Notice: Wiley Online Library will be unavailable on Saturday 30th July 2016 from 08:00-11:00 BST / 03:00-06:00 EST / 15:00-18:00 SGT for essential maintenance. Apologies for the inconvenience.
Brassinosteroids (BRs) are essential regulators of plant architecture. Understanding how BRs control plant height and leaf angle would facilitate development of new plant type varieties by biotechnology. A number of mutants involved in BR biosynthesis have been isolated but many of them lack detailed genetic analysis. Here, we report the isolation and characterization of a severe dwarf mutant, chromosome segment deleted dwarf 1 (csdd1), which was deficient in BR biosynthesis in rice.
We isolated the mutant by screening a tissue culture-derived population, cloned the gene by mapping, and confirmed its function by complementary and RNAi experiments, combined with physiological and chemical analysis.
We showed that the severe dwarf phenotype was caused by a complete deletion of a cytochrome P450 gene, CYP90D2/D2, which was further confirmed in two independent T-DNA insertion lines in different genetic backgrounds and by RNA interference. Our chemical analysis suggested that CYP90D2/D2 might catalyze C-3 dehydrogenation step in BR biosynthesis.
We have demonstrated that the CYP90D2/D2 gene plays a more important role than previously reported. Allelic mutations of CYP90D2/D2 confer varying degrees of dwarfism and leaf angle, thus providing useful information for molecular breeding in grain crop plants.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Plant height is an important agronomic trait in breeding for grain yield. In the 1960s and 1970s, the widespread introgression of semi-dwarfism in rice and wheat varieties allowed substantial increase of grain yield, a process known as the ‘green revolution’. Various factors regulate plant height, of which GAs and BRs are two major growth regulators. Two green revolution genes – rice semi-dwarf1 (sd1) and wheat Reduced height1 (Rht1) – are involved in GA biosynthesis and signaling, respectively (Peng et al., 1999; Ashikari et al., 2002; Sasaki et al., 2002; Spielmeyer et al., 2002), which have been successfully used in crop breeding worldwide.
BRs are a group of plant steroids that are involved in a wide range of plant developmental process, including promotion of cell division and elongation, vascular system differentiation, stem elongation, root and leaf development, skotomorphogenesis, and grain filling (Fujioka & Yokota, 2003; Wu et al., 2008). Consistent with this, BR mutants usually display developmental defects, including dwarfed stature with shortened stems, curled dark-green leaves and reduced fertility or sterility. When grown in darkness, they show varying degrees of de-etiolation (DET), including short hypocotyl, open cotyledons and premature emergence of primary leaves in Arabidopsis (Chory et al., 1991), or inhibited elongation of mesocotyl and lower internodes in rice (Yamamuro et al., 2000). However, rice BR-related mutants with mild morphological changes display the semi-dwarf phenotype with erect leaves. Erect leaves enhance light capture for photosynthesis by allowing more light to penetrate to the lower leaves, thus enabling planting at a higher density with a higher leaf area index (Sinclair & Sheehy, 1999). Recent studies have revealed that use of such mutants with higher planting density can enhance biomass and grain yield (Sakamoto et al., 2005; Morinaka et al., 2006), providing evidence that control of BR metabolism may be a target for crop improvement.
Isolation and characterization of genes encoding various BR biosynthetic enzymes in plants have increased our knowledge of the BR biosynthetic pathway (Pereira-Netto, 2007; Ohnishi et al., 2009), which is important not only for understanding regulation of BR activity, but also for genetic manipulation of BR levels in crop improvement. Twenty-one genes involved in the BR biosynthetic pathway have been cloned so far (Takahashi et al., 1995; Li et al., 1996; Choe et al., 1998, 1999, 2000; Schultz et al., 2001; Hong et al., 2002, 2003, 2005; Tao et al., 2004; Choe, 2006; Sakamoto & Matsuoka, 2006). Genetic studies, combined with chemical feeding or enzymatic reaction assays, have assigned most of the genes to different catalytic reactions. The C-22α hydroxylation is catalyzed by the DWARF4 C-22α hydroxylase (Fujita et al., 2006). The dwf4 mutant of Arabidopsis has a severe dwarf phenotype possibly because there is only one copy of DWARF4 in the genome (Azpiroz et al., 1998; Choe et al., 1998). Rice has two DWARF4 orthologs, CYP90B2/OsDWARF4 and CYP724B1/OsDWARF4L1/D11; severe dwarfism was observed only in the double mutant osdwarf4-1/d11-4 (Sakamoto et al., 2005; Tanabe et al., 2005). CYP90A/CPD, previously proposed to function as C-23α hydroxylase (Szekeres et al., 1996), recently has been shown to catalyze C-3 oxidation (Ohnishi et al., 2006, 2012). Arabidopsis has one copy of CPD and a cpd mutant was also severely dwarfed (Szekeres et al., 1996). CYP90A3/OsCPD1 and CYP90A4/OsCPD2 are two rice genes closely related to the Arabidopsis gene CYP90A1/CPD, suggesting that CYP90A3/OsCPD1 and CYP90A4/OsCPD2 may function as C-3 oxidases in rice; a null mutant of OsCPD1 did not show any abnormal phenotypes, implying that OsCPD1 and OsCPD2 may be functionally redundant (Sakamoto & Matsuoka, 2006). CYP85A C-6 oxidases catalyze the C-6 and Baeyer-Villiger oxidation. Arabidopsis has two CYP85A genes, CYP85A1 and CYP85A2. The cyp85a1 or cyp85a2 single null mutants showed a normal or a mild dwarf phenotype, whereas a double mutant cyp85a1/cyp85a2 displayed a severe dwarf phenotype (Kim et al., 2005b; Kwon et al., 2005; Nomura et al., 2005). CYP85A1/OsDWARF/BRD1 is the only C-6 oxidase gene in rice; and severe dwarfism was observed in null mutants of this gene (Hong et al., 2002; Mori et al., 2002). Two CYP85A genes exist in tomato, with CYP85A1/DWARF expressed in vegetable tissues and fruits, whereas CYP85A3 preferentially expressed in developing fruits (Nomura et al., 2005). Consistent with their expression profiles, a transposon-tagged mutant of CYP85A1/DWARF was severely dwarfed (Bishop et al., 1996), suggesting that there is no functional overlap between CYP85A1/DWARF and CYP85A3 in vegetative tissues. Therefore, it is most likely that disruption of a reaction in BR biosynthesis, if catalyzed by only one enzyme encoded by a single gene, can result in a BR-deficient severely dwarfed mutant. In this context, CYP90B1/DWARF4 and CYP90A1/CPD in Arabidopsis, CYP85A1/OsDWARF/BRD1 in rice and CYP85A1/DWARF in tomato, are considered as key genes in regulating stem elongation.
Studies in Arabidopsis have shown that CYP90C1/ROT3 and CYP90D1 participate in C-23α dehydrogenation (Ohnishi et al., 2006). Neither the cyp90c1 nor the cyp90d1 mutant alone exhibited a typical BR-deficient phenotype, whereas the double mutant cyp90c1/cyp90d1 was severely dwarfed (Tsuge et al., 1996; Kim et al., 1998, 2005a; Ohnishi et al., 2006), indicating that CYP90C1/ROT3 and CYP90D1 are functionally redundant C-23α hydroxylases. Two CYP90D1 homologs, CYP90D2/D2 and CYP90D3, exist in rice. The two allelic D2 mutants, d2-1 and d2-2, had a mild dwarf phenotype, and D2 was previously proposed to function in catalyzing the C-3 dehydrogenation based on quantitative analysis and feeding assays of BR intermediates in those mutants (Hong et al., 2003). CYP90D3 has high identity to D2 and it was suggested that the mild phenotype in d2-1 and d2-2 might be due to the contribution of CYP90D3 activity. Recently, Sakamoto et al. (2012) showed that CYP90D2 and CYP90D3 catalyze C-23 hydroxylation of brassinosteroids based on in vitro biochemical assays. However, there has been no genetic evidence that CYP90D3 is functional and has redundancy with CYP90D2. Furthermore, characterization of more severe D2 and CYP90D3 mutants, such as using enzyme assays, would provide better understanding on catalytic activities of D2 and CYP90D3 in rice.
In this study we report the isolation and characterization of three rice allelic mutants of D2. All of the three mutants, as well as D2-RNAi plants either in the wild -type or d2-1 background, were severely dwarfed. We provide genetic evidence that the three mutations were null and that the severe dwarfism was independent of genetic backgrounds. Our preliminary biochemical analysis suggests that D2 might catalyze the C-3 dehydrogenation. We conclude that D2 encodes a key enzyme involved in BR biosynthesis and plays crucial roles in regulation of internode elongation, leaf development and vascular system differentiation in rice.
Materials and Methods
Authentic BRs for quantitative analysis and enzyme assays were kindly provided by Prof. Takao Yokota (Department of Bioscience, Teikyo University, Utsunomiya, Japan). 24-epibrassinolide (24-epiBL) for the feeding experiment was purchased from Toronto Research Chemicals Inc. (Toronto, Ontario, Canada).
Plant materials and growth conditions
Five rice (Oryza sativa L.) dwarf mutants, csdd1, d2-1, d2-2, d2-3 and d2-4 were analyzed in this study. The wild-type of csdd1 is Kitaake, a japonica cultivar. d2-1 and d2-2 possessed alleles of the CYP90D2/D2 gene in the T65 background and were kindly provided by Prof. Makoto Matsuoka (BioScience and Biotechnology Center, Nagoya University, Chikusa, Nagoya, Japan). The d2-3 and d2-4 mutants involved T-DNA insertions in the CYP90D2/D2 gene, was isolated from the POSTECH rice T-DNA insertion sequence database (Jeon et al., 2000; Jeong et al., 2006). The wild-types of d2-3 and d2-4 were Hwayong and Dongjin, respectively.
Rice plants were cultivated in an experimental plot under natural growing conditions. For light and transmission electron microscopy, pigment determination and BL feeding experiments, seedlings were grown in a controlled growth chamber at 30°C in daytime and 26°C at night under long-day conditions (16 h : 8 h, light (c. 150 μmol photons m−2 s−1) : dark).
Light and transmission electron microscopy
For light microscopy, internodes and leaves of csdd1, d2-3, d2-4 and wild-type plants were harvested and fixed in FAA solution, followed by a series of dehydration and infiltration steps before embedding in paraplast. After sectioning, 8-μm sections were dewaxed with xylene, rehydrated, stained with 1% toludine blue and observed with a Leica DM5000 B microscope (Leica Microsystems, Wetzlar, Germany).
For transmission electron microscopy, leaf samples harvested from 2-wk-old csdd1 and wild-type seedlings were fixed in 2% glutaraldehyde solution and further fixed with 1% osmium oxide solution. Samples were subsequently dehydrated through a graded ethanol series and embedded in Spurr's resin. After sectioning, samples were stained with uranyl acetate and observed with a JEOL JEM-1230 electron microscope (JEOL, Tokyo, Japan).
Leaves of csdd1 and wild-type plants were harvested; Chl and carotene were extracted from the samples and measured with a DU 800 UV/Vis spectrophotometer (Beckman Coulter, Fullerton, CA, USA) based on the method of Arnon (1949).
Map-based cloning of the csdd1 locus
For rough mapping, an F2 population was generated by crossing csdd1 plants with indica variety 93-11. For fine mapping, a BC1F2 population was obtained by backcrossing F2 dwarf plants with 93-11. InDel markers developed for fine mapping of the csdd1 locus are listed in Supporting Information Table S1. Primers used for vector construction are listed in Table S2.
Seeds of heterozygous csdd1/+ plants and wild-type were surface-sterilized and germinated in distilled water for 10 d. Then, csdd1/csdd1 and wild-type seedlings were transferred to half-strength Murashige and Skoog medium with or without 24-epibrassinolide (24-epiBL), and cultivated for 7 wk. The media were changed twice per week.
RNA extraction, RT-RCR and real-time PCR analysis
Total RNA was isolated from various organs of plants or whole seedlings using the RNAprep Pure Plant Kit (TIANGEN, Beijing, China). First strand cDNAs were synthesized from 2 μg total RNA using the PrimeScript 1st Strand cDNA Synthesis Kit (Thakara, Dalian, China).
For RT-PCR analyses, PCR was performed at 28 cycles. Real-time PCR was performed using SYBR Premix Ex Taq (Thakara) and run on an Applied Biosystems 7900HT Real-Time PCR System (Applied Biosystems, Foster City, CA, USA) following the manufacturer's instructions. PCR primers used in the RT-RCR and real-time PCR analysis are listed in Tables S3 and S4, respectively.
Quantitative analysis of endogenous BRs in rice
Quantitative analysis of endogenous BRs in rice was modified from Kim et al. (2005b). Four-week-old shoots of the homozygous mutant csdd1 (54.39 g FW), d2-3 (85.30 g FW), d2-4 (80.00 g FW) and corresponding wild-type plants (140.00 g FW each) were harvested, lyophilized and extracted three times with 500 ml of 90% methanol. Deuterium-labeled teasterone (TE), 3-dehydroteasterone (3DT), 3-dehydro-6-deoxoteasterone (6-deoxo3DT), typhasterol (TY), 6-deoxotyphasterol (6-deoxoTY), castasterone (CS), 6-deoxocastasterone (6-deoxoCS) and brassinolide (BL) were added as internal standards for quantitative analysis of the extracts (200 ng each). Evaporated extracts were partitioned three times between water and chloroform. The chloroform-soluble fractions were concentrated in vacuo and partitioned three times between 80% methanol and n-hexane. The concentrated 80% methanol extracts were repartitioned three times between ethyl acetate and phosphate buffer, pH 7.8. The ethyl acetate-soluble residues were subjected to silica gel chromatography. The column was eluted with 150 ml of chloroform containing 0–10% (v/v) methanol. The 1–7% (v/v) methanol fractions were combined, concentrated in vacuo, and subsequently purified by passage through a SepPak C18 Plus silica cartridge column (Waters, Seoul, South Korea) and eluted with 5 ml of 50–90% (v/v) methanol in water. The fractions eluted with 70–90% (v/v) methanol in water were dried and dissolved in a small amount of methanol, and then subjected to reverse phase HPLC (SenshuPak C18, 10 × 150 mm). The column was eluted at a flow rate of 2.5 ml min−1 using acetonitrile-water gradients: 0–20 min, 45% acetonitrile; 20–40 min, 45–100% acetonitrile; 40–70 min, 100% acetonitrile. Fractions were collected every minute. The fractions corresponding to authentic TE, 3DT, 6-deoxo3DT, TY, 6-deoxoTY, CS, 6-deoxoCS and BL were selected, and analyzed by a capillary GC-MS: a Hewlett-Packard 5973 mass spectrometer (electron impact ionization, 70 electron voltage; Palo Alto, CA, USA) connected to a 6890 gas chromatograph fitted with a fused silica capillary column (HP-5, 0.25 mm × 30 m, 0.25-μm film thickness). Oven temperature was maintained at 175°C for 2 min, elevated to 280°C at a rate of 40°C min−1, and then maintained at 280°C. Helium was used as the carrier gas at a flow rate of 1 ml min−1, and samples were introduced using an on-column injection mode. Methaneboronation was performed by heating samples dissolved in pyridine containing methaneboronic acid (2 mg ml−1) at 80°C for 30 min.
The enzyme assay of rice was carried out using the methods of Kim et al. (2004). Four-week-old shoots of homozygous d2-4 and wild-type were harvested, ground, homogenized and centrifuged at 8000 g for 10 min, and the resulting supernatants were re-centrifuged at 20 000 g for 30 min. Cold acetone was then added to the obtained supernatants (final volume 40%), and the acetone precipitates were re-suspended in 0.1 M sodium phosphate buffer (pH 7.4) containing 1.5 mM 2-mercaptoethanol and 30% (v/v) glycerol for crude enzyme solution. The protein concentration of the enzyme solution was determined with a microassay from Bio-Rad (Cambridge, MA, USA) using bovine serum albumin as a standard.
An enzyme assay for the conversion of TE to 3DT, TE to TY and 3DT to TY was initiated by the addition of TE or 3DT (5 μg each) to the crude enzyme extract in the presence of NADPH. After incubation at 37°C for 60 min, deuterium-labeled 3DT or TY was added for quantitative analysis. Enzyme products were extracted with ethyl acetate (1.2 ml × 3). The obtained ethyl acetate soluble fractions were loaded on a Sep-Pak C18 cartridge column eluted with 50% and 90% methanol (5 ml each). The 90% methanol fractions were further purified by reverse phase HPLC (Nova Pak, C18, 8 × 100 mm). The column was eluted at a flow rate of 2.5 ml min−1 using acetonitrile-water gradients: 0–15 min, 45% acetonitrile; 15–30 min, 30–100% acetonitrile; 40–70 min, 100% acetonitrile. Fractions were collected every minute and analyzed by GC-MS described above. The specific enzyme activity was calculated as ng product μg−1 protein min−1.
Isolation and characterization of the dwarf mutant csdd1
We generated a population of transgenic lines in our efforts to evaluate genes of various origins in japonica rice at large scale. A mutant with severe dwarfism, named chromosome segment deletion dwarf 1 (csdd1, see below: csdd1 had a 168-kb deletion on chromosome 1), was identified in the T1 population. The mutant was most likely a tissue culture-induced mutant because our PCR analysis did not show any association of T-DNA with the mutant phenotype. csdd1 was characterized by slightly rolled and twisted dark-green leaves and sterility, and was much shorter (c. 23% of the wild-type height) due to unelongated internodes (Fig. 1a–d; Table S5). The culms were thicker and displayed smaller leaf angles than wild-type plants (Fig. 1e). Pigment analysis and electron microscopy showed that the dark-green leaves of the mutant had higher chlorophyll contents and probably contained more chloroplasts per cell and more grana per chloroplast (Fig. S1a–f). The size of chloroplasts in the mutant looked larger than in wild-type plant (Fig. S1c,d).
In order to investigate the cause of developmental defects in the culms and leaves of csdd1, we performed light microscopy on wild-type and mutant plants. In contrast to wild-type, cells in the internodes of csdd1 were smaller, and failed not only to form well-organized cell files, but also to elongate even in adult plants, causing the severe dwarfism in csdd1 plants (Fig. 1f,g). Cross-sections of culms showed that csdd1 had smaller parenchyma cells, but much more layers of parenchyma and sclerenchyma cells as well as vascular bundles than wild-type (Fig. 1h,i). In the unelongated internodes of csdd1, abnormal vascular bundles were randomly scattered among parenchyma cells, and the structure of vascular bundles was changed (Fig. 1i,k). There seemed to be increased amount of phloem compared to the xylem; and phloem cells were smaller than in wild-type plants. Smaller and immature primary tracheary elements were not so orderly arranged as in the vascular bundles of wild-type internodes (Fig. 1j,k). Clearly, csdd1 had an abnormal vascular system.
Cross-sections of leaves showed that csdd1 had only one layer of mesophyll cells, which invaginated toward the gas chambers at the adaxial base of the midrib (Fig. 1l,m); and bulliform cells in the csdd1 blades were fewer and not well-developed compared to those of wild-type (Fig. 1n,o). These changes might cause the rolled and twisted morphology of the csdd1 leaves.
csdd1 had a 168-kb deletion on chromosome 1
We used pollen from wild-type plants to pollinate the sterile mutant and produced F1 seeds. F1 plants have a wild-type phenotype and F2 progeny exhibit a 3 : 1 ratio (wild-type : dwarf, 174 : 55; χ23 : 1 = 0.118; P >0.75), indicating that the dwarf phenotype segregates as a single recessive locus. Using an F2 population derived from a cross between csdd1 and 93-11, an indica variety whose genome sequence is publicly available (Yu et al., 2002), the mutated locus was mapped to the short arm of chromosome 1 between the SSR markers RM283 and RM5496 (Fig. 2a). Because F1 plants of the japonica/indica cross had relatively low seed set (< 5%), we backcrossed F2 dwarf plants with 93-11 to increase the size of the mapping population. New InDel markers were also developed based on the sequences of 93-11 and Nipponbare (Table S1). Using 2332 BC1F2 dwarf individuals, we finally narrowed down the locus to a 219-kb interval between the InDel markers ID11 and ID99 (Fig. 2b).
We further designed primers to develop markers between ID11 and ID99. However, most of the primers failed to amplify any fragment from csdd1 DNA. By chromosome walking with a series of primers and in combination with sequencing, we found a 168-kb deletion in csdd1 (Fig. 2c–e). The Primers 1F and 1R amplified a fragment from WT but not from csdd1 whereas the Primers 2F and 2R amplified a fragment from the mutant but not from WT, providing a simple detection of the deletion (Fig. 2c,d; Table S6). Thus, we named the mutant as chromosome segment deleted dwarf 1 (csdd1).
The D2 gene is responsible for the dwarf phenotype
There are 23 annotated genes located in the deleted segment (http://rice.plantbiology.msu.edu/, Fig. 2c; Table S7), of which CYP90D2/D2 and the cytokinin dehydrogenase gene have been reported to regulate plant height and grain number, respectively (Hong et al., 2003; Ashikari et al., 2005). In BR biosynthesis, D2 was previously proposed to catalyze C-3 dehydrogenation, but recently it was shown to catalyze C-23 hydroxylation (Hong et al., 2003; Sakamoto et al., 2012). Two allelic mutations of D2, d2-1 and d2-2, led to c. 30% and c. 20% reduction in plant stature, respectively (Hong et al., 2003; Table S5). The other 22 genes included in this deletion have not been reported or predicted as BR-related genes. Thus, the D2 gene was considered the first candidate responsible for the dwarfism in csdd1.
When grown in darkness, rice BR-deficient mutants usually show a typical de-etiolated phenotype, including inhibited elongation of both mesocotyl and lower internodes (Mori et al., 2002; Hong et al., 2003), and some BR-deficient mutants are more sensitive to BRs treatment (Hong et al., 2002; Tanabe et al., 2005). We grew csdd1 seedlings in darkness and observed that elongation of both mesocotyl and the first internode from the bottom was significantly reduced (Fig. S2a,b; Methods S1). We also grew csdd1 plants in hydroponic solutions containing different concentrations of 24-epibrassinolide (24-epiBL) for 7 wk and measured their plant height and length of the upmost fully expanded leaf blade. The results showed that the csdd1 plants were more sensitive to the BL treatments than the WT plants and that the difference in plant height between csdd1 and WT plants became smaller as BL concentration increased (Fig. 3a,b). The leaf blade length of csdd1 was not significantly different from WT when the concentration was increased to 500 nM (Fig. 3c). In BR-deficient mutants, BR signaling and BR-regulated genes are usually downregulated while BR biosynthetic genes are upregulated (Sakamoto et al., 2005; Nakamura et al., 2006). We found that all the four BR signaling and BR-regulated genes measured were downregulated and three of the six BR biosynthetic genes were upregulated in csdd1 (Fig. S3a,b). As expected, D2 transcription was not detectable in the mutant (Fig. S3b). Taken together, these results suggested that the dwarfism in csdd1 was most likely due to deficiency in BRs, which might have resulted from deletion of D2 from the genome.
We used two primer pairs, Primer 1F/Primer 1R and Primer 2F/Primer 2R (Fig. 2c), to genotype individual callus lines induced from seeds harvested from heterozygous plants and isolated homozygous csdd1 calli. A 9952-bp genome fragment of D2, including a region 1950 bp upstream of the ATG start codon and a region 104 bp downstream of the stop codon, was introduced into the homozygous csdd1 calli by Agrobacterium-mediated transformation (Methods S2). All nine primary transgenic plants (T0) were normal whereas plants transformed with the empty vector were still dwarfs (Fig. 4a–d). T1 lines from nine T0 plants were analyzed and the normal phenotype was always associated with the transgene (data not shown). To investigate whether other 22 genes had any contribution to the severe dwarfism, we analyzed their expression in both seedlings and adult plants by RT-PCR and identified 14 genes which showed expression at various levels (Fig. S4a,b). Six of them, including the cytokinin dehydrogenase gene, which we thought could affect plant architecture, were introduced into cssd1. However, none of them rescue the dwarf phenotype (Fig. S5). Taken together, those results suggest that the absence of D2 contributed the most to the dwarfism in cssd1 and that other genes had no effect on plant height or the effect, if any, was minor and covered by the major effect from lack of D2.
In order to further confirm if D2 was the only gene responsible for the dwarfism, a D2-RNAi vector was constructed and introduced to wild-type plants (Fig. 5a). Nineteen T0D2-RNAi plants were regenerated and they showed dwarf phenotypes varying in severity (Figs 5b, S6). In the severe dwarf plants, the internodes were shortened dramatically and the plants were sterile (Figs 5c, S6). Expression analysis revealed that D2 expression was significantly decreased in the weak dwarf plants such as in D2-RNAi-2 and D2-RNAi-5 and was not detectable in the severe dwarf plants such as in D2-RNAi-15 (Fig. 5d). Taking all these results together, we concluded that it was loss-of-function of D2 that caused the severe dwarfism in cssd1.
The severe dwarf phenotype caused by loss-of-function of D2 is independent of genetic background
Two allelic mutations of D2, d2-1 and d2-2, were reported to cause erect leaves and reduce plant height to c. 70% and c. 80%, respectively, of wild-type; but the mutant plants were fertile although seed size was smaller (Hong et al., 2003; Table S5). In addition, the mutations had an effect mainly on elongation of the second internode but very little effect on other internodes. We grew d2-1 and d2-2 plants and observed exactly the same phenotype (Fig. S7; Table S5). By contrast, csdd1 was c. 23% of wild-type in height and sterile, with no elongated internode (Fig. 1a,c,d; Table S5).
d2-1 had a base change in the first exon, resulting in a premature stop codon which truncated the protein from 490 amino acids to 82 amino acids; d2-2 had a base change in the fourth exon resulting in one amino acid substitution (Pro to Ser) (Hong et al., 2003). We re-sequenced d2-1 and d2-2 and confirmed the two base changes. If d2-1 could be considered as a null mutant, it was interesting to consider why there was dramatic difference in severity of dwarfism between csdd1 and d2-1. Kitaake, a japonica variety from which csdd1 was isolated, is short and flowers early. To see if there was any effect of genetic background on degree of dwarfism, we crossed csdd1 with T65, the wild-type of d2-1 and d2-2, followed by four backcrosses with T65. The BC4F2csdd1 plants (csdd1T65), which contained the homozygous 168-kb deletion, had a severe dwarf phenotype, as seen in the Kitaake background, but clearly different from d2-1 and d2-2 (Fig. 6a). Indeed, homozygous csdd1 in BC1F2, BC2F2 and BC3F2 always showed severe dwarfism with sterility, regardless of backcross generations (Fig. S8a–c). The csdd1 deletion was also introduced to the indica variety 93-11. As expected, the severe dwarf phenotype was seen in all three backcrosses (Only a BC3F2csdd1 plant shown in Fig. 6b).
We searched the rice Tos17 and T-DNA tagging databases publicly available for lines with D2 interrupted. Two lines, 2A-10747 (japonica variety Hwayong) and 1A-25015 (japonica variety Dongjin), were identified having D2 inserted by T-DNA (http://www.postech.ac.kr/life/pfg/risd/index.html). We named 2A-10747 and 1A-25015 as d2-3 and d2-4, respectively (Fig. 6c). Our detailed analysis of the two lines showed that T-DNA was inserted in the fourth intron in d2-3 and in the seventh intron in d2-4 (Fig. 6d) and the dwarf seedlings were always associated with T-DNA insertion (Fig. S9a–c). Expression of D2 in d2-3 and d2-4 was not detectable (Fig. 6e). All plants homozygous for either of the two insertions were severe dwarfs with unelongated internodes and sterility (Fig. 6c; Table S5). Histological study indicated that d2-3 and d2-4 had unelongated cells in the unelongated internodes (Fig. S10a–d). We grew d2-3 and d2-4 seedlings in darkness and found that they showed a typical de-etiolated phenotype of BR mutants (Fig. S11a,b; Methods S1).
Taking all the results together, we concluded that loss-of-function of D2, either caused by deletion from the genome or interruption of its transcription by T-DNA insertion, resulted in a severe dwarf phenotype with sterility at least in the five genetic backgrounds used.
There may be another mechanism regulating plant height in d2-1
The rice cytochrome P450 protein CYP90D3 is homologous to D2 (66% identity and 79% similarity) but CYP90D3 is expressed at a much lower level than D2 (Hong et al., 2003). d2-1, a loss-of-function allelic mutant, did not display a severe dwarf phenotype possibly because of the functional redundancy of D2 and CYP90D3. However, csdd1T65 plants showed severe dwarf phenotypes with unelongated internodes. But leaf development is less affected than that of internodes in csdd1T65 plants with slightly rolled and twisted leaves (Fig. 6a), which suggests that CYP90D3 may be functional in the leaves rather than internodes of rice. The d2-1/csdd1 F1 plants from a cross between csdd1 and d2-1 showed strong resemblance to d2-1: semi-dwarf but fertile, with the second internode not elongated in particular (Fig. 7a), indicating dominancy of d2-1 over csdd1. Those observations suggested that d2-1 might function somehow. To investigate this possibility, we transformed a D2-RNAi construct into d2-1 and produced 22 T0 transgenic plants. Those plants exhibited varying degrees of dwarfism (Fig. S12a,b). The extremely dwarf plants neither had any elongated internodes nor produced any panicles even after being grown > 10 months, which was associated with the fact that transcripts of D2 were significantly decreased or not detectable in those plants (Fig. 7b). We concluded that the double-level mutations, RNAi plus a premature stop caused by a single base substitution in d2-1, could cause even more severe dwarfism.
D2 was annotated having an alternative splice site in the third intron, resulting in the second transcript (D2b) which contains part of intron 3 and exons 4–8 (http://www.gramene.org/, Fig. 7c). The deduced protein of this transcript consists of 243 amino acids, a truncated version of D2 with N-terminal deleted. We were able to detect the D2b transcript in wild-type plants and d2-1, but not in csdd1, d2-3 and d2-4 (Fig. 7d), suggesting that D2b might contribute to the weak phenotype of d2-1. We cloned and introduced D2b cDNA under control of the maize ubiquitin promoter into csdd1 and d2-3, respectively. However, none of the transgenic plants displayed any rescued phenotype although D2b transcripts were present in the transgenic plants as detected by RT-PCR (data not shown). By contrast, when D2 full-length cDNA D2a was introduced to csdd1 and d2-3, respectively, the severe dwarf phenotype was rescued (Figs 4a,b, S13). Thus, there may be an unknown mechanism in d2-1, rather than D2b, involved in BR biosynthesis.
BRs content significantly reduced in D2 mutants irrespective of genetic backgrounds
D2 was proposed to catalyze the C-3 dehydrogenation in BR biosynthesis (Hong et al., 2003). However, the most recent study concluded a role of D2 in C-23 hydroxylation (Sakamoto et al., 2012). D2 has the closest homology to CYP90D1 and CYP90C1 in Arabidopsis (54% and 46% identity, respectively). CYP90D1 and CYP90C1 were reported to catalyze the C-23α hydroxylation (Ohnishi et al., 2006) and CYP90A1/CPD to catalyze the C-3 dehydrogenation (Ohnishi et al., 2012) in Arabidopsis, implying that D2 may catalyze the C-23α hydroxylation in rice. To better understand the catalytic activity of D2 in rice, we analyzed the content of TE and 6-deoxo3DT, and downstream BR intermediates in the three D2 mutant alleles (csdd1, d2-3, d2-4) and compared to that in the corresponding WT plants (Fig. S14a). 6-deoxo3DT, 3DT and the downstream BRs were always lower in the three mutants than in the controls, whereas TE was accumulated in the mutants. Those preliminary results are in favor of a role of D2 in C-3 dehydrogenation. However, we can not exclude the possibility that reduced 6-deoxo3DT and 6-deoxoTY was due to deficient C-23 hydroxylation of 22-hydroxy-3-one and 3-epi-6-deoxoCT in the shortcut pathway. Further analyses are needed in order to make a clear conclusion.
We further performed in vitro enzyme activity assays using crude protein extracts from d2-4 and its WT plants (Fig. S14b). When TE was used as a substrate, the conversion rate of TE to 3DT and subsequently to TY with the d2-4 extract was significantly lower than that with the WT extract. When 3DT was used as a substrate, however, there was no difference in the conversion rate of 3DT to TY between the d2-4 and WT extracts. Those results indicate that the ability of the extract from d2-4 to catalyze the conversion from TE to 3DT was compromised, again suggesting a role of D2 in C-3 dehydrogenation. Given a different role of D2 demonstrated recently (Sakamoto et al., 2012), it will be interesting to test other substrates, such as 22-hydroxy-3-one and 3-epi-6-deoxoCT, in further D2 activity assays.
We have provided genetic and chemical evidence that CYP90D2/D2 plays a key role in the control of plant architecture by regulating BR biosynthesis in rice. First, csdd1 was characteristic of conventional BR-deficient mutants and was responsible for BL treatment. Second, a deletion of D2-containing genomic fragment or a complete interruption of D2 transcription by T-DNA insertions was responsible for the severe dwarfism independent of genetic backgrounds. Third, RNA interference targeted to D2 resulted in varying severities of dwarfism associated with reduction of D2 transcript level. Fourth, the biochemical analyses revealed reduced BR intermediates and the enzyme assays indicated reduced activity in catalyzing BR-related reactions in D2 mutants.
The severe dwarf phenotype in d2-3 and d2-4 was caused by insertions of the T-DNA that carries an enhancer trap design (Jeong et al., 2002). The enhancer sequences from the CaMV 35S promoter can potentially influence expression of genes, especially those nearby T-DNA insertion sites. However, the dwarfism in d2-3 and d2-4 is unlikely to be attributed to enhanced expression of other genes. First, the two insertions completely interfered expression of D2. Second, six genes nearby D2, including the cytokinin dehydrogenase gene, were introduced into cssd1 but none of them rescued the dwarf phenotype, indicating that their expression level has no visible influence on plant height, at least in the absence of D2. Third, a wild-type D2 sequence fully rescued the d2-3 phenotype. In addition, our quantitative RT-PCR analysis did not detect a significant expression level change in d2-3 and d2-4 of LOC_Os01 g10030 and LOC_Os01 g10050 (data not shown), the two genes flanking the left and right of D2, respectively, consistent with the observation that the enhancer trap design does not always increase expression level of genes nearby T-DNA insertions in rice (Jeong et al., 2002).
Cytochrome P450 enzymes involved in BR metabolism contain conserved functional domains, including membrane anchor, proline-rich, dioxygen binding, steroid binding and heme binding domain (Nebert & Gonzalez, 1987; Kalb & Loper, 1988; Choe et al., 1998). Genetic studies on allelic mutations have revealed a requirement of some of the domains for full function of an enzyme. CYP85A1/BRD1 has one copy in the rice genome; so any phenotypic change in a BRD1 mutant would explain the importance of the mutated domain or site. The brd1-1 allele had a 113-bp deletion, which created a premature stop codon, resulting in a truncated protein with the deletion of the heme binding domain (Hong et al., 2002). The brd1-1 plant showed an extreme dwarf phenotype, indicating that the heme binding domain is essential. A similar BRD1 mutant, brd1, with the heme binding domain disrupted was also extremely dwarfed (Mori et al., 2002). The brd1-2 allele, with an extreme dwarf phenotype almost identical to brd1-1, had a change from glycine to valine at residue 111, a conserved site across CYP85 members, demonstrating a crucial role of glycine at the site for the enzymatic activity (Hong et al., 2002). CYP90B1/DWF4 has a single copy in the Arabidopsis genome. The severe dwarf mutant, dwf4-3, carried a premature stop codon that leads to a loss of dioxygen binding, steroid binding and heme binding domain (Choe et al., 1998). The CYP90D2/D2 mutant d2-1 harbored a premature stop codon in exon 1, resulting in a truncated protein with only the membrane anchor and proline-rich domain (Hong et al., 2003), suggesting that d2-1 could be a null allele. We confirmed the mutation in d2-1 by sequencing. Surprisingly, the d2-1 was semi-dwarf, with only the second internode unelongated. By contrast, csdd1, d2-3 and d2-4 had a severe dwarf phenotype with all internodes unelongated. In addition, RNA interference of d2-1 resulted in severe dwarfism. We suspect that there is another mechanism contributing to the semi-dwarf phenotype in d2-1.
D2 has a homologous gene, CYP90D3, in rice genome. D2 and CYP90D3 share 66% identity and both recently were shown to catalyze C-23 hydroxylation of brassinosteroids based on in vitro biochemical assays (Sakamoto et al., 2012), implying that the semi-dwarf phenotype of d2-1 might be due to activity of CYP90D3. However, csdd1, d2-3 and d2-4 showed severe BR-deficient phenotypes in the three genetic backgrounds, respectively. One possibility is that D3 may play a subsidiary role in those three varieties, but it does play an important role in the d2-1 background, T65. The transcription level of CYP90D3 in T65 is low but detectable (Hong et al., 2003). When the csdd1 deletion was introgressed into T65, however, the introgression line (csdd1T65) showed the csdd1 phenotype with unelongated internodes but slightly rolled and twisted leaves, not semi-dwarf. Morphological abnormalities of leaves are less severe than those of the internodes, suggesting that CYP90D3 is likely to be functional in leaves rather than internodes of rice. Alternatively, loss of the other 22 genes, together with D2, may cause the severely dwarfed phenotype in T65. Introgression of d2-3 and d2-4 alleles to T65 will clarify this possibility.
Annotation of D2 predicts an alternative transcript that encodes a protein without the membrane anchor and the proline–rich domain (http://www.gramene.org/). Indeed, our RT-PCR analysis confirmed the existence of two transcripts, D2a and D2b, in the wild-type and d2-1 (Fig. 7d). In d2-1, the D2a transcript carried a premature stop codon which is supported to produce a shortened D2a protein, but the D2b transcript was not affected. It is possible that the D2b protein may be able to work as a functional enzyme protein but in low activity, hence partially complementing the loss-of-function mutation of D2a. However, the fact that expressing D2b cDNA in csdd1 and d2-3 did not rescue their severe dwarf phenotype does not support this possibility, suggesting that the membrane anchor and the proline-rich domain are essential for the functioning of D2.
It is noteworthy that the single base substitution in d2-1 introduced a premature stop codon but did not affect D2 transcription (Fig. 7d). Although an introduced premature stop codon usually leads to the production of a truncated protein, translation reinitiation can take place in eukaryotes if the downstream start codon is close to the premature stop codon (Nishimura et al., 2005; Paulsen et al., 2006). There are several potential methionine start codons downstream of the premature stop codon and in frame with the first start codon (Fig. 8). The premature stop codon in d2-1 is supposed to produce an 82-aa protein containing the membrane anchor and the proline-rich domain. It is possible that translation reinitiates from the second start codon (methionine 131), producing a protein containing the dioxygen binding, steroid binding and heme binding domain. The proteins produced by the two separate translations may form a complex which contains all domains and may contribute to the mild phenotype in d2-1. There may be another possibility that the second protein without an N-terminal of D2 alone is partially functional. Further investigation on d2-1 is needed to understand the molecular mechanism whereby d2-1 does not display a severe dwarf phenotype.
In BR biosynthesis, the assessment of the function of D2 and CYP90D3 has been controversial over the last decade because there were reports favoring the implication of CYP90D2 in the C-3 dehydrogenation or in the C-23 hydroxylation steps. In the beginning, D2 was proposed to catalyze C-3 dehydrogenation in BR biosynthesis based on quantitative analysis of the endogenous BR level and lamina joint bending assays of BR intermediates in those mutants (Hong et al., 2003). After that, the closest orthologs of D2, CYP90D1 and CYP90C1, were reported to catalyze the C-23α hydroxylation in Arabidopsis, implying that D2 may catalyze C-23α hydroxylation in rice (Ohnishi et al., 2006). Recently, Sakamoto et al. (2012) expressed D2 and CYP90D3 protein in a baculovirus/insect cell system, and analyzed their catalytic activities and substrate specificity based on detailed and elaborate in vitro biochemical assays. Their results provided strong evidence that D2 and CYP90D3 catalyze C-23 hydroxylation of various 22-hydroxylated brassinosteroids in vitro, with preference in the shortcuts of the BR biosynthesis pathway. This is in agreement with the theory that CYP90D1 and CYP90C1, the Arabidopsis orthologs of D2, function as C-23 hydroxylases. Although our preliminary quantitative analysis of endogenous BRs level in csdd1, d2-3 and d2-4 mutants provides results tending to support the assumption that D2 might catalyze the C-3 dehydrogenation, our data cannot exclude the possibility that D2 catalyzes C-23 hydroxylation. It is noteworthy that previous studies and our analyses on the content of endogenous BRs in D2 mutants mainly focused on the intermediates of the C-6 oxidation pathway (Hong et al., 2003; Sakamoto et al., 2012; Fig. S14a). Further analyses on the content of more intermediates such as 22-hydroxy-3-one and 3-epi-6-deoxoCT in the shortcut pathway of C-23 hydroxylation, combined with enzyme activity assays with labeled intermediates using D2 mutants, are needed to forge a solid conclusion.
Our in vitro enzyme activity assays using crude protein extracts from d2-4 and its WT shoots showed that catalytic activity of the C-3 dehydrogenation step converting TE to 3DT was also compromised in d2-4. Considering that the BR intermediates are extremely low in plants, we do not know if there is an effect of high concentration of the substrates added in the assays on enzyme activity. Nevertheless, further enzyme activity assays with other substrates such as 22-hydroxy-3-one and 3-epi-6-deoxoCT are needed.
The CS level in csdd1, d2-3 and d2-4, reduced to 48%, 44% and 38% (Fig. S14a), respectively, which is inconsistent with the severely dwarfed phenotype. CYP90D3 is a close homologous gene of D2 in rice and the transcription level of CYP90D3 in T65 is low but detectable (Hong et al., 2003). The leaf abnormality of the three D2 mutants is less severe. It is possible that CYP90D3 functions mainly in the leaf. We used the aboveground parts (shoots) of 4-wk-old seedlings for analysis and as the shoot apices comprised a very small part of the harvest, the CS level detected was mainly from leaves. Previous studies revealed that endogenous BRs cannot be transported over long distances in plants (Symons & Reid, 2004; Montoya et al., 2005; Symons et al., 2008). Hence, in these mutants, CS produced in leaves could not be transported into internodes to promote internode elongation.
We owe special thanks to: Professor G. An (Department of Plant Systems Biotech, Kyung Hee University, Yongin, Korea) for providing the d2-3 and d2-4 lines; Professor Takao Yokota (Department of Bioscience, Teikyo University, Utsunomiya, Japan) for providing the authentic BRs; Professor M. Matsuoka (BioScience and Biotechnology Center, Nagoya University, Chikusa, Nagoya, Japan) for providing the d2-1 and d2-2 seeds, and Dr T. Lu (Biotechnology Research Institute, Chinese Academy of Agricultural Sciences, Beijing, China) for providing the rice transformation vector pCUbi1390. This research was supported by grants from the National Key Transformation Program (2009ZX08009-104B), National Natural Science Foundation of China (30871498), the earmarked fund for Modern Agro-industry Technology Research System, Jiangsu Science and Technology Development Program (BK2010016, BE2012303), the Jiangsu PAPD Program and by the Next-Generation Biogreen 21 program (No. PJ007967 to S-K.K.), Rural Development Administration, Republic of Korea.