Dominant and pleiotropic effects of a GAI gene in wheat results from a lack of interaction between DELLA and GID1.

Dominance, semidominance, and recessiveness are important modes of Mendelian inheritance. The phytohormone gibberellin (GA) regulates many plant growth and developmental processes. The previously cloned semidominant GA-insensitive (GAI) genes Reduced height1 (Rht1) and Rht2 in wheat (Triticum aestivum) were the basis of the Green Revolution. However, no completely dominant GAI gene has been cloned. Here, we report the molecular characterization of Rht-B1c, a dominant GAI allele in wheat that confers more extreme characteristics than its incompletely dominant alleles. Rht-B1c is caused by a terminal repeat retrotransposons in miniature insertion in the DELLA domain. Yeast two-hybrid assays showed that Rht-B1c protein fails to interact with GA-INSENSITIVE DWARF1 (GID1), thereby blocking GA responses and resulting in extreme dwarfism and pleiotropic effects. By contrast, Rht-B1b protein only reduces interaction with GID1. Furthermore, we analyzed its functions using near-isogenic lines and examined its molecular mechanisms in transgenic rice. These results indicated that the affinity between GID1 and DELLA proteins is key to regulation of the stability of DELLA proteins, and differential interactions determine dominant and semidominant gene responses to GA.

Reduced height (Rht) genes in wheat (Triticum aestivum) are used to increase lodging resistance and harvest index (Gale and Youssefian, 1985). Twenty-one genes with major effects on plant height in wheat have been reported and assigned designations from Rht1 to Rht21 (McIntosh et al., 2008). Two of these genes, Rht1 (semidominant) and Rht3 (completely dominant), are allelic at the renamed Rht-B1 locus (Bö rner et al., 1996;McIntosh et al., 2008). Norin 10, carrying Rht1 and Rht2, is an important semidwarf source that contributed to increased wheat production in the 1960s and 1970s and events known as the Green Revolution (Borlaug, 1983;Gale and Youssefian, 1985;Peng et al., 1999).
In 1999, an Rht gene from Norin 10 wheat was isolated and shown to be an ortholog of an Arabidopsis GA-INSENSITIVE (GAI) gene (Peng et al., 1999). It encoded a DELLA protein. A base substitution difference, C for T, between the sequences of the Rht-B1a tall wild type and the Rht-B1b semidwarf allele led to a stop codon and the termination of translation within the DELLA region, resulting in a truncated Rht protein lacking part of the DELLA domain (Peng et al., 1999). The DELLA protein belongs to a plant-specific family of transcription factors known as GRAS (for GAI, REPRESSOR OF GA1-3 [RGA], and SCARECROW [SCR]) and possesses a variable N-terminal region and a considerably conserved C-terminal GRAS domain (Pysh et al., 1999). Because of its variable N terminus, DELLA has a special function in the GA signal transduction pathway. A DELLA box located at the N terminus acts as the GA signal for repressors of GA signaling Silverstone et al., 2001;Itoh et al., 2002;Fu and Harberd, 2003;Jiang and Fu, 2007). Another important factor in GA signaling is the GA receptor GA-INSENSITIVE DWARF1 (GID1), which was first isolated through the analysis of a rice (Oryza sativa) mutant. GID1 binds to SLENDER RICE1 (SLR1), a rice DELLA protein, in a GA-dependent manner in yeast cells (Ueguchi-Tanaka et al., 2005).
Later papers reported that GA-dependent interactions between GID1 and DELLA protein also occur in Arabidopsis (Arabidopsis thaliana; Griffiths et al., 2006;Nakajima et al., 2006;Willige et al., 2007). They concluded that the DELLA domain of RGA and GAI is necessary for interaction with AtGID1a (Griffiths et al., 2006;Willige et al., 2007). Meanwhile, in rice, Ueguchi-Tanaka et al. (2007), using yeast two-hybrid (Y2H), gel filtration, and bimolecular fluorescence complementation methods, confirmed that DELLA regions are essential and sufficient for GA-dependent GID1, with a DELLA protein-SLR1 interaction. However, GA is not always necessary for the interaction between GID1 and DELLA proteins (Yamamoto et al., 2010). Whereas there has been clear evidence for GID1-mediated GA signaling in Arabidopsis, rice, and nonflowering plants (mosses and ferns; Ueguchi-Tanaka et al., 2005Griffiths et al., 2006;Nakajima et al., 2006;Hirano et al., 2007;Willige et al., 2007;Yasumura et al., 2007;Yamamoto et al., 2010), there has been no report about any interaction between GID1 and DELLA proteins in wheat.
Tom Thumb, a well-known wheat landrace originally from Tibet, China, is characterized by several specific characters: (1) strongly reduced plant height of about 30 cm, much shorter than other dwarf sources in wheat except for Aibian 1; (2) resistance to preharvest sprouting, a serious problem in many wheat-producing regions of the world; (3) late maturity; (4) shriveled seed with reduced fecundity; and (5) larger tiller angle and looser plant type (Morris et al., 1972;Gale 1982, 1983). Genetic analysis showed that Tom Thumb possessed a single completely dominant reduced-height and GAI gene, initially named Rht3, but later revised to Rht-B1c (Morris et al., 1972;Gale and Marshall, 1973;Bö rner et al., 1996;McIntosh et al., 2008). Rht-B1c was located on the short arm of chromosome 4B at a distance of less than 15 centimorgan from the centromere and at the same locus (or closely linked) as Rht-B1b (Gale and Marshall, 1976;McVittie et al., 1978).
As introduced above, several partially dominant genes confer a dwarf phenotype in different plant species. However, the molecular characteristic of Rht-B1c, a dominant and pleiotropic GAI allele, has not yet been reported, and the molecular mechanisms of this GA-insensitive growth response still remain unclear. In this study, we isolated a completely dominant allele from a multiallelic gene in wheat, deciphered its mode of action, and discovered differences between complete and partial dominance.

A Genomic Segment with Large Effects on Multiple Traits
To better understand the effect of Rht-B1c on agronomic traits, we developed a pair of near-isogenic lines (NILs) using Yanzhan 1 (YZ1), a commercial variety that carries Rht-B1a, as a recurrent parent, and Tom Thumb (Rht-B1c) as the donor parent (Fig. 1A). Field examination under natural long-day (Hebei; 115°41#E, 41°41#N) and short-day (Yunnan; 101°41#E, 25°42#N) conditions showed that five agronomic traits (plant height, resistance to preharvest sprouting, tiller angle, fertility and fecundity, and heading date) were significantly affected by the presence of Rht-B1c (Fig.  1). The plant height of the NIL-Rht-B1c was 30 cm, or about 50% (P # 0.01) shorter than NIL-Rht-B1a (YZ1; Fig. 1, A and B). NIL-Rht-B1c had more shriveled seed and lower setting percentage than NIL-Rht-B1a (Fig. 1C). The level of preharvest sprouting of NIL-Rht-B1c was 13.7% and 17.8% in Hebei and Yunnan, respectively, significantly lower than NIL-Rht-B1a (64.7% and 66.7%, respectively; P # 0.01; Fig. 1D). The mean tiller angle of NIL-Rht-B1c plants was obviously larger than that of NIL-Rht-B1a plants (Fig. 1E). There were clear differences in heading date between NIL-Rht-B1a and NIL-Rht-B1c, the latter being 10 d later than NIL-Rht-B1a in Yunnan and 5 d later in Hebei (Fig. 1F). All five traits were investigated in two segregating populations. One was a fifth backcross (Tom Thumb/ 6*YZ1) F2 population, the other was an F2 population derived from a recombinant inbred line (RIL; Rht-B1c) crossed with Yumai 18, which lacks Rht-B1c. These results showed that all traits cosegregated; that is, all short plants showed late maturity, resistance to preharvest sprouting, and displayed larger tiller angles (Supplemental Table S1), obviously reflecting pleiotropic effects of the same dwarfing gene, namely Rht-B1c. Assuming an allelic series at the Rht-B1 locus (Gale and Marshall, 1976), we set out to clone Rht-B1c based on the known sequences of Rht-B1a and Rht-B1b. We amplified template DNAs extracted from Chinese Spring (Rht-B1a), Norin 10 (Rht-B1b), and Tom Thumb (Rht-B1c). We obtained the Rht-B1a and Rht-B1b sequences, which were as expected (i.e. a T-for-C substitution in Rht-B1b; Supplemental Fig. S1). However, we could not get the full-length Rht-B1c sequence from Tom Thumb using this normal PCR procedure.
We next modified the PCR amplification system, such as increasing denaturizing temperature (94°C-98°C) and prolonging extension time (2-4 min), and used a new PCR enzyme (Kod Fx). We obtained a larger PCR-amplified fragment from Tom Thumb. Strikingly, we found that the fragment was about 4 kb, which was approximately twice of the length of the fragment among other alleles ( Fig. 2A). Furthermore, a pair of Rht-B1 primers was designed to amplify the full-length cDNA of Rht-B1c, and the PCR products were sequenced. Compared with Rht-B1a, we found that there was a 2,026-bp insertion at position 147 bp numbered from the ATG start code in Rht-B1c ( Fig. 2B; Supplemental Fig. S1). In the process of transcription, the N-terminal DELLA domain of Rht-B1c exists as a 90-bp insertion, which completely matched the 3# end of the genomic 2,026-bp insertion fragment (Fig. 2B). Interestingly, this is consistent with other alleles in the C-terminal GRAS domain. Rht-B1a (and Rht-B1b) is an intronless gene, whereas Rht-B1c consists of two exons and one intron.
At the posttranslational level, this insertion fragment resulted in an additional 30 amino acid residues in the DELLA domain of the predicted Rht-B1c protein (Fig. 3A). Further detailed sequence alignment analysis showed that the 30 amino acids were inserted between residues Lys (K) and Val (V) of DELLAAL-GYKV, which function as a GA receptor recognition sequence (Murase et al., 2008).
To assess whether this 30-amino acid insertion affected three-dimensional (3D) protein structures of Rht-B1c, we then used the swissmodel workspace (http://swissmodel.expasy.org/) to construct homology protein structure models of Rht-B1a and Rht-B1c based on the template of Arabidopsis GA 3 -GID1A-DELLA (Fig. 3B). As predicted, the DELLA domain of Rht-B1a formed five a-helices (aA, aB, aC, aD, and aE), and aA, aB, aC, and aD were consistent with Arabidopsis GAI. However, Rht-B1c possessed an additional b-sheet, which resulted from the 30-amino acid insertion. In addition, aB of Rht-B1c was shorter than that of Rht-B1a. Thus, the 3D protein structure of Rht-B1c was severely damaged by the supererogatory b-sheet. We also constructed a 3D model based on template GA 4 -GID1A-DELLA and found that there was another a-helix (aF) caused by the 30-amino acid insertion located within loop A-B of the DELLA domain of Rht-B1c (Fig. 3C). These data suggested that the DELLAALGYKV motif was split into regions I (DELLAALGYK) and II (V) by insertion of a b-sheet or an a-helix, and this insertion disrupts the DELLA domain structure.
More detailed analysis of the insertion sequence revealed flanking 5-bp nucleotides containing AG (located upstream of the 5# insertion sequence) and GTG (located downstream of the 3# insertion sequence) to form intact terminal repeat retrotransposons in miniature (TRIM), for which the total length is 2,031 bp. This TRIM element contains a conserved 503bp terminal direct repeat sequence with no substitution or deletion and flanks a 1,015-nucleotide internal domain containing a 19-bp primer-binding site and a 15-bp polypurine tract. The primer-binding site is located immediately downstream of the 3# terminal direct repeat complementary to Met tRNA, and the polypurine tract is located immediately upstream of the 5# terminal direct repeat (Fig. 4A). None of the elements characterized to date encodes mobilityrelated proteins in the internal domain (Fig. 4A). The intact element is flanked by 5-bp direct repeats AGGTG, representing the target site duplications generated after insertion. BLASTed against the Triticeae Repeat Sequence Database (release 10; http://wheat. pw.usda.gov/ITMI/Repeats/), there was low similarity with known elements. Therefore, we deduced that the insertion sequence was a new nonautonomous long terminal repeat retrotransposon known as TRIM, designated as CAAS-TRIM. To further investigate the origin of the inserted TRIM fragment, we searched the draft genome sequence database of Chinese Spring (http://www.cerealsdb.uk.net/search_reads.htm). The homologous sequence obtained from the Chinese Spring database was assembled for about 5,000 bp ( Fig. 4A), and the TRIM region of 2,026 bp was identical to the TRIM sequence found in Tom Thumb, implying a high degree of conservation. To map the TRIM in the genome of Chinese Spring, a pair of primers (TRIM-F/TRIM-R) spanning TRIM was designed based on the TRIM flanking sequences (Fig. 4A) and used to amplify the target fragments in Lumai 14 and Hanxuan 10, parents of a doubled haploid mapping population. A fragment with a length close to the expected 2,862 bp was detected in Lumai 14, confirming that there was a TRIM insertion in the target region. A fragment with a length close to 836 bp was detected in Hanxuan 10 ( Fig. 4B). Sequencing results confirmed that there was no TRIM insertion in the target region of Hanxuan 10. The insertion/deletion polymorphic marker was used to genotype the Hanxuan 10/Lumai 14 doubled haploid population. Using available mapping data for the same population, linkage analysis showed that TRIM was located on chromosome 6B and flanked by simple sequence repeat markers Xwmc341 and Xwmc182 at 7.9 and 7.5 centimorgan, respectively (Fig. 4C). To detect the distribution of CAAS-TRIM-6B in wheat germplasm, 150 diverse wheat accessions were screened with the CAAS-TRIM-6B marker (Supplemental Table S2). Nine lines, including Chinese Spring, Lumai 14, and Taizhong 23, carried the CAAS-TRIM-6B element (Fig. 4B). However, in most tested lines without a CAAS-TRIM on chromosome 6B, including Langdon (tetraploid durum wheat [Triticum durum]) and Tom Thumb, the element was apparently present on another chromosome, such as 4B in Tom Thumb. TRIM appears to be an active element in the wheat genome capable of random movement to other locations.

Verification of the Pleiotropic Effects of Rht-B1c in Wheat NILs and Transgenic Rice
We developed a specific PCR-based marker (Rht-B1cM) to detect the presence of the 2,026-bp insertion in Rht-B1c, allowing it to be distinguished from other Rht-B1 alleles. Its forward and reverse primers span the 2,026-bp insertion and produced different PCR-amplified fragments with different alleles (Supplemental Fig.  S2A). Tom Thumb, which contains the Rht-B1c allele, gave an amplification product of the expected size (2,300 bp) with the Rht-B1cM-specific primer, whereas the amplified fragment from Chinese Spring without Rht-B1c was about 250 bp (Supplemental Fig. S2B). PCR results showed that Rht-B1cM could be used to detect Rht-B1c in wheat. Rht-B1cM was then employed to test the segregating populations (fifth backcross [Tom Thumb/6*YZ1] F2 population and another F2 population derived from RIL [Rht-B1c] crossed with Yumai 18). The results showed that Rht-B1cM was cosegregated with the target traits (dwarfing, later flowing, larger tiller angles, etc.) for Tom Thumb and also confirmed that Rht-B1c conferred a number of morphological traits.
To further confirm the function of Rht-B1c, we generated constructs with the full-length Rht-B1b and Rht-B1c genes in the vector pCAMBIA1300 under the control of their own wheat promoters. We introduced the constructs into rice cv Nipponbare, which is tall and has small tiller angles, by Agrobacterium tumefaciensmediated transformation. We obtained 25 independent transgenic T0 rice plants with Rht-B1b constructs; all were shorter in stature than Nipponbare. We also obtained approximately 70 transgenic T0 plants with Rht-B1c; most of them showed the expected dwarf phenotypes ( Fig. 5; Table I). Compared with Nipponbare, the average height of pRht-B1b:Rht-B1b heterozygous transgenic plants was reduced by 23.8%, whereas the average height of pRht-B1c:Rht-B1c T0 transgenics at 8.6 cm was reduced by 80.9% (Table I). This effect of Rht-B1c was more extreme in transgenic rice than in wheat, where the reduction was about 50%. In order to answer whether GA responses were correlated with the dwarfism, we analyzed the GA response in those transgenic rice plants expressing Rht-B1b and Rht-B1c (Supplemental Table S3). The plant height of Nipponbare treated with exogenous GA was significantly higher (4.76 6 1.34 cm; P # 0.01) than that of the nontreated control. Conversely, those transgenic rice plants expressing Rht-B1b and Rht-B1c showed GAinsensitive responses (Rht-B1b, 0.72 6 0.18 cm; Rht-B1c, 0.68 6 0.21 cm).
Furthermore, pRht-B1c:Rht-B1c transgenic plants also showed increased tiller spread with wider tiller angles, whereas pRht-B1b:Rht-B1b transgenics, like Nipponbare, had a compact plant architecture with narrower tiller angles (Fig. 5). pRht-B1c:Rht-B1c transgenics had delayed heading dates, whereas pRht-B1b:Rht-B1b transgenics and Nipponbare had similar heading dates. These results confirmed the pleiotropic functions of Rht-B1c in reducing plant height, increasing tiller angle,  and delaying heading date. The above results also confirmed that the cloned gene was Rht-B1c. Reverse transcription-PCR analysis revealed no significant difference in transcription level between Rht-B1b and Rht-B1c (Supplemental Fig. S2C), suggesting that the difference in plant height between Rht-B1b and Rht-B1c was not due to a difference in expression level.
Comparative Responses of Rht-B1b and Rht-B1c to Exogenous GA 3 Rht-B1b and Rht-B1c have different mutations leading to different of functions. In order to explore the differences, we compared the physiological effects of GA 3 , a-amylase activity, and TaGID1 protein interaction with Rht-B1b and Rht-B1c.
The seedling length responses of lines with Rht-B1a, Rht-B1b, and Rht-B1c to exogenous GA 3 treatment were assessed on 10-d-old seedlings sprayed with 50 mL L 21 exogenous GA 3 after germination (Fig. 6A). The average length of NIL-Rht-B1a seedlings to such treatment was significantly higher (30%; P # 0.01) than that of the nontreated control. Seedlings with Rht-B1b differed from the control at P # 0.05, with an increase of 11%. GA treatment of Rht-B1c seedlings resulted in a 5% increase that was not statistically significant (Fig. 6B). The GA responses of the three genotypes were significantly different at P # 0.01 (Fig. 6C). These results indicate that Rht-B1a has a normal GA response, Rht-B1b has a reduced GA response, and Rht-B1c is GA insensitive.
The seed aleurone layer provides an effective and convenient system for the study of molecular mechanisms involved in GA-regulated gene expression, and exogenous GA 3 causes rapid increases in a-amylase expression in isolated aleurone layers (Gubler et al., 1995). The differences in response to GA 3 in terms of a-amylase activity among aleurones of Rht-B1a, Rht-B1b, and Rht-B1c individuals were determined by an agar plate assay using embryoless seed halves (Fig. 6, D and E). Controls with Rht-B1a, Rht-B1b, and Rht-B1c produced scarcely any a-amylase in the absence of exogenous GA 3 (Fig. 6D). Embryoless half-seeds from Rht-B1c genotypes produced a very small amount of a-amylase on the GA 3 plates, whereas those from Rht-B1a and Rht-B1b plants secreted the enzyme, as evidenced by the clear zones on the plates containing GA 3 ; the clear zones of Rht-B1a were larger than those for Rht-B1b, indicating higher activity (Fig. 6E). These results indicated that the a-amylase activity in wildtype (Rht-B1a) plants was higher than those in the Rht-B1b and Rht-B1c variants. Because Rht-B1c genotypes were more insensitive than Rht-B1b to GA 3 , much less a-amylase was produced in the seeds, resulting in further resistance to preharvest sprouting.
In order to detect the levels of interaction between TaGID1 protein with Rht-B1a, Rht-B1b, and Rht-B1c, we used full-length Rht-B1a, Rht-B1b, and Rht-B1c cDNA as prey constructs, and TaGID1 full-length cDNA was amplified from Chinese Spring cDNA as the bait construct. Interaction activity between Rht-B1a and Rht-B1b with TaGID1 protein in both plate and liquid assays showed that GA-dependent interaction occurred between the proteins (Fig. 6F). Interestingly, positive colonies of Rht-B1b on fresh synthetic dextrose/ 2Trp/2Leu/2His/2Ade master plates grew for longer times than those of Rht-B1a before colonies appeared. Positive colonies of Rht-B1a were observed on plates after about 36 h, whereas for Rht-B1b, the corresponding time was more than 60 h. Liquid assays showed similar results to the plate assays; the highest activity was detected with Rht-B1a-TaGID1, which was about twice that for Rht-B1b-TaGID1 in the presence of GA (Fig. 6F, right). We next examined the biochemical characteristics of the DELLA protein containing the 30amino acid insert in the Lys-49 codon (Rht-B1c) in terms of its interaction with TaGID1 protein. In agreement with the absence of colonies produced by Rht-B1c-TaGID1 on the same plates even after much prolonged periods (Fig. 6F), the liquid assay also showed that GAdependent interaction between TaGID1 and Rht-B1c was abolished. These results thus showed that a nucleotide substitution in Rht-B1a to generate Rht-B1b led to a partial decrease in the binding activity, whereas the 90bp insertion into Rht-B1a to produce Rht-B1c resulted in a drastic decrease in the presence of GA.

Dominance/Recessiveness of GAI Depends on the Level of Interaction between GAI and GID1
The previous studies have shown that the semidominant GAI genes Rht1 and Rht2 in wheat were the  (Peng et al., 1999). However, the completely dominant GAI gene has not yet been cloned. In this study, we demonstrated that the completely dominant Rht-B1c, an ortholog of Arabidopsis GAI, was caused by a 30amino acid insertion in the DELLA domain and disrupted the GA-dependent interaction with TaGID1. According to the "relief-of-restraint" model, DELLA proteins restrain plant growth while GA relieves DELLA repression by overcoming DELLA-mediated growth restraint Richards et al., 2001;Silverstone et al., 2001;Harberd, 2003;Jiang and Fu, 2007;Hou et al., 2010;Gao et al., 2011). Binding of bioactive GA to GID1 promotes the interaction of GID1 with the DELLA proteins, the subsequent polyubiquitination of the DELLAs via the E3 ubiquitin-ligase SCF SLY1/GID2 , and the eventual destruction of DELLA proteins in the 26S proteasome (McGinnis et al., 2003;Sasaki et al., 2003;Dill et al., 2004;Fu et al., 2004;Ueguchi-Tanaka et al., 2005;Griffiths et al., 2006;Jiang and Fu, 2007;Murase et al., 2008;Harberd et al., 2009;Hou et al., 2010). Thus, DELLA proteins restrain plant growth while GA promotes growth by targeting the DELLA proteins for degradation (Silverstone et al., 2001;Fu et al., 2002;Gao et al., 2011).
Previous studies have suggested that the DeLLaF-LxYxV motif is a GA receptor GID1 recognition sequence (Murase et al., 2008), and the deletion of this motif results in a GA-insensitive dwarfism, such as in gai and rga-D17 (Peng et al., 1997;Itoh et al., 2002). In our experiments, we found that there were no, weak, and strong interactions between TaG-ID1 and Rht-B1c, Rht-B1b, and Rht-B1a, respectively. Furthermore, we found that the 30-amino acid insertion into the DELLA domain causes the reduction of affinity between GID1 and Rht-B1c, which in turn abolished the GA-mediated degradation of DELLA protein and consequently resulted in DELLA protein accumulation and GA-insensitive dwarfism. These observations suggested that Rht-B1c might not form Figure 6. Loss of interaction activity with TaGID1 results in GA insensitivity. A, Morphologies of 10-d-old Rht-B1a, Rht-B1b, and Rht-B1c wheat seedlings treated with 50 mL L 21 GA 3 (right) or without GA 3 (left). Bar = 5 cm. B, Comparison of seedling lengths of 10d-old Rht-B1a, Rht-B1b, and Rht-B1c plants treated with (gray bars) or without (black bars) 50 mL L 21 GA 3 . Data were measured as means of 30 seedlings at each time with three replicates. Significance is as follows: * P # 0.05, ** P # 0.01. C, Relative variances of Rht-B1a, Rht-B1b, and Rht-B1c seedlings. The formula for variance is as follows: relative variance = (length of seedling with GA 3 2 length of seedling without GA 3 )/length of seedling without GA 3 3 100%. Significance is as follows: ** P # 0.01. D and E, GAmediated induction of a-amylase activity in Rht-B1a, Rht-B1b, and Rht-B1c seeds. A concentration of 1 mM GA 3 was applied to the plates. Sector I, Rht-B1a; sector II, Rht-B1b; sector III, Rht-B1c. F, Interaction between TaG-ID1 and Rht-B1. Y2H assays were performed using TaGID1 as bait and Rht-B1a, Rht-B1b, and Rht-B1c as prey. Left, growth of yeast strain AH109 transformants on 2His2Ade plates; right, b-galactosidase activity detected in liquid assay with yeast strain AH109 transformants. To measure relative lacZ activity, three independent experiments were carried out. the Rht-B1c-TaGID1 complex. Therefore, the affinity of the interaction between DELLA and GID1 controls DELLA protein accumulation, which determines completely insensitive, semi-insensitive, and sensitive responses to GA in Rht-B1c, Rht-B1b, and Rht-B1a, respectively.

Pleiotropism of GAI Is Associated with Its Dominance
Our results showed that Rht-B1c affects several traits, including strongly reduced plant height, preharvest sprouting tolerance, extreme prostrate habit, late flowering, and partial male sterility. However, Rht-B1b, a semidominant GAI, affects only plant height. GA acts as a major hormone in plants and regulates many processes of growth and productivity. It has been reported that GA controls plant height (Richards et al., 2001;Jiang and Fu, 2007;Davière et al., 2008;Gao et al., 2008;Shimada et al., 2008), seed germination and dormancy Finch-Savage and Leubner-Metzger, 2006;Razem et al., 2006;Seo et al., 2006), fertility and flowering (Achard et al., 2004(Achard et al., , 2007Cheng et al., 2004;Kaneko et al., 2004;Chhun et al., 2007;Lin et al., 2011), and prostrate habit (Cui et al., 2005;Appleford et al., 2007;Wolbang et al., 2007). Our results showed that plants with Rht-B1c are completely insensitive to GA, whereas those with Rht-B1b and Rht-B1a are weakly insensitive and sensitive, respectively. We hypothesized that responses to endogenous GA for different traits are different. For example, plant height might be more sensitive than other traits, such that small reductions in endogenous GAs result in reduced height without significantly affecting the other traits. Thus, only dramatic reductions in endogenous GAs lead to multiple phenotypic changes. Further studies are required to examine these differential responses to GA.

Structural Characteristics and Function of CAAS-TRIM-6B
In this work, we determined that the Rht-B1c mutation resulted from an insertion of a highly conserved TRIM element, CAAS-TRIM, belonging to the long terminal repeat retrotransposon class. TRIM is believed to be constantly active (Witte et al., 2001). Sabot et al. (2006) suggested that TRIM elements were the most active retrotransposons in the wheat genome based on an activity index (Sabot et al., 2006). According to Witte et al. (2001), TRIM elements have significant functions in the reorganization of plant genomes by affecting the promoter, coding region, and intronexon structure of genes (Witte et al., 2001). In this report, a TRIM-CAAS insertion in the DELLA domain of Rht-B1a resulted in a structural change of the gene. TRIM elements also have positive roles in creating new functions in the Brassica genome (Yang et al., 2007). High conservatism of terminal direct repeat sequences showed that CAAS-TRIM is a recently inserted element. However, we have not been able to determine the mobilization mechanism of CAAS-TRIM, because it does not appear to encode its own proteins, such as gag, pol, and env. The retrotransposons Morgane and LARDs similarly lack all or most of their open reading frames, indicating that they are probably parasitic on autonomous retrotransposon families that in turn are parasitic on the genome (Witte et al., 2001;Kalendar et al., 2004;Antonius-Klemola et al., 2006;Sabot et al., 2006;Tanskanen et al., 2007). We detected CAAS-TRIM in 150 wheat accessions and found that it acts as a highly active element capable of being inserted in different locations of the wheat genome. When it is inserted into a gene, such as Rht-B1a, the gene structure and function may be altered. The function is similar to that of rice retrotransposon Tos17 and maize (Zea mays) transposable elements Ac/ Ds and En/Spm (Walbot, 1992;Wisman et al., 1998;Parinov et al., 1999;Speulman et al., 1999;Tissier et al., 1999;Hirochika, 2001). Therefore, it can be used as a tool to study wheat gene function by the detection of genes in which it is inserted. Among 150 accessions in which the element was detected in this study, its position was located on chromosome 6B in only nine. Detection of the insertion position in the other accessions may lead to the discovery of genes with altered functions. We assume that insertion is random and, therefore, that its insertion into Rht-B1a was an extremely rare event, probably explaining why Tom Thumb is a unique short variant among hundreds of thousands of wheat accessions in the world.

Potential Application of Rht-B1c in Crop Improvement
The cloning of Rht-B1c will accelerate its application in agriculture. Our work confirms that Rht-B1c is a GAI gene. GAI genes are highly conserved across plant species, including Arabidopsis (GAI), Brassica rapa (dwf2), grape (Vitis vinifera; VvGAI1), rice (SLR1), maize (D8), barley (Hordeum vulgare; SLN1), and wheat (Rht-B1b and Rht-D1b;Peng et al., 1999;Ikeda et al., 2001;Boss and Thomas, 2002;Chandler et al., 2002;Muangprom and Osborn, 2004). We transferred Rht-B1c to rice and proved that it functions in that species, where its effect was even more extreme than in wheat. Hence, it should function in other species. Its ability to confer large reductions in plant height could be used in tall plants, such as rye (Secale cereale), maize, and sorghum (Sorghum bicolor). Its ability to reduce height and increase tiller angles and to delay heading date might be useful in turf grass improvement. Among its several pleiotropic functions, some are advantageous for plant improvement and some are disadvantageous. To utilize favorable functions and avoid unfavorable ones, tissue-specific promoters may offer opportunities. For example, resistance to sprouting is an important attribute of Rht-B1c, but to exploit this function and avoid unwanted functions, an endosperm-specific promoter could be used when constructing a vector for transformation.

Plant Materials, Growth Conditions, and Field Trials
Materials included wheat (Triticum aestivum) lines Tom Thumb, Norin 10, and Chinese Spring. For developing NILs for the Rht-B1c allele, a cross was made between cv YZ1 with genotype Rht-B1a/Rht-D1b as a recurrent parent and Tom Thumb with genotype Rht-B1c as donor. A segregating population of 408 F2 plants derived from a cross between Yumai 18 and Rht-B1c-RIL (selected from a RIL population derived from a cross between Tom Thumb and YZ1), a previously mapped random set of 150 Hanxuan 10 3 Lumai 14 doubled haploid lines, and 150 diverse wheat accessions from different countries (Supplemental Table S2) were also used. The NILs and parents were planted in long-day (Hebei; 115°41#E, 41°41#N) and short-day (Yunnan; 101°41#E, 25°42#N) environments. The F2 segregating population, doubled haploid mapping population, and 150 diverse wheat accessions were planted in Beijing (116°21#E, 39°54#N). Plant heights were measured as means of five individuals in each plot after flowering. Tiller angles were measured between the vertical and most extreme intact tiller on each plant at the mature stage. One hundred visually undamaged seeds per wheat genotype were used in germination tests with two replicates 10 d before harvest.

GA Treatment Used in Seedlings and Assays of a-Amylase Activity
For GA sensitivity, seeds of lines with Rht-B1a, Rht-B1b, and Rht-B1c were germinated in water. Seedlings were sprayed daily with an aqueous solution containing or lacking 50 mL L 21 GA 3 for 10 d. For assays of a-amylase activity, embryoless half-seeds of wheat were surface sterilized for 30 min with 3% NaClO and washed several times with sterile distilled water. The half-seeds were placed on an agar medium (pH 5.3) containing 0.2% starch, 2% agarose, 10 mM sodium acetate, and 2 mM CaCl 2 with or without 1 mM GA 3 . The plates were incubated for 60 to 72 h at 30°C in darkness, following which they were exposed to iodine vapor to detect whether starch had been digested by a-amylase.

Plasmid Construction and Rice Transformation
Full-length Rht-B1b and Rht-B1c cDNAs were produced by reverse transcription-PCR from total RNA from Norin 10 and Tom Thumb. For transgenesis, full-length Rht-B1b and Rht-B1c cDNAs, including about 4 kb of the promoter and NOS terminator regions, were inserted into binary vector pCAMBIA1300 to generate the plasmids pRht-B1b:Rht-B1b and pRht-B1c:Rht-B1c.
The binary vectors were introduced into Agrobacterium tumefaciens strain EHA105 by electroporation. Rice transformation was performed as described by Hiei et al. (1994). Transgenic rice plants were selected on medium containing 50 mg L 21 hygromycin. Transgenic rice plants were grown in a greenhouse in Beijing (116°21#E, 39°54#N) and in the field in Hainan (110°0#E, 18°24#N).
Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers JN857970 (gDNA of Rht-B1c) and JN857971 (cDNA of Rht-B1c).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S2. Specific marker for Rht-B1c in wheat and mRNA levels for Rht-B1b and Rht-B1c in transgenic rice.
Supplemental Table S1. Agronomic traits of NILs and the F2 segregating population planted under natural conditions.
Supplemental Table S2. One hundred fifty accessions representing a broad range of wheat germplasm.
Supplemental Table S3. Increasing heights of different transgenic rice plants treated with GA 3 .