Characterization of the final two genes of the gibberellin biosynthesis gene cluster of Gibberella fujikuroi: des and P450-3 encode GA4 desaturase and the 13-hydroxylase, respectively.

Recently, six genes of the gibberellin (GA) biosynthesis gene cluster in Gibberella fujikuroi were cloned and the functions of five of these genes were determined. Here we describe the function of the sixth gene, P450-3, and the cloning and functional analysis of a seventh gene, orf3, located at the left border of the gene cluster. We have thereby defined the complete GA biosynthesis gene cluster in this fungus. The predicted amino acid sequence of orf3 revealed no close homology to known proteins. High performance liquid chromatography and gas chromatography-mass spectrometry analyses of the culture fluid of knock-out mutants identified GA1 and GA4, rather than GA3 and GA7, as the major C19-GA products, suggesting that orf3 encodes the GA4 1,2-desaturase. This was confirmed by transformation of the SG139 mutant, which lacks the GA biosynthesis gene cluster, with the desaturase gene renamed des. The transformants converted GA4 to GA7, and also metabolized GA9 (3-deoxyGA4) to GA120 (1,2-didehydroGA9), but the 2alpha-hydroxylated compound GA40 was the major product in this case. We demonstrate also by gene disruption that P450-3, one of the four cytochrome P450 monooxygenase genes in the GA gene cluster, encodes the 13-hydroxylase, which catalyzes the conversion of GA7 to GA3, in the last step of the pathway. This enzyme also catalyzes the 13-hydroxylation of GA4 to GA1. Disruption of the des gene in an UV-induced P450-3 mutant produced a double mutant lacking both desaturase and 13-hydroxylase activities that accumulated high amounts of the commercially important GA4. The des and P450-3 genes differ in their regulation by nitrogen metabolite repression. In common with the other five GA biosynthesis genes, expression of the desaturase gene is repressed by high amounts of nitrogen in the culture medium, whereas P450-3 is the only gene in the cluster not repressed by nitrogen.

Gibberellins (GAs) 1 are plant hormones that are produced by all higher plants and some fungi. The rice pathogen Gibberella fujikuroi, mating population C (anamorph Fusarium fujikuroi), produces high amounts of gibberellic acid (GA 3 ) and some other GAs and is used for commercial production of these agriculturally important compounds.
GAs are diterpenoids and are synthesized in G. fujikuroi via the mevalonate pathway (1). Most of the genes of the early isoprenoid pathway have been cloned from G. fujikuroi, viz 3-hydroxy-3-methylglutaryl-CoA reductase (2), farnesyl diphosphate synthase (3), and two geranylgeranyl diphosphate (GGPP) synthase genes, one of which (ggs1) is involved in general isoprenoid biosynthesis (4) and a second (ggs2) of which is specific for GA biosynthesis (5). Biosynthesis of the major metabolite GA 3 (gibberellic acid) from GGPP requires 13 steps (Fig. 1). GGPP is converted to ent-kaurene via ent-copalyldiphosphate in a two-step cyclization reaction (6). ent-Kaurene is metabolized to GAs in G. fujikuroi by a series of oxidation reactions catalyzed by cytochrome P450 monooxygenases, whereas in plants, P450 monooxygenases and 2-oxoglutaratedependent dioxygenases are involved (7). In contrast to plants in which cyclization of GGPP is catalyzed by two enzymes, ent-copalyl diphosphate synthase and ent-kaurene synthase, in the fungi G. fujikuroi and Phaeosphaeria, both steps are catalyzed by a bifunctional ent-copalyl diphosphate synthase/entkaurene synthase enzyme (8 -10).
In this paper, we characterize the remaining two genes of the cluster and demonstrate that they are responsible for the last two steps in the biosynthesis of GA 3 . We describe the isolation and functional characterization of the seventh gene, orf3, which is located at the left border of the cluster. By gene knock-out and expression in the absence of the other GAbiosynthetic genes, we show that orf3 encodes the desaturase that converts GA 4 to GA 7 . Furthermore, we demonstrate that the fourth P450 monooxygenase gene, P450 -3, encodes the 13-hydroxylase that converts GA 7 to the end product, GA 3 . The identification of the last two GA-biosynthetic genes and the production of single or double deletion mutants have enabled us to construct strains producing GA 7 , GA 1 or the commercially important GA 4 . Expression studies with the desaturase and P450 -3 genes in combination with biochemical analysis revealed significant differences in their regulation by nitrogen.

EXPERIMENTAL PROCEDURES
Fungal Strains and Culture Conditions-G. fujikuroi m567, a wildtype strain from rice, was provided by the Fungal Culture Collection (Weimar, Germany). Mutant 6314 producing GA 7 as main product was obtained by UV mutagenesis of strain m567. The wild-type strain G. fujikuroi IMI 58289 and the GA-defective mutant strain SG139 (15) were kindly provided by J. Avalos (Sevilla, Spain). SG139 has completely lost the GA-biosynthetic gene cluster as demonstrated by Southern blotting and PCR analysis. Culture conditions for preparing mycelia for DNA or RNA isolation were as described previously (14). For analysis of GA production, the fungus was grown in the complex optimized GA production medium, OPM (14), or in 20% ICI medium (16) for 7-10 days at 28°C on a rotary shaker (200 rpm).
Bacterial Strains and Plasmids-Escherichia coli strain Top10 (Invitrogen) was used for plasmid propagation. Vector pUC19 was used to clone DNA fragments carrying the G. fujikuroi orf3 gene or parts of it. The plasmid porf3-Sal carries a 3.4-kb fragment with the entire gene. For inactivation of orf3, an internal BamHI/XbaI fragment of plasmid porf3-Sal, was replaced by the hygromycin resistance cassette from pGPC1 (17). For inactivation of P450 -3, three different gene disruption vectors, pP450 -3-GD1, pP450 -3-GD2, pP450 -3-GD3, were constructed by cloning internal fragments of the gene into the vectors pUCH2-8 (18) or pAN7-1 (19), both of which carry the hygromycin B resistance cassette. The internal fragments of P450 -3 were first amplified by PCR, cloned into the PCR cloning vector pPCR2.1 (Invitrogen), cut with KpnI/XhoI (pP450 -3-GD2 and pP450 -3-GD3), and then cloned into pUCH2-8. For pP450 -3-GD1, the fragment was cut with HindIII/XbaI and cloned into pAN7-1. For gene expression of orf3 in the GA-deficient mutant SG139, the 3.4-kb SalI fragment from plasmid porf3-Sal was cloned into pUCH2-8, resulting in plasmid porf3-GC. For expression of P450 -3 in SG139, a 2.3-kb XbaI fragment with the entire gene was cloned into pAN7-1, resulting in the vector p450 -3-GC. In addition, the cosmid clone, cos5 carrying the entire P450 -3 gene and 40 kb to the right of P450 -3 was used for complementation of SG139. The Uni-Zap XR vector of the orf3 cDNA clones allowed in vivo excision and recircularization of cloned inserts to form a phagemid in pBluescript SK(Ϫ) carrying the insert. This subcloning step was performed according the manufacturer's protocol (Stratagene, La Jolla, CA).
Southern and Northern Blot Analysis-DNA and RNA isolation and Southern and Northern analyses were as described previously (14). The G. fujikuroi small subunit of rDNA was used as a control for RNA transfer.
Screening of G. fujikuroi cDNA and Genomic EMBL3 Libraries-The expression library (UniZap TM XR vector, Stratagene) was constructed from RNA isolated from mycelium grown under optimal conditions for GA 3 formation (4). Approximately 30,000 recombinant phages were plated at ϳ7,500 plaques/150-mm diameter Petri dishes and transferred to nylon membranes. For screening of the genomic library (20), ϳ40,000 recombinant phages were plated and transferred to membranes. The hybridization was performed at high stringency (65°C), and the blots were washed at the same temperature in 2ϫ SSC, 0.1% SDS, followed by 0.1ϫ SSC, 0,1% SDS. Positive recombinant clones were used for a second round of plaque purification.
Transformation of G. fujikuroi-Preparation of protoplasts and transformations were carried out as described previously (20). 10 7 protoplasts (100 l) of the strain IMI 58289 or the mutant 6314 were transformed with 10 g of the SalI-fragment from the gene replacement vector pORF3-GR or one of the circular disruption vectors pP450 -3-GD1, pP450 -3-GD2, and pP450 -3-GD3. Transformed protoplasts were regenerated at 28°C in a complete regeneration agar (0.7 M sucrose, 0.05% yeast extract, 0.1% (NH 4 ) 2 SO 4 containing 120 g/ml hygromycin B (Calbiochem)) for 6 -7 days. Single conidial cultures were established from hygromycin B-resistant transformants and used for DNA isolation and Southern blot analysis. For expression of orf3 and P450-3 in the GA-deficient mutant SG139, 10 7 protoplasts were transformed with 10 g of the complementation vectors porf3-GC, pP450 -3-GC, or the cosmid clone GA-cos5, carrying the entire P450 -3 gene and an ϳ40-kb genomic region to the right of P450 -3 probably not involved in GA biosynthesis.
Gibberellin Analysis-For analysis of GA formation, the wild-type strain IMI 58289 and orf3-disrupted mutants were cultivated in 100-ml Erlenmeyer flasks containing 100 ml of OPM medium. The cultures were incubated for 7-10 days on a rotary shaker (200 rpm) at 28°C. GA 3 , GA 1 , GA 4 , and GA 7 were analyzed by HPLC according to Barendse et al. (22) using a Merck HPLC system with a UV detector and a LiChrospher 100 RP-18 column (5-m particle size; 250 ϫ 4 mm; Merck Eurolab GmbH, Darmstadt, Germany). GA 3 , GA 4 , and GA 7 were also analyzed by thin layer chromatography using a mixture of ethyl acetate/chloroform/acetic acid (60:40:5). GC-MS analysis of total culture filtrates or HPLC-purified fractions was performed on derivatized samples (11) using a Trio2A mass spectrometer (Micromass, Manchester, United Kingdom) connected to an Agilent 5890 gas chromatograph. Samples in hexane (1 l) were injected splitless into a 25 m ϫ 0.2 mm ϫ 0.25 m (film thickness) BPX-5 WCOT column (Scientific Glass Engineering) at 50°C. After 2 min, the oven temperature was increased at 13°C⅐min Ϫ1 to 180°C and then at 3°C⅐min Ϫ1 to 300°C. The He inlet pressure was 90 kPa, and the injector, transfer line, and MS source temperatures were 280, 280, and 200°C, respectively. Spectra were obtained at 70 eV, scanning at 2 Hz from 750 -45 atomic mass unit. GAs were identified by comparison of their mass spectra with published data (23).

Cloning and Expression of orf3-
pUC19. Sequence analysis revealed a 1029-bp open reading frame (designated orf3) transcribed in the same orientation as P450 -4. The cDNA sequence data indicated that the open reading frame is not interrupted by any introns. The orf3 gene is deposited under the GenBank TM accession number AJ417493. Scrutiny of the GenBank TM data base for similar sequences revealed only a weak homology with ORF8 of the two-component 7␣-cephem-methoxylase from Nocardia lactamdurans (26), providing little indication of its possible function in GA biosynthesis. In the 5Ј non-coding region of orf3, 10 putative GATA motifs for binding of the major nitrogen regulatory protein, AREA, were identified, indicating that expression of orf3 may be regulated by nitrogen repression as are most of the other GA-biosynthetic genes (5,12,13).
Functional Analysis of orf3-A gene replacement vector pORF3-GR was obtained by replacing the central BamHI/XbaI fragment of plasmid porf3-Sal carrying the 3.4-kb SalI fragment of orf3 (Fig. 2B) with the hygromycin B resistance cassette from the vector pGPC1 (17). The wild-type strain IMI58289 was transformed with the 4.3-kb SalI-fragment from pORF3-GR. TLC and HPLC analysis revealed that 5 of 42 transformants (T18, T21, T23, T55, and T60) were not able to produce GA 3 , the final product of the GA-biosynthetic pathway in G. fujikuroi, or its immediate precursor, GA 7 . Southern blot analysis confirmed that the transformants have lost the 6.3-kb PstI wild-type band because of homologous integration of the replacement cassette into the orf3 locus (data not shown). GC-MS analysis of culture filtrates of transformant T23 and wild-type indicated that GA 1 rather than GA 3 was the major C 19 -GA in the orf3-deficient line (Fig. 3). Fujenoic acid and GA 13 (Fig. 1) were major metabolites in both lines. Three of the transformants were shown by HPLC analysis to contain GA 1 and GA 4 in ratios from 3:1 to 5:1 after cultivation in the synthetic 20% ICI medium or in OPM for 10 days (Table I).
Since the mutants produced no GA 3 or GA 7 and accumulated GA 4 and GA 1 , the mutation appears to block the 1,2-desaturation of GA 4 to GA 7 (Fig. 1), indicating that orf3 encodes GA 4 desaturase.
To confirm the function of the orf3 expression product, we transformed the deletion mutant SG139, which has lost the entire GA gene cluster, with the gene complementation vector porf3-GC carrying the wild-type orf3 gene. From 20 analyzed transformants, 12 were shown by PCR to contain a 2.1-kb fragment expected after correct integration of the gene into the SG139 genome. Analysis of five transformants by Northern hybridization confirmed that the gene was expressed in strain SG139 despite the loss of the entire GA gene cluster (data not shown). Four strains SG139-T1, SG139-T2, SG139-T4, and -T5, which expressed orf3 at levels comparable with the wild-type, were incubated with the radiolabeled substrates [ 14 C]GA 4 (3␤hydroxylated) or [ 14 C]GA 9 (non-hydroxylated) in medium buffered at pH 3.0. GC-MS analysis of the HPLC-purified product from incubation of [ 14 C]GA 4 with transformants T1, T2, T4, and T5 identified [ 14 C]GA 7 as sole product (shown for T1 in Table II), whereas SG139 did not metabolize this substrate (data not shown). Thus, the desaturase gene is expressed as an active enzyme in SG139. In addition, the orf3 transformants efficiently utilized the non-hydroxylated substrate [ 14 C]GA 9 , producing two more polar fractions of radioactivity on HPLC in 93% yield. The more polar of these (65%) was found by GC-MS to contain [ 14 C]GA 40 (2␣-hydroxyGA 9 ), whereas the less polar fraction (28%) contained GA 120 (1,2-didehydroGA 9 ) (Table II and Fig. 4) and three isomers of a diene dicarboxylic acid derivative assumed to be formed from GA 120 by rearrangement during work-up and derivatization. The enzyme will thus accept both 3␤-hydroxylated and non-hydroxylated substrates, although in the latter case, there is an apparent loss of regiospecificity.
Functional Analysis of P450 -3-To determine the function of the remaining gene of the cluster, P450 -3, the wild-type strain IMI58289 was transformed with the gene disruption vector pP450 -3-GD1, which contained the hygromycin-resistant gene, hph (Fig. 2C). After analysis of 100 hygromycinresistant transformants by TLC, all were found to produce high amounts of GAs after 7 days of cultivation in OPM but three lines, T52, T59, and T61, did not produce GA 3 , although its precursors, GA 4 and GA 7 , were present. To be sure that this change in GA profile was the result of P450 -3 gene inactivation, we constructed two additional disruption vectors pP450 -3-GD2 and pP450 -3-GD3. In pP450 -3-GD3, a mutation in the heme-binding domain was introduced via PCR (Fig. 2C, primer  P450 -3-R3). We analyzed 100 hygromycin-resistant transformants for both vectors by TLC. One of the transformants from pP450 -3-GD2 (T92) and three of the transformants from pP450 -3-GD3 (T11, T49, and T55) produced GA 7 as final product instead of GA 3 , which was confirmed by HPLC. GC-MS analysis of the culture fluid from transformant ⌬P450 -3-T55 confirmed that GA 7 was the major product and that no GA 3 was produced (Fig. 3). Southern blot analysis (data not shown) revealed that transformants with all of the three vectors had lost the 4.3-kb EcoRI wild-type band for P450 -3 and produced smaller hybridizing bands. Therefore, the loss of the ability to produce GA 3 in these transformants is due to the disruption of gene P450 -3, suggesting that it encodes the 13-hydroxylase catalyzing the conversion of GA 7 to GA 3 . Furthermore, the Total ion currents are shown for extracts as methyl esters trimethylsilyl ethers. Components were identified by comparison of their mass spectra and GC retention times with published data (23) as follows: peak 1, GA 9 ; peak 2, GA 25 ; peak 3, GA 14 ; peak 4, 7␤-hydroxykaurenolide; peak 5, GA 4 ; peak 6, GA 7 ; peak 7, fujenoic acid and GA 13 ; peak 8, 7␤,18-dihydroxykaurenolide; peak 9, GA 3 ; and peak 10, GA 1 . transformants produced no GA 1 , another 13-hydroxylated GA, which is produced in small amounts (ϳ1.3% total GAs) in the wild-type. Thus, P450 -3 is responsible for 13-hydroxylation of both GA 7 and GA 4 (Fig. 4).
To confirm the function of P450 -3, complementation of SG139 with the complementation vector pP450 -3-GC and cosmid clone cos5, both carrying the entire gene P450 -3, was performed. However, Northern analysis indicated that none of the resulting transformants expressed the gene despite the correct integration of the gene into the SG139 genome (data not shown). Furthermore, none of the transformants from both complementation vectors were able to metabolize [ 14 C]GA 7 or [ 14 C]GA 4 to radiolabeled GA 3 and GA 1 , respectively. Several intermediates of the GA-biosynthetic pathway were tested as possible inductors of P450 -3 expression by adding them to cultures of the P450 -3 transformants at 350 -500 M. We found that ent-kaurenoic acid, GA 14 , GA 4 , or GA 7 were ineffective in inducing GA 13-hydroxylase activity in the SG139-P450 -3 transformants, which remained unable to convert [ 14 C]GA 7 or [ 14 C]GA 4 under these conditions. The lack of P450 -3 expression in the SG139 ϩ P450 -3 transformants is in contrast with the efficient expression of P450 -1 (13), P450 -2 (14), P450 -4 (12), and des in SG139. Interestingly, complementation of SG139 with a cosmid carrying the entire gene cluster (cos1) fully restored 13-hydroxylase activity.
Characterization of the 13-Hydroxylase Mutant 6314 -The identification of P450 -3 as the 13-hydroxylase that catalyzes the conversion of GA 7 to GA 3 allowed us to classify the GA 7overproducing mutant 6314 as a putative P450 -3 mutant. This strain was isolated after mutagenesis of the highly GA 3 -producing wild-type strain m567 and screening for GA-overproducing mutants. 2 To identify the nature of the mutation, the coding region of the P450 -3 gene from the mutant was amplified by PCR using the primer pairs P450 -3-MF and P450 -3-MR as well as P450 -3-MF and P450 -3/1. Nine independent clones were sequenced from both strands. A point mutation was found at position 844, resulting in an amino acid substitution from arginine to tryptophan at position 221 (Fig. 5). This substitution results in a dramatic change in the GA product spectrum, indicating greatly reduced 13-hydroxylase activity, whereas the wild-type strain m567 produced ϳ84% GA 3 , 1-2% GA 4 and ϳ14% GA 7 . As determined by HPLC, the mutant 6314 produced 3% GA 3 , 3% GA 4 , and 94% GA 7 under the same 2 B. Tudzynski, unpublished data.  (2), 386 (7), 358 (12), 300 (12) (Fig. 6). GA 1 , which was found in small amounts in strain m567, was not detectable in the mutant. Complementation of mutant 6314 with the P450 -3 gene restored GA 3 synthesis as demonstrated by GC-MS analysis of the culture fluid (data not shown), thus confirming that the mutation was in the 13-hydroxylase. Interestingly, the P450 -3 gene was expressed efficiently in 6314, which contains all of the GA-biosynthesis genes, in contrast to its lack of expression in SG139 in which the gene cluster is missing. Creation of a des/P450 -3 Double Mutant-Wild-type strains of G. fujikuroi produce mainly GA 3 , whereas certain horticultural applications require specifically GA 4 (27). Therefore, strains that produce GA 4 in the absence of GA 3 and GA 7 would have considerable utility. To this end, we prepared double mutants lacking both desaturase and 13-hydroxylase activities by transforming the mutant strain 6314 with the SalI fragment of vector pORF3-GR. HPLC analysis of 35 hygromycin B-resistant colonies by HPLC revealed that three, 6314⌬ORF3-T1, 6314⌬ORF3-T2, and 6314⌬ORF3-T8, had lost their ability to produce GA 7 and instead accumulated large amounts of GA 4 ( Fig. 6 and Table I). In contrast to the des mutants isolated from the wild-type IMI58289, they do not produce detectable amounts of GA 1 , confirming that P450 -3 catalyzes the 13hydroxylation of GA 4 as well as GA 7 . Thus, the targeted mutation of des, P450 -3, or both allows the creation of strains producing high amounts of GA 1 , GA 7 , or GA 4 , respectively.
Expression of P450 -3 Is Not Repressed by Nitrogen-All genes of the GA-biosynthetic gene cluster analyzed so far are highly expressed under conditions of nitrogen limitation, whereas only a low level of transcripts was detected in media with high amounts of ammonium or glutamine (5, 12, 13). Expression of the newly isolated gene des showed the same pattern. Transcription of des was investigated in mycelia grown for 15, 24, 38, 48 and 60 h in media with a high N (100% ICI) or low N (10% ICI) content (Fig. 7A). Northern blot analysis of total RNA revealed a single band of ϳ1.0 kb in 10% ICI medium, the transcript level increasing in abundance with culture time in a similar manner to P450 -4 (Fig. 7A), cps/ks and ggs2 (data not shown). No transcript was observed for des or P450 -4 in mycelia grown in 100% ICI medium. In contrast to the other six genes of the cluster, P450 -3 appears to be expressed more highly in medium with high N than with low N (Fig. 7A, C). After 15 h in culture in 10% ICI medium, when some nitrogen still remains, only very low levels of expression of des, P450 -4 (Fig. 7A) and the other four genes (data not shown) can be detected, whereas P450 -3 is highly expressed. With depletion of nitrogen, transcript abundance for des and P450 -4 increases, whereas P450 -3 transcript levels decrease (Fig. 7A). In addition, transcript abundance for P450 -3 is not reduced in areA mutants confirming that AREA does not control expression of P450 -3, whereas des expression (Fig. 7B) and that of the other genes (data not shown) require the active nitrogen regulator, AREA. The pH of the culture medium in the range 3.5 to 8.0 does not affect the expression of des, P450 -3 (Fig. 7C) or any of the other GA-biosynthetic genes (data not shown).
To confirm that the differential regulation of des and P450 -3 expression can be observed at the biochemical level, we investigated metabolism of [ 14 C]GA 4 or [ 14 C]GA 7 in P450 -2 disruption mutants. These mutants are blocked for the oxidative removal of C-20 from GA 14 and are therefore not able to produce GA 4 , GA 7 , GA 1 and GA 3 (Fig. 1) (14). The experiments with the gene-disruption mutants ⌬P450 -2-T35 and T39, demonstrated clear differential inhibition of the desaturase and 13-hydroxylase activities by N. Under high N conditions (100% ICI), [ 14 C]GA 4 was mainly converted to [ 14 C]GA 1 , together with a small amount of [ 14 C]GA 3 (Table III) (Table III). On the other hand, [ 14 C]GA 7 was converted efficiently to [ 14 C]GA 3 in both high and low N conditions, thus confirming that the 13-hydroxylase shows no significant inhibition by high N concentration. In contrast, the above results demonstrate that the desaturase is strongly inhibited by N, in agreement with the expression analysis on the des and P450 -3 genes described above (Fig. 7A). The product distribution obtained from [ 14 C]GA 4 under high N conditions, FIG. 4. Reactions catalyzed by DES  and P450 -3. The substrates GA 4 and GA 9 are formed from the 3␤-hydroxylation and non-3␤-hydroxylation pathways, respectively. The pathway to GA 3 is denoted by thick arrows. 13-hydroxylase mutant (6314). The latter contains a point mutation resulting in an amino acid (R3 W) substitution.

FIG. 5. Partial nucleotide and amino acid sequences of P450 -3 for a wild-type (m567) and
is consistent with the spectrum of GAs found in des disruption mutants, which accumulate mainly GA 1 (Table I). DISCUSSION We describe the cloning of the seventh and final gene of the GA biosynthesis gene cluster in G. fujikuroi and demonstrate that it encodes GA 4 desaturase (DES). The gene is located between the recently identified membrane transporter gene smt (25) and the ent-kaurene oxidase gene P450 -4 (12) at the left side of the cluster. The gene is not related to any other GA-biosynthetic gene from G. fujikuroi from higher plants, but it has low homology with ORF8 in the cephamycin C cluster of N. lactamdurans (26). This protein forms part of the twoprotein 7␣-cephem-methoxylase, which has both hydroxylase and methyltransferase activities. However, ORF8 has no apparent activity in isolation and did not assist us in determining the function of DES, which was demonstrated by a combination of gene replacement and biochemical analysis. Transformants of the SG139 mutant containing the wild-type des gene in the absence of the other GA biosynthesis genes converted [ 14 C]GA 4 to [ 14 C]GA 7 and also metabolized the non-hydroxylated substrate [ 14 C]GA 9 to [ 14 C]GA 120 (1,2-didehydroGA 9 ). However, GA 40 (2␣-hydroxyGA 9 ) was the major product from GA 9 , showing that 2␣-hydroxylation is an additional activity of the desaturase. This enzyme would thus account for the presence of 2␣-hydroxylated GAs in G. fujikuroi cultures (28). No 2␣-hydroxylated products were detected in incubations of the des transformants with the 3␤-hydroxylated substrate GA 4 , suggesting that the presence of the 3␤-hydroxyl group directs 1,2-desaturation in favor of 2␣-hydroxylation. However, the occurrence of 2␣,3␤-dihydroxylated GAs in fungal cultures (28) indicates that 2␣-hydroxylation of GA 4 may occur if at a low rate. Although we have no information on the mechanism of the desaturation, the formation of a hydroxylated product implies a requirement for oxygen.
In plants, oxidation of ring A of the GAs is catalyzed by 2-oxoglutarate-dependent dioxygenases, many of which are (1,2-desaturase), and P450 -3 (13-hydroxylase) was compared at various times of cultivation in media with a high (100% ICI) or low (10% ICI) N content. B, expression of des and P450 -3 was compared in a wild-type, an areA-disruption mutant (⌬areA-T5), and the areA-disruption mutant complemented with the areA wild-type gene (areA-T5/1) following resuspension of mycelia into medium without N (0% ICI). C, expression of des and P450 -3 was determined in media at different pH values with high (100% ICI) or no (0% ICI) N content. Transcript abundance was determined by Northern blot analysis of total RNA probed with the appropriate cDNA. As control for RNA loading, the blots were hybridized with G. fujikuroi ribosomal RNA (rRNA). multifunctional, catalyzing desaturation, and hydroxylation reactions (29). For example, a GA 3-oxidase from Phaseolus vulgaris catalyzes dehydrogenation of GA 20 at C-2,3 and 2␤hydroxylation in addition to its major 3␤-hydroxylase activity (30). Furthermore, the same multifunctional dioxygenase is probably responsible for GA 3 synthesis from the 2,3-didehydro intermediate (GA 5 ) by rearrangement of the double bond from C-2,3 to C-1,2 followed by 3␤-hydroxylation (31). Thus, in this case, desaturation at C-1,2, which results in the loss of the 1␤,2␤-H atoms (32), is a side reaction of the main pathway to GA 1 . In contrast, in G. fujikuroi direct 1,2-desaturation by loss of the 1␣, 2␣-H atoms (33) is the major reaction and is accompanied by 2␣-hydroxylation. It is of interest that immature seeds of Prunus persica contain a number of 3-deoxy 1,2-didehydro GAs including GA 120 that are probably formed by direct 1,2-desaturation (34). The seeds also contain 1␣-hydroxy GAs, which could be formed as byproducts of the desaturation. Nothing is known regarding the nature of the putative desaturase/ 1␣-hydroxylase in this case.
We have also shown by gene disruption that P450 -3, the fourth cytochrome P450 monooxygenase gene in the GA gene cluster, encodes the 13-hydroxylase, which catalyzes the final reaction in the biosynthesis of GA 3 . The same enzyme catalyzes also the 13-hydroxylation of GA 4 to form GA 1 . However, little GA 1 (2-3% total GA content) is produced in these two wild-type strains of G. fujikuroi, suggesting either that the fungal cultures contain much more GA 4 desaturase than 13-hydroxylase activity or that GA 4 has a higher affinity for the desaturase than for the 13-hydroxylase. Removal of desaturase activity by disruption of the des gene resulted in approximately a 100-fold accumulation of GA 1 but only a 2-fold accumulation of GA 4 . The mutant strain could thus provide the means to produce large quantities of GA 1 for agricultural or experimental purposes. Similarly, the double mutant lacking both desaturase and 13hydroxylase activities could be used for the production of the commercially important GA 4 .
There was an unexpected difference in the regulation of the des and P450 -3 genes. Expression of des is high under conditions of low N that promote GA production but very low in the growth phase when the N content is high. These expression patterns correspond to those of ggs2, cps/ks, P450 -1, P450 -2, and P450 -4, which are all under control of the major nitrogen regulator, AREA (35,36). In contrast, the expression pattern of P450 -3 differs from that of the other six GA genes. Northern analysis with mycelia grown in medium with a high or low N content showed that P450 -3 is expressed under both conditions, and in fact, more transcript was present under high N conditions. Thus, P450 -3 is the only gene in the GA cluster for which transcript is detectable in 100% ICI medium, which contains 4.8 g⅐liter Ϫ1 NH 4 NO 3 . These results are in agreement with those from incubations of 14 C-labeled GA 4 and GA 7 with P450 -2 knock-out mutants, which do not contain these GAs and later metabolites because the conversion of GA 14 to GA 4 is blocked (14). In cultures growing in low N, both desaturase and 13-hydroxylase activities were present so that GA 3 was the major product formed from GA 4 , whereas in high N, more GA 1 was produced than GA 3 due to low desaturase activity (Table  III). Furthermore, GA 7 was converted to GA 3 regardless of the N content of the medium. The lack of repression of P450 -3 expression by N is consistent with the absence of the double GATA sequence elements in its promoter. These elements, which bind the AREA transcriptional regulator (37), are present in the promoters of the other six GA-biosynthesis genes.
In addition to its different regulation by N, P450 -3 is the only gene in the cluster that cannot be expressed in the deletion mutant SG139, which lacks the gene cluster. In contrast, it was effectively expressed in 6314, which has a point mutation in P450 -3, as well as in SG139 complemented with cosmid 1 containing the entire GA gene cluster. We have not yet investigated how many of the other genes need to be present and active for P450 -3 to be expressed but have addressed the possibility that its expression requires induction by a GAbiosynthetic intermediate by applying precursors to SG139-P450 -3 transformants. However, the presence of ent-kaurenoic acid, GA 14 , GA 4 , or GA 7 , which are the end products of P450 -4, P450 -1, P450 -2, and DES, respectively, did not improve expression. The fusion of the P450 -3 gene to a strong and/or N-repressible promoter will determine whether or not the low expression of the gene under N-limiting conditions is responsible for the failure to complement P450 -3 in SG139 or whether other elements of the gene cluster are needed for gene expression and enzyme activity.
The functional characterization of des and P450 -3 completes the analysis of the GA gene cluster in G. fujikuroi. Disruption of smt (25) and an alcohol dehydrogenase gene 2 at the left border of the cluster and of orf1 and orf2 at the right of P450 -3 2 did not affect the production of GA 3 or its regulation. Thus, the biosynthesis of the final product, GA 3 , in 13 steps beginning with the formation of GGPP by the pathway-specific geranylgeranyl-diphosphate synthase GGS2 requires only seven enzymes, many of which are multifunctional. In contrast to many other fungal secondary metabolite gene clusters, e.g. the aflatoxin gene cluster in Aspergillus parasiticus (38) and the trichothecene gene cluster in Fusarium sporotrichoides (39), the GA gene cluster does not contain a pathway-specific regulatory gene. We have evidence that, in addition to the general regulator AREA, a regulatory gene that controls expression of all of the seven GA-biosynthesis genes exists and must be located elsewhere in the genome.
We are now able to compare GA biosynthesis in G. fujikuroi with that in higher plants at the chemical, biochemical, and gene levels. Although the fungus and plants produce structurally identical GAs and the early enzymatic steps up to the formation of GA 12 -aldehyde are similar, the pathways thereafter differ fundamentally as do the character of enzymes involved and the regulation of their genes. An important difference is that the early 3␤-hydroxylation and 20-oxidation steps in the fungus are catalyzed by cytochrome P450 monooxygenases, whereas soluble dioxygenases are responsible for these reactions in higher plants. Another major difference is that 13-hydroxylation occurs early (at the stage of GA 12 ) in plants, whereas it is the last step in the fungus. Even in the first part of the pathways in which cytochrome P450s participate in the fungus and plants, the equivalent enzymes in G. fujikuroi and plants have low levels of amino acid identity, e.g. only 10% for ent-kaurene oxidase of the fungus (P450 -4) and A. thaliana (40).
In higher plants, GAs play a key role in development and their concentration is maintained at a very low level by a number of endogenous and environmental factors including feedback regulation (7,41). In contrast, wild-type strains of G. fujikuroi produce approximately 1 g⅐liter Ϫ1 GA 3 under laboratory conditions. The absence of feedback regulation in the fungus (14) suggests that GAs do not have an important role in growth and differentiation in this species. On the other hand, as discussed above, the production of fungal GAs in common with that of many fungal secondary metabolites is regulated by N repression. However, the molecular mechanism of the AREAmediated regulation including N sensing, signal transduction, and modulation of activity of the transcription factor AREA is still to be investigated.