Synergistic Derepression of Gibberellin Signaling by Removing RGA and GAI Function in Arabidopsis thaliana
- Alyssa Dill and
- Tai-ping Sun⇓
- Developmental, Cell and Molecular Biology Group, Department of Biology, Duke University, Durham, North Carolina 27708-1000
- Corresponding author: Tai-ping Sun, Department of Biology, Box 91000, Duke University, Durham, NC 27708-1000. E-mail: tps{at}acpub.duke.edu
Abstract
RGA and GAI are negative regulators of the gibberellin (GA) signal transduction pathway in Arabidopsis thaliana. These genes may have partially redundant functions because they are highly homologous, and plants containing single null mutations at these loci are phenotypically similar to wild type. Previously, rga loss-of-function mutations were shown to partially suppress defects of the GA-deficient ga1-3 mutant. Phenotypes rescued include abaxial trichome initiation, rosette radius, flowering time, stem elongation, and apical dominance. Here we present work showing that the rga-24 and gai-t6 null mutations have a synergistic effect on plant growth. Although gai-t6 alone has little effect, when combined with rga-24, they completely rescued the above defects of ga1-3 to wild-type or GA-overdose phenotype. However, seed germination and flower development defects were not restored. Additionally, rga-24 and rga-24/gai-t6 but not gai-t6 alone caused increased feedback inhibition of expression of a GA biosynthetic gene in both the ga1-3 and wild-type backgrounds. These results demonstrate that RGA and GAI have partially redundant functions in maintaining the repressive state of the GA-signaling pathway, but RGA plays a more dominant role than GAI. Removing both RGA and GAI function allows for complete derepression of many aspects of GA signaling.
THE plant hormone gibberellin (GA) plays an important role in many aspects of plant growth and development (reviewed in Hooley 1994; Rosset al. 1997). For instance, GA promotes seed germination, leaf expansion, and stem elongation. In a number of species, GA also regulates flowering time and is necessary for flower and fruit development. The essential role of GA in plant growth is clearly illustrated by the phenotype of severe GA biosynthetic mutants (Phillips 1998). In Arabidopsis thaliana, these mutants, for example, ga1-3, fail to germinate without exogenously applied GA, produce small, dark green leaves, and are malesterile dwarfs (Koornneef and van der Veen 1980). These defects are rescued completely by GA treatment. Progress is being made to identify components involved in the GA signal transduction pathway using both pharmacological and genetic approaches (reviewed in Bethke and Jones 1998; Thorntonet al. 1999; Lovegrove and Hooley 2000; Sun 2000). Biochemical studies using cereal aleurone cells provide evidence that the GA receptor is localized on the plasma membrane (Hooleyet al. 1991; Gilroy and Jones 1994), but the nature of the receptor remains elusive. Second messengers and transcription factors implicated in GA signaling include Ca2+, calmodulin, heterotrimeric G-proteins, protein kinases, and GAMYB (Bethke and Jones 1998; Lovegrove and Hooley 2000).
GA-signaling mutants have been isolated from various species and fall into two phenotypic categories, GA-unresponsive dwarf mutants and slender mutants (Thorntonet al. 1999; Sun 2000). The GA-unresponsive mutants resemble the dwarf mutants that are defective in GA biosynthesis, but fail to respond to exogenous GA application. Recessive mutants in this class are likely to be impaired in positive regulators of GA signaling, e.g., dwarf 1 (d1) in rice (Mitsunagaet al. 1994), gse in barley (Chandler and Robertson 1999), and sleepy1 in Arabidopsis (Steberet al. 1998). The D1 gene has been shown to encode the putative α-subunit of the heterotrimeric G-protein (Ashikariet al. 1999; Ueguchi-Tanakaet al. 2000). A number of semidominant GA-unresponsive dwarf mutants have also been isolated, including gai-1 in Arabidopsis (Koornneefet al. 1985), D8 and D9 in maize (Phinney 1956), and the Rht mutants of wheat (Borneret al. 1996). Isolation of intragenic revertants and recent cloning of these genes showed that they encode negative regulators of GA response (Penget al. 1999; see below). Another GA-unresponsive dwarf mutant is the semidominant shi mutant in Arabidopsis (Fridborget al. 1999). The phenotype in this plant is due to overexpression of its product and SHI is therefore also a putative negative regulator of GA response.
The recessive slender mutants resemble wild-type plants that have been treated with excess GA, suggesting that they are defective in negative regulators of GA signaling. Mutants in this category include la crys from pea (Pottset al. 1985), sln in barley (Lanahan and Ho 1988), and spindly (spy) from Arabidopsis (Jacobsen and Olszewski 1993). SPY is predicted to encode an O-linked N-acetylglucosamine (GlcNAc) transferase (OGT; Thorntonet al. 1999), which regulates target protein function by glycosylation of serine or threonine residues (Hart 1997).
We have identified another negative regulator of GA signaling, RGA, by screening for Arabidopsis mutants that were able to suppress the GA-deficient phenotype of ga1-3 (Silverstoneet al. 1997). The homozygous rga/ga1-3 double mutants, while still nongerminating and male sterile, have larger leaves and a semidwarf stature. Cloning of RGA revealed that RGA and GAI are 82% identical at the amino acid level and have hallmarks of transcriptional regulators, such as a nuclear localization signal, homopolymeric serine and threonine sequences, leucine heptad repeats, and an SH2-like domain (Peng et al. 1997, 1999; Silverstoneet al. 1998). Additionally, a green fluorescent protein-RGA fusion protein has been shown to localize to the nucleus in a transient assay in onion cells (Silverstoneet al. 1998) and in stably transformed Arabidopsis (Silverstoneet al. 2001).
RGA and GAI are members of the GRAS (GAI, RGA and SCARECROW) family of regulatory proteins (Pyshet al. 1999). In Arabidopsis, at least 38 GRAS family members are present, and all of them contain highly conserved central (VHIID) and C-terminal (RVER) regions. Their N termini, however, are more divergent. Interestingly, RGA and GAI have a conserved sequence near their N termini, termed the DELLA domain, after an amino acid motif contained therein (Silverstoneet al. 1998). This DELLA region may be required for the inactivation of GAI and RGA by the GA signal, because a 17-amino-acid deletion in this domain in either protein causes a GA-insensitive dwarf phenotype (Penget al. 1997; A. Dill and T.-p. Sun, unpublished results). It has been hypothesized that deleting the DELLA sequences turns the mutant protein into a constitutive repressor of GA signaling (Penget al. 1997).
Recently, the functional orthologs of RGA and GAI in wheat (Rht), maize (d8), barley (SLN; P. M. Chandler, A. Marion-Poll, F. Gubler, personal communication), and rice (SLR; Ogawaet al. 2000; Ikedaet al. 2001) have been isolated. Mutant studies revealed that these genes also function as repressors of GA signaling, indicating that RGA/GAI function is conserved among dicots and monocots (Penget al. 1999; Ikedaet al. 2001). The most notable example is the semidwarf wheat cultivars, a crucial component of the “Green Revolution,” which all contain deletions in the DELLA region of an Rht gene.
Because a loss-of-function mutation at either the RGA or GAI locus in wild-type GA background does not cause any obvious phenotype, these two homologous genes may have partially or completely redundant functions in GA signaling (Penget al. 1997; Silverstoneet al. 1998). To test this hypothesis, we generated homozygous rga/gai double mutant lines in the wild-type and ga1-3 backgrounds and showed that rga and gai null alleles have synergistic effects on a number of GA-mediated processes. GA signaling can control GA biosynthesis through feedback mechanisms (reviewed in Bethke and Jones 1998; Hedden and Phillips 2000; Yamaguchi and Kamiya 2000). By RNA blot analysis, we demonstrated that loss of RGA and GAI function also affects the feedback regulation of the GA biosynthetic gene GA4.
MATERIALS AND METHODS
Isolation of mutant lines: We isolated rga-24, rga-24/gai-t6, gai-t6/ga1-3, and rga-24/gai-t6/ga1-3 homozygous mutant lines from crosses between rga-24/ga1-3, gai-t6, ga1-3, and wild type in the Landsberg erecta (Ler) ecotype. gai-t6 and rga-24/gai-t6 also contained tt1-1. gai-t6 was a gift from Nicholas Harberd. Allele-specific primers were designed to identify homozygous mutants from segregating F2 populations of different crosses. Primers 219 (5′-GTTGATAGACATTTTCAATGA-3′) and 220 (5′-GGTCATCAGTAGAGACTAA-3′), which flank the 8.4-kb deletion in rga-24, amplify a 4-kb region in rga-24 but fail to amplify RGA DNA because the distance between these primers is too great for amplification of the product under standard conditions. RGA was identified using primers 219 and 212 (5′-GGTGATTTTCACGGTGGTTG-3′), which amplify a 3.3-kb region in RGA but fail to amplify rga-24 because the sequence in primer 212 is deleted in rga-24. To detect gai-t6 we used primer 304 (5′-TCGGTACGGGATTTTCGCAT-3′), and primer 300 (5′-CTAGATCCGACATTGAAGGA-3′), which are located in the Ds insertion and the GAI coding sequence, respectively. Together, these primers amplify a 720-bp fragment in gai-t6, but not in the GAI allele. GAI was identified using primers 300 and 302 (5′-AGCATCAAGATCAGCTAAAG-3′), which flank the Ds insertion in gai-t6 and therefore only amplify a 1.2-kb product in GAI but not in gai-t6. The GA1 and ga1-3 alleles were verified using PCR primers as previously described (Silverstoneet al. 1997). All the PCR reactions were done using AmpliTaq (Perkin-Elmer, Norwalk, CT) except for genotyping RGA and rga-24, for which Accu Taq LA (Sigma, St. Louis) was used.
Plant growth conditions: For all experiments except for the examination of GA4 mRNA levels and germination, seeds were imbibed for 3 days at 4° and then sown on soil. Seeds in the ga1-3 background were imbibed in 100 μm GA3 and then rinsed thoroughly with water before sowing on soil. Sterilization of seed surface in bleach solution weakens the seed coat and could result in up to 5% germination in nongerminating ga1-3 seeds (Silverstoneet al. 1997). To avoid this effect, seeds for germination percentage measurements were gently washed in 0.02% Triton X-100 to remove most contamination, rinsed with sterile water, and then spread on three layers of moist Whatman filter paper. Long day (LD) plants were grown at 22° with 16-hr light and 8-hr dark cycles supplied under a light intensity of 140 μE. All experiments except flowering time, abaxial trichome initiation, rosette radius, and GA4 mRNA levels were carried out under LD conditions. Flowering time, abaxial trichome initiation, and rosette radius were studied under short day (SD) conditions of 8-hr light and 16-hr dark cycles at 22° with a light intensity of 160 μE. SD lighting was supplied by a 3:1 mixture of cool-white:wide spectrum fluorescent bulbs (General Electric).
Measuring flowering time, rosette radius, and germination: The flowering time was scored when the flower bud was first visible without manipulation or magnification. Rosette radius was obtained by measuring the longest rosette leaf of each plant. Germination percentage was scored after 7 days of incubation in LD conditions. A seed was considered germinated if the radicle protruded from the seed coat.
GA4 mRNA quantitation: Seeds were washed with 95% ethanol for 1 min, sterilized in bleach for 2 min, and rinsed thoroughly with sterile water. Seeds were then imbibed for 3 days at 4° in either 50 μm GA4 (ga1-3-containing lines) or water (GA1-containing lines) and then washed five times with sterile water before plating on agar media containing Murashige-Skoog media and 2% sucrose in 100 × 15 mm plates. The plates were incubated at 22° under continuous light for 13 days with a light intensity of 100 μE. The seedlings were then either harvested or treated with 3 ml of 100 μm GA3 per plate for 8 hr and then harvested. Total mRNA was isolated and GA4 mRNA detected using an antisense GA4 RNA probe as described by Yamaguchi et al. (1998). As a loading control 18S RNA levels on the same blot were examined using 5′-32P-labeled oligonucleotide (5′-TGAAGGGATGCCTCCAC-3′). The blot was prehybridized for 2 hr at 42° in 10× Denhardt's, 5× SSPE, 1% SDS, and 100 μg/ml salmon sperm DNA. 32P-labeled 18S oligonucleotides were then added, with the final concentration of oligonucleotides at 15 nm and 5 × 105 cpm/ml, and hybridized overnight at 42°. The filters were washed four times (10 min each wash) in 6× SSC and 0.1% SDS at 48° and analyzed using a PhosphorImager as previously described (Silverstoneet al. 1998).
RESULTS
To investigate the role of RGA and GAI in GA signaling, we generated homozygous single and double rga-24/gai-t6 mutant lines in the wild-type and ga1-3 backgrounds by genetic crosses. Both rga-24 and gai-t6 are null alleles due to a deletion spanning the RGA locus (Silverstoneet al. 1998) and a Ds insertion in the GAI coding region (Penget al. 1997), respectively. Allele-specific PCR markers for RGA and GAI loci were designed to identify homozygous mutant and wild-type alleles. A number of GA-controlled developmental phenotypes in the mutants were examined, including seed germination, stem elongation, juvenile-to-adult transition in leaf development, apical dominance, flowering time, and flower development. Our previous mutant analysis suggested that RGA plays an important role in most of these processes, except seed germination and flower development (Silverstoneet al. 1997). Because RGA and GAI are likely to have overlapping functions in GA signaling, we tested whether gai-t6 has a similar effect as rga-24 and whether rga-24 and gai-t6 showed additive interactions in these GA-mediated processes.
Phenotypes of mutant lines: Figure 1 shows the phenotypes of the digenic (double homozygous) and trigenic (triple homozygous) mutants, along with wild-type Ler and ga1-3. The rga-24 and gai-t6 single mutants (not shown) had a similar phenotype as Ler. The gai-t6/ga1-3 mutant did not bolt at all. This is in contrast to rga-24/ga1-3, which had partially restored stem height (59% of Ler, Figure 2). However, the combination of rga-24 and gai-t6 had a dramatic effect on stem growth in both wild-type and GA-deficient ga1-3 backgrounds. Compared to the nonbolting ga1-3, the final stem height of the trigenic rga-24/gai-t6/ga1-3 mutant was remarkable (even taller than Ler by 32%). The digenic rga-24/ gai-t6 mutant was 35% taller than Ler (Figures 1 and 2). This synergistic effect of rga and gai-t6 indicates that RGA and GAI are the major repressors regulating GA-induced stem growth in Arabidopsis.
Phase change and flowering time: The appearance of trichomes on the lower surface of the leaf indicates the developmental transition from juvenile-to-adult stages. GA is essential for this transition because the ga1-3 mutant does not produce abaxial trichomes at all (Chien and Sussex 1996). Previously, we showed that the rga/ga1-3 mutants initiate abaxial trichomes, although later than wild type (Silverstoneet al. 1997). This indicated that RGA inhibits the GA-induced transition from juvenile-to-adult stages during leaf development. Here, we examined the timing of the phase change under SD conditions because the plants would have a longer juvenile phase. We found that rga-24 alone caused the abaxial trichomes to initiate much earlier than the control in both the ga1-3 and wild-type GA backgrounds (Figure 3A). The gai-t6 mutant initiated abaxial trichomes only very slightly earlier than Ler, whereas gai-t6/ga1-3 did not produce any abaxial trichomes (Figure 3A). However, GAI does play an important role in suppressing the juvenile-to-adult phase transition because the trigenic mutant rga-24/gai-t6/ga1-3 initiated abaxial trichomes even earlier than Ler (Figure 3A). These results indicate that both RGA and GAI inhibit the juvenile-to-adult developmental stage transition, although RGA plays a more dominant role than GAI.
GAs are important for promoting flowering in Arabidopsis and are required for flower initiation in SD (Wilsonet al. 1992). We showed that rga partially rescues the flowering time defect of ga1-3, suggesting that RGA suppresses flowering (Silverstoneet al. 1997). In this study we scored flowering time in both days to flower and leaves to flower. The rga-24 and gai-t6 single mutants flowered slightly earlier than Ler (8.8 days and 6.5 leaves for rga-24 and 3.2 days and 3 leaves for gai-t6; Figure 3, A and B). In the GA-deficient background, both rga-24 and gai-t6 rescued the nonflowering defect of ga1-3 although rga-24/ga1-3 flowered 63.7 days (42%) and 23.2 leaves (30%) earlier than gai-t6/ga1-3. The flowering time of the trigenic rga-24/gai-t6/ga1-3 mutant was even earlier than Ler. This plant flowered 10 days (16%) earlier with 10 fewer leaves (30%) than Ler, which was comparable to rga-24/gai-t6 and rga-24.
The earlier flowering time of plants containing rga-24 or rga-24/gai-t6 alleles appears to correlate with a shorter juvenile stage of the plants (Figure 3A). In contrast, the length of the adult stage was similar in all lines that underwent the phase transition. Therefore, the main role of RGA and GAI in repressing flowering is by delaying the juvenile-to-adult transition.
Phenotypes of plants. All plants are 52 days old except rga-24/gai-t6, which is 37 days old.
Leaf expansion and apical dominance: GAs are known to promote leaf expansion. We examined the effect of rga-24 and gai-t6 on rosette size by measuring the maximum radius of each line. Figure 4A shows that rga-24 partially rescued the leaf expansion defect of ga1-3 whereas gai-t6 alone had no effect. The rosette radius of the trigenic mutant is similar to Ler.
Final heights of plants. The final height of 10 plants per line was measured and the means ±SE are shown. Some error bars are too small to be seen.
The greatly reduced apical dominance in ga1-3 is partially rescued by rga (Silverstoneet al. 1997). The axillary branch numbers of different mutant lines are shown in Figure 4B. The gai-t6 mutation alone in either a wild-type or GA-deficient background did not affect this phenotype at all. However, in combination with rga-24, apical dominance was restored to wild type (Figure 4B). These results show that RGA and GAI both function to repress GA-induced leaf expansion and apical dominance and that RGA plays a more dominant role than GAI in these processes.
Germination, fertility, and flower morphology: GAs are vital for both germination and flower development as reflected in the nongerminating, male-sterile phenotype of ga1-3. We showed previously that mutations at the RGA locus were unable to rescue these phenotypes. Therefore, we tested whether gai-t6 or rga/gai-t6 have any effect on these processes. Neither gai-t6 nor rga-24/ gai-t6 increases the germination percentage or restores the fertility of ga1-3 (Table 1). The digenic rga-24/ga1-3, gai-t6/ga1-3 (not shown) and trigenic rga-24/gai-t6/ga1-3 mutants all had male-sterile flowers with rudimentary petals, which are similar to those of ga1-3 (Figure 5B). These results suggest that RGA and GAI may not control seed germination or flower development. However, we noticed that the carpels of the rga-24/gai-t6/ga1-3 mutant flower are longer than those in ga1-3 flowers (Figure 5B). We also found that the digenic rga-24/gai-t6 mutant had much reduced fertility compared to that of wild-type Ler (Table 1). In fact, rga-24/gai-t6 had reduced pollen levels compared to Ler (not shown) and the stamen filaments were shorter than the carpels (Figure 5B). This developmental defect is probably due to an elevated level of GA signaling, which mimics the effect of GA overdose on flower development. Although GA is required for stamen development, an overdose of GA on wild-type Arabidopsis plants also results in reduced fertility (Jacobsen and Olszewski 1993). These data suggest that RGA and GAI may play only a minor role in flower development.
The effect of rga-24 and gai-t6 null alleles on flowering time in SD. (A) Number of juvenile, adult, and cauline leaves produced by the primary inflorescence stem of each genotype after bolting. Juvenile leaves do not have abaxial trichomes whereas adult leaves do. (B) Number of days from sowing until floral buds are clearly visible. The values plotted are the means ±SE of 10–20 plants. Some error bars are too small to be seen. indicates that ga1-3 produced only juvenile leaves and did not flower after 170 days. SE of ga1-3 in A represents the variation in total leaf number of plants on day 170.
Feedback control of GA biosynthesis: In addition to the above phenotypes, we also investigated the effect of the rga and/or gai mutations on the feedback regulation of GA biosynthesis, which is affected by the activity of the GA response pathway (Bethke and Jones 1998). The GA4 gene in Arabidopsis encodes GA 3β-hydroxylase, which catalyzes the conversion of GA precursors to bioactive GAs (Chianget al. 1995). The transcript level of this gene is upregulated in the GA-deficient background, and the elevated GA4 mRNA level can be reduced by application of GA (Chianget al. 1995; Cowlinget al. 1998; Silverstoneet al. 1998; Yamaguchiet al. 1998). In addition, the gain-of-function gai-1 mutant, which shows constitutively repressed GA signaling, has an elevated GA4 mRNA level compared to that in wild type (Cowlinget al. 1998). These results indicate that decreased GA-signaling results in increased levels of GA4 mRNA.
The effect of rga-24 and gai-t6 null alleles on rosette radius and apical dominance. (A) Rosette radius of SD-grown plants as labeled. (B) Number of axillary inflorescence stems. Means ±SE were measured for 10–20 plants per line. Some error bars are too small to be seen.
Germination percentage and fertility in wild type and homozygous GA biosynthesis or response mutants
Previously, we found that the digenic rga/ga1-3 mutant has an intermediate level of GA4 mRNA compared to Ler and ga1-3 (Silverstoneet al. 2001), probably because loss of RGA function resulted in partially derepressed GA signaling. On the basis of this hypothesis, we predicted that removing both RGA and GAI function in the ga1-3 background may lead to a completely derepressed GA response, and this, in turn, may lower GA4 gene expression even further. RNA gel blot analysis showed that the trigenic rga-24/gai-t6/ga1-3 mutant, in the absence of exogenously applied GA, accumulated a lower level of GA4 mRNA than that in Ler (Figure 6). This experiment was repeated four times and the average level of GA4 mRNA in each line is shown in Table 2. As with the phenotypes we studied, removing RGA function caused a partial reduction (by 25%) in the level of GA4 mRNA but removing GAI function alone did not. In fact, gai-t6 and gai-t6/ga1-3 had higher levels of GA4 mRNA than Ler and ga1-3, respectively. GA3 treatment lowered GA4 mRNA levels in all lines to approximately the same level (Figure 6 and Table 2).
Inflorescence and flowers of wild-type and mutant plants. Inflorescences (A) and individual flowers (B) of plants as labeled.
The levels of GA4 mRNA in rga and gai mutants in wild-type and ga1-3 backgrounds. Shown is an autoradiogram of RNA blots containing 9 μg of total RNA isolated from Ler and various mutants with (+GA3) or without (−GA3) treatment. The blot was hybridized with a labeled GA4 antisense RNA probe and then reprobed with a labeled 18S rDNA probe.
Relative GA4 mRNA levels in Ler and homozygous mutant lines
DISCUSSION
RGA and GAI are not completely redundant: Our data illustrate that RGA and GAI interact synergistically to repress a set of GA-induced growth processes. Phenotypes affected by rga and gai null mutations include leaf expansion, stem elongation, juvenile-to-adult phase change in leaf development, vegetative-to-reproductive transition, and apical dominance. Removing both RGA and GAI function leads to the complete derepression of the above phenotypes because the trigenic rga-24/gai-t6/ga1-3 mutant shows a wild-type or even GA-overdose phenotype. Although RGA and GAI interact synergistically in repressing GA signaling, RGA alone is a more active repressor than GAI by itself. This is evident from the observation that rga-24 alone partially rescues the above defects in ga1-3 whereas gai-t6 individually has little or no effect.
The rga-24 and gai-t6 single mutants did have subtle phenotypes when compared to wild-type Ler, further supporting that RGA and GAI do not have completely redundant functions. Both rga-24 and gai-t6 flowered slightly earlier than wild type, and rga-24 underwent the transition from juvenile-to-adult phase 46% earlier than Ler.
The effects of single and double rga-24 and gai-t6 mutations are more evident in the GA-deficient ga1-3 background than in Ler. This result supports the hypothesis that RGA and GAI are more active repressors of GA signaling in GA-deficient conditions than in wild-type background and that GA derepresses the GA-signaling pathway by inactivating RGA and GAI. In the wild-type GA background, however, RGA and GAI are still functional, although they have a lower activity in modulating GA signaling. This is evident by the subtle phenotypes associated with loss of RGA and/or GAI in wild-type GA background.
Are RGL genes involved in controlling seed germination or flower development? Neither rga-24 nor gai-t6 singly or in combination rescued the germination and flower development defects of ga1-3, suggesting that additional gene(s) must modulate these GA-regulated phenotypes. Candidates that may control these processes are RGL (for RGA-LIKE; Sanchez-Fernandezet al. 1998), RGA1-LIKE (GenBank accession no. AC009895), and RGA-LIKE PROTEIN (GenBank accession no. AL39-1150), all of which are DELLA domain-containing GRAS family members with 56–60% amino acid sequence identity to RGA and GAI. The current names of these RGA/GAI homologs are confusing, and we propose renaming these genes as RGL1, RGL2, and RGL3, respectively. Isolation of knockout mutants that are defective in these genes using a reverse genetic approach will help to test their role in GA-regulated germination and/or flower development. The function of the RGL genes may be better revealed in multiple mutant backgrounds because of functional redundancy. Also, their phenotype might be more visible in the ga1-3 background if RGLs, like RGA and GAI, are more active in the GA-deficient condition. If these genes do play a role in GA signaling, differential expression of their products in various tissues and/or varying the sensitivity to the GA signal for each gene would allow fine-tuning of the GA response.
Gene duplications and the RGA, GAI, and RGL genes: Recent sequence analysis has revealed a large number of gene duplications in the Arabidopsis genome (Visionet al. 2000). RGA and GAI, on chromosomes 1 and 2, are within duplicated block 10 that was estimated to have taken place ~100 million years ago (mya; age class C). RGL2 and RGL3 are located in duplicated block 71 in age class E (170 mya) on chromosomes 3 and 5, respectively. RGL1 (chromosome 1) is in duplicated block 37 in age class F (200 mya). But there is no paralogous gene in the other block 37 on chromosome 5. Our data illustrate that RGA and GAI have similar, but not completely redundant functions. After the duplication event, they clearly evolved to have slightly different roles in controlling GA signaling. This demonstrates that gene duplications could provide ways for the organisms to evolve more sophisticated regulatory mechanisms in controlling cellular processes. Future studies on the RGL genes will reveal whether RGL2 and RGL3 have more similar biological function to each other than to RGA, GAI, and RGL1.
The RGA/GAI orthologs in rice (SLR) and in barley (SLN) appear to function as the single major repressor for GA-mediated stem growth because recessive slr and sln mutations result in constitutive GA response in rice and barley, respectively (Lanahan and Ho 1988; Ikedaet al. 2001; P. M. Chandler, A. Marion-Poll, and F. Gubler, personal communication). These results suggest that the RGA/GAI orthologs are not redundant in rice and barley. The oldest gene duplication in Arabidopsis (age class F) identified by Vision et al. (2000) occurred at approximately the same time as the divergence of monocots and dicots. If future studies indicate that most dicots with diploid genomes contain multiple RGA/GAI orthologs, whereas monocots have only a single RGA/GAI ortholog, then it is possible that the first duplication event for RGA/GAI occurred after the divergence of monocots and dicots. However, subsequent loss of duplicated genes in monocots after divergence from dicots could also account for such results.
Feedback regulation of GA biosynthesis by GA response: Because RGA and GAI have overlapping functions, removing RGA alone only partially derepresses GA signaling. We showed that rga-24 partially reduced expression of the GA biosynthetic gene GA4. When both RGA and GAI were inactivated, the GA4 mRNA level in rga-24/gai-t6/ga1-3 was further decreased to be even lower than Ler. These results support the current model that increased activity in GA response can downregulate GA biosynthesis by a feedback mechanism. GA treatment only slightly decreased GA4 mRNA level in the trigenic rga-24/gai-t6/ga1-3 mutant, indicating that feedback inhibition was almost at its maximum. On the basis of the above model, we would expect that gai-t6 by itself may not affect GA4 gene expression because it has little effect in all GA-regulated events. However, we were surprised to find that gai-t6 alone caused increased GA4 mRNA levels in both wild-type and ga1-3 backgrounds. One possible explanation is that RGA and/or RGLs may be activated further and overly compensate for the loss of GAI function in repressing GA signaling, and consequently may affect the feedback control of GA biosynthesis.
Model of GA-signaling pathway in Arabidopsis. SPY inhibits all GA responses by activating RGA, GAI, and perhaps RGL1, RGL2, and RGL3, through GlcNAc modification. GA signal derepresses its signal transduction pathway by inactivating all repressor proteins. The activity of GA response regulates GA biosynthesis via a feedback mechanism. The extent of GA-mediated growth is determined by the balance between the amount of the GA signal and the levels and/or activities of the repressors. Arrows and T-bars indicate positive and inhibitory effects, respectively. Weaker vs. stronger effects are represented by dotted and solid lines, respectively. RGA is in boldface, indicating that RGA alone is a more active repressor than GAI by itself.
A working model of GA signal transduction pathway in Arabidopsis: In this article, we demonstrate that RGA and GAI synergistically repress a number of GA-controlled processes. Our previous studies, using the trigenic mutant rga/spy/ga1-3, showed that SPY additively interacts with RGA (Silverstoneet al. 1997). It was postulated that SPY may activate RGA and GAI by GlcNAc modification, because SPY is probably an OGT (reviewed in Harberdet al. 1998; Thorntonet al. 1999; Sun 2000). The ground state of the GA-signaling pathway is repressive, and growth and development occur through derepression of the pathway by the GA signal, which inactivates SPY, RGA, and GAI (Figure 7). To achieve differential growth, quantitative differences in the GA signal can determine the degree of derepression of the pathway in different cells. Therefore, a proper balance between the bioactive GA level and activity in GA signaling is important, and this is consistent with our finding that rga and gai affect not only GA response, but also regulation of GA biosynthesis.
Unlike the rga and gai null alleles, the spy mutation also completely restores seed germination and partially rescues the defect in flower development of ga1-3 (Jacobsen and Olszewski 1993; Silverstoneet al. 1997). Therefore, we propose that SPY may also activate RGL1, RGL2, or RGL3 to inhibit GA-induced seed germination and flower development (Figure 7). Future studies will be needed to verify this model and to place additional putative activators (e.g., SLEEPY, G-proteins, and GA-MYB) and a repressor (SHI) in the pathway.
Acknowledgments
We thank Nicholas Harberd for the gai-t6 seeds, Aron Silverstone, Christine Fleet, and Shinjiro Yamaguchi for technical advice and helpful discussions. We also thank Daphne Stam and Aron Silverstone for critical reading of the manuscript. This work was funded by National Science Foundation grants IBN-9723171 and IBN-0078003.
Footnotes
-
Communicating editor: V. L. Chandler
- Received February 28, 2001.
- Accepted July 12, 2001.
- Copyright © 2001 by the Genetics Society of America