second, we examined α -amylase induction by GA in BU1:OX-4 and BU1F:RNAi-6 seeds by using the starch plates. For example, slender rice, a constitutive GA response mutant, produces and secretes amylase from embryoless half seeds without GA addition and a plaque zone around half seeds is observe on agar plates after staining by iodine (Ikeda et al., 2001). Production of amylase from embryoless half seeds of OX-4 and RNAi was observed only on GA applied agar plates (clear zones), but not on control plates. There was no significant difference in sensitivity to GA relative to the WT (Supplemental Fig. S6C). These results indicate that BU1 is not directly involved in GA signaling. Our study suggests that many non-DNA binding bHLH proteins may play an important role not only in GA signaling but also in some plant hormone signaling similar to the function of human Id proteins, although little is known about functions of HLH proteins in plants.

Plant Material and Growth Condition
All transgenic and WT plants were generated from O. sativa japonica cultivar Nipponbare, except for the d61-2 mutant, d1 mutant and their WT, which were obtained from japonica cultivar Taichung 65 (T65). All transgenic plants were grown in isolated green houses at 28°C. were sown on half-strength MS medium supplemented with 2 μ M paclobutrazol (Wako pure chemicals) or its equal volume of solvent (DMSO), and grown at 25°C under long-day conditions. After a week, the lengths of the 2nd leaf sheath were measured.
The method described by Sekimata et al. (2001) was used as reference in this experiment.

Oligo DNA Microarray Analysis
We used a rice 22K oligo DNA microarray kit (G2554A; Agilent Technology, Santa Clara, CA) containing 21,938 oligonucleotides based on the sequence data of the rice full-length cDNA project (Kikuchi et al. 2003;Yazaki et al., 2004). Total RNA was extracted using Isogen (Wako Pure Chemicals) and further purified with an RNeasy Plant Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. RNA amplification, labeling, hybridization, scanning and image analysis were performed according to the manufacturer's instructions (Agilent Technology). The microarray results were filtered to select candidate clones with P-value log ratios of less than 0.01.
A 2.0-fold expression cutoff was applied, and the cases in which dye swapped replications passed this cutoff were scored as differential expression.

Quantitative Real-Time RT-PCR Analysis
After RNA extraction, first-strand cDNAs were synthesized from equal amounts of total RNA (1 μ g / reaction) with a PrimeScript II 1st strand cDNA Synthesis Kit

Lamina Joint Bending Assay
Detailed method is described in Fujioka et al. (1998). Seeds of transgenic plants and WT were dehusked, sterilized and grown on half-strength MS medium at 25°C under long-day conditions. After a week (when 3rd leaf began to emerge), 1 μ l of ethanol:DMSO (9:1, v/v) solution containing 0, 10, 100, 1000 ng BL was spotted on the lamina joint of the 2nd leaf of seedlings. After 3 days, the angles of the lamina joint were measured.

Quantification of Endogenous BRs
Mature shoots (BU1:OX, 50g fresh weight; Vector control , 50g fresh weight) grown in soil in an isolated green house for two months were used for the experiment.
The detailed method is described in Mori et al. (2002).

Subcellular Localization of BU1 Protein
The BU1 cDNA fragment was cloned into a multi-cloning site in frame at the C-terminus of EGFP in the pSAT6-EGFP-C1 vector (Tzifara et al., 2005). This construct was introduced into coleoptiles of 5-day old rice seedlings by particle bombardment method (Bio-Rad Biolistic PDS-1000/He Particle Delivery System, Bio-Rad, Hercules, CA;Heiser, 1992;Sanford et al., 1993). After incubation at 28°C for 24h, GFP fluorescence was observed using a fluorescent microscope.

α -amylase Induction Assay
Five embryoless half seeds per plate were dehusked and sterilized. These half seeds were placed on starch plates (0.2% (w/v) starch and 2% (w/v) agar) with or without 2 μ M of GA 3 . These plates were incubated for 5 days at 28°C in darkness. For detection of secreted amylase, iodine vapor was applied to plates (Ueguchi-Tanaka et al., 2000).
Purple zones indicated unhydrolyzed starch, while clear zones (plaques, not turned to purple) indicated starch hydrolyzed by the amylase present in the half seeds.

Supplemental Data
The following materials are available in the online version of this article.   was shorter, but its two lower internodes were longer than WT. Bar = 50 cm.     OsBRI1mutant (d61-2) and RGA1 mutant (d1). In addition, BU1 is the primary response (upregulated) gene to BR signaling. Therefore, BU1 may be a novel mediator of a distinct pathway from OsBZR1, because OsBZR1 functions as the transcription repressor. Moreover, regulation of downstream genes by BU1 may be through a complex mechanism like the regulation of human Id protein.