Exogenous GA3 Application Can Compensate the Morphogenetic Effects of the GA-Responsive Dwarfing Gene Rht12 in Bread Wheat

The most common dwarfing genes in wheat, Rht-B1b and Rht-D1b, classified as gibberellin-insensitive (GAI) dwarfing genes due to their reduced response to exogenous GA, have been verified as encoding negative regulators of gibberellin signaling. In contrast, the response of gibberellin-responsive (GAR) dwarfing genes, such as Rht12, to exogenous GA is still unclear and the role of them, if any, in GA biosynthesis or signaling is unknown. The responses of Rht12 to exogenous GA3 were investigated on seedling vigour, spike phenological development, plant height and other agronomic traits, using F2∶3 and F3∶4 lines derived from a cross between Ningchun45 and Karcagi-12 in three experiments. The application of exogenous GA3 significantly increased coleoptile length and seedling leaf 1 length and area. While there was no significant difference between the dwarf and the tall lines at the seedling stage in the responsiveness to GA3, plant height was significantly increased, by 41 cm (53%) averaged across the three experiments, in the GA3-treated Rht12 dwarf lines. Plant height of the tall lines was not affected significantly by GA3 treatment (<10 cm increased). Plant biomass and seed size of the GA3-treated dwarf lines was significantly increased compared with untreated dwarf plants while there was no such difference in the tall lines. GA3-treated Rht12 dwarf plants with the dominant Vrn-B1 developed faster than untreated plants and reached double ridge stage 57 days, 11 days and 50 days earlier and finally flowered earlier by almost 7 days while the GA3-treated tall lines flowering only 1–2 days earlier than the untreated tall lines. Thus, it is clear that exogenous GA3 can break the masking effect of Rht12 on Vrn-B1 and also restore other characters of Rht12 to normal. It suggested that Rht12 mutants may be deficient in GA biosynthesis rather than in GA signal transduction like the GA-insensitive dwarfs.


Introduction
Gibberellins (GAs) are a major class of plant hormones that regulate plant growth and development, from seed germination and stem elongation to fruit-set and growth [1][2][3][4]. It is important for plants to produce and maintain optimal levels of bioactive GAs to ensure normal growth and development. Mutants with impaired GA biosynthesis or response show typical GA-deficient phenotypes, such as dark green leaves, dwarfism and lateflowering, while elevated exogenous GA dose or increased signaling can cause excessive plant growth and earlier flowering [5][6][7][8]. Mutants deficient in GA biosynthesis can be rescued by exogenously applied GAs but this is not possible if the mutation is in the GA signaling pathway [7,8].
The deployment of genes influencing plant height through the GA pathway was a major factor in the success of the Green Revolution, which created high-yielding cultivars of rice and wheat with shorter and sturdier culms [9]. In contrast to the recessive, semi-dwarf sd-1 Green Revolution allele in rice, which is a loss-of-function mutation in one of the major GA biosynthetic genes (GA20ox2) [10][11][12], the reduced height Rht-B1b (Rht1) and Rht-D1b (Rht2) Green Revolution alleles in wheat are semidominant gain-of-function mutations causing impaired GA signaling and thus conferring dwarfism through constitutive repression of cell division and elongation [13]. The wheat Green Revolution genes are orthologues of the Arabidopsis GA-insensitive (gai), the rice slender1 (slr1) [14] or the rice gai [15], the barley slender1 (sln1) [16,17] and the maize dwarf-8 (d8) [13] genes. However, in addition to reducing plant stature, Rht-D1b and Rht-B1b also reduce seedling vigour and coleoptile length, and may reduce crop water-use efficiency [18][19][20][21] and performance in some unfavorable environments [22][23][24]. So, opportunities exist for replacing Rht-B1b and Rht-D1b in wheat with alternative dwarfing genes, such as the GA-responsive dwarfing genes (Rht4, Rht5, Rht9, Rht12, Rht13, Rht14, Rht15, Rht16 or Rht18). These genes have been reported to reduce plant height without compromising early plant growth [25]. Even though there are several GA-responsive dwarfing genes in wheat, their molecular characteristics remain obscure [26,27] and the mechanisms by which they resulted in a reduction of plant height is not well understood. Fortunately, the metabolic pathways of gibberellin biosynthesis, deactivation and signaling have become relatively clear and many of the genes involved have been identified [28,29], which lays the foundation for analysis of GA-responsive dwarfing genes in wheat.
Generally, dwarfing genes in wheat are classified into two categories, GA-responsive (GAR) and GA-insensitive (GAI), reflecting the relative magnitude of their responses to application of exogenous GAs [25,30]. GA-responsive dwarfing genes show significantly enhanced growth response to exogenous GAs (probably have mutations in GA biosynthesis pathway) while GA-insensitive dwarfing genes show very little response to exogenous GAs (probably have mutations in GA signaling pathway, such as Rht-D1b and Rht-B1b) [25]. This classification has usually been conducted at the seedling stage, for example, based on the response of coleoptile length or the first seedling leaf elongation rate to exogenous GAs [25,30]. There is less information available on the response of the GAR dwarfing genes to exogenous GAs at later growth stages.
Rht12, a dominant dwarfing gene from the gamma ray-induced mutant Karcagi 522M7K of winter wheat (here referred to as Karkagi- 12), has been classified as a GA-responsive dwarf gene [31]. We have previously reported the effects of Rth12 on wheat using a F 2:3 population derived from a cross between Ningchun 45 (a tall spring wheat cultivar) and Karcagi-12 in two field experiments [32]. Rht12 significantly decreased stem length (43%,48% for peduncle) and leaf length (25%,30% for flag leaf), while the thickness of the internode walls and width of the leaves were increased. Additionally, the Rht12 dwarf lines showed very dark green leaves compared to tall lines. Rht12 significantly decreased plant height, by around 40%, while seedling vigour, coleoptile length and root traits at the seedling stage were not affected adversely. Rht12 lines had significantly increased floret fertility and grain number and achieved a higher harvest index (due to the lower plant biomass) than the tall genotypes. However, Rht12 extended the duration of the spike development phase, especially the duration from sowing to double ridge, and delayed anthesis date by around 5 days. Even the dominant Vrn-B1 allele could not compensate for these effects on phenological development, which may hamper the direct utilization of Rht12 in wheat breeding. Another negative effect of Rht12 on yield components was that grain size was reduced significantly. Similarly, other studies have found that Rht12 had a substantial effect on reducing plant height without altering early vigour and significantly increased spikelet fertility, harvest index, and lodging resistance but these were usually accompanied by delayed ear emergence and reduced grain weight. [31,[33][34][35].
Although Rht12 has been classified as a GA-responsive dwarfing gene [25,33], a comprehensive understanding on the response of Rht12 to exogenous GAs is lacking. Thus, the role of Rht12, if any, in GA biosynthesis or signaling is still unclear. Moreover, it has been recently found that the effects of the dwarfing gene Rht8, which had been considered as 'GA-responsive', was possibly not due to defective GA metabolism or signaling because the wild type and Rht8 lines responded with a very similar increase in final plant height (15% and 13%, respectively; P,0.05) with GA 3 application. It has been proposed that the effects of Rht8 are possibly due to reduced sensitivity to brassinosteroids [26]. So, can exogenous GA 3 restore the characters of Rht12 (such as plant height, flowing time or seed size) to normal? Is there a significant difference between Rht12 lines and the tall lines in GA 3 -responsiveness?
To date none of the dwarfing genes have been cloned except a few GA-insensitive dwarf genes [27]. More information on these genes is needed for a better understanding of how the GAR dwarf genes act on plant growth and what their roles are in GA biosynthesis or signaling. The aim of this paper was to extend the work of Chen et al. [32] by examining the response of Rht12 to exogenous GA 3 on plant development, agronomic traits and yield components and to investigate the possible role of Rht12 in GA biosynthesis or signaling. The response to exogenous GA 3 of Rht12 dwarf and tall lines were compared to determine if Rht12 was responsive to exogenous GA 3 during the whole life cycle of wheat and if Rht12 dwarf lines had a notably greater responsiveness to exogenous GA 3 than tall lines.

General Description
The experiments were carried out in the experimental field of the Institute of Water Saving Agriculture in Arid Regions of China, Northwest A&F University, Yangling, Shaanxi, China, during two growing seasons of 2011-2012 and 2012-2013. Supplemental irrigation was provided as needed to avoid water stress. Weeds were manually removed where necessary, and fungicides and insecticides were applied to prevent diseases and insect damage. Weather data were recorded at an automated weather station at the site.

Plant Material
A cross was made using Ningchun 45 as the female and Karcagi-12 as pollen donor in May, 2009 as reported by Chen et al [32]. Karcagi-12 (Triticum aestivum L.) is the mutant carrying the dominant GA-responsive dwarfing gene Rht12 and shows strong winter habit with the recessive loci for all three Vrn-1 genes. Ningchun 45 (Triticum aestivum L.) is a tall Chinese spring wheat cultivar which carries the dominant gene Vrn-B1 and lacks any known dwarfing genes detectable by molecular markers.
The two parents and 24 random F 2:3 homozygous lines (of a total of 57 homozygous lines identified) were used in the experiments to evaluate the effects of exogenous GA 3 on Rht12. Among the F 2:3 homozygous lines there were 7, 7, 5 and 5 lines with the genotypes RRBB, RRbb, rrBB, and rrbb, respectively. There were two sowing dates in 2011-2012 growing season: October 6, 2011 (Autumn Sowing, AS) and February 6, 2012 (Spring Sowing, SS). In 2012-2013 growing season, the 24 F 3:4 lines were sown on October 6, 2012 (Autumn Sowing, AS). The lines and parents were sown in plots of four rows that were 2 m long and 25 cm apart, with seeds spaced 5 cm apart within rows. The parents and the 14 dwarf and 10 tall lines were randomly arranged to avoid competitive effects with two replications.
Exogenous Gibberellic Acid (GA 3 ) Treatments GA 3 was applied on the dwarf and the tall genotypes and the two parents. GA 3 solution (100 mM or 35 mg/L) was sprayed using a small aerosol or smeared on the leaves and culm surface by cotton at several stages of development: the 5 leaf stage (Z15), tillering stage (Z21), stem elongation stage (Z31), early booting stage (Z41), early heading stage (Z51), early anthesis stage (Z61) and early kernel and milk development stage (Z71) [25,26,36]. For each plant, 1-2 mL GA 3 solution was applied at the first 2 developmental stages and 3-5 mL at the later 5 stages. Control plants were treated with the same solution without GA 3 [26,37].

Coleoptile Length and Seedling Root Characters
Coleoptile length was measured from the seed to the tip of the coleoptile with a ruler after germination in a darkened growth chamber irrigated with water or GA 3 solution (100 mM) at 20uC after 200uCd using the method described [22,38]. The characteristics of seedling leaf 1 and seedling root were assessed using a 'Cigar' method as described previously [32]. Seedling vigour was evaluated as the area of the first leaf and coleoptile length.

Spike Development and Fertility
Beginning at the 5 leaf stage (Z15), three plants were randomly taken from each plot every 3 days to observe spike differentiation. The main shoot was dissected to determine the timing of double ridge formation (DR), as described [39] using a digital Stereo Microscope (Nikon, SMZ1500). The timing of heading (Z55) and anthesis (Z65) was visually determined when 50% of the plants in the plot had reached the stages.
At anthesis, five plants were harvested in the central row of each plot, and then the number of fertile florets in the main shoot spike was counted. Florets were considered fertile when the stigmatic branches spread wide, with either pollen grains present on them or with green anthers [40].

Plant Height, Leaf and Internode Character
Seedling height was measured in 10 plants for each plot as the distance from the soil surface to the ligule of the last fully emerged leaf, and the elongation rate of different lines was analyzed weekly from Z12 (two leaf stage) until heading. Seedling dry weight was calculated from 3 plants of each line every week to determine the plant growth rate. The number of leaves emerging on the main shoot was determined on 5 plants per plot [41]. At maturity, ten random plants for each plot were measured to get the mean plant height as the distance from the soil surface to the top of the ear (awns excluded).
The lengths of internodes (mm) were measured from the midpoint of their subtending nodes. Stem diameter (mm) was taken at the middle of each internode with digital calipers (TESA Etalon). Internodes were then cut at their centre point and the thickness (mm) of stem wall was obtained using digital calipers, with two measurements on opposite sides of the stem, then a mean value was calculated [42]. The characters of the internodes were measured when they fully developed.
At maturity, lodging was scored as the fraction of the area affected and the severity of lodging in those areas noted, on a scale of 0 for a standing crop to 90 for a crop flat on the ground [36].

Yield Components and Yield
At maturity, the main shoot ears of 10 plants of each line were measured for assessment of spike length, spikelets per spike, grains per spike and fertile shoots per plant. Due to the frequent sampling prior to maturity about 20 to 40 plants remained in each plot for investigating the average above-ground biomass per plant, average yield per plant, harvest index and 1000-grain weight.

Data Analysis
For each parameter measured, the mean value for each homozygous class (7 RRBB, 7 RRbb, 5 rrBB and 5 rrbb lines) was calculated. Because the previous study [32] found that there was no significant difference between BB and bb genotypes on many traits in either dwarf or tall lines, statistical evaluation of the data was carried out as two categories (14 RR and 10 rr lines) by ANOVA analysis with multiple comparisons (LSD test at the 0.05 level) using the statistical package SPSS18.0. 0uC was chosen as base temperature, whenever thermal time was used to estimate developmental progress.

Seedling Vigour
In the F 2:3 and F 3:4 lines, no significant difference was observed for the length or width of seedling leaf 1 between the tall and dwarf groups in the absence of exogenous GA 3 . In the treatments with exogenous GA 3 , the area of seedling leaf 1 significantly increased in both the dwarf and tall lines. The Rht12 dwarf lines had a small but significant increase in leaf length over the tall lines, however, there was no notable difference in its response to exogenous GA 3 between dwarf and tall lines ( Table 1; Fig. 1). Additionally, coleoptile lengths of the dwarf and tall lines in the GA 3 treatments were both significantly increased compared to that of the control treatments in both years (29% and 41% in the dwarf lines and 13% and 24% in the tall lines, respectively) ( Table 1). There was no significant difference in coleoptile length of Rht12 genotypes compared with the corresponding tall lines, either with or without additional GA 3 . Summarizing, in comparison with the wild-type tall, Rht12 had no major effect on the growth of seedling leaf 1 or coleoptile length and also on their responsiveness to exogenous GA 3 .
Exogenous GA 3 had negative effects on seminal root growth in both dwarf and tall lines. The total root length and root dry mass of the tall and Rht12 dwarf lines was significantly decreased in the GA 3 treatments and the dry mass ratio of root to shoot was also significantly reduced due to the increased shoot dry mass as well ( Fig. 1).

Spike Development
For the three experiments in the two years, spike development was significantly delayed and the duration of the spike development phase was extended in the Rht12 dwarf lines compared to that of tall lines. There was no significant difference between RRBB and RRbb groups in these effects. Compared to the tall lines with winter growth habit (rrbb), time to double ridge of the Rht12 dwarf lines either with Vrn-B1 or with vrn-B1 was delayed by 16 days (45uCd), 5 days (89uCd) and 23 days (105uCd) in the three experiments respectively (Table 2). Thus, the effect of the dominant vernalization gene Vrn-B1 was masked in the Rht12 dwarf lines. In contrast, the tall lines without Rht12 but with Vrn-B1 (rrBB) reached double ridge about 64 days (30uCd), 4 days (57uCd) and 48 days (55uCd) earlier than those with vrn-B1 (rrbb) in the three experiments, respectively, indicating that tall plants having dominant Vrn-B1 needed less time to undergo vernalization. Application of exogenous GA 3 significantly affected the progress of spike development and shortened the duration of the pre-anthesis phase in the dwarf lines (Table 2). It especially shortened the time to double ridge. The RRBB-GA lines (Rht12 dwarf lines with Vrn-B1, with GA 3 application) showed 39 days (18uCd), 5 days (70uCd) and 24 days (25uCd) shorter SW-DR than the RRbb-GA lines (Rht12 dwarf lines with vrn-B1, with GA 3 application) in the three experiments, respectively. It seemed that exogenous GA 3 relieved the epistatic effects of Rht12 on Vrn-B1 in the dwarf lines, such that RRBB-GA plants displayed a spring-like phenotype while RRbb-GA still showed a winter phenotype. Moreover, tall lines with GA 3 treatment also reached double ridge quicker, by about 2-4 days than the untreated tall plants, though this effect was not as significant as in the Rht12 dwarf lines ( Table 2). The flowering of Rht12 dwarf lines was delayed by about 5-8 days compared to tall lines while there was no significant difference between BB and bb lines in either year (Table 2). However, the GA 3 -treated Rht12 dwarf lines flowered earlier by almost 7 days compared to those without GA 3 treatment, while the GA 3 -treated tall lines flowered only 1-2 days earlier than the tall lines without GA 3 treatment. It was therefore clear that GA 3 had a greater effect on spike development in the Rht12 dwarf genotypes than in the tall genotypes. Application of GA 3 broke the masking effect of Rht12 on Vrn-B1, indicating either that Rht12 lines had serious defects in the GA metabolic pathway or that exogenous GA 3 could in some other way compensate for the negative effects on plant development by Rht12.
Due to a longer duration to double ridge, Rht12 dwarf lines developed more leaves than the equivalent tall genotypes.
However, due to the shorter SW-DR of the GA 3 -treated Rht12 dwarf plants, less leaves were produced in the GA 3 -treated Rht12 dwarf plants than that in the untreated ones (Table 2). Despite this, the number of elongated internodes was not changed. In this study, the dwarf parent Karcagi-12 generated the largest number of leaves in all experiments due to its longest duration of SW-DR while the GA 3 -treated Karcagi-12 had almost 4 leaves less than the untreated ones.

Plant Height and Associated Traits
The culm elongated faster in tall lines than in Rht12 dwarf lines from seedling stage, with the tall lines reaching jointing stage earlier and producing longer internodes and more biomass than the Rht12 dwarf lines (Fig. 2). This difference was sustained to maturity in both years. However, Rht12 dwarf lines achieved a higher resistance to lodging through shorter internode length and increased wall thickness without altering internode diameter compared with the tall lines. In particular, the lengths of the first and second internodes at the base of the stem, was significantly decreased in the Rht12 dwarf lines while their wall thickness was increased by 0.24 mm (31%) and 0.22 mm (34%) in the AS experiment, 0.23 mm (24%) and 0.23 mm (29%) in the SS experiment in the 2011-2012 growing season and 0.13 mm (15%) and 0.17 mm (22%) in the 2012-2013 growing season, respectively (Table 3). These shorter and thicker internodes might confer a greater resistance to lodging in the dwarf plants. However, after exogenous GA 3 application, the Rht12 dwarf plants showed a faster stem elongation rate, similar to that of tall plants, compared with the untreated Rht12 dwarf plants (Fig. 2). The lengths of the first and second internodes at the base of the stem were significantly increased in the GA 3 -treated Rht12 dwarf plants compared with the untreated dwarf plants: by 4.1 cm (117%) and 6.9 cm (115%) in the AS experiment, 6.3 cm (252%) and 7.1 cm (131%) in the SS experiment of the 2011-2012 growing season and 4.3 cm (165%) and 4.8 cm (87%) in the 2012-2013 growing season, respectively. Also, the diameter and wall thickness of those internodes were significantly reduced in the GA 3 -treated Rht12 dwarf lines and thus these now 'tall' GA 3 dwarf plants had serious lodging, as did the true tall lines (Table 3  respectively (Table 3 and Fig. 4). In contrast, the internode characters and plant height of the tall lines were not affected significantly by GA 3 treatment (,20% and ,10% increase in the length of the first and the second internode at the base of the stem, less than 10 cm (,10%) increase in plant height), suggesting that the tall plants were not as sensitive as the Rht12 dwarf plants to exogenous GA 3 due to the possible difference in the GA metabolic pathways of the two height categories or the tall lines were likely saturated at endogenous levels of GA. Despite this increase in response to application of endogenous GA, the plant height of the RR-GA lines remained lower than those of the tall lines.

Spike Characters and Floret Fertility
Spike length, number of spikelets per spike and number of florets initiated per spike were all decreased by Rht12 compared with the tall lines. However, Rht12 significantly increased the number of fertile florets per spike and achieved a higher fertility than the tall lines (Table 4). If the Rht12 dwarf lines could maintain higher fertility and increase the number of florets initiated per spike after applying exogenous GA 3 , more grains would be produced resulting in increased yield. Unfortunately, in the AS experiments of both years, exogenous GA 3 increased spike length and also number of spikelets per spike in the Rht12 dwarf lines (Fig. 3B) but the number of florets initiated per spike and fertile florets per spike were significantly decreased and the fertility was reduced compared with untreated dwarf ones (Table 4). This lower fertility might have been due to there being less dry matter partitioned to spikes to support floret survival during the late reproductive phase as more available assimilates were transported to the larger vegetative organs in the now 'tall' GA 3 -treated Rht12 dwarf plants. There was no significant difference on spike characters between the GA 3 -treated and untreated plants in the true tall lines though the number of florets initiated per spike was slightly reduced in the GA 3 -treated tall plants.   Possibly due to slower development of the vegetative phase (SW-DR), which shortened the period of time available under favorable conditions prior to flowering, the Rht12 dwarf lines produced fewer (23%) fertile florets per spike than the tall lines in the SS experiment (see [32]). Exogenous GA 3 accelerated plant development in dwarf lines and provided more time under favorable conditions to produce fertile florets in the Rht12 dwarf plants. The fertility of GA 3 -treated Rht12 dwarf lines was 21% higher than that of the untreated ones in the SS experiment (Table 4).

Yield and Yield Components
For the AS experiments in both years, grain numbers per spike were significantly increased whereas 1000-grain weight and plant biomass were significantly decreased in Rht12 dwarf lines compared to the tall lines. The increased grain numbers and the larger number of fertile ears per plant of the dwarf lines resulted in there being no significant difference in plant yield between the dwarf and tall lines. Additionally, since plant biomass was significantly decreased in the Rht12 dwarf lines, there was a net increase in harvest index (Table 5). Exogenous GA 3 increased plant height of Rht12 dwarf lines as well as plant biomass but it reduced the number of fertile florets, which resulted in a lower grain number per spike and lower plant yield (Table 5). This was despite exogenous GA 3 increasing seed size in the Rht12 dwarf lines (Fig. 3C). The 1000-grain weight of the GA 3 -treated Rht12 dwarf plants was significantly increased, by 9.2 g (28%), 3.5 g (12%) and 6.5 g (18%), compared with untreated dwarf plants in the three experiments, respectively (Table 5). This indicated that GA 3 could partially compensate for the substantial negative effect of Rht12 on yield components, although the increased 1000-grain weight of the GA 3 -treated Rht12 dwarf lines was still less than that of the tall lines. Additionally, the number of efficient spikes per plant and plant yield were decreased in the GA 3 -treated Rht12 dwarf lines compared with the untreated ones. Thus the harvest index of the GA 3 -treated Rht12 dwarf lines was significantly reduced, by 25% and 18% in the two AS experiments, compared with that of the untreated plants.  In the SS experiment, the Rht12 dwarf lines performed poorly for yield components due to their slow development rate. Grain number, 1000-grain weight, tiller number, plant biomass and plant yield of the Rht12 dwarf lines were all lower than for the tall lines. GA 3 -treated Rht12 dwarf lines had a faster development rate and flowered earlier than the untreated ones, which contributed to achieving higher plant yield. Grain number, seed size, tiller number, plant biomass and plant yield were all increased in the GA 3 -treated Rht12 dwarf lines compared with the untreated ones. Harvest index of the GA 3 -treated Rht12 dwarf lines was reduced due to the significantly increased plant height and biomass (Table 5).
In all three experiments in the two years, plant biomass and height and 1000-grain weight of the GA 3 -treated Rht12 dwarf lines were all increased compared with the untreated ones. These traits of Rht12 dwarf lines showed a consistent response to exogenous GA 3 application such that they performed like tall genotypes, indicating that the shorter plant stature and smaller seed size caused by Rht12 was likely to be a result of a deficiency of endogenous GAs.
Actually, GA 3 treatment like the dosage here were harmful to the tall genotype (which probably with sufficient GAs). Their endogenous hormone metabolic balance might be disturbed by the excess exogenous GA 3 application. As observed in this study, some variant genotypes were also founded in the GA 3 experiment, such as curling flag leaf, abnormal node shape and ''closed-flowered'' with stamens inside the hard glumes of variants. Some yield components of the GA 3 treated tall lines performed poorly than Table 3. Length, diameter and wall thickness of the first and second internode at the base of the stem of Rht12 dwarf and tall genotypes with and without GA 3 treatment in the three experiments.
Year that of the untreated tall ones. For example, grain number, number of efficient spikes per plant and plant yield all reduced in the GA 3 treated tall lines though this decrease was not significant as that in the GA 3 treated Rht12 dwarf lines, whereas seed size was not changed significantly in the GA 3 treated tall lines compared with the untreated tall ones (Table 5). GA 3 application resulted in lower yield for tall lines. This was associated with fewer grains per spike and fewer fertile spikes per plant. Plant biomass and HI were unchanged by application of GA 3 to tall lines.

Effects of Exogenous GA 3 on the Two Parents Karcagi-12 and Ningchun45
Exogenous GA 3 had similar effects on the dwarf parent Karcagi-12 as on the dwarf F 2:3 and F 3:4 genotypes described above. Plant height, plant biomass and seed size were significantly increased by GA 3 application while harvest index was decreased. In particular, plant height was increased by 70% compared with the untreated plants ( Table 6). Plants of GA 3 -treated Karcagi-12 reached double ridge stage earlier than untreated plants, by 25 days (90uCd), 14 days (222uCd) and 32 days (200uCd) in the three experiments, respectively. However, exogenous GA 3 had smaller effects on the tall parent Ningchun45 compared with Karcagi-12 (Table 6). This difference between the two parents suggested that Karcagi-12 may need additional GAs to promote its development and to obtain a 'normal' phenotype and that defects probably exist in GA metabolic pathways in Karcagi-12.

Discussion
This study is part of a series of experiments carried out to obtain a better understanding of the GA-responsive dwarfing gene Rht12.
In the previous studies, the effects of Rht12 on plant development and agronomic traits were analyzed comprehensively [32][33][34][35]. However, a clear understanding of the response of Rht12 to exogenous GA 3 is lacking. Moreover, the role, if any, of Rht12 in GA biosynthesis or signalling or both remains unknown. Here, contrasting homozygous lines with or without Rht12 and Vrn-B1 genes (RRBB, RRbb, rrBB and rrbb) were selected in a F 2 segregating population and assessed as F 2:3 and F 3:4 lines to evaluate the response of Rht12 to exogenous GA 3 .
Previous studies showed that Rht12 had no negative effects on seedling vigour [34,43]. Here we observed that there was no significant difference between dwarf and tall lines in area of the first seedling leaf or coleoptile length, in response to exogenous GA 3 . A similar result was reported by Ellis et al. (2004) who found that the maximum first leaf elongation rate (LER max ) of Rht12 dwarf lines and their response to GA was not significantly different to the corresponding tall lines though the dwarf lines had a small but significant increase in LER max (,23%) over the tall lines (,37%) in the presence of GA. However, there was no notable difference on seedling vigour between the dwarf and the tall lines in either the presence or absence of GA. Thus, Rht12 was classified as a late-acting dwarfing gene [25]. Additionally, the GAinsensitive dwarfing genes, such as Rht-B1b and Rht-D1b, which are characterized by a lack of GA response and verified to be involved in GA signaling, cause a significant reduction in seedling vigour [13,19,25]. Moreover, another class of dwarfing genes (which includes Rht16 and Rht18) had reduced seedling vigour compared with the corresponding tall lines while their relative GA response was significantly greater (50,80%) than the tall lines [25]. It was predicted that these Rht mutants were deficient in GA biosynthesis because the reduction in leaf elongation rate could (at least partially) be reversed by the application of GA [25]. Quantifying GA and its precursors in these mutants would test this hypothesis and could pinpoint the biochemical block leading to the reduction in GA. In fact, GA-responsive dwarf mutants identified in many species often result from mutations in genes encoding GA biosynthetic enzymes [1]. In particular, the major semi-dwarfing gene in rice (sd-1) has been shown to result from a deficiency in a late step of GA biosynthesis [12].
Even though Rht12 had no effect on coleoptile length, the area of seedling leaf 1 or responsiveness to exogenous GA at the very early seedling stage, it was found that Rht12 had a great effect on spike development, tillering, leaf and stem length (especially flag leaf length and peduncle length), harvest index and many other agronomic traits [25,32]. In this study, the application of Table 5. Yield components and harvest index of Rht12 dwarf and tall genotypes with and without GA 3 treatment in the three experiments.
Year and Expt Genotype Rht12 has been found to have significantly reduced grain size compared with other wheat dwarf genes [44,45]. The reason for this is still not clear [32]. It may simply be due to the delayed flowering of Rht12 lines into a more stressful period. Alternatively, the possible endogenous GA deficiency of the Rht12 lines might be the main reason for producing smaller seeds. In this study, exogenous GA 3 significantly increased grain size, although this may have been because seed-set was decreased by GA 3 treatment. Wang et al (2001) found that the seed-set of the third floret was significantly reduced by GA 3 application though it did not affect seed-set of the first and second florets [46]. This may not have been a direct effect of GA 3 application. Poor seed set at the third floret with GA 3 application may have been due to increased competition for assimilates from the strongly elongating stem. Lower grain number of the GA-treated lines might also contribute to the larger grain size [47], although there are many GA effects that could enhance grain size. For instance: GA application elongated both the lemma and palea of the florets especially in the third and fourth florets in the spikelet [46] and may increase cell length in the pericarp [48]; the increased size of vegetative organs may provide more assimilates available for producing large grains; induced early flowering may extend the period of time available with favorable conditions prior to harvest for grain development [33]. However, the reduction in grain size of different dwarf genes may be through different modes of action from each other [18,34,45] and further studies are needed to determine the possible relation between Rht genes and grain size as well as GA and grain size.
Even though wheat dwarf genes have been studied for many years, only the GA-insensitive dwarf genes Rht-B1b and Rht-D1b have been cloned [27]. However, there is still limited information on GA-responsive dwarf genes. Rht12 is located on chromosome 5AL, linked to Xgwm291 by 5.4 cM. High-resolution mapping should be initiated for eventual map-based cloning of Rht12. Alternatively, due to the possibility that Rht12 is involved in GA metabolic pathways, it would be instructive to use the quantitative RT-PCR strategy to investigate the expression patterns of the known GA metabolism genes (such as TaKAO, Ta20ox, Ta13ox, Ta3ox and Ta2ox) to look for potential candidate genes for Rht12. If the expression of a gene is significantly decreased or even nonexpressed in the dwarf plant compared with that of the corresponding tall lines (transcript abundance probably correlated with plant height), this gene could be a candidate gene for Table 6. Duration of pre-anthesis developmental phases and yield components in the two parents with and without GA 3 application.
Year and Expt  sequence analysis. After alignment of sequences of the dwarf plant and the wild type or tall plants, the mutation site may be found in the gene. Indeed, there has been more information collected on genes involved in GA signal transduction than in GA biosynthesis related to wheat Rht genes [13,27,49,50], though there have been many successful examples of GA biosynthesis variations correlated with dwarf phenotypes in Arabidopsis and rice [7,11,12,51]. Thus, more information of the genes involved in GA biosynthesis in wheat is needed for a better understand of how these genes act on plant growth and result in dwarfism. On the other hand, the manipulation of GA biosynthesis or perception may be a good target for regulating crop height [49,52]. However, it is clear that analysis of the endogenous GAs in Rht12 mutant is a priority.
In conclusion, the Rht12 dwarf lines grown in this study showed compact and dark green leaves, dwarfism and a late-flowering phenotype compared with the tall lines. All of these are typical features of GA-deficient mutants [8,52,53]. Moreover, these effects could be rescued by exogenous GA 3 treatment, suggesting that this possible defect/mutation was not involved in GA signal transduction like the GA-insensitive dwarfs, but probably in the GA biosynthesis pathway [15,25,53,54]. This study also confirms that reduced plant height could be dissociated from any effects on the early stage of growth [25]. Rht12 caused a strong form of dwarfism and yet had no effect on coleoptiles length or seedling leaf growth. This type of late-acting dwarf gene in wheat therefore offers the opportunity to reduce plant stature without compromising on early growth and crop establishment [25,55]. Since Rht12 lines posses this feature, it is probable that they do not have a deficiency in GA at the early seedling stage. Why this changes later in development is an interesting question.