Introduction

The fungus Fusarium fujikuroi has played a crucial role in gibberellin (GA) research for more than 100 years. GAs were first identified as secondary products of the rice pathogenic fungus causing overgrowth symptoms (“bakanae” disease) in rice seedlings (see Phinney 1983 for review). The compound GA3 was first isolated by Japanese scientists (Cross et al. 1959). The ability of GA3 to restore normal growth of plant dwarf mutants (Lang 1956), and the occurrence of GA3-like substances in higher plants (Radley 1956) prompted the suggestion that GAs are natural plant hormones regulating growth and development in higher plants.

The biosynthesis of GAs has been investigated for many years—as soon as the structure of GA3 was known—both in the fungus and in higher plants. Birch and coworkers determined incorporation patterns from 14C-labelled acetate and mevalonate into GA3 in fungal cultures and concluded that GAs were diterpenoid compounds (Birch et al. 1958, 1959). At that time, several GA-deficient mutants of F. fujikuroi were available in which GA biosynthesis is blocked at different points of the pathway. Such mutants played an important role in the discovery of major steps of GA biosynthesis. Potential intermediates were isolated from cultures of these mutants and used to demonstrate their incorporation into GA3 without the requirement for isotopically labelled substrates. The most famous mutant obtained by UV mutagenesis, B1-41a, has a genetic block in ent-kaurene oxidation (Bearder et al. 1974). Other mutants, such as R9 (Bearder 1983) and SG123 and SG133 (Avalos et al. 1997, 1999), have a defect in C-13 hydroxylation and secrete mainly GA7 instead of GA3. These feeding experiments allowed determination of the basic biosynthetic pathway to GA3 in the 1960s and early 1970s. The development of combined gas chromatography-mass spectrometry (GC-MS) for GA analysis at about this time was an essential pre-requisite for identification of intermediates and products of fungal and plant GA biosynthesis (Binks et al. 1969; MacMillan et al. 1967).

GA research in higher plants profited from successful experiments with cultures of F. fujikuroi, but advanced more slowly. The plant biosynthetic pathway was established mostly by using cell-free systems from developing seeds, particularly those from Marah macrocarpus (Graebe et al. 1965), Cucurbita maxima (pumpkin) (Graebe et al. 1974), and Pisum sativum (garden pea) (Kamiya and Graebe 1983). This work enabled the determination of GA pathways in these organisms, characterisation of the enzymes involved and, finally, cloning of the corresponding genes in the 1990s (reviewed in MacMillan 1997). Almost all GA biosynthetic genes in Arabidopsis thaliana have now been identified following the sequencing of the Arabidopsis genome. However, despite the initial identification of GAs, and determination of the main biosynthetic steps first in the fungus F. fujikuroi, there had been little advance in the fungal system for almost 20 years. The breakthrough came with the discovery that the genes for the GA biosynthetic pathway were clustered on the F. fujikuroi genome (Tudzynski and Hölter 1998). Since then, all pathway genes have been cloned, and the character of the corresponding enzymes has been identified. On the basis of recent data from the fungus and from higher plants, it is now possible to compare GA biosynthetic genes and genome organisation, the character of the encoded enzymes, pathway steps, and regulation of gene expression in both systems. The profound differences between F. fujikuroi and higher plants on the one hand, and the widespread occurrence of GAs in plants, different fungi, and some bacteria on the other, raise questions regarding the evolutionary origin of GA biosynthetic genes.

This article updates a previous review (Tudzynski 1999), and concentrates on recent progress in characterisation of the fungal GA pathway, regulation of gene expression, and evolutionary aspects of GA biosynthetic genes.

The G. fujikuroi species complex and GA production capability

Fusarium fujikuroi belongs to the Gibberella fujikuroi (Sawada) species complex, which contains species from the Fusarium sections Liseola and Elegans. This species complex is composed of at least nine reproductively isolated biological species (mating populations) denoted by the letters A through I, and 32 additional asexual species (Britz et al. 1999; Kerényi et al. 1999; Leslie 1995; Marasas 2001; Nirenberg and O’Donnell 1998; Nirenberg et al. 1998; O’Donnell et al. 1998; Zeller et al. 2003). The elevation of the mating populations to distinct species was supported by DNA sequence analysis of multiple unlinked loci (Nirenber and O’Donnell 1998; Nirenberg et al. 1998; O’Donnell et al. 1998; Steenkamp et al. 1999; Yun et al. 2000).

All these Fusarium species are important fungal pathogens of various crops such as maize, rice, barley, sugarcane, pine, mango, pineapple, sorghum, and many more (Leslie and Plattner 1991; Leslie 1995, 1996; Leslie et al. 2004). Moreover, these species differ in their ability to produce secondary metabolites such as fumonisins (Desjardins et al. 1995, 2000; Kedera et al. 1999; Leslie et al. 1992; Proctor et al. 1999), fusaric acid (Bacon et al. 1996); beauvericin (Logrieco et al. 1998; Torres et al. 2001), moniliformin (Leslie et al. 1996; Desjardins et al. 2000), fusarins (Wiebe and Bjeldanes 1981; Song et al. 2004), the pigments neurosporaxanthin and bikaverin (Linnemannstöns et al. 2002a,b), and GAs (El-Bahrawi 1977; Tudzynski and Hölter 1998; Desjardins et al. 2000).

So far, the ability to produce GAs and to cause “bakanae” disease has been confirmed only for rice isolates belonging to the species F. fujikuroi (sexual stage: G. fujikuroi MP-C). However, some strains of MP-A (anamorph F. verticillioides) and MP-D (anamorph F. proliferatum) have also been isolated from rice in different geographic regions (Desjardins et al. 1997). It is not yet clear if these Fusarium species are also associated with the symptoms of “bakanae” disease, or are present only as saprophytes. Only now, since the genes in the GA biosynthetic pathway have been cloned and characterized from F. fujikuroi (G. fujikuroi MP-C), can we determine if members of the other MPs of the G. fujikuroi species complex also contain these genes, and if they are organised in a similar gene cluster. Recently, it was found that all MPs of the G. fujikuroi species complex, except for F. verticillioides (MP-A) and F. circinatum (MP-H), contain the entire GA biosynthetic gene cluster but are not able to produce GAs due to several mutations in the coding and non-coding regions of the GA biosynthetic genes (S. Malonek and B. Tudzynski, unpublished). Experiments aimed at restoring GA production capability in F. verticillioides and F. proliferatum by complementing the mutated or missing pathway genes with the corresponding genes from F. fujikuroi are underway.

The GA biosynthetic pathway in F. fujikuroi

The GA biosynthetic pathways in plants and F. fujikuroi have been summarised in many reviews, most recently by MacMillan (1997), Tudzynski (1999), and Hedden et al. (2001).

GAs, like other diterpenoids, are produced from hydroxymethylglutaryl (HMG) coenzyme A via mevalonic acid, isopentenyl diphosphate, geranyldiphosphate (GDP), farnesyl diphosphate (FDP) and geranylgeranyl diphosphate (GGDP), which is a precursor not only for GAs, but also for the carotenoid neurosporaxanthin (Domenech et al. 1996; Linnemannstöns et al. 2002a) and ubiquinones. Ent-kaurene, the first GA-specific intermediate, is produced in two cyclisation steps from GGDP via ent-copalyl diphosphate (CPP). Sequential oxidation of ent-kaurene at C-19 via ent-kaurenol and ent-kaurenal yields ent-kaurenoic acid, which is further oxidised to ent-7α-hydroxykaurenoic acid. A final oxidation at C-6β, resulting in contraction of ring B, leads to formation of GA12-aldehyde (Fig. 1). These first steps of the pathway are identical in the fungus and in higher plants. After GA12-aldehyde, the pathways in higher plants and F. fujikuroi differ. In F. fujikuroi, GA12-aldehyde is first 3β-hydroxylated to GA14-aldehyde, which is then oxidized at C-7 to form GA14 (Hedden et al. 1974; Urrutia et al. 2001). GA14 is then converted to the 19-carbon gibberellin GA4 by 20-oxidation. GA4, the first biologically active GA, is desaturated to GA7, which is then converted to GA3 by late 13-hydroxylation. GA1 is formed in a minor side reaction by 13-hydroxylation of GA4 (Fig. 1).

Fig. 1
figure 1

Gibberellin (GA)-biosynthetic pathways in Fusarium fujikuroi, indicating reactions of the main (3β-hydroxylated) and the minor (non-hydroxylated) pathways as well as side reactions catalysed by the enzyme P450-1. All reactions affected in the cpr-Gf mutant and dependent on cytochrome P450 oxidoreductase (CPR) are marked. Dotted arrows Proposed sequence of reactions. GGPP Geranylgeranyl diphosphate, CPPent-copalyl diphosphate

In plants, GA12-aldehyde is converted to GA12, which is either oxidised at C-20 to form the 19-carbon gibberellin, GA9, or is first 13-hydroxylated to GA53, which is then oxidized at C-20 to yield GA20 (Fig. 2). GA9 and GA20 are formed in parallel pathways, both involving oxidation of C-20 to alcohol and aldehyde, and the final formation of biological active 19-carbon GAs by loss of C-20. In plants, the oxidation and removal of C-20 is catalysed by a multifunctional GA 20-oxidase (Lange et al. 1994; Phillips et al. 1995). At the end of the pathway, both GA9 and GA20 are converted to GA4 and GA1, respectively, by introduction of a 3β-hydroxyl group.

Fig. 2
figure 2

Biosynthetic pathways to GA1 and GA3 in higher plants and the fungus F. fujikuroi. Higher plant enzymes and reactions are shown on the right, fungal enzymes on the left (modified from Hedden et al. 2001)

Thus, a major difference between the GA pathways in F. fujikuroi and plants is the stage at which the hydroxyl groups are introduced. In the fungus, GA12-aldehyde is 3β-hydroxylated to GA14-aldehyde, whereas in plants GA12-aldehyde is converted to GA12, which is then 13-hydroxylated to GA53 (Fig. 2). In fungi, 13-hydroxylation takes place only in the final step to form GA3 from GA7, whereas in plants the final step is the 3β-hydroxylation of GA9 and GA20 to GA4 and GA1, respectively (Fig. 2).

GA biosynthetic genes in F. fujikuroi

The main isoprenoid genes and enzymes involved in GA biosynthesis are highly similar in plants and the fungus. In both systems mevalonic acid is converted to GGDP via HMG, GDP and FDP. The corresponding genes are highly conserved in plants, animals and fungi, and were cloned easily from F. fujikuroi (Table 1). All three genes, encoding HMG-CoA reductase (hmg; Woitek et al. 1997), FDP synthase (fpps; Homann et al. 1996), and GGDP synthase (ggs1; Mende et al. 1997), are constitutively expressed: light, pH, and the concentration of nitrogen or carbon sources do not affect their transcription level (Mende et al. 1997, Woitek et al. 1997, Homann et al. 1996).

Table 1 Genes of the gibberellin (GA) biosynthetic gene cluster and adjacent genes: function and regulation. HMG-CoA Hydroxymethylglutaryl coenzyme A, GDPP geranylgeranyl diphosphate

In contrast to the central isoprenoid pathway genes, the genes encoding enzymes that catalyse later steps in the pathway have only a weak similarity to those in plants. Therefore, only one gene encoding the ent-copalyldiphosphate/ent-kaurene synthase CPS/KS has been cloned by PCR on the basis of sequence data from plant genes (Tudzynski et al. 1998). The homologous cps/ks gene had already been cloned previously from another GA-producing fungus, Phaeosphaeria sp. L487 (Kawaide et al. 1997).

The next pathway-specific gene of F. fujikuroi, P450-1, was cloned via a differential cDNA screening approach under conditions of nitrogen starvation/nitrogen excess. A second P450 monooxygenase gene, P450-2, has been identified on the same genomic fragment (Tudzynski and Hölter 1998), suggesting the possibility of clustering of some or all GA pathway genes. It is well known that genes involved in biosynthesis of secondary metabolites, such as aflatoxins and penicillin, are often organised in gene clusters in filamentous fungi (Keller and Hohn 1997). In order to identify more potential GA biosynthetic genes upstream and downstream of P450-1 and P450-2, a chromosome walking approach was used. Altogether, a genomic region of about 40 kb was cloned and sequenced in F. fujikuroi m567 (Tudzynski and Hölter 1998; Voss et al. 2001; Tudzynski et al. 2003; B. Tudzynski, unpublished results). An overview of all genes cloned is given in Table 1 and Fig. 3.

Fig. 3
figure 3

The GA biosynthetic gene cluster in F. fujikuroi. The arrows indicate the direction of transcription

The next gene downstream of P450-2 is a gene with 60% sequence identity to the F. fujikuroi GGDP synthase-encoding gene ggs1, which is involved in the central isoprenoid pathway (Mende et al. 1997). It has been suggested that ggs2 in F. fujikuroi might encode a GGDP synthase specifically responsible for supplying GGDP for GA production. The gene ggs2 is transcribed from a bidirectional promoter in the opposite direction from the already identified cps/ks gene encoding the key enzyme of the GA pathway.

The conversion of GGDP to ent-kaurene requires two cyclisation reactions. In contrast to higher plants, in which these two steps are catalysed by two enzymes, CPS and KS, the two-step cyclisation of GGDP is catalysed by only one bifunctional enzyme, the so called “fungal-type” CPS enzyme, in both Phaeosphaeria and F. fujikuroi (Kawaide et al. 1997; Tudzynski et al. 1998).

A third P450 monooxygenase gene, P450-3, two open reading frames without sequence similarity to known genes (orf1 and orf2), and a gene encoding a membrane transporter (mfs) are located downstream of the cps/ks gene (Hölter and Tudzynski 1998; B. Tudzynski, unpublished results) (Table 1, Fig. 3).

Upstream of P450-1, a fourth P450 monooxygenase gene (P450-4), a gene without any similarity to other fungal genes (orf3, later named des), and genes encoding a sugar membrane transporter (smt, Voß et al. 2001), an ankyrine domain-containing enzyme (ank, T. Voss and B. Tudzynski, unpublished) as well as an aldehyde (ald-DH) and an alcohol dehydrogenase (alc-DH) genes (T. Voss and B. Tudzynski, unpublished) have been found.

Functional characterisation of genes involved in GA biosynthesis

Disruption of the F. fujikuroi cps/ks gene results in total loss of GA production, confirming the prediction that this gene encodes the key enzyme of the GA biosynthetic pathway, the bifunctional CPS/KS (Tudzynski et al. 1998). In order to find out if the genes upstream and downstream of cps/ks also encode GA biosynthetic enzymes, all genes in the cluster have been disrupted and the intermediates that accumulate in the resulting mutants have been identified.

The P450-1 and P450-4 gene complex

Two of the four P450 monooxygenase genes, P450-1 and P450-4, are closely linked in the gene cluster, sharing the same promoter and being transcribed in opposite directions (Fig. 3). The corresponding deduced proteins have only 33% amino acid identity (Rojas et al. 2001; Tudzynski et al. 2001). Disruption of P450-1 and P450-4 resulted in strains producing no GAs. ΔP450-4 mutants accumulate ent-kaurene as the only intermediate in the GA biosynthetic pathway. After feeding 14C-labelled ent-kaurene, ent-kaurenol, ent-kaurenal, and ent-kaurenoic acid to cultures of ΔP450-4 mutants, only ent-kaurenoic acid was efficiently converted to GAs (Tudzynski et al. 2001). On the other hand, ΔP450-1 mutants accumulate ent-kaurenoic acid, but no later intermediates. To define precisely the metabolic steps catalysed by P450-4 and P450-1, wild-type gene copies were transformed into the GA-deficient mutant strain SG139. This mutant was first described as a regulatory mutant as none of the expected intermediates were detected (Barrero et al. 1999). However, it was later shown by Southern blot and PCR-analysis, that this mutant strain completely lacks the entire GA gene cluster (B. Tudzynski, unpublished), making this strain an excellent tool for functional analysis of single GA genes. Both the P450-1 and P450-4 genes were highly expressed in SG139 under nitrogen starvation conditions. SG139, which failed to metabolise ent-[14C]kaurene and ent-[14C]kaurenoic acid, converted these substrates after transformation with the P450-4 gene (Tudzynski et al. 2001), and metabolised labelled ent-kaurenoic acid, ent-7α-hydroxykaurenoic acid, GA12-aldehyde and GA12 to the major product [14C]GA14 after transformation of P450-1 (Rojas et al. 2001). These results clearly demonstrate that both P450-1 and P450-4 are multifunctional monooxygenases catalysing several steps in the GA pathway. P450-4 encodes ent-kaurene oxidase, catalysing the oxidation of ent-kaurene to ent-kaurenoic acid via ent-kaurenol and ent-kaurenal (see Fig. 1).

P450-1 was shown not only to catalyse the 7β-hydroxylation of ent-kaurenoic acid, but also to convert ent-kaurenoic acid to GA14, thus catalysing four steps in the GA pathway: hydroxylation at C-7, oxidation at C-6 resulting in contraction of ring B, 3β-hydroxylation, and finally oxidation at C-7 (see Fig. 1). Intermediates in this sequence, ent-7β-hydroxykaurenoic acid and GA12-aldehyde, are converted by this enzyme predominantly to GA14. Furthermore, it has been shown that P450-1, as well as catalysing multiple reactions in the GA biosynthetic pathway, may be involved in two branch pathways—from ent-kaurenoic acid to the kaurenolides and from ent-7β-hydroxykaurenoic acid to seco-ring B products (Rojas et al. 2001, 2004) (see Fig. 1).

Sequential oxidations at a single carbon atom as, for example, the three-step oxidation of ent-kaurene to ent-kaurenoic acid by P450-4 (Tudzynski et al. 2001), or the loss of C-14 in steroid biosynthesis (Lamb et al. 1990), have been described. However, the reactions catalysed by P450-1 involve oxidations at four different carbon atoms: C-6, C-7, C-3, and C-18, most of which are nonadjacent. Prior to the description of P450-1, the only previously described examples of a single P450 monooxygenase catalysing oxidations of multiple carbons were a mammalian corticoid oxidase (Sun et al. 1995) and a cortisol hydroxylase of the fungus Curvularia lunata (Suzuki et al. 1993). Interestingly, in plants the oxidation of ent-kaurene to ent-kaurenoic acid, and the conversion of ent-kaurenoic acid to GA12, are also catalysed by P450 monooxygenases (Helliwell et al. 1998, 2001). However, 3β-hydroxylation, which is one of the activities of P450-1 in F. fujikuroi, is catalysed by 2-oxoglutarate-dependent dioxygenases in plants and occurs late in the pathway (Hedden et al. 2001).

The ggs2-cps/ks complex

cps/ks was the first gene shown to be involved in the GA pathway in F. fujikuroi (Tudzynski et al. 1998). Adjacent to cps/ks, a second F. fujikuroi GGDP synthase-encoding gene, ggs2, has been found (Fig. 3) (Tudzynski and Hölter 1998). At that time, the two-step cyclisation of GGDP to ent-kaurene had been defined as the first GA pathway-specific step. However, deletion of ggs2 revealed a total block of GA biosynthesis, demonstrating that GGS2 catalyses formation of GGDP specifically for GA biosynthesis (B. Tudzynski, unpublished). The close linkage of ggs2 and cps/ks—they share the same promoter—hints at strict co-regulation of both genes catalysing early steps in the GA pathway leading to ent-kaurene. Interestingly, the recently identified carotenoid biosynthesis gene cluster in F. fujikuroi does not contain a pathway-specific ggs gene (Linnemannstöns et al. 2002a), whereas the paxillin gene cluster in Penicillium paxilli contains a second copy of GGDP synthase (paxG) besides the weakly expressed gene ggs1 outside the gene cluster (Young et al. 2001). Deletion of the pathway-specific ggs gene, paxG, resulted in total loss of paxillin formation. These results demonstrate that the ggs1 homologues in P. paxilli and F. fujikuroi are unable to compensate for loss of function of paxG and ggs2, respectively. The data provide genetic support for the suggestion of cellular partitioning and different metabolic functions of both GGS1 enzymes (primary metabolism) and PaxG and GGS2 (secondary metabolism) in the two fungi. The concept of metabolic channels had already been proposed for isoprenoid biosynthesis in plants (Chappel 1995).

P450-2

P450-2 is located in the GA biosynthetic gene cluster between P450-1 and ggs2 (Fig. 3). Disruption of P450-2 results in mutants that produce only GA14 according to TLC, HPLC, and GC-MS analysis (Tudzynski et al. 2002). In contrast, wild-type F. fujikuroi produces mainly GA4, GA7, GA13, and GA3. This result indicates that P450-2 activity is required for oxidation of GA14 at C-20, and the final removal of C-20, through which 20-carbon GAs are converted to biologically active 19-carbon-GAs. Interestingly, in all higher plants analysed so far, the removal of C-20 by progressive oxidation of the C-20 methyl group is catalysed by a 2-oxoglutarate-dependent dioxygenase. To confirm this unexpected result, i.e. that in F. fujikuroi this reaction is indeed catalysed by a P450 monooxygenase, a copy of the P450-2 gene was introduced into the cluster-deletion mutant SG139. In most of the transformants, P450-2 was highly expressed under optimum conditions (nitrogen starvation). The labelled GA precursor [14C]GA14 was efficiently metabolised to [14C]GA4, whereas the non-3β-hydroxylated [14C]GA12 was converted to [14C]GA9 and smaller amounts of [14C]GA25, the tricarboxylic acid product (Tudzynski et al. 2002). These results confirm the suggestion that P450-2 is responsible for oxidation of GA14 at C-20 and the final removal of the 20-carbon in F. fujikuroi.

In contrast, incubation of both labelled substrates with the mutant SG139 gave no conversion, whereas the wild-type converted these substrates mainly to [14C]GA3 and, in the minor pathway, to [14C]GA9 and [14C]GA25 (non-3β-hydroxylated GAs).

Thus, the P450-2 protein has a function analogous to that of the soluble GA 20-oxidases (dioxygenases) in plants (Lange et al. 1994; Hedden and Phillips 2000a). However, in contrast to the activities of P450-4 and P450-1 (Urrutia et al. 2001), it has so far not been possible to demonstrate in vitro activity of the fungal GA20 oxidase, P450-2.

The two final genes of the GA pathway

Only recently, a gene upstream of P450-4, called orf 3, was cloned and functionally characterised (Fig. 3). Sequence analysis revealed a 1,029 bp open reading frame uninterrupted by introns (Tudzynski et al. 2003). The amino acid sequence has only a weak similarity with a 7α-cephem-methoxylase from Nocardia lactamdurans (Coque et al. 1995), providing no indication of its possible function in GA biosynthesis. Knock-out mutants are unable to produce GA7 and GA3, the final products of the GA-biosynthetic pathway in F. fujikuroi. GC-MS analysis of culture filtrates of wild-type and mutants indicated that GA1 and GA4 rather than GA3 were the major C19-GAs (Fig. 4). The ratio of GA1 and GA4 was 3:1 to 5:1 depending on the culture medium. Therefore, the mutation appears to block the 1.2 desaturation of GA4 to GA7 (Fig. 4), indicating that the new gene (des) encodes GA4 desaturase (Tudzynski et al. 2003).

Fig. 4
figure 4

Reactions catalysed by the enzymes P450-3 (13-hydroxylase) and DES (GA4 1.2 desaturase) in the GA biosynthetic pathway of F. fujikuroi (modified from Tudzynski et al. 2003)

To confirm the function of the desaturase, the des gene was expressed in the deletion mutant SG139. Transformants carrying des as the only GA-biosynthetic gene were incubated with the radiolabelled substrates [14C]GA4 (3β-hydroxylated) or [14C]GA9 (non-hydroxylated). GC-MS analysis of the HPLC-purified product from incubation with labelled GA4 identified [14C]GA7 as the sole product, whereas labelled GA9 was converted to [14C]GA40 (2α-hydroxy GA9) and [14C]GA120 (1.2-didehydro GA9). Therefore, the desaturase accepts both 3β-hydroxylated and non-hydroxylated substrates (Tudzynski et al. 2003).

The remaining gene downstream of cps/ks is P450-3, one of the four cytochrome P450 monooxygenase genes of the cluster (Fig. 3). Gene disruption resulted in loss of the ability to produce GA3, suggesting that P450-3 encodes the missing 13-hydroxylase catalysing conversion of GA7 to GA3. Furthermore, ΔP450-3 mutants produced not GA1, but another 13-hydroxylated GA, which is produced in small amounts (1-2% total GAs) in the wild-type, but is the main product in Δdes mutants (Tudzynski et al. 2003) (Fig. 4).

In contrast to the other three P450 monooxygenase genes and des, P450-3 was not expressed in the cluster deletion mutant SG139. None of the transformants were able to metabolise [14C]GA7 or [14C]GA4 to labelled GA3 or GA1, respectively. Feeding the transformants with several precursors did not induce P450-3 gene expression. On the other hand, the 13-hydroxylase UV mutant 6314, which was previously isolated as a GA7-overproducing mutant (B. Tudzynski, unpublished), could be successfully complemented with a copy of the P430-3 gene, restoring the ability to produce GA3. Sequence analysis of the mutant P450-3 gene copy revealed a point mutation at nucleotide position 844 resulting in a single amino acid substitution from arginine to tryptophan at position 221 (Tudzynski et al. 2003). These results suggest that P450-3, in contrast to the other genes, requires the existence of the GA gene cluster for its expression.

The functional characterisation of des and P450-3 completed analysis of the GA gene cluster in F. fujikuroi. Disruption of the border genes smt upstream of des (Voß et al. 2001) and of orf1 and orf2 downstream of P450-3 (see Fig. 3) (B. Tudzynski, unpublished) did not affect production of GAs or its regulation. Therefore, the biosynthesis of the final product, GA3, in 15 steps beginning with the formation of GGDP by the pathway-specific GGDP synthase GGS2 requires only seven enzymes, many of which are multifunctional.

Cytochrome P450 oxidoreductase affects GA biosynthesis in F. fujikuroi

Eukaryotic non-mitochondrial cytochrome P450 monooxygenases are membrane proteins that require association with a NADPH-cytochrome P450 oxidoreductase (CPR) for activity. CPRs facilitate the transfer of electrons from NADPH via FAD and FMN to the prosthetic heme group of the P450 monooxygenases. Only one CPR-encoding gene, but about 40 different P450 monooxygenases, have been identified in the sequenced genomes of filamentous fungi such as Fusarium graminearum, Magnaporthe grisea, Aspergillus nidulans and Neurospora crassa.

The F. fujikuroi CPR would be expected to have a strong influence on GA biosynthesis since four P450 monooxygenases (P450-1P450-4) are involved in this pathway. Thus, loss of CPR activity should affect the rates of several GA biosynthetic steps. A phenotype similar to that expected for cpr mutants has been described for strain SG138. This UV-induced mutant has lost most oxidation steps catalysed by P450 monooxygenases and does not produce any bioactive GAs (Barrero et al. 2001).

The cpr gene has been cloned from F. fujikuroi, and its role in GA biosynthesis and primary metabolism studied (Malonek et al. 2004). Knock-out mutants, as well as UV mutant SG138, showed a reduction in growth rate on agar plates—more significantly on minimal medium. The mutants did not produce GA4, GA7, or GA3, the last three products of the pathway, whereas complementation of the Δcpr mutant with the wild-type gene resulted in full restoration of GA production (Fig. 5). However, small amounts of non-hydroxylated GAs, such as GA15 and GA24, and the high level of ent-kaurene found in Δcpr mutants indicate low residual activities of ent-kaurene oxidase (P450-4), and GA7 oxidase (one of the P450-1 activities), in the absence of the reductase, suggesting the participation of a second electron transport protein, e.g. cytochrome b5, in GA biosynthesis (Malonek et al. 2004). On the other hand, the 3β-hydroxylase activity of the multifunctional enzyme P450-1 and production of biologically active 19-carbon GAs by GA 20-oxidase (P450-2) appears to be completely blocked (see Fig. 1).

Fig. 5
figure 5

Gas chromatography-mass spectrometry (GC-MS) analysis of culture filtrates of the wild-type (IMI58289), cpr-disruption mutant (ΔCPR-T20) and transformant KT-1, in which T20 has been complemented with the F. fujikuroi cpr gene. Total ion currents are shown for extracts as methyl esters trimethylsilyl ethers. Peaks: 1ent-kaurene, 2ent-kaurenoic acid, 3 GA9, 4 GA25, 5 GA24, 6 GA14 and 7β-hydroxykaurenolide, 7 GA4, 8 GA7, 9 fujenoic acid, 10 GA13, 11 GA36, 12 GA3 isolactone, 13 7β, 18-dihydroxykaurenolide, 14 GA1, 15 GA3, 16ent-kaurenol, 17 GA15. Unlabeled peaks represent compounds unrelated to GA biosynthesis. The peak at the same retention time as ent-kaurene in the KT-1 extract contains no ent-kaurene. Modified from Malonek et al. 2004

Sequence comparison between the wild-type cpr gene and the cpr copy of the mutant strain SG138 confirmed the suggestion that loss of almost all oxidative enzymatic activities of the GA-specific P450 monooxygenases in this mutant is due to a point mutation in the cpr gene in SG138 (Malonek et al. 2004). Interestingly, GA production capability could be restored not only by transformation with a wild-type copy of F. fujikuroi cpr, but also by the Aspergillus niger gene cprA (van den Brink et al. 1995). This result indicates that CPRs act non-specifically as general electron donors for P450 monooxygenases from different pathways, e.g. in detoxification of benzoate by the P450 enzyme benzoate-p-hydroxylase (Malonek et al. 2004).

Regulation of GA biosynthesis in F. fujikuroi

The conditions for optimal production of GAs by F. fujikuroi are of biotechnological interest and have been studied intensively over the last 50 years (Borrow et al. 1955, 1964; Brückner and Blechschmidt 1991; Candau et al. 1992; Darken et al. 1959; Geissman et al. 1966; Avalos et al. 1997, 1999; Tudzynski 1999). The most obvious regulatory principle is the strong repression of GA production by high amounts of nitrogen (e.g. ammonium, glutamine, glutamate, asparagine, nitrate) in the culture medium (Muñoz and Agosin 1993; Brückner and Blechschmidt 1991).

The study of the molecular mechanism of nitrogen regulation began after the GA biosynthetic genes were cloned. Although GAs do not contain any nitrogen in their structure, it has been found that AREA (NIT2), the major nitrogen regulatory protein in filamentous fungi (Caddick 1994; Feng et al. 1995), controls expression of GA biosynthetic genes. Replacement of the F. fujikuroi homologue of areA and nit-2, areA-Gf, resulted in a significant reduction in GA production yields (Tudzynski et al. 1999). Complementation of ΔareA-Gf mutants with the wild-type gene led to full recovery of GA production.

Recent studies have shown that the expression level of six of the seven GA biosynthetic genes is drastically reduced in mutants lacking areA-Gf (Mihlan et al. 2003). Only the expression of P450-3, catalysing the final 13-hydroxylation of GA7 to form GA3, is not repressed by high amounts of nitrogen (Fig. 6). Electrophoretic gel mobility shift assays (EMSA) with promoter fragments of GA biosynthetic genes revealed binding of the zinc finger peptide of F. fujikuroi AREA-Gf to all GATA sequence elements in vitro. However, mutation and deletion of those elements in the bidirectional promoter of P450-1/P450-4 and its fusion to the Escherichia coli β-glucuronidase (uidA) gene clearly demonstrated that only two double GATAs out of the nine GATA elements play an important role in transcription regulation in vivo (Mihlan et al. 2003).

Fig. 6
figure 6

Time-course of expression of GA biosynthetic genes in 100% ICI medium (high in nitrogen) and 10% ICI (low in nitrogen). R Repressing conditions, I inducing conditions

In A. nidulans and N. crassa, the activity of AREA (NIT2) is regulated post-translationally by direct binding of a second regulatory protein, NMR (Jarai and Marzluf 1990; Andrianopoulos et al. 1998). However, deletion of the homologous gene nmr-Gf in F. fujikuroi did not result in the expected repression of GA biosynthetic genes. Overexpression of nmr-Gf led to slightly more repression of AREA-regulated genes (Mihlan et al. 2003). However, the limited role of NMR-Gf in F. fujikuroi suggests that other regulatory proteins must exist in this fungus that are involved in nitrogen metabolite repression, e.g. by interaction with AREA.

Interestingly, besides areA-Gf, the glutamine synthetase (GS) of F. fujikuroi seems to be involved in nitrogen regulation in general, and unexpectedly also in the control of N-regulated secondary metabolite pathways. Deletion of the GS-encoding gene glnA-Gf resulted in total loss of GA- and bikaverin (red pigment)-biosynthesis even under nitrogen starvation conditions (Teichert et al. 2004).

Some secondary metabolite gene clusters in fungi, e.g. the aflatoxin gene cluster in Aspergillus parasiticus, contain pathway-specific regulatory genes (Keller and Hohn 1997). The GA biosynthetic gene cluster in F. fujikuroi does not include a gene encoding a pathway-specific transcription factor. Point mutations in the P450-4 gene promoter revealed a 30 bp region important for a high transcription level of P450-4. This sequence element could be the binding motif for a putative GA pathway-specific transcription factor that should be cloned by a one-hybrid approach in the near future.

Biotechnological importance of cloning and functional analysis of GA biosynthetic genes

Gibberellins play important roles in plant development, controlling such processes as seed germination, internode elongation, flower development, and fruit set in higher plants (Crozier 1983). Therefore, cloning and characterisation of GA biosynthetic genes in higher plants and F. fujikuroi is not only of scientific, but also of biotechnological interest.

The spectacular increases in wheat and rice yields during the ‘Green Revolution’, were enabled by the introduction of dwarfing traits into the plants (Hedden 2003). Now, identification of the genes responsible for these traits shows that they interfere with the action or production of gibberellins. Mutation and overexpression of plant genes encoding GA-biosynthetic enzymes have been shown to affect plant growth and developmental processes (Hedden and Phillips 2000b; Hedden 2003). Thus, GA 20-oxidase is a regulatory enzyme controlled by developmental and environmental stimuli (Phillips et al. 1995) and, as such, is a prime target for genetic manipulation. In Arabidopsis, overexpression of GA 20-oxidase genes produced a GA-overproduction phenotype with elongated hypocotyls and stem and early flowering (Huang et al. 1998). Among the GA 20-oxidases known to date, only that from developing pumpkin seeds has been shown to produce biologically inactive GAs as the major products; the pumpkin enzyme converts the aldehyde intermediates GA24 and GA19 to the tricarboxylic acids, GA25 and GA17, respectively (Lange 1998). Overexpression of this GA 20-oxidase gene in transgenic plants could, therefore, result in a reduction of biologically active GAs by diverting the pathway to tricarboxylic acids. Thus, the GA 20-oxidase gene is a potentially useful tool for controlling plant growth (Niki et al. 2001).

In addition to the modification of GA biosynthesis in crop plants to improve agriculturally important traits, modification of GAs in F. fujikuroi for commercial production is also of increasing importance. The cloning and characterisation of genes encoding the final two enzymes in the GA biosynthetic pathway in F. fujikuroi is of special biotechnological interest. Certain agricultural and horticultural applications specifically require GA4 (Hedden et al. 1993; Almqvist 2003; Tudzynski and Sharon 2002). Therefore, strains producing GA4 in the absence of GA3 and GA7 would have considerable utility. Double mutants lacking both desaturase and 3β-hydroxylase activities were constructed and shown to produce almost exclusively GA4 in high amounts. Deletion of P450-3 resulted in accumulation of GA7, whereas GA4 desaturase (des) mutants produce a mixture of GA1 and GA4 (Fig. 7) (B. Tudzynski, unpublished results).

Fig. 7
figure 7

GA production profiles in the wild-type strain IMI 58289, the ΔP450-3 mutant, the ΔP450-3/Δdes double mutant and the Δdes mutant (P. Linnemannstöns and B. Tudzynski, unpublished results)

Intermediates of GA biosynthesis, such as ent-kaurene, ent-kaurenoic acid, GA12-aldehyde, and GA14, are not yet commercially available and are synthesised in small amounts from pumpkin endosperms for scientific experiments. By using F. fujikuroi mutants with defined genetic blocks at specific steps of the pathway, large scale biotechnological production of these intermediates is now possible. Thus, ΔP450-4, ΔP450-1, and ΔP450-2 mutants accumulate ent-kaurene, ent-kaurenoic acid, and GA14, respectively, in high amounts.

In the last 10 years, several attempts have been made to improve secondary metabolism in microorganisms by genetic engineering (Newbert et al. 1997; Perez-Llarena et al. 1997; Kosalkova et al. 2001). Amplification of entire gene clusters, genes encoding enzymes that catalyse rate-limiting steps, or of genes encoding pathway-specific transcription factors, are possible strategies for strain improvement. A prerequisite for successful application of recombinant DNA technology to improve strain performance is identification of the optimal genetic change. Thus, in Penicillium chrysogenum, the elevated copy number of the entire penicillin gene cluster and increased gene transcription levels largely explains the high penicillin titres of production strains resulting from classical mutagenesis and selection-based strain development programs over the last 50 years (Newbert et al. 1997). Using a single copy low-producing strain, the effect of overexpressing the penicillin biosynthetic genes has been confirmed by molecular approaches. Transformants in which the penicillin gene cluster was amplified showed the largest increase in penicillin production—up to 176% (Theilgaard et al. 2001).

In F. fujikuroi, amplification of a genomic fragment carrying the ggs2 and cps/ks genes encoding the first pathway-specific enzymes, did not result in significant increase of GA production. However, amplification of all GA biosynthetic genes by transforming a cosmid carrying the entire GA gene cluster resulted in a significant increase (up to 300%) of GA production yields (P. Linnemannstöns and B. Tudzynski, unpublished).

Another promising approach for development of GA-overproducing strains is the blocking of competing pathways, such as the bikaverin and carotenoid biosynthetic pathways. The genes encoding the key enzymes of both pathways have been deleted in F. fujikuroi, but without any significant effect on GA production capability (Linnemannstöns et al. 2002a,b). However, although disruption of pks4 (encoding the bikaverin-specific polyketide synthase PKS4) did not improve GA production, the mutants should be useful in large scale production of GAs since the absence of the red pigment will simplify purification of GAs from the fermentation medium (Linnemannstöns et al. 2002b).

Another important approach to strain improvement will be modulation of gene expression after detailed study of regulation principles. Knowledge of the transcription factors involved, and their specific binding motifs in the promoters of GA biosynthetic genes, will allow manipulations such as overexpression of pathway-specific transcription factors, amplification of sequence elements in the promoters, and/or fusion of strong promoters to GA biosynthetic genes.

Furthermore, it has been shown in many studies that complex culture media containing plant components, such as plant oils, plant meals, or corn steep liquor, are much better as GA production media than synthetic media with glucose and defined nitrogen sources (Darken et al. 1959; Fuska et al. 1961; Muromtsev and Agnistova 1984). The identification of the inducing substances in these complex plant materials and confirmation of their positive effect on gene expression would help to further optimise the composition of production media.

Evolution of GA biosynthetic genes

The identity of GAs produced by higher plants and F. fujikuroi, as well as the identity of the first pathway steps as far as GA12-aldehyde in both systems, led to the suggestion that fungi acquired the GA pathway genes from plants by horizontal gene transfer (Chapman and Regan 1980). However, after cloning of all pathway genes in plants and F. fujikuroi it became clear that there are dramatic differences in the character of enzymes involved, the organisation of GA biosynthetic genes in the two genomes, and the pathway stages at which hydroxyl groups are introduced. As shown above, the final stages of GA biosynthesis in plants, comprising the removal of C-20 by GA 20-oxidases (GA20ox) as well as the introduction of the 3β-hydroxyl group by GA 3-oxidases (GA3ox), are carried out by soluble 2-oxoglutarate-dependent dioxygenases that are assumed to be localised in the cytosol (Hedden and Phillips 2000a; Hedden et al. 2001). In F. fujikuroi, the same activities are the responsibility of cytochrome P450 monooxygenases (P450-1, P450-2), demonstrating that GA biosynthesis has evolved independently in plants and F. fujikuroi. If this is the case, the question of the origin of the GA biosynthesis gene cluster in F. fujikuroi arises. Our recent data show that seven of the nine sexually fertile species of the Gibberella fujikuroi species complex contain all seven (e.g. F. proliferatum, MP-D; F. subglutinans, MP-E), and two species (F. verticilloides, MP-A; F. circinatum, MP-H) contain only one or two GA cluster genes. However, none of these species except F. fujikuroi (MP-C) are able to produce GAs (S. Malonek and B. Tudzynski, unpublished).

Recently, the entire genome of F. graminearum has been sequenced. Despite the high degree of sequence identity between primary metabolism genes of F. graminearum and F. fujikuroi, there are dramatic differences in secondary metabolite production by the two species. Most surprisingly, F. graminearum does not contain any obvious GA biosynthetic genes. On the other hand, a number of other fungal species not related to Fusarium produce GAs at high concentrations (milligram quantities per litre of culture fluid; Rademacher 1994). Several species of the ascomycete Sphaceloma produce GA4, but not GA3, and therefore seem to lack 1.2 desaturase and 13-hydroxylase activities. A species of Phaeosphaeria produces GA1 via a pathway in which 3β-hydroxylation occurs after C19-GA formation, as is the case in higher plants (Kawaide et al. 1995). However, similar to F. fujikuroi but unlike in plants, 13-hydroxylation occurs late in the pathway. Thus, the pathway to GA1 includes GA9 as an intermediate, which is 3β-hydroxylated to GA4 and then 13-hydroxylated to GA1 (Kawaide et al. 1995).

Genomic libraries of two Sphaceloma species were recently constructed, and analysis of the GA biosynthesis genes is underway C. Bömke and B. Tudzynski, unpublished). The characterisation of these genes and their organisation in the genomes of both fungi will facilitate the study of the evolutionary origin of GA biosynthetic genes in fungi.

In addition to plants and fungi, some bacteria (e.g. Acetobacter diazotropicus, Azospirillum lipoferum; Rhizobium phaseoli; and Bacillus pumilus) produce GAs (reviewed in MacMillan 2001). However, the relevant biosynthetic genes have not yet been cloned. Thus, Azospirillum spp., inoculated on rice GA-deficient mutants, metabolised GA20 to GA1 and reversed dwarfism, showing an in vivo capacity to perform 3β-hydroxylation. The bacterial 3β-hydroxylase may be a 2-oxoglutarate-dependent dioxygenase similar to those of plants (Cassan et al. 2001).

The evolution of the ability to synthesise specialised metabolites is likely to have been the key for survival and diversification of different plant, fungal and bacterial species. For further study of the phylogenetic origin of the GA gene cluster in the F. fujikuroi species complex, it will be interesting to compare the GA genes of F. fujikuroi with those in other MPs, but also with those of Sphaceloma and Phaeosphaeria in order to find out if there is a general pattern of inheritance of this secondary gene cluster during evolution. It is still not clear if the GA gene cluster evolved from one ancestral species, with divergence during speciation resulting in the inability of most of ascomycete species to produce GAs. The general question is why organisms retain such large gene units, like gene clusters, without their having obvious functionality in the organism. If such cooperating genes are conditionally dispensable, but have an adaptive value in colonising certain ecological niches, the incipient cluster could be maintained by positive selection in those environments (Lawrence 1997). However, to answer this question, it will be necessary to investigate the role of GA production for the fungus itself, e.g. in its effects on host plants, in more detail.

Summary and conclusion

In parallel with progress in higher plants and bacteria, advances have been made in understanding GAs in the fungus F. fujikuroi. The clustering of GA biosynthetic genes has facilitated their rapid isolation, sequencing, and functional characterisation following identification of the first gene. Knock-out mutants of all pathway genes are now available and allow the production of high amounts of GA intermediates and precursors of GA3, such as ent-kaurene, ent-kaurenoic acid, GA14, GA4, GA7 and GA1. The profound differences in GA biosynthesis in plants and F. fujikuroi identified in these studies strongly suggest that GA biosynthesis pathways evolved independently in these two groups of organisms. These data raise questions regarding the evolutionary origin of GA biosynthetic genes in Fusarium, since Fusarium species outside the G. fujikuroi species complex, such as F. oxysporum, or F. graminearum, do not have GA biosynthetic genes. On the other hand, fungi from other taxonomic groups, such as Sphaceloma manihoticola and Phaeosphaeria sp. have been shown to produce GAs. However, the sequence, character and organisation of GA genes in the genomes of these fungi, as well as regulation of gene expression compared with F. fujikuroi, remain to be studied.

The availability of all pathway genes in F. fujikuroi now allows a detailed analysis of the regulation of gene expression on a molecular level. One aspect of future research will be a thorough study of nitrogen metabolite repression, its mechanism, and components involved in addition to the general regulator AREA. On the other hand, in addition to nitrogen regulation, it seems likely that GA biosynthesis is also under the control of pathway-specific transcription factors that have yet to be identified. A detailed knowledge of the mechanisms of gene regulation and identification of the transcription factors involved are important prerequisites for further strain improvement programs by metabolic engineering.