To obtain brassinosteroid-related mutants, we performed large-scale screening of rice mutant collections that were produced by using a retrotransposon (Tos17), chemical mutagens, and γ-ray irradiation. Three candidate lines (d61-1N, d61-11, and d61-12) were obtained from a Nipponbare library of tissue culture-induced mutations [19]. Seeds of wild-type Nipponbare and the mutants were sterilized in 1% NaClO for 30 min and sown on Murashige and Skoog agar medium. For the initial phytohormone treatment, seedlings were grown in a growth chamber at 28˚C under continuous light for 2 weeks, and then transplanted into medium containing 10 nM brassinolide. For the subsequent field experiments, seedlings were grown for 1 month in a greenhouse, and then transplanted into a paddy field. For the gene expression analysis, we selected 10-day-old seedlings that exhibited uniform growth and adapted them to hydroponic culture for 2 days before treatment. Brassinolide treatment (100 nM) was carried out by adding the pure chemical to the culture medium. Whole seedlings were harvested 3 h after the treatment and used for RNA isolation as described below.

For the mapping of each mutant, we performed linkage analysis using an F₂ population of about 2000 plants derived from crosses between the Nipponbare-derived mutant (japonica) and wild-type Kasalath (indica). The mutation sites of all three candidate lines were mapped onto the long arm of chromosome 1, tightly linked to the d61 locus (OsBRI1 gene). The nucleotide sequence of the OsBRI1 gene (from positions –4825 to +4571, taking the translation initiation site as +1) from these three lines, Nipponbare, and Taichung 65 was determined by using a dideoxynucleotide chain-termination method using an automated sequencing system (ABI377; Applied Biosystems, Foster City, CA, USA), and was analyzed by using the Lasergene software (DNAStar, Inc., Madison, WI, USA).

For complementation of the d61-1N mutant, we amplified the wildtype (Nipponbare) genomic sequence of D61 (OsBRI1) from –4825 to +4571 (taking the translation initiation site as +1) by PCR and cloned into pBluescript II SK (Stratagene, La Jolla, CA, USA). The sequence of the amplified fragment was determined as described above, and the fragment was then cloned into the hygromycin-resistant binary vector pCAMBIA 1300. The resulting construct was introduced into Agrobacterium tumefaciens strain EHA105, and Agrobacterium-mediated transformation of the d61-1N mutant was performed as described [20]. Transgenic plants were selected on medium containing 50 mg·L^–1 hygromycin.

Total RNA was extracted from whole seedlings by using an RNeasy Plant Mini Kit (Qiagen, Valencia, CA, USA). Single-strand cDNAs were synthesized by using the Advantage RT-for-PCR Kit (Clontech, Palo Alto, CA, USA). Quantitative RT-PCR was performed with an iCycler iQ real-time PCR system (Bio-Rad Laboratories, Hercules, CA, USA). The primer sequences were 5’‑GTAGCCAG CTTGATCTCATCTC‑3’ and 5’‑GGGACGACTCTAC TGCATCA‑3’ for BU1, 5’‑TGATCCATTCCTGTACC CTG‑3’ and 5’‑TACCTTCTTCCTCCCATCTG‑3’ for CYP85A1, 5’‑CAGCTACTTGGCTATCTTGAAGCTC AGC‑3’ and 5’‑CCATTCTTGTTGAAGGTGTACTCC GTGC‑3’ for OsBRI1, and 5’‑CGCCAGTTTGGTC GCTCTCGATTTCG‑3’ and 5’‑TCAGGAGCTCCGTG CTCTTCTGGTAC‑3’ for Histone H3. These primers specifically amplified the target gene sequences. Expression levels were normalized against the values obtained for Histone H3, which was used as an internal reference gene.

As the result of a large-scale screening of rice mutant collections, we identified three lines that showed the morphological characteristics of brassinosteroid-related mutants, namely dwarf plant stature and erect leaves. These mutants were obtained from a Nipponbare library of tissue culture-induced mutations [19]. The most wellknown physiological effect of brassinosteroids on growth and development of rice is the increasing of leaf angle (leaf lamina inclination), which has been used as a sensitive bioassay for brassinosteroids [21-24]. When wildtype seedlings were treated with brassinolide, the leaf angles increased (Figure 1(a)). However, exogenously applied brassinolide did not affect the leaf angle of the mutants (Figures 1(b)-(d)), suggesting that these mutants are brassinosteroid insensitive. Previously isolated brassinosteroid-insensitive rice mutants have been categorized into two groups on the basis of leaf morphology and gross morphology. Plants exhibiting a group-1 phenotype formed only abnormal leaves with stiff, tortuous blades, and their leaf sheaths were scarcely developed. These plants did not flower, exhibit internode elongation, or bear seeds. These phenotypes were observed in d61-3, d61-4, d61-5, and d61-6 [16]. Plants exhibiting a group-2 phenotype showed a range of semi-dwarf phenotypes (about 70% - 95% height of wild-type). The leaves of these plants were erect, but the other abnormal phenotypes such as twisted leaf blades were not observed. Reproductive development of group-2 phenotype plants seems normal. These phenotypes were observed in

d61-1, d61-2, d61-7, d61-8, d61-9, dlt, and lic-1 [15, 17,18,25]. Because the gross morphology of newly isolated three mutants were ranged between group-1 and group-2 (discuss below), we hypothesized that these mutants had defects in one or more novel brassinosteroid signaling components or had novel defects in one or more previously identified brassinosteroid signaling components.

For the mapping of each mutant, linkage analysis was performed using an F₂ population of ~2000 plants derived from crosses between the mutant (japonica) and Kasalath (indica) varieties. The mutation sites of all three mutant lines were mapped onto the long arm of chromosome 1, tightly linked to the d61 locus (OsBRI1 gene). OsBRI1 encodes the rice ortholog of Arabidopsis BRI1, a leucine-rich-repeat receptor-like kinase that functions as a brassinosteroid receptor [15]. Sequence analysis of OsBRI1 revealed that one of the three lines had a single nucleotide substitution (ACC to ATC) that resulted in a substitution of amino acid residue Thr-988 (in the kinase domain) with Ile. Because the same nucleotide substitution was found in the previously characterized d61-1 mutant, in this paper, we designated this new mutant allele as d61-1N (d61-1 from wild-type Nipponbare) and renamed d61-1 as d61-1T (d61-1 from wild-type Taichung 65). The other two lines (d61-11 and d61-12) had single nucleotide substitutions (GGG to GAG and GGG to AGG) that resulted in substitution of amino acid residue Gly-643 in the leucine-rich repeats with Glu and Arg, respectively (Figure 2(a)). The mutant phenotype of d61-1N was complemented by the introduction of a 9.4- kb genomic segment containing the OsBRI1 gene (from positions –4825 to +4571, taking the translation initiation site as +1; Figure 2(b)). Based on these results, we conclude that abnormal phenotype of d61-1N is caused by a defect in the OsBRI1 gene, and considered that all three lines identified in this study contain novel d61 mutant alleles.

The previously identified brassinosteroid-insensitive d61 alleles could be categorized into two groups (weak and strong alleles) by their leaf and gross morphology [16,25]. The culm length of Taichung 65, the strain from which d61-1T was derived (hereafter, “original strain”), was about 110 cm, whereas that of weak allele d61-1T was about 75 cm (Figures 3(a) and (b)). In wild-type rice, the leaf blade bends away from the vertical axis of the leaf sheath toward the abaxial side, whereas almost all of the leaves of d61-1T were erect (Figure 3(c)). The other weak d61 alleles showed similar semi-dwarf phenotypes, with plant heights ranging between 60 and 100 cm [25]. On the other hand, strong alleles such as d61-4 formed only abnormal leaves with twisted, stiff blades, and the leaf sheath was scarcely developed (Figure 3(d)). These plants did not flower, show internode elongation, or bear any seeds, as described previously [16].

The three newly obtained lines had similar phenotypes, including severely dwarfed stature and completely erect leaves. The culm length of Nipponbare, the original strain of d61-1N, was about 90 cm, whereas those of d61-1N, d61-11, and d61-12 were about 25 cm (Figures 3(e)-(h)). Although the ratio of leaf sheath length to leaf blade length was reduced in the mutants, other abnormal leaf morphologies such as the twisted leaf blade observed in the d61 strong alleles (such as d61-4; Figure 3(d)) were not observed. In addition to the mutant phenotypes

of the vegetative organs of the Nipponbare-derived d61 mutants, abnormal morphology was also found in the grains. Although the grains of d61-1T were indistinguishable from those of its original strain, Taichung 65 (Figures 3(i) and (j)), those of d61-1N, d61-11, and d61-12 were visibly shorter and smaller than those of their original strain, Nipponbare (Figures 3(k)-(n)).

The phenotypes of d61-1N, d61-11, and d61-12 were less severe than those of the d61 strong alleles (such as d61-4): all three lines showed culm elongation, flowering, and normal leaf blade morphology. On the other hand, the phenotypes of the three Nipponbare-derived d61 mutants were more severe than those of the d61 weak alleles (such as d61-1T): all three lines had severely dwarfed stature and small grains, neither of which was observed in d61-1T. Among the previously identified brassinosteroid-related rice mutants, only the brassinosteroid-deficient d11 mutants showed the small-grain phenotype, although their culm length reached 60% of that of their original strains [26]. Based on these observations, we conclude that d61-1N, d61-11, and d61-12 have unique brassinosteroid-related abnormal phenotypes and hypothesize that these mutants have novel defects in the function of brassinosteroid receptor OsBRI1.

Next, we compared the pattern of internode elongation between d61-1N and d61-1T to characterize the mutant phenotypes in detail. In rice, the junction of the culm and the spike is referred to as the neck of the spike. We numbered the internodes of each culm from top to bottom; i.e., the internode just below the neck of the spike was called the first internode. Dwarf mutants of rice have been categorized into six groups based on the elongation pattern of the upper four to five internodes [27], and we attempted to place the new mutants into these categories. d61-1T showed specific inhibition of second-internode elongation; therefore, it was grouped into the dm type (Figures 4(a), (b), (f) and (g)). In contrast, the elongation of all internodes below the first internode was severely suppressed in d61-1N, which clearly fit into the d6 type (Figures 4(c), (f) and (g)). dm-type internode elongation patterns were also observed in the brassinosteroid-deficient d2 and d11 mutants [26,28], suggesting that a stunted second internode is one of the characteristic phenotypes of brassinosteroid-related rice mutants. Although the reason why the elongation of second internode is specifically inhibited in these mutants has not yet been clarified, it is possible that brassinosteroids produced in the developing spike at a particular stage are necessary for second-internode elongation because internode differentiation occurs successively from top (first internode) to bottom, and synchronously with spike development. On the other hand, the second through fourth internodes were stunted in the d6-type mutant d61-1N, indicating that incomplete internode differentiation occurred for a prolonged period during culm development.

In many japonica varieties, the basal rachis internode of the spike (just above the neck) scarcely elongates; however, it was elongated in both d61-1N and d61-1T (Figures 4(a) and (c)-(g)). This is very interesting because it indicates that partial suppression of brassinosteroid receptor function caused two inverse phenomena: reduced culm internode elongation and increased rachis internode elongation. Internode elongation depends on the intercalary meristem, whereas elongation of the rachis internode depends on the rib meristem of the shoot apical meristem [29], suggesting that brassinosteroids have opposite effects on the activity of these two meristems.

Note that the mutation in d61-1N was identical to that in d61-1T, although the phenotypes of d61-1N and d61-1T were quite different (Figure 4). Because d61-1N and d61-1T were derived from two different japonica varieties (Nipponbare and Taichung 65, respectively), our results suggest that the genetic background of a mutant affects the severity of the mutant phenotype. Interestingly, all of the previously identified d61 weak alleles

(d61-1, d61-2, d61-7, d61-8, and d61-9) were obtained from a Taichung 65 mutant library, whereas the strong alleles include both Nipponbare-derived (d61-3 and d61- 6) and Taichung 65-derived mutants (d61-4 and d61-5; Figure 2(a)). These results indicate that severe defects in brassinosteroid receptor function induce the same abnormal phenotypes (group-1 phenotype described above) in mutants regardless of their genetic background. On the other hand, partial suppression of brassinosteroid receptor function resulted in different phenotypes in two different japonica varieties: the d61-1 mutant phenotype was less severe in the Taichung 65 background than in the Nipponbare background.

We want to point out that differences in phenotype were also observed for Nipponbare– and Taichung 65–derived brassinosteroid-deficient mutants. ebisu dwarf (d2) mutants are caused by defects in the rice brassinosteroid biosynthetic enzyme gene CYP90D2/D2 [28]. CYP90D2/ D2 encodes a cytochrome P450 monooxygenase that functions in brassinosteroid C-23 hydroxylase [7]. We have isolated three new d2 alleles derived from a Nipponbare mutant library (d2-3, d2-4, and d2-6), and showed that they produced more severe dwarf phenotypes in the Nipponbare genetic background than did d2-1, a previously characterized null allele from a Taichung 65 mutant library [28], produced in the Taichung 65 genetic background [30].

The phenotype of Taichung 65-derived null allele d2-1 was semi-dwarf: plants reached ~55% (about 60 cm) of the total culm length of wild-type Taichung 65 (about 110 cm; Figures 5(a), (b) and (e)). Interestingly, the total culm length of the Nipponbare–derived null allele d2-6 (about 15 cm) was only about 16% of that of wild-type Nipponbare (about 95 cm; Figures 5(c), (d) and (e)). We also compared the pattern of internode elongation between d2-6 and d2-1. As was the case with d61, a stunted second internode (dm-type pattern) and elongated basal rachis internode were observed in the Taichung 65-derived allele d2-1, whereas the elongation of internodes II and below was severely suppressed in the Nipponbarederived allele d2-6 (d6-type pattern; Figures 5(e) and (f)). These results indicate that in two cases (d2 and d61-1), the severity of phenotype conferred by mutants at the same locus differed between the two genotypes, whereas the phenotypes of mutants at two different loci were similar when evaluated within the same genetic background. In other words, the dwarf phenotypes conferred by d61 weak alleles and d2 null alleles are strongly affected by the genetic background in which they are evaluated.

To compare the brassinosteroid response between the Nipponbareand Taichung 65-derived mutants, we monitored the effect of brassinolide treatment on the expression of BU1 and CYP85A1. BU1 encodes a helix-loophelix protein that participates in the regulation of rice lamina inclination [31]. BU1 is thought to be a primary brassinosteroid-responsive gene, because brassinosteroids increase BU1 expression even in the presence of the protein synthesis inhibitor cycloheximide. CYP85A1 encodes a cytochrome P450 monooxygenase that functions in brassinosteroid C-6 oxidation, and its expression is regulated by a homeostatic system that controls bioactive brassinosteroid levels [8].

The level of BU1 mRNA in Taichung 65 seedlings was increased by brassinolide treatment to 2.4 times that in untreated plants (Figure 6(a)). Similarly, BU1 expression was increased by brassinolide treatment in the

Taichung 65-derived d2-1 mutant, to 3.1 times that in untreated mutant plants. The level in the treated d2-1 mutant plants was the same as that in treated wild-type Taichung 65, although the steady-state level of BU1 mRNA in the absence of brassinolide treatment was lower in d2-1 than in Taichung 65 (about 78% of that in Taichung 65; Figure 6(a)). This expression pattern of BU1 in d2-1 plants can be explained by brassinosteroid deficiency. In d61-1T, however, BU1 expression was only about 74% of that in untreated Taichung 65, regardless of brassinolide treatment (Figure 6(a)).

Expression of BU1 in the Nipponbare seedlings was also increased by brassinolide treatment both in the wild-type (to 2.0 times that in untreated plants) and in the Nipponbare-derived d2-6 mutant (to 2.6 times that in untreated mutant plants; Figure 6(a)), although the steady-state level of BU1 mRNA in the absence of brassinolide treatment was lower in d2-6 than in Nipponbare (about 72% of that in Nipponbare; Figure 6(a)). In d61-1N, BU1 expression was about 68% of that in Nipponbare in the absence of exogenous brassinolide and was not affected by brassinolide treatment (Figure 6(a)).

The relative increase in BU1 expression caused by brassinolide treatment was less in Nipponbare (to 2.0 times) and the Nipponbare-derived d2-6 mutant (to 2.6 times) than in Taichung 65 (to 2.4 times) and the Taichung 65-derived d2-1 mutant (to 3.1 times). In addition, the steady-state level of BU1 mRNA in the absence of brassinolide treatment was lower in d61-1N (68% of that in Nipponbare) than in d61-1T (74% of that in untreated Taichung 65). These results support our hypothesis that the response to brassinosteroid is weaker in the Nipponbare background.

The level of CYP85A1 mRNA in brassinolide-treated Taichung 65 seedlings decreased to about 35% of the untreated control level owing to feedback down-regulation by the homeostatic system (Figure 6(b)). Similarly, CYP85A1 expression decreased in brassinolide-treated d2-1 plants (to 43% of that in untreated mutant plants), although the steady-state level of CYP85A1 mRNA in the

absence of brassinolide treatment was higher in d2-1 than in wild-type Taichung 65 owing to feed-forward up-regulation by the homeostatic system (about 1.2 times that in Taichung 65; Figure 6(b)). This expression pattern of CYP85A1 in d2-1 plants can also be explained by brassinosteroid deficiency. In d61-1T, CYP85A1 expression was slightly higher than wild-type in the absence of brassinolide treatment (1.1 times that in Taichung 65) and unaffected by brassinolide treatment (Figure 6(b)).

Expression of CYP85A1 in brassinolide-treated Nipponbare seedlings decreased to about 44% of the untreated control level (Figure 6(b)). Interestingly, the level of CYP85A1 mRNA in untreated d2-6 was lower than that of Nipponbare (about 85%), and was further decreased by brassinolide treatment (Figure 6(b)). Because the bioactive brassinosteroid castasterone accumulated in d2-6 mutant [30], we hypothesize that the expression of CYP85A1 may also be regulated directly by its catalytic product, castasterone. On the other hand, the level of CYP85A1 mRNA in d61-1N was 1.3 times that of wild-type Nipponbare and was not affected by brassinolide treatment (Figure 6(b)).

The differences in expression between the wild-type and d61-1 mutant in each genotype indicate that the effect of feed-forward up-regulation on CYP85A1 expression was greater in d61-1N (Nipponbare background) than in d61-1T (Taichung 65 background), even though both d61-1N and d61-1T harbor the same amino acid change in the brassinosteroid receptor OsBRI1.

The expression analysis of BU1 and CYP85A1 supports our hypothesis that the response to brassinosteroid is weaker in the Nipponbare background than in the Taichung 65 background. Next, we monitored the effect of brassinolide treatment on the expression of OsBRI1 in the Nipponbareand Taichung 65-derived mutants. The level of OsBRI1 mRNA in Taichung 65 seedlings was decreased by brassinolide treatment to about 70% of that in untreated plants (Figure 6(c)). Similarly, OsBRI1 expression was decreased by brassinolide treatment in the Taichung 65-derived d2-1 mutant to about 67% of that in untreated mutant plants, although the steady-state level of OsBRI1 mRNA in the absence of brassinolide treatment in d2-1 was 1.1 times that in Taichung 65 (Figure 6(c)). This expression pattern of OsBRI1 in d2-1 plants can be explained by the regulation of feedback mechanism caused by brassinosteroid deficiency. In d61-1T, OsBRI1 expression was about 1.1 times that in Taichung 65 and was not affected by brassinolide treatment (Figure 6(c)).

Expression of OsBRI1 in the Nipponbare seedlings was decreased by brassinolide treatment to about 56% that in untreated plants (Figure 6(c)). Interestingly, however, expression of OsBRI1 in the untreated d61-1N and d2-6 mutants was about 80% and 89% of that in untreated Nipponbare, respectively (Figure 6(c)). These results are clearly opposite those observed for the Taichung 65-derived d61-1T and d2-1 mutants, which had higher levels of OsBRI1 than wild-type Taichung 65, and indicate that feed-forward up-regulation of OsBRI1 did not occur in the Nipponbare-derived d61-1N and d2-6 mutants. Feed-forward up-regulation of OsBRI1 (and the resulting increase in the amount of OsBRI1 receptor) is considered to partially compensate for the deficiency of brassinosteroids in biosynthetic mutants and for the insensitivity to brassinosteroids in receptor mutants, even though the function of OsBRI1 is partially suppressed in these mutants. Thus, a defect in regulation of OsBRI1 would explain why the response to brassinosteroid is weaker in the Nipponbare background and why the Nipponbare-derived d61-1N and d2-6 mutants showed more severe phenotypes than the Taichung 65-derived d61-1T and d2-1 mutants.

Because the expression of OsBRI1 was regulated differently in the Nipponbare and Taichung 65 genetic backgrounds, we hypothesized that Nipponbare has a mutant OsBRI1 gene and that the expression of OsBRI1 is misregulated in the Nipponbare genetic background. To test this hypothesis, we compared the nucleotide sequence of a 9.4-kb genomic segment containing the OsBRI1 gene (from positions –4825 to +4571, taking the translation initiation site as +1) from Nipponbare and Taichung 65. We found two nucleotide substitutions at positions –2269 (G to A) and –848 (G to C), and an insertion of a Tourist-like miniature inverted-repeat transposable element (MITE) of 430 bp, miniature Ping (mPing) [32-34], in the promoter region of OsBRI1 from the Nipponbare genome (Figure 6(d)). The insertion was flanked on both ends by a three-nucleotide sequence, TAA, which is the typical target sequence for duplication upon mPing transposition. These structural features indicate that mPing transposed into the promoter region of OsBRI1 in Nipponbare. Ohmori et al. [35] reported that an insertion of Ping (an autonomous transposon that can activate mPing transposition) into the fourth intron of the DROOPING LEAF (DL) gene decreased the transcript level of DL. In our case, the Nipponbare OsBRI1 promoter contains an mPing insertion, and the expression of OsBRI1 in Nipponbare seems to be misregulated by the homeostatic system that controls the levels of brassinosteroid contents and signals. We also compared the nucleotide sequence of the CYP90D2 (D2) and CYP85A1 genes from Nipponbare and Taichung 65, and found no nucleotide substitution between these two cultivars. Based on these results, we conclude that the severity of the mutant phenotypes caused by the mutations in brassinosteroid-related genes can be explained by the insertion of mPing into the promoter of the Nipponbare OsBRI1 gene.

Nipponbare is a model rice genotype being sequenced by the International Rice Genome Sequencing Project Consortium; however, based on the results here, it can be categorized as an OsBRI1 mutant having reduced sensitivity to brassinosteroid. The insertion of mPing in the promoter region of OsBRI1 can be detected by PCR, and we found this insertion in several other rice cultivars we tested (Figure 6(e)), indicating that this mutation might be present in other rice cultivars investigated for brassinosteroid function. Although brassinosteroid-related mutants have not been isolated from any of these mPinginserted cultivars other than Nipponbare, we speculate that the expression of OsBRI1 is misregulated in these cultivars. These findings indicate that possible changes in sensitivity to brassinosteroids caused by this mPing insertion must be considered when the effects of brassinosteroids on the regulation of various growth and developmental processes in rice are examined.