Ectopic Expression of Pumpkin Gibberellin Oxidases Alters Gibberellin Biosynthesis and Development of Transgenic Arabidopsis Plants

Abstract

Immature pumpkin (Cucurbita maxima) seeds contain gibberellin (GA) oxidases with unique catalytic properties resulting in GAs of unknown function for plant growth and development. Overexpression of pumpkin GA 7-oxidase (CmGA7ox) in Arabidopsis (Arabidopsis thaliana) resulted in seedlings with elongated roots, taller plants that flower earlier with only a little increase in bioactive GA₄ levels compared to control plants. In the same way, overexpression of the pumpkin GA 3-oxidase1 (CmGA3ox1) resulted in a GA overdose phenotype with increased levels of endogenous GA₄. This indicates that, in Arabidopsis, 7-oxidation and 3-oxidation are rate-limiting steps in GA plant hormone biosynthesis that control plant development. With an opposite effect, overexpression of pumpkin seed-specific GA 20-oxidase1 (CmGA20ox1) in Arabidopsis resulted in dwarfed plants that flower late with reduced levels of GA₄ and increased levels of physiological inactive GA₁₇ and GA₂₅ and unexpected GA₃₄ levels. Severe dwarfed plants were obtained by overexpression of the pumpkin GA 2-oxidase1 (CmGA2ox1) in Arabidopsis. This dramatic change in phenotype was accompanied by a considerable decrease in the levels of bioactive GA₄ and an increase in the corresponding inactivation product GA₃₄ in comparison to control plants. In this study, we demonstrate the potential of four pumpkin GA oxidase-encoding genes to modulate the GA plant hormone pool and alter plant stature and development.

The gibberellin (GA) plant hormones are known for playing an important role in many aspects of plant growth and development, including germination, stem growth, flowering, and fruit development (Hedden and Proebsting, 1999; Richards et al., 2001; Olszewski et al., 2002). The GA biosynthetic pathway has been characterized and the genes encoding most of the GA biosynthetic enzymes have been cloned in Arabidopsis (Arabidopsis thaliana) and other species (Lange, 1998; Hedden and Phillips, 2000a). GA biosynthesis in plants can be divided into three major parts according to the type of enzymes involved and their subcellular localization (Lange, 1998; Hedden and Phillips, 2000a). The first part takes place in plastids and results in the formation of ent-kaurene from the precursor geranylgeranyl diphosphate by the action of diterpene cyclases. This part of the pathway is common to all plant systems that have been studied. The second part of the pathway takes place at the endoplasmic reticulum and cytochrome-P450-dependent monooxygenases are involved in the conversion of ent-kaurene to GA₁₂-aldehyde and GA₁₂. The final part of GA biosynthesis involves soluble 2-oxoglutarate-dependent dioxygenases and leads to the production of the GA plant hormone and inactive GAs. GA dioxygenases are often multifunctional with broad substrate specificity, resulting in many side reactions and numerous GAs (Hedden and Kamiya, 1997; Lange, 1998; Hedden and Phillips, 2000a).

In Arabidopsis, two pathways are described diverging from GA₁₂ to GA plant hormones: a non-13-hydroxylation pathway leading to GA₄ and a 13-hydroxylation pathway leading to GA₁ (Fig. 1). These steps are catalyzed by GA 20-oxidase and GA 3-oxidase enzymes, each encoded by small multigene families (Sponsel and Hedden, 2004). Eight GA-2 oxidases have been identified in Arabidopsis with specificity for either C₁₉- (Thomas et al., 1999) or C₂₀-GAs (Schomburg et al., 2003).

Figure 1.

The third part of the GA biosynthetic pathway leading to the formation of bioactive GAs in Arabidopsis. Biosynthetic steps resulting from the overexpression of pumpkin GA oxidases. a, 7-Oxidase (CmGA7ox). b, Seed-specific 20-oxidase1 (CmGA20ox1). c, Seed-specific 3-oxidase1 (CmGA3ox1). d, 2-Oxidase1 (CmGA2ox1). Main pathways are indicated by thick arrows. The boxed GAs indicate bioactive GAs. Metabolic relationships are discussed in the text.

Developing pumpkin (Cucurbita maxima) seeds express a set of GA oxidases with unique catalytic properties that, to our knowledge, have not been identified in other plant species to date and that synthesize GAs of unknown function in plant development (Fig. 1; Lange, 1998). A multifunctional GA 7-oxidase from pumpkin (CmGA7ox) oxidizes GA₁₂-aldehyde to GA₁₂ and, less efficiently, GA₁₂ to GA₁₄ (Fig. 1; Lange, 1997; Frisse et al., 2003). The seed-specific GA 20-oxidase1 from pumpkin (CmGA20ox1) is, like other GA 20-oxidases, multifunctional with broad substrate specificity but produces mainly C₂₀-GAs (e.g. GA₂₅) instead of C₁₉-GAs (e.g. GA₉; Fig. 1; Lange, 1994, 1998; Lange et al., 1994; Frisse et al., 2003). Unlike most plant species that have been investigated so far in pumpkin, GA 3-oxidase1 (CmGA3ox1, formerly named 2β,3β-hydroxylase; Lange et al., 1997) is a bifunctional enzyme. In addition to its 3-oxidation catalytic properties that can lead to the formation of GA plant hormones (e.g. GA₄; Fig. 1), it also exhibits a 2-oxidation catalytic function (Lange et al., 1997). Moreover, 3-oxidases from other plant species act mainly on C₁₉-GAs, but CmGA3ox1 prefers C₂₀-GAs as the substrate (Fig. 1; Lange et al., 1997; Hedden, 1999). Finally, the pumpkin GA 2-oxidase1 (CmGA2ox1) shares very high sequence identity with an unidentified dioxygenase from Marah macrocarpus and the recombinant protein uses C₁₉-GAs as the substrate (Fig. 1; MacMillan et al., 1997; Frisse et al., 2003).

The multiple roles of GAs and the large number of enzymes and genes involved in the biosynthetic pathway suggest that regulation of GA levels in planta is likely to be rather complex (Hedden and Phillips, 2000a). Transgenic plants overexpressing genes of the GA biosynthetic pathway have been produced to investigate their effects on GA biosynthesis, GA homeostasis, and plant morphology. Furthermore, this approach can be of benefit for controlling plant stature in agriculture and horticulture (Phillips, 2004).

Up-regulation of early steps of the pathway has been achieved by overexpressing AtCPS and AtKS in Arabidopsis and resulted in accumulation of early intermediates of the biosynthetic pathway, but caused no changes in plant morphology and levels of active GAs, showing the ability of plants to maintain GA homeostasis (Fleet et al., 2003). However, GA overdose morphologies were obtained by overexpression of GA 20-oxidases in Arabidopsis (Huang et al., 1998; Coles et al., 1999) and potato (Solanum tuberosum; Carrera et al., 2000). Similarly, overexpression of Arabidopsis GA 20-oxidase in hybrid aspen (Populus tremula × Populus tremuloides; Eriksson et al., 2000) and overexpression of a GA 20-oxidase from citrus or Arabidopsis in tobacco (Nicotiana tabacum) plants (Vidal et al., 2001; Biemelt et al., 2004) resulted in elongated phenotypes associated with GA overproduction. In hybrid aspen and Arabidopsis, overexpression of an Arabidopsis GA 3-oxidase resulted in no major changes in morphology (Israelsson et al., 2004; Phillips, 2004) and in the case of hybrid aspen, the authors suggest that 20-oxidation is the limiting biosynthetic step for GA-controlled shoot elongation.

The green revolution that originated an increased yield in cereal crop cultivars resulted from the introduction of dwarfed varieties (Peng et al., 1999; Monna et al., 2002; Sasaki et al., 2002; Spielmeyer et al., 2002). Ectopic overexpression of the seed-specific GA 20-oxidase1 from pumpkin (CmGA20ox1), which produces mainly inactive GA products, might result in a reduction of bioactive GAs by diverting the pathway to the tricarboxylic acids (Fig. 1) and therefore originate dwarfed phenotypes. This has been achieved successfully in lettuce (Lactuca sativa) plants, where CmGA20ox1 was introduced under a very strong promoter cassette and dwarfed lettuce plants were obtained (Niki et al., 2001). However, in Arabidopsis and Solanum dulcamara, this strategy to reduce GA content and produce dwarfed plants was not achieved, weakening the usefulness of this approach to alter GA levels and reduce plant height in other plant species (Xu et al., 1999; Curtis et al., 2000). The overexpression of CmGA20ox1 resulted in semidwarfed plants in S. dulcamara and slight reduction in plant height in Arabidopsis, suggesting that a feedback control of the endogenous GA 20-oxidases (S. dulcamara and Arabidopsis) and GA 3-oxidases (Arabidopsis) compensates for the effect of the CmGA20ox1 transgene (Xu et al., 1999; Curtis et al., 2000). GA 2-oxidases are catabolic enzymes and can potentially be used to decrease GA levels and create dwarfed phenotypes. Overexpression of GA 2-oxidase genes in Arabidopsis, tobacco, rice (Oryza sativa), and poplar (Populus spp.) resulted in the expected dwarfed phenotypes (Busov et al., 2003; Sakamoto et al., 2003; Schomburg et al., 2003; Biemelt et al., 2004), but the developmental role of GA 2-oxidases in plants is not well understood.

To our knowledge, no attempt had been made to overexpress the multifunctional CmGA7ox from pumpkin and investigate its potential regulatory function in controlling the levels of bioactive GAs. Here we discuss the feasibility of increasing bioactive GAs and altering plant morphology, changing the flux through the pathway by overexpressing CmGA7ox in Arabidopsis. Moreover, we also show that overexpression of the bifunctional CmGA3ox1 in Arabidopsis results in increased plant height and increased GA₄ levels despite the enzyme's preferences for oxidizing C₂₀-GAs instead of C₁₉-GAs.

In this work, we obtained dwarfed Arabidopsis plants and diverted GA precursors into inactive forms by overexpressing the CmGA20ox1 gene using a very strong promoter cassette (Niki et al., 2001) and confirmed the importance of this approach in controlling plant stature. In Arabidopsis, GA 2-oxidase is encoded by a gene family (Thomas et al., 1999; Hedden and Phillips, 2000a; Schomburg et al., 2003), but, up to now, only one active GA 2-oxidase gene has been cloned from pumpkin (CmGA2ox1; Frisse et al., 2003). GA 2-oxidases have been overexpressed in Arabidopsis (Hedden and Phillips, 2000b; Singh et al., 2002; Schomburg et al., 2003). To understand the role of CmGA2ox1 in plant development, we overexpressed the corresponding cDNA in Arabidopsis, resulting in extreme dwarfed phenotypes and reduction of active GA levels.

RESULTS

Generation of Transgenic Arabidopsis Plants Expressing Pumpkin GA Oxidases

Arabidopsis plants were transformed with constructs containing sense (S) copies of CmGA20ox1 and S or antisense (AS) copies of CmGA7ox, CmGA3ox1, or CmGA2ox1. Transgenic plants were selected in the presence of kanamycin and the integration of the pumpkin GA-oxidase S or AS copies of 10 lines were analyzed by PCR (data not shown). The AS lines were used together with the wild-type plants as controls. From the 10 lines, three homozygous (T4) lines showing altered phenotypes were chosen and expression of pumpkin GA oxidases was estimated by quantitative reverse transcription (RT)-PCR (Table I).

Table I.

Levels of pumpkin GA oxidase mRNAs in Arabidopsis overexpression lines as determined by quantitative RT-PCR

n.d., Not detectable.

	Transcript Levels (μg/g Total RNA)
	CmGA7ox			CmGA20ox1			CmGA3ox1			CmGA2ox1
Lines:	S13.1	S8.9	S12.8	S10.8	S2.2	S17.2	S1.3	S19.4	S17.7	S9.7	S5.5	S12.9
	60	80	100	n.d.	n.d.	10	20	100	1,000	20	90	125

Transgenic lines expressing CmGA7ox (S13.1, S8.9, and S12.8) and CmGA3ox1 (S1.3, S19.4, and S17.7) were slender, while transgenic lines expressing CmGA20ox1 (S17.2) and CmGA2ox1 (S9.7, S5.5, and S12.9) were dwarfed. Lines S10.8 and S2.2 expressed semidwarfed phenotypes in spite of the fact that, in both lines, no CmGA20ox1 transcripts were detectable (Table I; data not shown). Transcripts of the four pumpkin GA oxidase-encoding genes were found in neither wild-type Arabidopsis plants nor AS lines (data not shown). Lines with the highest expression levels of each pumpkin GA oxidase gene always showed the strongest phenotype and were chosen for further investigation (Fig. 2; Table II).

Figure 2.

Overexpression of pumpkin GA oxidases in Arabidopsis: effects on plant development. A, Phenotypes of 14-d-old seedlings grown in Murashige and Skoog medium. Wild-type seedlings (left) compared to seedlings expressing S copies of CmGA7ox (line S12.8, middle) or expressing S copies of CmGA3ox1 (line S17.7, right). Bar = 1 cm. B, Phenotypes of 7-week-old plants transferred to soil after 28 d in Murashige and Skoog medium. Nontransformed plants (wild type, left) compared to AS and S lines transformed with CmGA7ox (middle) or CmGA3ox1 (right). C, Phenotypes of 7-week-old plants transferred to soil after 28 d in Murashige and Skoog medium containing 10⁻⁶ m GA₃. Nontransformed plants (wild type, left) compared to different AS and S lines transformed with CmGA20ox1 (middle) or CmGA2ox1 (right).

Table II.

Phenotypic characteristics of wild-type and transgenic Arabidopsis plants expressing pumpkin GA oxidases

Results are shown as mean ± se (n = 5).

	Wild Type	CmGA7ox		CmGA3ox1		Wild Typea	CmGA20ox1a	CmGA2ox1a
		AS15.9	S12.8	AS5.9	S17.7		S17.2	AS7.7	S12.9
Final heightb (cm)	35.0 ± 2.4	36.7 ± 1.8	50.8 ± 0.6	30.7 ± 1.0	52.6 ± 2.0	38.9 ± 2.4	11 ± 0.8	39.3 ± 2.2	4.84 ± 0.4
Internode lengthbc (cm)	1.38 ± 0.1	1.70 ± 0.3	2.42 ± 0.1	1.30 ± 0.1	2.44 ± 0.2	1.40 ± 0.1	0.90 ± 0.1	1.90 ± 0.1	0.68 ± 0.1
Siliquesbd (no.)	22.0 ± 1.8	27.2 ± 0.7	43.8 ± 0.7	22.4 ± 1.0	54.2 ± 1.5	36.4 ± 0.5	10.4 ± 2.6	40.2 ± 1.0	6.8 ± 1.8
Flowering timee (d)	42.6 ± 1.0	41.4 ± 0.5	34.2 ± 1.5	42.6 ± 0.5	30 ± 1.6	41.2 ± 0.4	50.4 ± 1.3	42.0 ± 1.4	57.0 ± 0.9

^aPlants were transferred to soil after 28 d in Murashige and Skoog medium containing 10⁻⁶ m GA₃.

^bNine-week-old plants.

^cThe first internode of the main inflorescence.

^dNumber per plant.

^eWhen the first flower appears.

Arabidopsis Plants Overexpressing CmGA7ox or CmGA3ox1 Show Accelerated Development

Arabidopsis 14-d-old seedlings overexpressing CmGA7ox or CmGA3ox1 showed altered root shapes when compared to wild-type seedlings (Fig. 2A). The seedlings of CmGA7ox overexpressors showed one thin, long primary root with very few lateral roots, while the seedlings of CmGA3ox1 overexpressors developed many thick lateral roots. Seedlings overexpressing CmGA3ox1 also showed enlarged leaves and an increased number of trichomes when compared to seedlings overexpressing CmGA7ox or wild-type seedlings (Fig. 2A). Arabidopsis plants overexpressing CmGA7ox or CmGA3ox1 showed slender phenotypes when compared to plants transformed with AS copies of the respective genes or to wild-type plants (Fig. 2B; Table II). The strongest CmGA7ox overexpressing line, according to the RT-PCR results (S12.8; Table I), was chosen for phenotypic characterization. Line S12.8 had an increase of about 50% in final height, with a similar increase in internode length, developed twice as many siliques, and flowered earlier when compared to wild-type plants or to the AS line AS15.9 (Fig. 2B; Table II). Line S17.7 showed the strongest overexpression of CmGA3ox1 by RT-PCR (Table I) and developed longer shoots, longer internodes, and flowered much earlier than wild-type plants or plants transformed with AS copies of the respective gene (line AS5.9). However, compared to line S12.8 (CmGA7ox overexpressor), in line S17.7 branching was more frequent and the total number of siliques increased (Fig. 2B; Table II).

Overexpression of CmGA20ox1 or CmGA2ox1 in Arabidopsis Results in Retarded Development

To get T4 homozygous Arabidopsis plants expressing CmGA20ox1 or CmGA2ox1, selection in the presence of kanamycin and correct segregations ratios could be obtained only in the presence of GA₃. Therefore, to compare phenotypes, seeds of CmGA20ox1 and CmGA2ox1 expressing lines, as well as seeds of the CmGA2ox1 AS line and a second set of wild-type plants, were all germinated in Murashige and Skoog medium containing 10⁻⁶ m GA₃. Arabidopsis plants expressing CmGA20ox1 or CmGA2ox1 were dwarfed with slightly darker green leaves compared to wild-type plants or plants transformed with AS copies of CmGA2ox1 (Fig. 2C). Line S17.2 showed the strongest expression of CmGA20ox1 and was therefore subjected to more extensive phenotype characterization. S17.2 plants showed reduced development, flowered later, and were much shorter, reaching only 28% of the final height of similarly grown wild-type plants (Fig. 2C; Table II). The number of siliques of the dwarfed plants was reduced to only 29%, compared to wild-type plants grown under the same conditions (Table II). Expression of CmGA2ox1 in Arabidopsis resulted in severely dwarfed phenotypes (Fig. 2C). Line S12.9 expressed the strongest transcript levels as determined by RT-PCR (Table I), and it reached only 12% of the final height of its respective AS line (AS7.7) grown under similar conditions (Table II). These plants showed a reduced development and flowered very late (Fig. 2C; Table II). The dwarfed plants had an extremely low number of siliques, only about 17% of their respective AS lines (Table II).

Analysis of Endogenous GA Levels in Arabidopsis Plants Overexpressing Pumpkin GA Oxidases

Endogenous GA levels were determined by combined gas chromatography-mass spectrometry-selected ion monitoring in the transgenic Arabidopsis lines expressing the highest levels of pumpkin GA oxidases to identify which steps of the GA biosynthetic pathway were affected in these plants (Table III). Similarly, endogenous GA levels were also determined in control Arabidopsis plants (plants expressing AS copies of pumpkin GA oxidases and wild-type plants). Arabidopsis slender plants expressing CmGA7ox had a 3- to 4-fold increase in GA₁₂ levels in comparison to control plants, but only a very slight increase in biologically active GA₄ and catabolic GA₃₄ levels. GA levels of the early 3-hydroxylated pathway (GA₁₄, GA₃₆, and GA₃₇) and of precursors of the 13-hydroxylated pathway (GA₅₃ and GA₄₄) decreased. No changes of levels for the other GAs were observed. In contrast, Arabidopsis slender plants expressing CmGA3ox1 showed increased GA₄ levels as well as increased levels of the corresponding inactivation product GA₃₄ and no changes in other GA levels when compared to their respective control plants (Fig. 1; Table III). The Arabidopsis dwarfed plants expressing CmGA20ox1 showed reduced levels of the bioactive GA₄, increased levels of the respective inactivation product GA₃₄, and increased levels of the tricarboxylic C₂₀-GAs, GA₂₅, and, particularly, GA₁₇, when compared to wild-type plants (Table III). The dwarfed plants also showed a slight decrease in GA₁₂-aldehyde and GA₁₂ and an increase in GA₃₆ (Table III). The level of bioactive GA₄ was considerably reduced in the severe dwarfed Arabidopsis plants expressing CmGA2ox1 accompanied by an increase in the respective inactivation product GA₃₄, when compared to their respective control wild-type plants or AS lines. The dwarfed plants showed also a slight decrease in GA₁₂ and GA₃₆ levels (Table III).

Table III.

GA-levels (ng/plant) in 7-week-old wild-type and transgenic Arabidopsis plants expressing S or AS copies of the CmGA7ox, CmGA20ox1, CmGA3ox1, or CmGA2ox1 gene

Measurements have been repeated at least once with similar results.

	Wild Type	Wild Typea	CmGA7ox		CmGA20ox1a	CmGA3ox1		CmGA2ox1a
			AS15.9	S12.8	S17.2	AS5.9	S17.7	AS7.7	S12.9
GA12-aldehyde	0.37	0.31	0.52	0.64	0.17	0.38	0.24	0.20	0.16
GA12	2.24	1.64	3.24	9.06	1.27	2.14	2.64	1.36	1.09
GA15	0.52	0.30	0.51	0.39	0.39	0.46	0.39	0.32	0.42
GA24	1.34	1.06	2.12	1.45	0.96	1.55	1.65	1.03	0.91
GA9	0.14	0.15	0.11	0.18	0.13	0.18	0.19	0.11	0.08
GA25	0.11	0.08	0.09	0.15	0.18	0.11	0.13	0.08	0.09
GA4	0.32	0.27	0.32	0.35	0.18	0.29	0.54	0.22	0.09
GA34	0.54	0.37	0.41	0.57	1.24	0.50	0.61	0.34	0.45
GA53	0.34	0.17	0.44	0.24	0.18	0.34	0.33	0.17	0.15
GA44	0.09	0.04	0.06	0.05	0.10	0.06	0.05	0.05	0.07
GA20	n.d.b	0.01	0.01	0.01	n.d.b	0.01	0.01	0.01	n.d.b
GA17	0.09	n.d.b	n.d.b	0.04	0.51	0.01	0.03	n.d.b	0.04
GA1	0.01	0.01	0.01	0.01	0.01	0.01	0.01	0.01	n.d.b
GA8	−c	−c	−c	−c	0.01	0.01	−c	0.01	0.01
GA14	0.01	n.d.b	0.07	n.d.b	n.d.b	n.d.b	n.d.b	n.d.b	n.d.b
GA36	0.92	0.48	1.32	0.82	0.87	1.13	0.91	0.55	0.31
GA37	0.03	n.d.b	0.01	n.d.b	0.01	0.01	0.01	n.d.b	n.d.b

^aPlants have been transferred to soil after 28 d in Murashige and Skoog medium containing 10⁻⁶ m GA₃.

^bNo dilution of internal standard.

^cNot determined.

DISCUSSION

Genes of the GA biosynthetic pathway have been overexpressed in plants to investigate their effects on GA biosynthesis, GA homeostasis, and plant morphology (Phillips, 2004). Developing pumpkin seeds express GA dioxygenases with unique catalytic properties resulting in GAs of unknown function for plant development. To investigate their potential role for modulation of GA biosynthesis, pushing the flux through the pathway, and possibly leading to a change in bioactive GAs, we overexpressed pumpkin GA oxidases in Arabidopsis.

Overexpression of ent-copalyl diphosphate synthase and ent-kaurene synthase alone or in combination in Arabidopsis resulted in high accumulation of ent-kaurene and ent-kaurenoic acid (1,000-fold more), GA₁₂ (about 10-fold), and GA₂₄ levels (about 4-fold) compared to wild-type plants (Fleet et al., 2003). Surprisingly, no changes in bioactive GA levels or plant morphology were observed. The authors propose that P450 ent-kaurenoic acid oxidase, producing GA₁₂, may be limiting for production of middle and later GA intermediates. In contrast, Arabidopsis plants expressing CmGA7ox (although in a different ecotype) show accelerated development, have longer stems and internodes, increase in number of siliques, and flower earlier than wild-type Arabidopsis plants (Fig. 2, B and C; Table II). All these phenotypic changes took place despite a 3- to 4-fold increase of GA₁₂ levels, with no considerable changes in GA₄ levels, the primary active GA in Arabidopsis (Talon et al., 1990; Cowling et al., 1998; Table III). This apparent discrepancy might be due to variations in GA levels between various types of tissues and during different stages of plant development. For instance, overexpression of GA 20-oxidase genes in Arabidopsis have been reported to result in GA overproduction phenotypes with no consistent differences in GA₄ and GA₁ levels between shoot tips of the transgenic lines and wild-type plants, while a 2- to 3-fold increase in GA₄ levels was observed when the rosette leaves of the transgenic lines were analyzed (Coles et al., 1999). In the case of 14-d-old Arabidopsis CmGA7ox overexpressors (using intact seedlings for the GA analysis), again GA₄ levels did not increase considerably compared to wild-type seedlings (data not shown). Therefore, in Arabidopsis, GA levels might be tightly regulated with only small changes in the GA plant hormone pool sufficient for modulating plant growth and development. No increase of early 3-hydroxylated GA levels (GA₁₄, GA₃₇, and GA₃₆) was observed, indicating that the 3-hydroxylation side activity of CmGA7ox that has been identified with recombinant CmGA7ox (Lange, 1997; Frisse et al., 2003) had no major impact in GA biosynthesis of the transgenic Arabidopsis line. However, the existence of yet unidentified GA biosynthetic pathways in Arabidopsis that play a role in plant development cannot be excluded.

In hybrid aspen, overexpression of a GA 3-oxidase from Arabidopsis resulted in no major changes in morphology and in only small changes of bioactive GA₁ and GA₄ levels (Israelsson et al., 2004). The authors conclude that in hybrid aspen, 20-oxidation rather than 3-oxidation is the limiting step in the formation of GA₁ and GA₄, and that expression of GA 3-oxidases alone does not increase the flux toward bioactive GAs. Moreover, Phillips (2004) reported that in Arabidopsis, overexpression of GA 3-oxidase does not affect plant development. However, our results demonstrate that ectopic overexpression of the seed-specific CmGA3ox1 in Arabidopsis leads to dramatic changes in plant growth and development (Fig. 2, A and B; Table II). Arabidopsis seedlings overexpressing CmGA3ox1 have elevated hypocotyl and leaf growth and an increased number of trichomes compared to wild-type plants, which are all known, typical GA effects (Perazza et al., 1998; Olszewski et al., 2002). Adult overexpressors of CmGA3ox1 develop similar to plants overexpressing CmGA7ox . They show a slender phenotype and flower earlier compared to wild-type plants or plants expressing AS copies of CmGA3ox1. Moreover, the number of siliques observed per plant was higher in CmGA3ox1 even as compared to CmGA7ox overexpressors, suggesting that flower formation and/or seed set are favored in these lines (Table II). In spite of the fact that CmGA3ox1 prefers C₂₀-GAs as the substrate, the phenotypic changes are accompanied by a 2-fold increase in bioactive GA₄ levels and a slight increase in the corresponding inactivation product GA₃₄ in the CmGA3ox1 overexpressors that would account for the obtained morphological changes (Fig. 1; Table III).

Root and shoot organs react differently to GA levels (e.g. for normal root growth, a much lower concentration of GAs are required than for normal shoot growth; Tanimoto, 1990). Root morphology of transgenic Arabidopsis seedlings overexpressing pumpkin CmGA7ox and CmGA3ox1 has changed dramatically (Fig. 2A). CmGA7ox overexpressors develop much longer primary roots compared to wild-type plants, indicating that a little increase in endogenous GA levels of these plants favors root elongation. However, CmGA3ox1 transgenic lines develop more lateral roots that are thicker compared to wild-type plants (Fig. 2A), indicating that a considerable increase of endogenous GA levels does not help root elongation but helps lateral root formation. Recently, Fu and Harberd (2003) found that GAs regulate root growth by repressing GAI (GA insensitive) and RGA (repressor of ga1-3), two DELLA proteins involved in GA signaling (Fleet and Sun, 2005). However, little is known about how GA signaling components modulate GA biosynthesis (Richards et al., 2001; Olszewski et al., 2002; Sun and Gubler, 2004; Thomas and Sun, 2004; Fleet and Sun, 2005). Moreover, other plant hormones (e.g. auxins) are known to interact with the components of the GA biosynthetic pathway that affects plant development (e.g. the development of lateral roots; Achard et al., 2003; Casimiro et al., 2003; Fu and Harberd, 2003; Fleet and Sun, 2005).

Seed-specific GA 20-oxidase1 from pumpkin (CmGA20ox1) encodes an enzyme with unique catalytic GA 20-oxidation properties: It is the only known GA 20-oxidase that produces mainly tricarboxylic C₂₀-GAs (e.g. GA₂₅ and GA₁₇) that have no known physiological function, rather than C₁₉-GAs (e.g. GA₉ and GA₂₀) that serve as precursors in GA plant hormone synthesis (Fig. 1; Lange, 1994, 1998; Lange et al., 1994; Frisse et al., 2003). Therefore, overexpression of CmGA20ox1 might be a useful strategy for reduction of bioactive GAs in planta (Hedden and Phillips, 2000b). This hypothesis has been tested already by several groups using different plant species. In lettuce, overexpression of CmGA20ox1 resulted in dwarfed plants with reduced levels of GA₁ and GA₄ and increased levels of GA₁₇ and GA₂₅ (Niki et al., 2001). However, in Arabidopsis and S. dulcamara, overexpression of CmGA20ox1 resulted in only a slight reduction of plant height associated with a semidwarfed phenotype (Xu et al., 1999; Curtis et al., 2000). In the case of Arabidopsis, CmGA20ox1 overexpressing plants showed reduced levels of GA₄ and no clear effect on GA₁ levels. In the case of S. dulcamara, GA₄ levels were unaltered in stems and increased in leaves and GA₁ levels were reduced. It was suggested that a feedback type of regulation, resulting in increased transcript levels of the endogenous GA 20-oxidase-encoding genes (Arabidopsis and S. dulcamara) and GA 3-oxidase-encoding gene (Arabidopsis), was responsible for the limited success in reducing plant height. We reinvestigated overexpression of CmGA20ox1 in Arabidopsis plants using a strong promoter cassette similar to the one that drove expression of CmGA20ox1 in lettuce (Niki et al., 2001) and obtained Arabidopsis plants with severe dwarfed phenotypes. As in transgenic lettuce, flowering was delayed and seed number was dramatically reduced in the dwarfed Arabidopsis plants (Table II). However, in contrast to what was reported for lettuce, we found that seed germination was affected and transgenic seeds were not able to germinate in the absence of applied GA₃ (data not shown). The transgenic Arabidopsis accumulated tricarboxylic GA₂₅ and GA₁₇ and had reduced levels of the bioactive GA₄ similar to the findings of Xu et al. (1999). In addition, increased endogenous levels of GA₃₄ indicate a surprising increase of 2-oxidation activity in transgenic plants.

Genetic manipulation of catabolic GA 2-oxidases offers another suitable strategy for modulating plant development (Hedden and Phillips, 2000b). Dwarfed phenotypes with decreased GA levels in planta have been achieved by overexpression of GA 2-oxidases in several plant species, including tobacco, rice, and poplar (Busov et al., 2003; Sakamoto et al., 2003; Biemelt et al., 2004). In Arabidopsis, GA 2-oxidases are encoded by a complex gene family (Thomas et al., 1999; Schomburg et al., 2003; Sponsel and Hedden, 2004). Overexpression of two of them that hydroxylate C₂₀- rather that C₁₉-GA precursors resulted in dwarfed phenotypes in Arabidopsis (Schomburg et al., 2003). Recombinant pumpkin CmGA2ox1 hydroxylates C₁₉-GAs and thus efficiently inactivates GA plant hormones, including GA₁ and GA₄ (Frisse et al., 2003). Expression studies indicate that CmGA2ox1 transcript levels are most abundant in roots of pumpkin plants (Lange et al., 2005). To test the influence of this gene on modulating Arabidopsis plant growth and its GA hormone pool, we used a strong promoter cassette (Niki et al., 2001) to express constitutively CmGA2ox1 in Arabidopsis. The transgenic plants express extremely dwarfed phenotypes and seeds of those lines are unable to germinate in the absence of exogenous applied GA. Compared to control plants, CmGA2ox1 overexpressors show severe reduced stem elongation and flower late, with dramatic decrease in the silique number (Fig. 2C; Table II). Rice plants overexpressing GA 2-oxidase constitutively, by the action of the actin promoter, showed severe dwarfism and failed to set grain. However, ectopic expression of the same gene in shoots under the control of the promoter of the GA biosynthesis gene, OsGA3ox2, resulted in semidwarfed phenotypes that are normal in flowering and grain development (Sakamoto et al., 2003). The dramatic changes observed in the phenotype of CmGA2ox1 overexpressors were accompanied by a severe reduction of the GA₄ levels and, as expected, in an increase of the corresponding inactivation product GA₃₄.

The four pumpkin GA oxidases utilized in this study offer a set of tools for manipulating GA biosynthesis and for regulation of plant development that might gain enormous benefits for designing optimized plant species important for agriculture and horticulture. Our results demonstrate that, by overexpression of the CmGA7ox and CmGA3ox1, it becomes possible to increase GA levels and elevate plant development in Arabidopsis. With an opposite effect, overexpression of CmGA20ox1 and CmGA2ox1 in Arabidopsis decreases GA levels, resulting in severe dwarfed phenotypes. Moreover, our results demonstrate the usefulness of overexpressing CmGA20ox1 under the control of a strong promoter cassette and thus offer an attractive alternative strategy for reducing GA content and modulating plant development. GA 20-oxidation steps have been shown to limit production of bioactive GAs with associated GA phenotypes (Huang et al., 1998; Coles et al., 1999; Carrera et al., 2000; Eriksson et al., 2000; Vidal et al., 2001; Biemelt et al., 2004). Our results indicate that, in addition, GA 7-oxidase and GA 3-oxidase also catalyze rate-limiting steps of the GA biosynthetic pathway in Arabidopsis. It is possible that local modulation of GA levels in the overexpressor lines account for the differences in plant development. As reviewed by Sponsel and Hedden (2004), expression studies on rice genes involved in GA biosynthesis and signaling suggest that bioactive GAs are produced close to or at their site of action. Further studies are necessary to understand the impact of the sites of GA biosynthesis, perception, and signaling as well as cross talk with other plant hormones.

MATERIALS AND METHODS

Plant Material and Growth Conditions

Arabidopsis (Arabidopsis thaliana) ecotype Columbia was used in all experiments. Seeds were sown on soil and stratified at 4°C for 2 to 3 d before transfer to a growth chamber under long-day conditions: 16-h light (approximately 120 μmol m⁻² s⁻¹) and 8-h dark. The temperature was kept at 22°C during the light and 20°C during the dark periods, respectively. For plate growth assays, seeds were sterilized and plated on 0.8% plant agar in 0.5× Murashige and Skoog medium (Duchefa) containing, when appropriate, 50 μg mL⁻¹ kanamycin and 10⁻⁶ m GA₃. The seeds were stratified and grown as above and transferred to soil after 2 to 4 weeks. For RT-PCR analysis, the rosette leaves of 8-week-old plants (for wild-type, CmGA7ox, and CmGA3-ox1 transgenic lines) or the rosette leaves of 9-week-old plants (for CmGA20ox1 and CmGA2ox1 transgenic lines) were collected and frozen immediately in liquid nitrogen. For GA quantification, the aerial part of 7-week-old wild-type and transgenic plants was harvested and frozen immediately in liquid nitrogen.

Plasmid Constructs and Plant Transformation

To enhance the CmGA20ox1 expression, a construct containing a strong promoter cassette and a translational enhancer (E12-35-Ω) was used as described by Mitsuhara et al. (1996) and Niki et al. (2001). The constructs for expression of S or AS copies of CmGA7ox, CmGA3ox1, or CmGA2ox1 were similar to the one used for expressing CmGA20ox1, but the S or AS copies of the different cDNAs were cloned at a unique EcoRI site of a multicloning site of the vector modified from pBE2113, as described by Mitsuhara et al. (1996), replacing CmGA20ox1.

The constructs carrying the S or AS copies of the different pumpkin (Cucurbita maxima) GA oxidases were introduced in Arabidopsis wild-type plants via Agrobacterium tumefaciens-mediated transformation using the floral-dip method (Clough and Bent, 1998). To identify transgenic plants, seeds from the dipped plants were grown on Murashige and Skoog medium plates supplemented with 50 μg mL⁻¹ kanamycin. The seeds of the kanamycin-resistant plants were further analyzed for kanamycin resistance and segregation. T2 seedlings of CmGA20ox1 and CmGA2ox1 S lines showed a segregation ratio of 3:1 only when GA₃ was added to the Murashige and Skoog plates. Therefore, T2 and further generation seeds of S lines of CmGA20-ox1 and S and AS CmGA2ox1 lines were all germinated in the presence of 10⁻⁶ m GA₃. A second set of wild-type plants was generated where the seeds were germinated in the presence of 10⁻⁶ m GA₃ and used as a control in the experiments involving CmGA20ox1 and CmGA2ox1 lines. The presence of every transgene was checked by PCR in 10 of the T2 S lines and five of the T2 AS lines that showed a segregation ratio of 3:1. Specific pumpkin GA oxidase primers, identical to the ones described below for the RT-PCR experiments, and a vector-specific primer 5′-CTACAACTACATCTAGAGG-3′ were used, respectively, as reverse and forward primers in the PCR experiments. After scoring at T3, three homozygous S lines and two homozygous AS lines for every gene were taken to generate T4 homozygous plants used for phenotype, biochemical, and molecular characterization.

Quantitative RT-PCR

Transcript levels of CmGA7ox, CmGA20ox1, and CmGA3ox1 (previously named 2β,3β-hydroxylase) were quantified as described previously by Lange et al. (1997), except that 50 ng of total RNA were reverse transcribed using a RevertAidH Minus first-strand cDNA synthesis kit (MBI Fermentas). For quantification of CmGA2ox1, three specific oligonucleotides were synthesized based on its cDNA sequence: CmGA2ox1 F (5′-CTCTGCAGCATTCTACTCTGGGATTCC-3′), CmGA2ox1 R (5′-GGCCCACCGAAGTAGATCATTGAAACC-3′), and CmGA2ox1 RT (5′-AGATGTTCGAATCC-3′). For preparation of the internal RNA standard, pBluescript SK plasmid containing the CmGA2ox1 cDNA was digested with HindIII that released a 448-bp fragment. The vector was religated and used for RNA synthesis. The annealing temperature used for PCR was 60°C.

Quantification of Endogenous GAs

For quantitative determination of endogenous GAs, frozen plant tissue from the aerial part (2 g fresh weight) was spiked with 17,17-d2-GA standards (2 ng each; from Professor L. Mander, Australian National University, Canberra, Australia) and pulverized under liquid nitrogen. Samples were extracted, purified, derivatized, and analyzed by gas chromatography-mass spectrometry-selected ion monitoring as described elsewhere (Lange et al., 2005).