- Copyright © 1999 American Society of Plant Physiologists
In his studies on the nature of inheritance, GregorMendel (1866) examined seven pairs of traits in pea, including one he called “difference in the length of the stem” (Fig. 1). The tall character dominated the dwarf, segregating three tall to one dwarf. When Mendel's work was rediscovered at the beginning of this century, Orland White introduced the term Le, for Length, to represent this trait and recognized that the Le/le pair of alleles controlled the presence or absence of a factor for tallness (White, 1917). As described below, this factor was later identified as GA, and dwarf peas that are homozygous for le became extremely important in establishing GAs as natural regulators of plant growth.
Isogenic pea seedlings (12 d old) containing wild-type (LE, right) or dwarfing (le/le, left) alleles of Mendel's height gene. They represent lines selected from an F7 segregant from a cross between line 58 (le, from Prof. Ian Murfet, University of Hobart, Australia) and I3 (Le, a selection from Alaska).
The term “gibberellin” was first used in 1935 to describe a substance produced by the fungus Gibberella fujikuroi that caused overgrowth symptoms in rice, which was termed the bakanae disease (Yabuta, 1935). This substance proved to be a mixture of GAs, with GA1 and GA3 being the active factors (Takahashi et al., 1955). The early work on GAs remained undiscovered outside of Japan until after World War II, when there was an explosion of interest among plant physiologists and horticulturists. P.W. Brian, working at the Imperial Chemical Industries' Akers Laboratories in the United Kingdom, induced dwarf pea plants to elongate as normal, tall plants by applying GA3, which he obtained from cultures of G. fujikuroi (Brian and Hemming, 1955). On the basis of these experiments, he proposed that GA was the natural growth factor that was deficient in the dwarf plants. This hypothesis was confirmed by Margaret Radley from the same laboratory, who reported that purified extracts from tall pea seedlings induced stem elongation in dwarf peas, thereby providing evidence that GAs were indeed naturally occurring in higher plants (Radley, 1956). Similar experiments were carried out in B.O. Phinney's laboratory at the University of California, Los Angeles, where dwarf maize was used as a bioassay to show that many plant extracts contained substances with GA-like activity (Phinney et al., 1957).
Nearly 30 years passed before the relationship between theLe alleles and GA content was confirmed on a firm chemical basis. During this time, the development of GC linked to MS facilitated the qualitative and quantitative analysis of GAs. Many GAs had been identified in plants, and their metabolic relationships (see below) were established using radiolabeled compounds. On the basis of the bioactivity of potential intermediates in the GA-biosynthetic pathway on GA-deficient dwarf mutants of maize, Phinney (1984) proposed that only GA1 was active in this species and that many of the other GAs were biosynthetic precursors of GA1 or were products of its catabolism. At approximately the same time, Ingram et al. (1984) demonstrated that thele pea mutant was defective in the conversion of GA20 to GA1 and suggested that the Le gene encodes the 3β-hydroxylase responsible for this conversion (see Fig. 2). This conclusion was supported by analyses of the GA content of tall (Le) and dwarf (le) pea; shoots of the latter contain approximately 10% of the GA1concentration in tall plants and have a highly elevated GA20 content (Ross et al., 1989). In 1997, in the final chapter of this story, the Le gene was cloned, allowing the genetic lesion in the mutant le gene to be identified and thus providing the molecular basis for Mendel's observed difference in stem length (Lester et al., 1997; Martin et al., 1997).
The GA-biosynthetic pathway from GGPP to the bioactive products GA4 and GA1, and their inactive catabolites formed by oxidation on C-2. Enzyme activities, with corresponding products of genes that are known sites of mutation and that have been cloned, are indicated. CPP, Copalyl diphosphate; KS,ent-kaurene synthase.
In this Update we will describe this and other recent successes in isolating genes involved in GA biosynthesis. These advances are providing new insights into GA biosynthesis and its regulation.
GA BIOSYNTHESIS
GAs are products of the diterpenoid pathway and their formation is initiated by cyclization of the common C20precursor GGPP. This intermediate is synthesized in plastids from isopentenyl diphosphate, shown recently to be formed in these organelles from glyceraldehyde 3phosphate and pyruvate rather than mevalonic acid, as was previously assumed (Lichtenthaler et al., 1997). However, it has not yet been demonstrated that GAs are products of the glyceraldehyde 3-phosphate/pyruvate pathway. In growing vegetative tissues, the cyclization of GGPP occurs in proplastids and results in the formation of a hydrocarbon, ent-kaurene, in a two-step process requiring the activity of two enzymes: CPS, which produces the intermediate copalyl diphosphate, and ent-kaurene synthase. GGPP is also the precursor of carotenoids and is incorporated into chlorophyll. These compounds are present at several orders of magnitude greater concentration than ent-kaurene and the GAs, so the first step on the GA pathway, catalyzed by CPS, needs to be tightly regulated.
ent-Kaurene is converted to the bioactive GAs by a series of oxidative reactions (Fig. 2) catalyzed by two types of enzyme. The early reactions, resulting in contraction of ring B from six to five C atoms to give the GA structure, occur on extraplastidic membranes. This requires the movement of ent-kaurene out of the plastid by an as-yet-unknown mechanism. The reactions are catalyzed by Cyt P450-dependent monooxygenases and, in the shoot tissues of most plants, give rise to GA12 and its 13-hydroxylated analog GA53. These intermediates are metabolized further by soluble dioxygenases, which use 2-oxoglutaric acid as a cosubstrate. Two dioxygenases are required to convert GA12 and GA53 by parallel pathways to the bioactive products GA4 and GA1, respectively. First, GA 20-oxidase converts C-20 from a methyl group to an aldehyde and then removes the C atom to form the characteristic γ-lactone of the C19GAs. Second, a hydroxyl group is introduced at the 3β position by a GA 3β-hydroxylase. A third 2-oxoglutarate-dependent dioxygenase that hydroxylates at the 2β position inactivates the GA molecule and thus ensures turnover of the active forms. The pathways (MacMillan, 1997) and biochemistry/molecular biology (Hedden and Kamiya, 1997;Lange, 1998) of GA biosynthesis have been reviewed recently elsewhere.
MUTANTS AND GENES
Approximately one-half of the genes involved in the biosynthesis of the biologically active GAs have now been isolated, using almost as many strategies as there are clones (Hedden and Kamiya, 1997; Lange, 1998). In some cases, such as the cloning of the Le gene (LE is now used) referred to above, the cloned loci are sites of known mutations that result in reduced GA content and dwarfism. Although GA-deficient mutants are known in many plant species, pea, Arabidopsis, and maize have been particularly well studied and, in each case, numerous loci for GA-biosynthesis genes have been identified. Several of these loci have now been cloned, confirming the nature of the encoded enzymes previously established by biochemical analysis. In Arabidopsis all five of the original mutant loci (ga1–ga5) that result in GA-responsive dwarfs (Koornneef and van der Veen, 1980) have now been cloned. In most cases growth of the Arabidopsis mutants was restored to normal by introducing the gene from wild-type plants, providing unequivocal evidence that the cloned gene corresponded to the mutant locus. All examples of the isolation of a previously known or suspected GA-biosynthesis gene are from pea, Arabidopsis, or maize.
Pea
Mutations in four loci, LS, LH,LE, and NA, are known to cause dwarfism and GA deficiency in pea. Biochemical evidence indicates that they encode, respectively, CPS (Ingram and Reid, 1987), ent-kaurene oxidase (Swain et al., 1997), GA 3β-hydroxylase (Ingram et al., 1984), and possibly GA12-aldehyde synthase (Ingram and Reid, 1987). In addition, the SLN locus controls both GA 2β hydroxylation and further oxidation at C-2 to form the so-called GA catabolites (see Fig. 2). It is not known if SLN encodes an enzyme that catalyzes both steps or if it controls a regulatory factor that modifies the activity of two enzymes (Ross et al., 1995). Two of these loci have now been cloned. A cDNA encoding CPS was isolated by screening a cDNA library with a DNA sequence obtained by PCR (Ait-Ali et al., 1997). The encoded protein had high homology with CPS from Arabidopsis (GA1) and from maize (An1). When the equivalent genomic sequences from the wild-type and the ls-1mutant were compared, a single base change was found at a splice site, such that ls-1 produced three incorrectly spliced mRNAs. Proteins encoded by these mutant mRNAs should all be truncated. However, ls-1 is not a null mutation, because a more severe allele, ls-3, has been identified (Reid et al., 1996). Thels-1 mutation reduces GA concentrations in shoots and in late-developing seeds to approximately 10% of those in wild-type plants, although it has a smaller effect on GA content during early seed development (Swain et al., 1995). Because the mutation does not affect flower formation or seed growth, it seems likely that other CPS genes are expressed in these organs. Although LS produces a transcript in young seeds (Ait-Ali et al., 1997), biochemical evidence suggests that the gene is expressed in the testa and not in the endosperm or embryo (Swain et al., 1995). However, in later seed development, LS is expressed in embryos. This complex pattern of gene expression during seed development has been noted with other GA-biosynthetic enzymes; different GA 20-oxidase genes are expressed in young and older seeds, corresponding to the biphasic pattern of GA production in developing seeds (Ait-Ali et al., 1997; Garcia-Martinez et al., 1997).
The LE locus was cloned simultaneously by two groups (Lester et al., 1997; Martin et al., 1997). After heterologous expression inEscherichia coli, it was shown to encode a GA 3β-hydroxylase, with a substrate preference for GA9 over GA20. Of the three mutant le alleles that are known, le-1 is the mutation described by Mendel. This mutation is due to a base substitution that introduces an amino acid change (Ala to Thr) close to an amino acid motif (His-Thr-Asp) that is known to bind Fe at the enzyme active site of the related dioxygenase, isopenicillinN synthase (Roach et al., 1995). The result is a decrease in enzyme activity that is at least in part due to a reduced affinity for the GA substrates (Martin et al., 1997). The le-2 mutation is caused by a base deletion that shifts the reading frame so that most of the protein is composed of a nonsense amino acid sequence. As might be anticipated, the mutant protein is completely inactive. The third mutation, le-3, causes an amino acid change (His to Tyr) that results in reduced enzyme activity. As expected, thele-2 mutation is the most severe of the three, but plants containing this mutation still possess small amounts of GA1 (Ross et al., 1989), indicating that other GA 3β-hydroxylase genes are expressed in pea. Indeed, the lemutations affect primarily stem extension and have no influence on flower development or seed and pod growth, which must rely on other 3β-hydroxylase genes for their source of GA.
Arabidopsis
Biochemical analysis of the mutants indicated that theGA1 and GA2 loci of Arabidopsis were involved in the conversion of GGPP to ent-kaurene (Zeevaart and Talon, 1992). Nine mutant ga1 alleles were isolated, one of which,ga1–3, contained a 5-kb deletion (Koornneef et al., 1983).Sun et al. (1992) took advantage of this deletion to clone theGA1 locus by genomic subtraction and, by expressing its cDNA in E. coli (Sun and Kamiya, 1994), showed that it encodes CPS. The predicted 93-kD protein contains a 50-amino acid N-terminal transit sequence that targets the protein to plastids. Sun and Kamiya, (1994) demonstrated that the encoded protein is indeed imported into chloroplasts and that the transit sequence is cleaved on entry into the plastid.
Koornneef and van der Veen (1980) recognized two types ofga1 mutants: those in which the seeds germinated to give male-sterile dwarf seedlings, and others that did not germinate unless treated with GA. The latter correspond to essentially null mutations with no active CPS produced from the GA1 gene and would be very highly GA deficient, whereas the former may have a functional enzyme with reduced activity. Mutants containing the severega1 alleles remain as rosettes unless treated with GA. When grown in a long-day photoperiod, they produce flower buds, although in the absence of GA the buds do not develop into viable flowers. In short days they do not produce flowers (Wilson et al., 1992). Despite the apparent absence of a functional GA1 product,ga1–3 plants contain low amounts of GAs (Talon et al., 1990), the formation of which would require another CPS-encoding gene expressed at very low levels in shoot and floral tissues, or the presence of a different terpene cyclase producing copalyl diphosphate or even ent-kaurene as a minor by-product.
The GA2 locus was cloned by heterologous screening with a cDNA from a pumpkin ent-kaurene synthase gene (Yamaguchi et al., 1998). The pumpkin cDNA had been obtained after purification of the enzyme from endosperm and the use of PCR with primers designed from partial amino acid sequence information (Yamaguchi et al., 1996). The function of the proteins encoded by the pumpkin cDNA and by cDNA corresponding to GA2 was confirmed by heterologous expression in E. coli, demonstrating that the expression products converted [3H]copalyl diphosphate toent-[3H]kaurene (Yamaguchi et al., 1996, 1998). The ga2–1 mutant, which has a severe (nongerminating) GA-deficient phenotype, contains a base substitution at the GA2 locus that results in the introduction of a premature stop codon and a highly truncated gene product (Yamaguchi et al., 1998). The full-length GA2 cDNA encodes a 90-kD protein with 70% similarity (52% identity) to the pumpkinent-kaurene synthase. Both proteins contain a potential N-terminal transit sequence for import into plastids, although such import has not been demonstrated.
On the basis of northern-blot analysis, GA2 is expressed in all tissues at relatively high levels (Yamaguchi et al., 1998). In contrast, expression of GA1 is at a much lower level and could be detected only by reverse-transcriptase PCR or the use of a GUS reporter gene driven by the GA1 promoter (Silverstone et al., 1997). Examination of transgenic Arabidopsis plants containing this reporter gene construct revealed that GA1 promoter activity was highest in rapidly growing tissues or in the vascular tissues of nongrowing organs, such as leaves. Thus, GA1(CPS) is a tightly regulated gene, which is consistent with the role of CPS as the first enzyme in the GA-biosynthetic pathway. The cell types that express GA1, i.e. dividing cells or cells of the vascular system, indicate that CPS is active in proplastids rather than in mature chloroplasts, as also shown by biochemical studies (Aach et al., 1997).
Helliwell et al. (1998) have recently used a combination of positional cloning and random sequencing of a bacterial artificial chromosome to identify a putative Cyt P450 gene that maps to the ArabidopsisGA3 locus. They also provided convincing evidence, based on the accumulation of ent-kaurene and the lack of a growth response of the ga3 mutant to appliedent-kaurene, that the mutant lacks ent-kaurene oxidase activity. Thus, GA3 encodes ent-kaurene oxidase, which is a Cyt P450-dependent monooxygenase. GA3contains six introns and an open reading frame of 1678 bp, encoding a protein of 58.1 kD. Two transcripts were detected due to alternative splicing sites at the downstream intron 6-exon boundary. The enzyme belongs to a new class of Cyt P450 and thus differs from the previously cloned Dwarf-3 gene of maize, which is also a Cyt P450 (Winkler and Helentjaris, 1995). Although Dwarf-3 is assumed to be involved in GA biosynthesis (Phinney, 1984), its function is unknown. Two mutant ga3 alleles were sequenced by Helliwell et al. (1998); both alleles contain single base substitutions that introduce in-frame stop codons.
Analysis of the GA content of the ga4 and ga5mutants indicated that they were defective in GA 3β-hydroxylase and GA 20-oxidase activity, respectively (Talon et al., 1990). This was confirmed by the cloning of these loci. The GA4 locus was isolated after it was tagged by a chance T-DNA insertion (Chiang et al., 1995). The gene contained an open reading frame with a single intron and encoded an enzyme of 40.2 kD. Based on its amino acid sequence, this enzyme is believed to belong to a group of dioxygenases, most of which use 2-oxoglutaric acid as a cosubstrate. By obtaining expression of its cDNA in E. coli and demonstrating that the recombinant protein converted GA9 to GA4 and GA20 to GA1, Williams et al. (1998) confirmed thatGA4 encodes a 2-oxoglutarate-dependent dioxygenase with GA 3β-hydroxylase activity. GA9 was the preferred substrate, with a K m value approximately 10-fold lower than that for GA20.
The GA5 locus was cloned by a strategy similar to that used for the isolation of GA2 described above. Xu et al. (1995)obtained it from a genomic library by screening with a pumpkin GA 20-oxidase cDNA that had been isolated using antibodies raised against the purified enzyme. The Arabidopsis gene contains two introns and an open reading frame of 1131 bp, encoding a 377-amino acid protein of 43.4 kD. Its function as a GA 20-oxidase was confirmed by demonstrating the ability of a fusion protein, obtained by expressing a cDNA clone inE. coli, to convert GA53 to GA44 and GA19, and GA19 to GA20. Xu et al. (1995) obtained proof of the identity of the genomic clone by mapping it to the GA5 locus and establishing that the homologous gene from the ga5 mutant contained a base substitution, relative to the wild-type Arabidopsis ecotype (Landsberg erecta) gene, that resulted in the introduction of a premature stop codon. At approximately the same time thatGA5 was cloned, Phillips et al. (1995) isolated two GA 20-oxidase cDNA clones from the ga1–2 mutant of Arabidopsis and discovered a third 20-oxidase sequence in a database of cDNA sequences. The three cDNAs encoded functionally similar enzymes, which converted GA12 to GA9 and GA53 to GA20, with GA12 being the preferred substrate. Phillips et al. (1995) showed that the cDNAs corresponded to genes that exhibited different patterns of expression; one gene, identical to GA5, was expressed in stems, another in the inflorescence and silique (fruit), and the third only in siliques.
In contrast to GA1, GA2, and GA3, there are no known mutant alleles of GA4 and GA5with the severe (nongerminating) phenotype. Thega4 and ga5 mutations are “leaky” semidwarfs that produce fertile flowers and normal siliques. The original ga4–1 allele (Koornneef and van der Veen, 1980) contains a base substitution that results in a change of Cys to Tyr (Chiang et al., 1995), which may not abolish enzyme activity completely. However, the ga4–2 mutant with the T-DNA insertion is unlikely to contain an active GA4 protein, although it is also a semidwarf. Furthermore, the ga5 mutant contains 10% to 30% of the C19-GA content of wild-type plants (Talon et al., 1990), despite apparently producing a truncated enzyme. It is now clear that several GA 20-oxidase genes are expressed in Arabidopsis and other species (Hedden and Kamiya, 1997); therefore, the loss of one enzyme can be partially compensated for by the activity of the others, perhaps via movement of GAs or their precursors between tissues. It seems likely that the GA 3β-hydroxylases are also encoded by multiple genes, whereas CPS, ent-kaurene synthase, andent-kaurene oxidase are predominantly the products of single genes in Arabidopsis, namely GA1, GA2, andGA3, respectively.
REGULATION OF GA BIOSYNTHESIS
GAs mediate many developmental and environmental responses in plants. Consequently, regulation of GA biosynthesis is necessarily complex. The isolation of GA-biosynthetic genes has enabled direct analysis of their expression in terms of transcript abundance, allowing new insights into regulatory mechanisms.
One such mechanism that plants use to maintain GA homeostasis (i.e. keeping the concentrations of bioactive GAs within certain limits) involves feedback regulation of GA biosynthesis, as illustrated in Figure 3. This effect was first observed in GA-response mutants, such as Rht3 in wheat,Dwarf8 in maize, and gai in Arabidopsis. Although they are dwarf mutants, they contain large amounts of bioactive GAs (for review, see Scott, 1990; Hedden and Kamiya, 1997). In contrast, slender mutants such as the la cry s mutant of pea grow as if they were treated with large quantities of GA; however, they actually contain reduced amounts of active GAs (Martin et al., 1996). Although these relationships initially seemed paradoxical (Scott, 1990) because stem elongation is normally controlled by the content of bioactive GA, they revealed a link between GA response and biosynthesis. Dwarf mutants unable to respond to GA were also unable to down-regulate GA biosynthesis, whereas slender mutants down-regulated GA biosynthesis strongly. Current evidence suggests that feedback regulation modifies expression of GA 20-oxidase and 3β-hydroxylase genes. Thus, transcript levels of the Arabidopsis GA5 and GA4 genes were highly elevated in GA-deficient plants and reduced when such plants were treated with GA (Chiang et al., 1995; Phillips et al., 1995; Xu et al., 1995). In addition, the gai response mutant, which accumulates GAs, also contained increased 20-oxidase transcript levels (Xu et al., 1995). Conversely, leaves of the slender la cry s mutant of pea contain reduced amounts of GA 20-oxidase mRNA relative to wild-type plants, which is consistent with a reduced GA1 content (Martin et al., 1996). Treatment of wild-type peas with bioactive GA also down-regulated the GA 20-oxidase message and GA1content, thereby mimicking the effect of la cry s (Martin et al., 1996). By comparing the activities of GA analogs, Cowling et al. (1998) demonstrated a close link between GA-induced growth and the down-regulation ofGA4 (GA 3β-hydroxylase) expression in Arabidopsis.
Scheme illustrating feedback regulation of GA biosynthesis. Expression of genes encoding GA 20-oxidase and 3β-hydroxylase is down-regulated by bioactive GAs. The semidominant mutations, Rht (wheat), Dwarf8 andDwarf9 (maize), and gai (Arabidopsis) that cause a reduction in GA responsiveness also reduce feedback regulation of GA biosynthesis.
Induction by photoperiod of the rapid stem elongation (bolting) that accompanies flowering in many rosette plants is mediated by GAs. Several studies indicate that 20-oxidation is the main step controlled by photoperiod in such plants (for review, see Hedden and Kamiya, 1997). Regulation of GA 20-oxidase gene expression by photoperiod was directly demonstrated in spinach, which is an obligate long-day plant (Wu et al., 1996). Higher expression of this gene in long days correlated with increased stem elongation, whereas expression was lower in short days, in which plants maintained rosettes. This pattern of 20-oxidase expression was established within 2 d after a change in photoperiod. In Arabidopsis, a facultative long-day plant, bolting is accelerated in long days and correlated with a slightly higher GA content and increased sensitivity to GAs (Xu et al., 1997). Furthermore, Xu et al. (1997) found that exposure to long days resulted in an accumulation of GA5 (20-oxidase) transcript but did not affect expression of GA4 (3β-hydroxylase).
CONCLUSIONS
Cloning the genes of the GA-biosynthetic pathway is opening a wide range of research opportunities in plant biology. ThisUpdate highlights the rapid progress in exciting but preliminary studies of how GA mediates developmental and environmental effects on growth. This progress is also providing new information on the sites of GA biosynthesis and on the function and structures of the biosynthetic enzymes, allowing genetic manipulation of specific steps in the pathways of transgenic plants. Furthermore, it has finally been possible to establish the molecular basis for many of the mutations that affect plant stature, some of which have been known for many years and are an important component of several crop varieties.
Footnotes
-
1 IACR receives grant-aided support from the Biotechnology and Biological Sciences Research Council of the United Kingdom. This paper is Oregon Agricultural Experiment Station technical paper no. 11,412.
-
* Corresponding author; e-mail peter.hedden{at}bbsrc.ac.uk; fax 44–1275–394281.
Abbreviations:
- CPS
- copalyl diphosphate synthase
- GGPP
- geranylgeranyl diphosphate
- Received September 29, 1998.
- Accepted October 28, 1998.