Gibberellin Mobilizes Distinct DELLA-Dependent Transcriptomes to Regulate Seed Germination and Floral Development in Arabidopsis^,

Abstract

Severe Arabidopsis (Arabidopsis thaliana) gibberellin (GA)-deficient mutant ga1-3 fails to germinate and is impaired in floral organ development. In contrast, the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant confers GA-independent seed germination and floral development. This fact suggests that GA-regulated transcriptomes for seed germination and floral development are DELLA dependent. However, it is currently not known if all GA-regulated genes are GA regulated in a DELLA-dependent fashion and if a similar set of DELLA-regulated genes is mobilized to repress both seed germination and floral development. Here, we compared the global gene expression patterns in the imbibed seeds and unopened flower buds of the ga1-3 mutant with that of the wild type and of the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant. We found that about one-half of total GA-regulated genes are apparently regulated in a DELLA-dependent fashion, suggesting that there might be a DELLA-independent or -partially-dependent component of GA-dependent gene regulation. A cross-comparison based on gene identity revealed that the GA-regulated DELLA-dependent transcriptomes in the imbibed seeds and flower buds are distinct from each other. Detailed ontology analysis showed that, on one hand, DELLAs differentially regulate the expression of different individual members of a gene family to run similar biochemical pathways in seeds and flower. Meanwhile, DELLAs control many functionally different genes to run specific pathways in seeds or flower buds to mark the two different developmental processes. Our data shown here not only confirm many previous reports but also single out some novel aspects of DELLA functions that are instructive to our future research.

Plant development is an orderly process that starts from seed germination to juvenility, maturity, flowering, and fruiting. The whole process is modulated by physical, chemical, and biological components in the environment as well as by several internal factors, including auxins, abscisic acid (ABA), cytokinins, ethylene, and GA. GA is essential for multiple processes of plant development, such as seed germination, stem elongation, and floral development (Richards et al., 2001; Olszewski et al., 2002; Peng and Harberd, 2002; Sun and Gubler, 2004). In Arabidopsis (Arabidopsis thaliana), the severe GA-deficient mutant ga1-3, which contains greatly reduced levels of bioactive GAs, is defective in seed germination, retarded in vegetative growth, and impaired in the development of floral organs (Koornneef and van der Veen, 1980; Wilson et al., 1992; Sun and Kamiya, 1994).

In recent years, significant progress has been made to understand the molecular mechanism of GA action. In brief, the binding of GA to its soluble receptor GIBBERELLIN INSENSITIVE DWARF 1 (OsGID1) or OsGID1-like (Ueguchi-Tanaka et al., 2005; Hartweck and Olszewski, 2006) triggers the degradation of plant growth repressor DELLA proteins (DELLAs) via the 26S proteasome pathway (Silverstone et al., 2001; Fu et al., 2002; Itoh et al., 2002; Hussain et al., 2005). The degradation process is mediated by the GA-specific F-box proteins OsGID2 (Sasaki et al., 2003) and AtSLY1 (McGinnis et al., 2003; Dill et al., 2004; Fu et al., 2004). The degradation of DELLAs will release the plants from the DELLA-mediated growth restraint (Harberd, 2003). DELLAs are named after a highly conserved motif at their N termini that is important for GA sensitivity (Peng et al., 1999; Boss and Thomas, 2002; Chandler et al., 2002), and they form a subfamily of the GRAS family of putative transcription regulators (Pysh et al., 1999; Richards et al., 2000). There are five DELLAs in Arabidopsis: GAI, RGA, RGL1, RGL2, and RGL3 (Dill and Sun, 2001; Lee et al., 2002; Wen and Chang, 2002; Hussain et al., 2005). Genetic studies have revealed that GAI and RGA (Peng et al., 1997; Silverstone et al., 1998) are involved in repressing stem elongation since loss-of-function of both GAI and RGA completely suppressed the dwarf phenotype of ga1-3 mutant (Dill and Sun, 2001; King et al., 2001). During floral development, RGA, RGL2, and RGL1 jointly repress petal and stamen development. Combinations of loss-of-function mutations of RGA, RGL1, and RGL2 suppressed the male sterile phenotype of the ga1-3 mutant (Cheng et al., 2004; Tyler et al., 2004; Yu et al., 2004). On the other hand, RGL2 is the key repressor of seed germination and this function is enhanced by GAI and RGA (Lee et al., 2002; Tyler et al., 2004; Cao et al., 2005).

The fact that, in the absence of exogenous GA, ga1-3 plants lacking the four DELLA proteins GAI, RGA, RGL1, and RGL2 (i.e. ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant line) can germinate, bolt, and produce fully developed fertile flowers as the wild-type control suggests that DELLAs are functional redundant factors and they act as the central signaling molecules in GA-mediated seed germination, stem elongation, and floral development pathways (Cheng et al., 2004; Tyler et al., 2004; Yu et al., 2004; Cao et al., 2005). However, as a group of putative transcription regulators (Pysh et al., 1999; Richards et al., 2000), the molecular mechanism of DELLAs repressing plant growth is largely unknown. For example, it is not known whether DELLAs simply control the expression of a similar set of genes to repress seed germination, stem elongation, and floral development, or whether they mobilize different subsets of genes in the genome to modulate these different processes. Meanwhile, it is of our great interest to know if all GA-regulated genes are GA regulated in a DELLA-dependent fashion.

One way to answer the above questions is to compare the gene expression patterns in the ga1-3 mutant to that in the plants of no DELLA activity in the ga1-3 background. The ga1-3 mutant fails to germinate and is retarded in floral development, suggesting that the transcriptome for germination and floral development in the ga1-3 mutant must be kept at a repressive state (Ogawa et al., 2003). On the other hand, the fact that the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant confers GA-independent seed germination and flowering suggests that, in this mutant line, the transcriptomes responsible for germination and floral development must have been constitutively activated. It is reasonable to speculate that genes normally up-regulated by GA would express at lower levels in ga1-3, and the stabilized high levels of DELLA repressors in ga1-3 would be responsible for a proportion of these lower expressed genes (Lee et al., 2002; Tyler et al., 2004). Therefore, the genes that are genuinely repressed, directly or indirectly, by DELLAs will be restored to wild-type levels or even higher in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant because this mutant line is absent from four DELLA proteins. Vice versa, genes activated by DELLAs might express at higher levels in ga1-3 and be brought back to wild-type levels or even lower in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant. Therefore, comparing the expression profile in ga1-3 with that in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant would help to identify the set of DELLA-dependent transcriptomes essential for seed germination and floral development. Because we also wished to compare the gene expression in ga1-3 showing phenotypic suppression by the quadruple DELLA knockout to gene expression in the wild-type plants, we chose to compare wild type and ga1-3 instead of ga1-3 treated with GA. We first identified GA-regulated (both up- and down-regulated) transcriptomes in both imbibed seeds and young flower buds by comparing the expression patterns between the ga1-3 mutant and the wild-type control. Then, we identified DELLA-dependent (both up- and down-regulated) transcriptomes by finding out the subgroup of GA-regulated genes with their expression restored to wild-type levels in the ga1-3 rga-t2 gai-t6 rgl1-1 rgl2-1 mutant. Data analysis showed that, in both imbibed seeds and young flower buds, approximately one-half of genes down- or up-regulated in ga1-3 were apparently restored to the wild-type level in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant, which suggests that (1) these GA-regulated DELLA-dependent genes are likely responsible for mediating seed germination or floral development and (2) there might be a DELLA-independent or -partially-dependent component of GA-dependent gene regulation despite the fact that the visible growth phenotype is at least substantially DELLA dependent. Surprisingly, regardless of the fact that GA triggers some similar cellular events during seed germination and floral development (e.g. GA induces epidermal cell elongation along both the hypocotyl of a germinating seed and the filament of a growing stamen; Cheng et al., 2004; Cao et al., 2005), the set of genes that are presumably regulated by DELLAs for seed germination overlaps little with and is largely distinct from the set of DELLA-regulated genes involved in floral development. This observation suggests that the GA-mediated seed germination and floral development are under the control of distinct DELLA-dependent transcriptomes.

RESULTS

Identification of DELLA-Dependent Transcriptomes for Seed Germination

Attempts were made to identify the DELLA-dependent transcriptome controlling seed germination by using oligonucleotide-based DNA microarray analysis (Affymetrix gene chip, carrying 23,000 genes). Seeds of the wild type, ga1-3, and ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 were imbibed at 4°C for 96 h under continuous white light. The cold treatment was included because it enhances both the biosynthesis of GA in seeds and the tissue sensitivity to GA so that it promotes and synchronizes seed germination (Ogawa et al., 2003; Yamauchi et al., 2004). Total RNA was separately extracted from these different seed samples and used for microarray analysis to compare their global gene expression profiles, as described in “Materials and Methods.” Three microarray replicates for each of the three genotypes in seeds were performed. To minimize the variation caused by individual hybridization, only genes with a logarithm base 2 of the signal ratio of wild type versus ga1-3 >1 (2-fold higher) or <−1 (2-fold lower) in all three replicates were referred to as GA-up-regulated (GA-up) or GA-down-regulated (GA-down), respectively. Data analysis using the above strict criteria identified a total of 541 genes as GA-up (Supplemental Table S1) and 571 genes as GA-down (Supplemental Table S2) in ga1-3 seeds when compared to the wild-type control.

We then compared the gene expression patterns between ga1-3 and the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant and found that, out of the 541 GA-up genes in ga1-3 seeds, mRNA levels of 360 genes (67%) were at least 2-fold higher in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant than that in ga1-3 (Table I; Supplemental Table S3), while the remaining 181 genes did not show significant changes in their expression (Supplemental Table S4), suggesting that these 360 genes are normally negatively regulated by DELLAs to repress seed germination. These 360 genes are considered to be DELLA down-regulated (DELLA-down) and the 181 genes to be DELLA-independent or -partially-dependent GA-regulated genes. Meanwhile, out of the 571 GA-down genes in ga1-3 seeds, mRNA levels of 251 genes (44%) were 2-fold lower in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant than that in ga1-3 (Table I; Supplemental Table S5), while the remaining 320 genes did not show significant changes in their expression (Supplemental Table S6), suggesting that these 251 genes are normally positively regulated by DELLAs to repress seed germination. These 251 genes are considered to be DELLA up-regulated (DELLA-up) and the 320 genes to be DELLA-independent or -partially-dependent GA-regulated genes. To confirm our microarray data, candidate genes were randomly chosen from the DELLA-down and the -up gene list, respectively, and were subjected to reverse transcription (RT)-PCR analysis using RNA samples independently prepared (Supplemental Table S7). The result showed that all 43 genes from the DELLA-down gene list were expressed at higher levels in seeds of both wild type and the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant than that in the ga1-3 seeds, while expression levels of 31/33 genes from the DELLA-up gene list were lower in both wild type and the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant than that in ga1-3, exhibiting patterns similar to that observed in the microarray analysis (Fig. 1, A and B). The high percentage of confirmation of microarray data by RT-PCR demonstrates that the microarray data we obtained are highly reproducible.

Table I.

Summary of GA- and DELLA-regulated transcriptomes

Criteria used for microarray data analysis are as described in “Materials and Methods.” Details are listed in Supplemental Tables S1 (GA-up in seed), S2 (GA-down in seed), S3 (DELLA-down in seed), S5 (DELLA-up in seed), S8 (GA-up in flower bud), S9 (DELLA-down in flower bud), S11 (GA-down in flower bud), and S12 (DELLA-up in flower).

		No. of Genes		No. of Genes
Seed	GA-up	541	DELLA-down	360
	GA-down	571	DELLA-up	251
Flower bud	GA-up	826	DELLA-down	360
	GA-down	422	DELLA-up	273

Figure 1.

RT-PCR confirmation of DELLA-down and DELLA-up genes in the imbibed seeds. A, DELLA-down genes. B, DELLA-up genes. RT-PCR analysis was repeated on three independent samples and a representative ethidium bromide gel picture is shown here. Corresponding gene locus identity (Gene ID) is provided. Two genes (At1g21680 and At3g22490) in B were confirmed in only one of the three repeats but not in other two repeats and were marked with an asterisk. Primer pairs for each individual gene are listed in Supplemental Table S7. penta: ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 penta mutant. ACT2 (ACTIN 2 gene) and UBQ10 (UBIQUITIN 10 gene) were used as the normalization controls.

Ontology Analysis of DELLA-Dependent Transcriptomes for Seed Germination

The 360 DELLA-down genes and 251 DELLA-up genes were subjected to gene ontology analysis using the tools and information provided by Affymetrix (NetAffx Gene Ontology Mining Tool), respectively. Among the 360 DELLA-down genes, 257 genes have each been assigned a putative molecular function based on amino acid homology, and the other 103 genes are recorded as functionally unassigned putative genes (NetAffx Gene Ontology Mining Tool; Supplemental Table S3). Ontology analysis showed that the largest group of DELLA-down genes belongs to the enzyme genes (total 162 genes, encoding hydrolase, transferase, and oxidoreductase, etc.) responsible for the biosynthesis and metabolism of carbohydrate, protein, nucleotide/nucleic acid, and lipid (Table II), suggesting the importance of mobilization of food reserves during seed germination. The second largest group of DELLA-down genes contains genes encoding proteins with binding activity to nucleic acid, nucleotide, ion, and protein binding (total 96 genes; Table II). Further examination of our dataset revealed that seven xyloglucan endo-1,4-β-d-glucanase genes, five expansin genes, six pectinesterase genes, two endo-1,4-β-glucanase genes, and one 1,4-β-mannan endohydrolase gene (Table III) were identified as DELLA-down genes. These genes encode well-known factors presumably associated with weakening of the tissue surrounding the embryo to facilitate the embryo growth and radicle protrusion (Bewley, 1997; Chen and Bradford, 2000; Chen et al., 2002), suggesting that derepressing DELLA function by GA is crucial for the expression of these important cell wall-modifying factors. Interestingly, three α-tubulin genes (TUA2, TUA4, and TUA6) and four β-tubulin genes (TUB1, TUB5, TUB6, and TUB7) were found as DELLA-down genes (Table III), suggesting that DELLA-mediated reorientation of cytoskeleton might be a key event following cell wall modification during seed germination (Yuan et al., 1994). In addition, ontology analysis showed that seven MYB family genes (e.g. MYB4, MYB25, MYB30, MYB34, MYB66), four bHLH family genes (e.g. SPATULA), and four putative zinc-finger family genes are also found as DELLA-down genes (Table III). SPATULA (At4g36930) has previously been shown to act as a repressor of seed germination, probably through repressing the expression of GA 3-oxidase (GA3ox; Penfield et al., 2005).

Table II.

Ontology analysis of DELLA-regulated genes based on molecular function assigned

In the imbibed seeds, 257/360 DELLA-down (DELLA-D) and 150/251 DELLA-up (DELLA-U) genes were assigned with putative molecular functions based on amino acid homology. In the young flower buds, 243/360 DELLA-D and 180/273 DELLA-U genes were assigned with molecular functions. Details are provided in Supplemental Tables S3, S5, S9, and S12.

Molecular Function			Imbibed Seeds		Unopened Flower Buds
				DELLA-D	DELLA-U	DELLA-D	DELLA-U
Catalytic activity	(Total)		(162)	(85)	(155)	(110)
	Hydrolase	(Subtotal)	(70)	(20)	(60)	(21)
	Acting on:	Glycosyl bonds	25	6	17	7
		Ester bonds	21	5	21	6
		Peptide bonds	13	6	16	3
		Acid anhydride	10	1	2	5
	Transferase		50	20	37	55
	Oxidoreductase		23	30	32	24
	Lyase		7	2	11	4
	Ligase		5	9	5	4
	Others		2	1	3	2
Binging activity	(Total)		(96)	(79)	(89)	(99)
Binding to:	Nucleic acid		35	25	36	30
	Ion		27	32	15	27
	Nucleotide		20	4	11	30
	Protein		19	12	10	18
	Oxygen		7	9	10	10
	Tetrapyrrole		6	8	6	9
	Lipid		3	3	7	2
	Carbohydrate		1	4	3	9
	Others		2	4	5	6
Transcription regulator activity	(Total)		(27)	(16)	(32)	(27)
	MYB		7	0	7	4
	Zinc finger		4	2	1	7
	bHLH		4	2	3	3
	MADS box		0	1	3	0
	WRKY genes		0	0	0	3
	Others		12	11	19	12
Transporter activity			32	12	24	15
Structure molecular			12	0	3	0
Antioxidant activity			4	1	4	0
Nutrient reservoir			2	4	1	1

Table III.

Cross-comparison of genes related to some important biochemical and biological processes in imbibed seeds and unopened young flower buds

Genes listed here are summarized from Supplemental Tables S3, S5, S9, and S12 based on information provided in Gene Title, Molecular Function, and Gene Description by Affymetrix. DELLA-D, DELLA-down genes; DELLA-U, DELLA-up genes.

Molecular Function	Imbibed Seeds		Unopened Flower Buds
		DELLA-D	DELLA-U	DELLA-D	DELLA-U
Cell growth and cell wall loosening
Xyloglucan endotransglycosylase/hydrolase	At2g06850	At5g57550		At4g37800
	At3g23730			At5g57560
	At4g03210
	At4g30280
	At4g30290
	At5g13870
	At5g57560
Pectinesterase	At1g02810
	At1g11580
	At3g10720
	At3g14310
	At4g02330
	At4g33220
Cellulose synthase			At4g18780
			At4g24000
			At5g17420
			At5g44030
Cellulase	At1g64390		At1g13130
	At1g70710		At3g26140
1,4-β-Mannan endohydrolase	At5g66460		At3g10890
Glycoside hydrolase			At3g16920
			At3g42950
Expansin	At1g69530	At2g18660	At1g20190	At2g18660
	At2g37640		At2g37640
	At2g40610		At2g40610
	At5g02260		At3g29030
	At5g05290
Tubulin α-chain	At1g04820
	At1g50010
	At4g14960
Tubulin β-chain	At1g20010
	At1g75780
	At2g29550
	At5g12250
Transcription factors
bHLH family proteins	At1g51070	At3g62090	At1g25330	At4g01460
	At1g63650	At5g46760	At1g59640	At5g46760
	At1g74500		At5g39860	At5g50915
	At4g36930
MYB family proteins	At1g01380		At1g17950	At1g06180
	At1g22640		At2g38090	At3g11280
	At2g39880		At3g01140	At5g44190
	At3g28910		At3g27810	At5g59780
	At5g14750		At3g27812
	At5g58900		At4g34990
	At5g60890		At5g40350
Zinc-finger family proteins	At1g14440	At2g31380	At5g25830	At1g13400
	At1g75710	At2g47890		At1g66140
	At2g24790			At1g68520
	At2g28200			At1g73870
				At2g01940
				At5g25160
MADS box family proteins		At1g77950	At2g45650
			At3g58780
			At4g09960
WRKY family proteins				At2g23320
				At3g56400
				At4g23810
AP2 domain containing protein	At2g40220	At1g53910	At1g15360	At1g25560
		At5g18450	At1g16060
			At5g67180
Squamosa promoter-binding protein			At1g27360	At1g53160
			At1g27370
			At3g15270
			At5g43270
Homeodomain transcription factor	At1g05230	At4g35550	At1g62990
	At3g60390		At2g17950
	At4g32880
	At5g15150
Protein phosphorylation
Protein kinase	At1g49580	At1g11050	At1g61590	At1g16260
	At3g08730	At1g70520	At5g57670	At1g21250
	At3g14370	At1g70530		At1g21270
	At5g28290	At2g35050		At1g29720
	At5g50000	At2g39360		At1g65190
	At5g67080	At2g45910		At1g66880
		At3g22750		At1g66920
		At5g03140		At1g69730
		At5g58350		At2g26980
				At2g32680
				At3g09830
				At3g23110
				At3g45640
				At3g45780
				At4g04540
				At5g25440
				At5g38210
				At5g40540
				At5g60900
Leu-rich repeat proteins	At1g10850	At1g09970	At4g18640	At1g09970
	At1g66150			At1g33560
	At2g25790			At1g35710
	At3g02880			At1g51805
	At3g56370			At1g56120
	At4g36180			At2g31880
	At5g43020			At3g11010
	At5g48940			At4g08850
	At5g51560			At5g48380
Receptor protein kinase	At5g60890			At1g75820
				At4g23130
				At4g23180
S-locus lectin protein kinase				At1g11350
				At2g19130
				At4g11900
				At4g27300
S-receptor kinase				At1g65790
Disease and stress response
Response to disease and pathogens	At1g18250	At2g43590	At1g55020	At1g33560
	At1g73620		At1g72260	At1g72930
	At1g80460		At3g11480	At2g32680
	At3g28910		At3g13650	At2g43570
			At3g16920	At2g43620
			At3g21240	At3g11010
			At4g23690	At3g20590
			At5g24780	At3g23110
				At3g50950
				At4g16990
				At4g19530
				At4g26090
				At5g45250
Water and salt stress	At1g20440	At1g01470	At1g05260	At1g33560
	At1g54410	At1g01470	At1g29395	At2g21620
	At3g08730	At1g72100	At1g52690
	At4g34240	At2g21490	At2g21490
	At4g39090	At2g38905	At5g24780
	At5g25610	At2g41280
		At2g42560
		At3g22490
		At3g22500
		At3g50980
		At3g53040
		At4g36600
		At5g52300
Oxidative stress	At5g64100		At3g59845	At3g45640
	At5g40150		At4g11290
	At5g39580		At4g30170
	At2g22420		At5g24780
			At5g51890
Cold	At3g08730	At2g38905	At1g05260	At5g57560
	At5g12250	At5g52300	At1g29395
	At5g57560
Heat		At3g46230	At5g67180
UV	At3g12610		At3g21240
			At4g13770
Toxin catabolism	At1g78370		At1g17190	At1g02930
	At2g30860
	At3g09270
	At3g43800
Multidrug transport	At1g71870		At3g26590
	At4g23030		At4g22790
			At5g49130
Wounding	At1g22640		At1g55020	At2g38870
			At3g11480
			At3g21240
			At5g24780
DNA damage response	At3g22880		At3g12710
	At4g02060		At5g44680
	At5g44680
Others	At2g23050	At1g03380	At1g11000	At1g31580
	At3g19820	At1g22070	At1g52040
		At3g58450	At2g43550
			At5g51060
Hormone response
ABA	At2g40220	At1g01470	At1g29395	At1g75750
	At4g34240	At1g72100	At1g52690	At2g26980
	At5g25610	At2g41280	At1g55020	At3g22060
		At2g42560	At5g59320	At3g45640
		At3g11050
		At3g22490
		At3g22500
		At3g53040
		At4g36600
		At5g52300
GA	At1g74670	At1g15550	At1g74670	At1g15550
	At5g14920		At1g78440	At1g22690
	At5g15230			At1g75750
				At4g25420
Auxin	At2g34680	At1g59500	At1g29510	At2g45210
	At3g07390	At1g60680	At1g44350	At3g60690
	At4g34760	At1g60710	At2g21220
	At5g57090	At2g28350	At3g15540
		At2g45210	At3g23050
		At3g02875	At3g25290
		At4g33670	At4g12410
			At4g13790
			At5g47530
Ethylene		At1g28360	At1g15360	At1g05010
		At3g58450		At1g28360
		At4g33670		At5g25190
		At5g10120
		At5g47220
BR	At3g50750			At1g75750
	At4g30610
Cytokinin		At1g28230		At2g26980

Among the 251 DELLA-up genes, 150 genes have each been assigned a putative molecular function, and 101 genes are recorded as expressed putative genes (Supplemental Table S5). As observed for the DELLA-down genes, the two largest groups of DELLA-up genes are genes encoding enzymes (total 85 genes) and proteins with binding activities (total 79 genes), although the total number of DELLA-up enzyme genes (85 genes) is far less than the DELLA-down enzyme genes (162 genes; Table II). Detailed ontology analysis showed that DELLA-up enzyme genes are mainly for encoding oxidoreductase (30 genes) and transferase (20 genes), while the majority of DELLA-down genes are for hydrolase (70 genes), transferase (50 genes), and oxidoreductase (23 genes; Table II). This result suggests that the activity of food metabolism is kept at a low level, while the biosynthetic pathways and energy production pathways are likely redirected to use a different set of enzymes in the imbibed ga1-3 seeds. Surprisingly, a significant number of genes related to phytohormonal response (e.g. response to ABA, auxin, and ethylene) and stress response/defense were identified as DELLA-up genes. These genes include 10 ABA-related genes such as responsive to desiccation 29B (RD29B, At5g52300; Uno et al., 2000), ferric iron binding gene ATFER2 (At3g11050; Petit et al., 2001), late embryogenesis abundant M10 (At2g41280; Raynal et al., 1999), etc. Seven auxin-related genes, including AUXIN RESPONSE FACTOR 10 (ARF10; At2g28350; Wang et al., 2005); IAA-LEUCINE RESISTANT 1 (ILR1; At3g02875; LeClere et al., 2002); IAA-amido synthase GH3.4 (At1g59500; Staswick et al., 2005); a putative auxin-regulated protein gene (At2g45210); and five ethylene-related genes, including ethylene responsive element binding factor ATERF2 (At5g47220; Fujimoto et al., 2000), ethylene insensitive 3 (EIN3; At5g10120; Riechmann et al., 2000), and ethylene response factor ERF12 (At1g28360; Ohta et al., 2001), were also identified (Table III; Supplemental Table S5). In addition, some stress-response genes, including two dehydrin genes (At3g58450, At5g17310) and one superoxide dismutase gene (At3g56350) responsive for the removal of superoxide radicals, are identified as DELLA-up genes (Supplemental Table S5). Compared with the relatively large number of DELLA-down genes in MYB family (seven genes), zinc-finger family (four genes), and bHLH family (four genes), only two zinc-finger family genes (At2g31380, At2g47890) and two bHLH genes (At5g46760, At3g62090) were identified as DELLA-up transcription factor genes, whereas none of the MYB family genes was identified as DELLA-up genes (Table III).

Identification of DELLA-Dependent Transcriptomes Expressed during Floral Development

Floral development consists of three distinct phases: floral identity determination (phase transition from vegetative meristem to an inflorescence meristem), floral organ initiation, and floral organ growth (Krizek and Fletcher, 2005). The development of floral organs, especially petals and stamens, is impaired in GA-deficient mutants, while retarded anthesis results in male sterility due to a lack of mature pollen (Wilson et al., 1992; Cheng et al., 2004). All of the floral phenotypes of the GA-deficient mutant ga1-3 are restored to normal in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant, suggesting that GA signaling through these four DELLAs is the major pathway for GA-mediated floral development (Cheng et al., 2004; Tyler et al., 2004; Yu et al., 2004). To identify DELLA-dependent transcriptomes essential for floral development, we also carried out the microarray assay using RNA samples extracted from the young and unopened flower buds of the wild-type control, the ga1-3 mutant, and the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant. Six microarray replicates for each of the three genotypes in seeds were performed. Only genes with a logarithm base 2 of the signal ratio of wild type versus ga1-3 >1 (2-fold higher) or <−1 (2-fold lower) in at least four replicates were referred to as GA-up or GA-down, respectively. Based on the above criteria, 826 genes were identified as GA-up in the ga1-3 young flower buds (Supplemental Table S8) when compared to that in the wild type. The transcript levels of 360 out of these 826 GA-up genes (44%) were at least 2-fold higher in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant than in ga1-3 (Supplemental Table S9), while the remaining 466 genes did not show significant changes in their expression (Supplemental Table S10). These 360 genes are supposed to be DELLA-down genes in the young flower buds, and the 466 genes should be DELLA-independent or -partially-dependent GA-regulated genes. Meanwhile, the transcripts of 422 genes were accumulated to higher levels in ga1-3 young flower buds than in the wild type (Supplemental Table S11). The transcript levels of 273 out of these 422 genes (65%) were at least 2-fold lower in the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant than in ga1-3 (Supplemental Table S12), while the remaining 149 genes did not show significant changes in their expression (Supplemental Table S13) These 273 genes are supposed to be DELLA-up genes in the flower buds, and the 149 genes to be DELLA-independent or -partially-dependent GA-regulated genes. RT-PCR analysis confirmed that all 38 DELLA-down genes and 19 out of 21 DELLA-up genes randomly examined exhibited the expected expression patterns (Supplemental Table S14; Fig. 2, A and B), demonstrating that the microarray data obtained here are highly reproducible.

Figure 2.

RT-PCR confirmation of DELLA-down and DELLA-up genes in the unopened young flower buds. A, DELLA-down genes. B, DELLA-up genes. RT-PCR analysis was repeated on three independent samples and a representative ethidium bromide gel picture is shown here. Corresponding gene locus identity (Gene ID) is provided. Two genes (At1g09970 and At2g04240) in B did not show obvious difference in expression and were marked with an asterisk. Primer pairs for each individual gene are listed in Supplemental Table S12. penta: ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 penta mutant. ACT2 (ACTIN 2 gene) was used as the normalization control.

Ontology Analysis of DELLA-Dependent Transcriptomes Expressed during Floral Development

Among the 360 DELLA-repressed genes, 243 genes have each been assigned a putative molecular function based on amino acid homology, and 117 are recorded as functionally unassigned putative genes (Supplemental Table S9). The majority of DELLA-down floral genes, as observed for the DELLA-down genes in the imbibed seeds, encode enzymes (total 155 genes) responsible for the metabolism of protein, carbohydrate, and lipid and encode proteins (total 89 genes) with binding activity to nucleic acid, nucleotide, ion, and protein binding, suggesting that the arrest of floral organ growth is coupled with low metabolic activities (Table II). Many types of transcription factors are known to control or regulate floral development (Krizek and Fletcher, 2005). Our microarray analysis identified seven MYB family genes, four squamosa promoter-binding protein genes, three bHLH family genes, three MADS box genes, and three AP2 domain-containing transcription factor genes as DELLA-down genes (Table III), suggesting these factors might be the link between DELLA-mediated GA signaling and floral development. Previous studies have shown that the impaired growth of petal and stamen filament in ga1-3 is mainly due to the arrest of cell elongation rather than cell division (Cheng et al., 2004). Accordingly, our microarray analysis showed that DELLAs repress the expression of genes responsible for the biogenesis and modification of cell wall components, including four cellulose synthase genes, four expansin genes, two cellulases, and one 1,4-β-mannan endohydrolase (Table III). GA 2-oxidase (responsible for the degradation of bioactive GAs) and a GAST1-like gene have previously been shown to be up-regulated by GA (Shi and Olszewski, 1998; Ogawa et al., 2003). The restoration of expression of these two genes in ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 suggests that GA regulates their expression through triggering the degradation of DELLA proteins, and therefore they are identified as DELLA-down genes (Table III). Interestingly, nine auxin-response genes, including auxin-responsive transcription factors IAA19 (At3g15540) and AUXIN RESISTANT 2 (AXR2; At3g23050; Liscum and Reed, 2002), putative IAA-amino acid hydrolase 6 (ILL6; At1g44350; LeClere et al., 2002), two auxin-responsive dopamine beta-monooxygenases (At5g47530 and At3g25290; Neuteboom et al., 1999), and four auxin-responsive genes (At4g13790, At1g29510, At4g12410, and At2g21220; http://www.godatabase.org/cgi-bin/amigo; Table III), are identified as DELLA-down genes in the young flower buds.

Among the 273 DELLA-up floral genes, 180 genes have each been assigned a putative molecular function, and 93 are recorded as expressed putative genes (Supplemental Table S12). Again, the two largest groups of DELLA-up genes consist of genes encoding proteins with catalytic activity (total 110 genes) or binding activity (total 99 genes; Supplemental Table S12). The majority of DELLA-up enzyme genes are transferase genes (total 55 genes) and oxidoreductase genes (total 24 genes) but not hydrolase genes, as observed in DELLA-down floral genes (Table II). GA biosynthesis is controlled by a negative feedback loop. The lower expression levels of three key GA biosynthesis genes (two GA 20-oxidase genes and one GA-3β-hydroxylase gene) in ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 suggest that GA down-regulates these genes through triggering the degradation of DELLA proteins, and therefore they are identified as DELLA-up genes (Table III; Ogawa et al., 2003). Further analysis showed that a great range of transcription factors, including six putative zinc-finger family genes (ZINC FINGER PROTEIN ZEP3, ZEP4, JAG, etc.; Tague and Goodman, 1995; Riechmann et al., 2000), four MYB family genes (MYB59, MYB13, At5g44190, and At3g11280; Riechmann et al., 2000), three putative bHLH family genes (At5g50915, At5g46760, and At4g01460; Heim et al., 2003), and three WRKY family genes (WRKY15, WRKY70, and WRKY53; Eulgem et al., 2000), belong to DELLA-up genes (Table III), suggesting that DELLAs mediate a complex genetic regulation network to repress floral development. Interestingly, while only two protein kinase genes (At5g57670 and At1g61590) and one Leu-rich repeat kinase gene (at4g18640) were identified as DELLA-down genes, a significant number of putative protein kinase genes (19 protein kinase genes, nine Leu-rich repeat kinase genes, four S-locus protein kinase genes, and three receptor protein kinase genes) are identified as DELLA-up genes in the young flower buds (Table III), suggesting that protein phosphorylation modification might play a key role in controlling floral organ growth (Morris and Walker, 2003). Surprisingly, DELLAs seem to play a crucial active role in defense against disease in the young flower buds because 13 disease resistance genes are identified as DELLA-up genes (Table III; Maleck et al., 2000).

DISCUSSION

DELLAs Regulate Distinct Transcriptomes to Control Seed Germination and Floral Development

Organ initiation, growth, and development are the result of precisely coordinated action of multiple genes. The combination of loss-of-function of RGL1, RGL2, RGA, and GAI suppressed the ga1-3 mutant phenotype, and the resultant ga1-3 rgl1-1 rgl2-1 rga-t2 gai-t6 mutant confers GA-independent seed germination and floral development, suggesting that the defective seed germination and floral organ development in ga1-3 likely result from alteration of the expression of a network of genes that are directly or indirectly regulated by DELLA activity. We are interested to know if a similar set of DELLA-regulated genes is used to control these two distinct developmental processes. For this purpose, we compared the gene identity of the 360 DELLA-down and 251 DELLA-up genes in the imbibed seeds with that of the 360 DELLA-down and 273 DELLA-up genes in the young flower buds, respectively. Surprisingly, only 21 DELLA-down genes and 15 DELLA-up genes were found to be shared between the two datasets (Supplemental Table S15). RT-PCR analysis confirmed that all 21 shared DELLA-down genes and 12 out of 14 shared DELLA-up genes examined showed the expected expression patterns in the imbibed seed (Fig. 3, A and B). In the young flower buds, 18 out of 21 shared DELLA-down genes and all 13 shared DELLA-up genes examined displayed the expected expression patterns (Supplemental Table S15). Among the 21 DELLA-down genes, only one GAST1-like (At1g74670) gene and two putative expansin genes (At2g37640 and At2g40610) are presumably related to GA response. Meanwhile, only GA-3β-hydroxylase gene (At1g15550) is a known GA-response gene (Ogawa et al., 2003) among the 15 DELLA-up genes (Supplemental Table S15). These data demonstrate that GA-mediated seed germination and floral development are controlled by distinct DELLA-dependent transcriptomes.

Figure 3.

RT-PCR confirmation of shared DELLA-down and DELLA-up genes in the imbibed seeds and young flower buds. A, Shared-DELLA-down genes. B, Shared-DELLA-up genes. RT-PCR analysis was repeated on three independent samples and a representative ethidium bromide gel picture is shown here. Corresponding gene locus identity (Gene ID) is provided. Three shared DELLA-down genes and two shared DELLA-up genes showed no difference in expression in the young flower buds and imbibed seeds, respectively, and these genes were marked with an asterisk. Primer pairs for each individual gene are listed in Supplemental Table S13. penta: ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 penta mutant. For the imbibed seeds, ACT2 (ACTIN 2 gene) and UBQ10 (UBIQUITIN 10 gene) were used as the normalization controls. For the young flower buds, ACT2 was used as the normalization control.

Since GA triggers some similar cellular events during seed germination and floral development (e.g. GA induces epidermal cell elongation both along the hypocotyl of a germinating seed and the filament of a growing stamen; Cheng et al., 2004; Cao et al., 2005), the obvious question to ask is how two distinct DELLA-dependent transcriptomes regulate similar cellular events. To address this question, we first subgrouped the DELLA-regulated genes, identified either in the imbibed seeds or young flower buds, based on their known or presumable molecular functions in planta. Next, we cross-compared genes in corresponding subgroups in the dataset for imbibed seeds and dataset for young flower buds.

Novel GAMYB Genes and Other Transcription Factors

DELLAs are putative transcription regulators. Presumably, they may directly regulate the expression of some GA-response genes. Unfortunately, there is currently no concrete evidence to prove this hypothesis. Alternatively, DELLAs regulate the expression of some downstream transcription factors, and these DELLA-regulated transcription factors then control the expression of GA-response genes. GAMYB genes are the best studied GA-regulated transcription factors, and previous studies have shown that GA regulates GAMYB through DELLA proteins SLN and SLR in barley (Hordeum vulgare) and rice (Oryza sativa), respectively (Gubler et al., 2002; Kaneko et al., 2003). In Arabidopsis, MYB33 and MYB65 are identified as GAMYB genes based on homology analysis. However, the MYB33 and MYB65 and their subfamily members are regulated at the posttranscription level by miRNA159 (Achard et al., 2004; Millar and Gubler, 2005). In fact, MYB33 and MYB65 are not identified among the DELLA-down or DELLA-up genes in our dataset. On the other hand, our data showed that MYB4, MYB25, MYB30, MYB34, MYB66, and two MYB homologs (At1g01380 and At5g58900) are the seven DELLA-down MYB genes involved in seed germination, while the other seven DELLA-down MYB genes (MYB24, MYB32, MYB52, MYB106, MYB21, MYB, and At2g38090) are involved in floral development (Table III), suggesting that DELLAs differentially regulate a different subset of MYB genes to repress seed germination and floral development. Interestingly, four MYBs (MYB59, MYB At5g44190, MYB At1g06180, and MYB At3g11280) were identified as DELLA-up genes in the young flower buds, while no DELLA-up MYB gene was found in the imbibed seeds (Table III). Therefore, these MYB genes may represent new types of GAMYBs, and future work will focus on studying the relationship between GA and these MYB genes. In addition to MYB genes, distinct DELLA-down or -up bHLH and zinc-finger family genes are also identified both in the imbibed seeds and young flower buds (Table III). It is interesting to note that four zinc-finger family genes were identified as DELLA-down genes in the imbibed seeds, while six other zinc-finger family genes were identified as DELLA-up genes in the young flower buds, indicating that the zinc-finger gene family is differentially regulated by DELLAs at different developmental stages. As expected, three types of transcription factors, namely, three MADS box family genes (AGL1, AGL6, and AGL11), three WRKY family genes (WRKY15, WRKY70, and WRKY53; Eulgem et al., 2000), and five squamosa promoter-binding protein-box family genes (SPL2, SPL5, SPL11, and SPL12), are found among the DELLA-regulated genes only for floral development (Table III; Krizek and Fletcher, 2005). Apparently, these transcription factors will target their own specific targets to fine-tuning the regulation initiated by DELLAs. One of the future tasks will be to find out the targets controlled by these transcription factors.

Protein Phosphorylation Might Represent a Major Pathway for DELLA Repression of Floral Organ Development

Protein phosphorylation and dephosphorylation is widely involved in signaling cascade to trigger the downstream cellular events. Six DELLA-down and nine DELLA-up protein kinase genes are identified in the imbibed seeds (Table III), suggesting that the change of phosphorylation status of signaling proteins is probably involved in causing the nongerminating phenotype of the ga1-3 mutant. Surprisingly, microarray analysis revealed that DELLAs regulate the protein kinase families in a unique way during floral development. About 21 DELLA-up protein kinase genes were identified in the young flower buds, while only two were found as DELLA-down genes (Table III). In addition, while the expression of nine Leu-rich repeat protein kinase genes was repressed by DELLAs in the imbibed seeds, a completely different set of nine Leu-rich repeat protein kinase genes was up-regulated by DELLAs in the young flower buds (Table III). Furthermore, five S-locus-related protein kinase genes were activated by DELLAs in the young flower buds only (Table III). Therefore, DELLAs differentially regulate the expression of different protein kinase genes to control seed germination and floral development. Combining all data, it seems that activating protein phosphorylation pathways might be a crucial step for DELLAs to repress floral development.

DELLAs Maintain the Low Metabolic Activity in the ga1-3 Mutant

Seed germination is an active process that needs to mobilize food reserves to provide sufficient energy and building blocks to sustain the dynamic cellular activities in the germinating seed. In contrast, a nongerminating seed normally maintains low metabolic activity (Bewley, 1997). In both imbibed seeds and young flower buds, a large number of genes encoding enzymes (especially hydrolase, transferase, and oxidoreductase) responsible for the metabolism of carbonhydrate, protein, and lipid are repressed by DELLAs (Table II; Supplemental Tables S5, S9, and S14). This fact suggests that the metabolic activities in both imbibed ga1-3 mutant seeds and young ga1-3 mutant flower buds are likely kept at a low level, and this low metabolic activity of mobilization of food reserves nicely correlates with the nongerminating and arrested floral development phenotypes displayed by the ga1-3 mutant. When compared to the wild type, the transcript levels of a large number of different transferase and oxidoreductase genes in ga1-3 were altered in the imbibed seeds and young flower buds, respectively (Table II; Supplemental Table S16), suggesting that biosynthetic and oxidative pathways are redirected to other pathways in the ga1-3 mutant. Cross-comparison showed that the identities (gene locus) of the DELLA-regulated (both -repressed and -activated) hydrolase genes, transferase genes, oxidoreductase, and other enzyme genes in the imbibed seeds are almost completely different from their respective counterparts in DELLA-regulated enzyme genes in the young flower buds (Table II; Supplemental Table S16). This fact strongly suggests that DELLAs differentially regulate different subsets of metabolic genes of similar molecular functions or different individual members of a same gene family to control seed germination and floral development.

Distinct Approaches Are Utilized to Control Cell Growth and Cell Wall Modification during Seed Germination and Floral Development

Prior to seed germination, a number of cell wall-modifying genes will be activated to loosen the cell wall and break the seed coat to facilitate the radicle protrusion. Similarly, during the period of the floral organ growth, factors will be produced to promote the elongation of epidermal cells of petal, stamen, and pistil. Five and four expansin genes were identified as DELLA-down genes in the imbibed seeds and young flower buds, respectively, and two of them (At2g37640 and At2g40610) are shared (Table III), suggesting that expansins are crucial for the cell elongation in both developmental processes. However, while seven xyloglucan endotransglycosylase/hydrolase and six pectinesterase genes are the major genes responsible for the cell wall loosening in the imbibed seeds, none of these two categories of genes was DELLA-down in the young flower buds (Table III). Instead, four cellulose synthase genes were found as DELLA-down genes only in the young flower buds but not in the imbibed seeds (Table III; Supplemental Tables S3 and S9). Interestingly, three α-tubulin genes (TUA2, TUA4, and TUA6) and four β-tubulin genes (TUB1, TUB5, TUB6, and TUB7) are also DELLA-down genes in the imbibed seeds but not in the young flower buds (Table III). These results suggest that the cell elongation activity during seed germination is probably mainly resulted from cell wall loosening coupled with cell reshaping, while the cell elongation during floral development is mainly due to the de novo biosynthesis of cellulose.

DELLAs Act as Convergence Point for Phytohormone Signaling

As expected, in both imbibed seeds and young flower buds, the GA-response gene GAST1 (At1g74670) and the key GA biosynthesis gene GA-3-β-hydroxylase (At1g15550) are identified as DELLA-down and -up genes, respectively (Shi and Olszewski, 1998; Ogawa et al., 2003). Recently, Ueguchi-Tanaka (Ueguchi-Tanaka et al., 2005) reported that OsGID1 in rice encodes a soluble GA-receptor with homology to the consensus sequence of the hormone-sensitive lipase (HSL) homologous family. A database search identified three OsGID1 homologs in Arabidopsis (Fig. 4), and all of them have recently been shown to bind GA (Nakajima et al., 2006). Interestingly, two of these OsGID1 homologs (At3g05120 and At3g63010) are identified as DELLA-up genes in the young flower buds and one (At3g05120) in the imbibed seeds, suggesting that these OsGID1 homologs are probably negatively regulated by GA. However, the fact that DELLA proteins are stabilized in the ga1-3 mutant suggests that GA is necessary to activate the GID1-like receptors to trigger the degradation of DELLA proteins. A total of 14 GDSL-type lipase genes, another type of lipase presumably related to defense (Akoh et al., 2004), are identified as DELLA-down genes (Table IV; Supplemental Fig. S1). Previous studies showed that ABA signaling through ABI1 and ethylene signaling through CTR1 enhance the stability of DELLAs (Achard et al., 2003, 2006), implying that a fraction of ABA- and ethylene-signaling response genes will probably be identified as DELLA-regulated genes in our dataset. Indeed, a number of ABA- and ethylene-response genes were identified as DELLA-up genes in both the imbibed seeds and young flower buds. For example, the expression of late-embryogenesis-abundant (LEA) proteins genes in the imbibed seeds is known to be responsive to ABA treatment (Ali-Benali et al., 2005; Kamisugi and Cuming, 2005; Bethke et al., 2006), and seven of these LEA protein genes were identified as DELLA-up genes (Table III). Also, five and three ethylene-related genes were found as DELLA-up genes in the imbibed seeds and young flower buds, respectively. These genes include genes for ethylene responsive element binding factor 1 (ERF2; At5g47220; McGrath et al., 2005), EIN3, ethylene responsive element binding factor (ERF12; At1g28360), ethylene-responsive element-binding family protein (At5g61600), and universal stress protein USP/ER6 (At3g58450; Table III; Supplemental Tables S5 and S12; Chang and Bleecker, 2004; Guo and Ecker, 2004). That low concentrations of auxin promote the destabilization of DELLAs (Fu and Harberd, 2003) fits well with the finding that four (the auxin efflux carrier EIR1, auxin-induced protein AIR12, AIR9, and At4g34760; Luschnig et al., 1998; Neuteboom et al., 1999) and nine auxin-response genes were identified as DELLA-down genes in the imbibed seeds and young flower buds, respectively. Interestingly, seven auxin-related genes, including genes encoding IAA-amino acid hydrolase ILR1 (At3g02875), auxin-regulated protein GH3 (At1g59500), auxin-induced protein IAA17/AXR3-1, and two auxin-induced proteins similar to auxin-induced atb2 (At1g60710 and At1g60680; Table III; Supplemental Table S5; Liscum and Reed, 2002), were identified as DELLA-up genes in the imbibed seeds, suggesting that the interaction between GA and auxin is probably more complicated than previously thought.

Figure 4.

Amino acid sequence alignment of rice GID1 with its three Arabidopsis homologs (At3g05120, At3g63010, and At5g27320) using the ClustalW program. Gene ID is provided on the left side of the figure.

In Addition to Protecting Plant from Adverse Environment, DELLAs Might Also Mediate Disease Resistance in Young Flower Buds

Recent studies have shown that DELLAs act as the integrator of environmental cues and endogenous phytohormonal signals to protect plants from the environmental stress (Lee et al., 2002; Achard et al., 2003, 2006; Cao et al., 2005). For both the imbibed seeds and young flower buds, genes responsive to oxidative stress were identified as DELLA-down genes (Table III). On the other hand, genes responsive to water stress (dehydration/desiccation) and genes responsible for toxin catabolism are more sensitive to DELLA regulation in the imbibed seeds than in the young flower buds. Interestingly, six water stress response genes were identified as DELLA-down genes and 13 water stress response genes were identified as DELLA-up genes in the imbibed seeds (Table III), suggesting there might be a switch of water stress response pathways during different stages of plant growth. In contrast, multidrug transport and wounding response genes were identified as DELLA-down genes mainly in the young flower buds (Table III). Most significantly, a great number of putative disease defense genes (13 genes) were identified as DELLA-up genes (Table III), implying that DELLAs are not only actively involved in protecting plant from different environmental stress but probably in mediating disease resistance as well, especially during stages of plant growth.

DELLA-Independent or -Partially-Dependent GA-Regulated Genes

Gene expression profiling data in Arabidopsis and rice has shown that a wide range of genes, including genes encoding enzymes and other factors that degrade the cell wall of endosperm and seed coat, are regulated by GA to stimulate the growth of the embryo, elongation of the embryo axis, and breakage of seed coat (Ogawa et al., 2003; Bethke et al., 2006). Ogawa et al. compared the expression profiles between GA-treated and untreated ga1-3 seeds at various time points after 48-h stratification in the dark and 24 h at 22°C in the light and identified a total of 230 GA-up genes and 127 GA-down genes using an Arabidopsis gene chip carrying approximately 8,200 genes (Ogawa et al., 2003). We cross-compared the 541 GA-up genes and 571 GA-down genes obtained in our experiment with the 230 GA-up genes and 127 GA-down genes identified by Ogawa et al., respectively, and found that 109 GA-up genes (approximately 47% of 230 genes obtained by Ogawa et al.) and 90 GA-down genes (approximately 71% of 127 genes obtained by Ogawa et al.) are shared in both datasets (Supplemental Tables S17 and S18). Given the differences in the experimental design, the high degree of overlap between the two datasets is quite impressive. More interestingly, further analysis showed that 91 out of the 109 shared GA-up genes were among the 360 DELLA-down genes (and thus are regulated in a DELLA-dependent fashion) in the imbibed seeds, while the remaining 18 shared GA-up genes were among the 181 DELLA-independent or -partially-dependent genes (Supplemental Table S17). Meanwhile, data analysis also showed that 56 out of 90 shared GA-down genes were among the 251 DELLA-up genes (and thus are DELLA dependent) in the imbibed seeds. However, the remaining 34 shared GA-down genes were among the 320 DELLA-independent or -partially-dependent genes (Supplemental Table S18). Because the GA-regulated genes in Ogawa's dataset were obtained by applying GA to the ga1-3 seeds, the above data support the hypothesis that there is probably an unknown DELLA-independent or -partially-dependent component essential for the regulation of some GA-dependent genes.

CONCLUSION

In this report, we identified GA-regulated (both GA-down and -up) transcriptomes in both imbibed seeds and young flower buds by comparing the expression patterns between the ga1-3 mutant and wild-type control. Then, we identified DELLA-dependent (both DELLA-down and -up) transcriptomes by finding out the subgroup of GA-regulated genes with their expression restored to the wild-type levels in the ga1-3 rga-t2 gai-t6 rgl1-1 rgl2-1 mutant. The high percentage of overlap between GA-regulated genes identified in our work and Ogawa's work, together with the high rate of confirmation of candidate genes by RT-PCR analysis, demonstrate that the datasets obtained are highly reproducible and reliable. The complete suppression of ga1-3 nongerminating and male sterile phenotypes by loss-of-function of RGA, GAI, RGL1, and RGL2 implies that GA-dependent gene regulation might be largely through the DELLA-dependent pathway. Interestingly, we observed that approximately half of total GA-regulated genes are regulated via the DELLA-dependent pathway, suggesting an unknown DELLA-independent component is probably essential for the regulation of other GA-dependent genes. However, because we have set strict criteria to identify the DELLA-regulated genes, we might have missed identifying some DELLA-regulated genes due to sample variations and also cannot exclude the possibility that a portion of the remaining GA-regulated genes might be partially regulated by DELLAs. Finally, we cross-compared the DELLA-dependent transcriptomes between imbibed seeds and young flower buds and surprisingly found that, based on gene identity (gene locus ID), the two DELLA-dependent transcriptomes are almost entirely distinct from each other. Ontology analysis revealed that a large number of genes with similar molecular and biochemical functions (e.g. genes for hydrolases, transferases, oxidoreductases, proteins with binding activity, MYBs, bHLHs, expansins, etc.) are repressed or up-regulated by DELLAs in both imbibed seeds and young flower buds. In fact, these groups of genes constitute the largest portion of DELLA-dependent transcriptomes in both imbibed seeds and young flower buds. This fact suggests that the many basic biochemical pathways are similarly mobilized during seed germination and floral development. However, specific factors participating in these pathways are different individual members from different gene families, suggesting that DELLAs differentially regulates the expression of these specific factors during seed germination and floral development. Meanwhile, detailed data analysis revealed that DELLAs also control the expression of many functionally completely different genes, including factors for cell wall loosening, stress and disease response, and protein phosphorylation modification, to run different pathways either specific for seed germination or for floral development, which signifies the differences between these two important biological processes. In conclusion, the data shown here not only confirm the results obtained from many previous reports but also single out some novel aspects of DELLA functions that will be instructive to our future research.

MATERIALS AND METHODS

Plant Growth Conditions, Genetic Nomenclature, and Plant Materials

Plants were grown as described previously (Lee et al., 2002). Arabidopsis (Arabidopsis thaliana) Landsberg erecta was used as the wild-type control. The ga1-3 mutant and the ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant are all in Landsberg erecta background and were obtained as described previously (Cheng et al., 2004; Cao et al., 2005).

RNA Samples from Seed and Young Flower Buds for Microarray Hybridization

Seeds were imbibed at 4°C on filter papers soaked in sterile water under continuous white light for 4 d. Total RNA was extracted from the imbibed seeds using RNAqueous RNA Isolation kit with Plant RNA Isolation Aid (Ambion). The residue DNA in total RNA was removed via a treatment with DNaseI, and total RNA was further purified with the RNAeasy Mini kit (Qiagen). Total RNA from young unopened flower buds of 28-d-old wild type, 22-d-old ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 mutant, and 38-d-old ga1-3 mutant was extracted using TRIzol (Gibco-BRL) and treated with DNaseI as described (Lee et al., 2002). cDNA synthesis, cDNA amplification, RNA probe labeling, GeneChip hybridization, washing, and staining were performed following the manufacturer's instruction (Affymetrix). GeneChip arrays were scanned on an Affymetrix probe array scanner. Data were preliminarily analyzed using the statistics software Microarray Suite version 5.0 (MAS5.0) from Affymetrix.

RT-PCR Analysis of Candidate Genes

Total RNA from imbibed seeds and young flower buds was extracted using the methods described above, respectively. Oligo-dT directed cDNA was synthesized from approximately 0.5 μg of total RNA in a 20-μL RT reaction following protocol supplied by the manufacturer (Invitrogen). The obtained cDNA were used as substrates for PCR assay. The primers used for PCR reactions are provided in Supplemental Tables S7, S14, and S15. Amplified PCR products were visualized and photographed under a UV translluminator.

Ontology Analysis and Cross-Comparing DELLA-Dependent Transcriptomes

We obtained the signal intensities of individual genes using the statistical algorithms on MAS5.0. The presence or absence of a reliable hybridization signal for each gene was determined by the detection call on MAS5.0. Genes were classified as GA responsive if the signal intensities deviated either positively or negatively 2-fold or more between ga1-3 and wild type. Genes for which transcripts were determined to be undetectable (absent or marginal present) in ga1-3 samples were eliminated from the list of up-regulated genes in ga1-3. Similarly, genes for which transcripts were determined to be undetectable (absent or marginal present) in wild-type samples were eliminated from the list of down-regulated genes in ga1-3. When the transcript was undetectable in only ga1-3 or wild-type sample, we gave the background signal intensity to the undetectable transcript. If the signal intensity from the other sample was greater by 2-fold or more relative to the background value, this gene was regarded as being GA regulated. A gene is regarded as DELLA-down if this gene is down-regulated in ga1-3, the signal intensity of ga1-3 was less by 2-fold or more relative to the intensity of ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1, and the signal intensity did not deviate negatively more than 2-fold between ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 and wild type. Similarly, a gene is regarded as DELLA-up if this gene is up-regulated in ga1-3, the signal intensity of ga1-3 was greater by 2-fold or more relative to the intensity of ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1, and the signal intensity did not deviate positively more than 2-fold between ga1-3 gai-t6 rga-t2 rgl1-1 rgl2-1 and wild type. In the seed samples, we classified the genes that were GA responsive in all three independent replicates as GA responsive. In the flower samples, genes that were GA responsive in any four of the six independent replicates were classified as GA responsive.

The Gene Ontology information was retrieved through the NetAffx Gene Ontology Mining Tool, based on the Molecular Function and biological process. Throughout the data sets, genes are identified by the AGI gene code, which was linked to Affymetrix Probe Set ID based on the gene annotation information in the NetAffx Analysis Center (https://www.affymetrix.com/analysis/netaffx/index.affx).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Amino acid sequence alignment of 13 GDSL-type lipase in Arabidopsis using ClustalW program.

• Supplemental Table S1. GA-up genes in seeds.

• Supplemental Table S2. GA-down genes in seeds.

• Supplemental Table S3. DELLA-down in seeds.

• Supplemental Table S4. GA-up but DELLA-independent genes in seeds.

• Supplemental Table S5. DELLA-up genes in seeds.

• Supplemental Table S6. GA-down but DELLA-independent genes in seeds.

• Supplemental Table S7. RT-PCR confirmation of DELLA-regulated genes in seeds.

• Supplemental Table S8. GA-up genes in flower buds.

• Supplemental Table S9. DELLA-down genes in flower buds.

• Supplemental Table S10. GA-up but DELLA-independent genes in flower buds.

• Supplemental Table S11. GA-down genes in flower buds.

• Supplemental Table S12. DELLA-up genes in flower buds.

• Supplemental Table S13. GA-down but DELLA-independent genes in flower buds.

• Supplemental Table S14. RT-PCR confirmation of DELLA-regulated genes in flower buds.

• Supplemental Table S15. Shared DELLA-regulated genes in seeds and flower buds.

• Supplemental Table S16. DELLA-regulated metabolic genes.

• Supplemental Table S17. Shared GA-up and DELLA-down genes with Ogawa's dataset.

• Supplemental Table S18. Shared GA-down and DELLA-up genes with Ogawa's dataset.

Table IV.

GDSL-type lipase genes regulated by DELLAs

DELLA-D, DELLA-down genes; DELLA-U, DELLA-up genes.

Imbibed Seeds		Unopened Flower Buds
DELLA-D	DELLA-U	DELLA-D	DELLA-U
At2g03980		At1g54790	At1g29670
At3g04290		At1g58430
At3g48460		At2g42990
At4g18970		At3g48460
At5g14450		At4g18970
At5g45670		At5g33370
At5g45950		At5g45960