Beta Hydroxylation of Glycolipids from Ustilago maydis and Pseudozyma flocculosa by an NADPH-Dependent β-Hydroxylase

Abstract

Flocculosin and ustilagic acid (UA), two highly similar antifungal cellobiose lipids, are respectively produced by Pseudozyma flocculosa, a biocontrol agent, and Ustilago maydis, a plant pathogen. Both glycolipids contain a short-chain fatty acid hydroxylated at the β position but differ in the long fatty acid, which is hydroxylated at the α position in UA and at the β position in flocculosin. In both organisms, the biosynthesis genes are arranged in large clusters. The functions of most genes have already been characterized, but those of the P. flocculosa fhd1 gene and its homolog from U. maydis, uhd1, have remained undefined. The deduced amino acid sequences of these genes show homology to those of short-chain dehydrogenases and reductases (SDR). We disrupted the uhd1 gene in U. maydis and analyzed the secreted UA. uhd1 deletion strains produced UA lacking the β-hydroxyl group of the short-chain fatty acid. To analyze the function of P. flocculosa Fhd1, the corresponding gene was used to complement U. maydis Δuhd1 mutants. Fhd1 was able to restore wild-type UA production, indicating that Fhd1 is responsible for β hydroxylation of the flocculosin short-chain fatty acid. We also investigated a P. flocculosa homolog of the U. maydis long-chain fatty-acid alpha hydroxylase Ahd1. The P. flocculosa ahd1 gene, which does not reside in the flocculosin gene cluster, was introduced into U. maydis Δahd1 mutant strains. P. flocculosa Ahd1 neither complemented the U. maydis Δahd1 phenotype nor resulted in the production of β-hydroxylated UA. This suggests that P. flocculosa Ahd1 is not involved in flocculosin hydroxylation.

INTRODUCTION

Extracellular glycolipids are produced by several microorganisms and are composed of different mono- or disaccharides that are either acylated or glycosidically linked to long-chain fatty acids or hydroxy fatty acids. Because of their amphipathic character, they act as biosurfactants; many glycolipids also have antibiotic properties (20). Because of their unique chemical properties, they have gained importance in the chemical and food industries and are also being exploited for environmental protection (6).

Ustilago maydis (de Candolle) Corda and Pseudozyma flocculosa (Traquair, Shaw & Jarvis) Boekhout & Traquair are two basidiomycete fungi that produce very similar glycolipids with antibiotic activities (8, 17, 18). While U. maydis is a phytopathogenic fungus that infects corn plants and causes smut diseases, P. flocculosa is a natural inhabitant of the phyllosphere that has been described as a biocontrol agent with particular activity against powdery-mildew pathogens (1, 19). Ustilagic acid (UA), the glycolipid produced by U. maydis, is a mixture of four different derivatives consisting of cellobiose o-glycosidically linked to 15,16-dihydroxypalmitic or 2,15,16-trihydroxypalmitic acid. In addition, the UA molecule is acetylated and esterified with a short-chain β-hydroxy fatty acid with a length of either C₆ or C₈. In contrast to UA, flocculosin, which is produced by P. flocculosa, has only one known isomer consisting of 3,15,16-trihydroxypalmitic acid linked to the cellobiose moiety. In addition, the molecule is diacetylated and decorated with β-hydroxyoctanoic acid.

It has been shown that UA biosynthesis is under the control of a gene cluster that includes all genes necessary for its production (27). The biosynthesis genes responsible for flocculosin production are also seemingly arranged within a common gene cluster, but at least one gene, the U. maydis ahd1 homolog, is missing or outside the borders currently defined for the cluster (Fig. 1) (26). Several genes from these clusters have been cloned and functionally analyzed. Among the proteins encoded by the genes, two cytochrome P450 monooxygenases (Cyp1 and Cyp2) (9, 27), an acyltransferase (Uat1 for U. maydis; Fat1 for P. flocculosa), several acetyltransferases (Uat2 for U. maydis; Fat2 and Fat3 for P. flocculosa) (26), and an enzyme involved in α hydroxylation of the long-chain fatty acid during UA biosynthesis (Ahd1) (27) have been identified. From this data, a hypothetical biosynthetic pathway for both glycolipids has been postulated (Fig. 2). During UA or flocculosin biosynthesis, several hydroxylation steps are required. The long-chain fatty acids of both glycolipids are hydroxylated at three different positions, two of which, the terminal and subterminal positions, are under the control of Cyp1 and Cyp2, respectively. While α hydroxylation of UA is catalyzed by Ahd1, it is still unknown which enzyme hydroxylates the β position of the long-chain fatty acid of flocculosin. The short-chain fatty acid requires another hydroxylation step to hydroxylate the β position, but it is not yet known which enzyme catalyzes this transfer.

Fig. 1.

UA and flocculosin biosynthesis gene clusters. The arrows indicate the positions of genes within the clusters and the orientations of transcription. The designations of genes analyzed within this work are underlined. Homologous genes are shown in the same color. For gene designations, see reference 26.

Fig. 2.

Biosynthetic pathway for UA and flocculosin. Enzyme names are provided to the right of the arrows (left, U. maydis enzyme; right, P. flocculosa enzyme). Cyp1 and Cyp2, cytochrome P450 monooxygenases; Ugt1/Fgt1, UDP glucose-dependent glycosyl transferases; Uat2/Fat2 and Fat3, acetyltransferases; Uat1/Fat1, acyltransferases; Ahd1, α-hydroxylase.

Here we report that Uhd1 from U. maydis and its homolog Fhd1 from P. flocculosa are necessary for β hydroxylation of the short-chain fatty acid. Both proteins show homology to members of the family of NADPH-dependent oxidoreductases. By deletion studies and heterologous expression of these genes, we were able to analyze their enzymatic functions during flocculosin and UA biosynthesis. We further demonstrate that Ahd1 from U. maydis has no functional homolog in P. flocculosa, thereby partly explaining the differences between the two organisms in hydroxylation of the long fatty-acid chain.

MATERIALS AND METHODS

Strains, plasmids, and culture conditions.

Escherichia coli strain TOP10 was used for all DNA manipulations. Construction of plasmids was performed using standard procedures (22). P. flocculosa (DAOM 196992) and U. maydis strains FB1 (2) and FB1Δahd1 (27) were maintained on potato dextrose agar (PDA; BD, Mississauga, Canada) at 4°C. To induce glycolipid production in U. maydis, strains were grown at 28°C in liquid YEPS medium (10 g/liter yeast extract, 20 g/liter peptone, 20 g/liter sucrose) to the logarithmic phase and then transferred into a nitrogen starvation medium containing 1.7 g/liter yeast nitrogen base (YNB; BD, Mississauga, Canada) and 5% glucose as a carbon source. Glycolipids were isolated after cultivation of cells for 4 days at 28°C on a rotary shaker set at 200 rpm.

Generation of U. maydis deletion strains.

U. maydis deletion strain FB1Δuhd1 (Mips Ustilago Maydis database [MUMDB] entry number for Uhd1, Um06466) (http://mips.helmholtz-muenchen.de/genre/proj/ustilago/) was generated according to the published protocol for PCR-based generation of gene replacement mutants (13) with minor modifications. Briefly, the 1-kb flanking regions of uhd1 were amplified by PCR and subjected to agarose gel electrophoresis. Primer sequences are listed in Table 1. After purification from the gel (Gel/PCR DNA fragments extraction kit; Avegene, Hamburg, Germany), the PCR products were digested with SfiI (New England BioLabs, Frankfurt, Germany) and ligated to the hygromycin phosphotransferase gene (hph), obtained by digesting plasmid pBS-hhn (13) with SfiI. Ligation was done using 1 U of T4 ligase (Roche Diagnostics, Mannheim, Germany) in a standard 10-μl reaction mixture at 14°C overnight. The ligation product was subjected to agarose gel electrophoresis, and, after purification from the gel, the 4-kb fragment was cloned into pCR2.1-TOPO (TOPO TA cloning kit; Invitrogen, Darmstadt, Germany). Prior to transformation into U. maydis, the plasmid was digested with NotI to separate the construct from the vector. Transformation of U. maydis was performed exactly as described by Brachmann et al. (4). For selection of transformants, PDA plates containing 200 mg/ml hygromycin were used. The mutants were confirmed by Southern analysis and verified by complementation.

Table 1.

Oligonucleotides used in the study

Oligonucleotide name	Sequence (5′ → 3′)a	Description
MC840	gtatgcggccgctccacacacagtggaggg	Amplification of the 5′ region of uhd1 for deletion construct
MC841	cacggcctgagtggcctgtgagcagttgtacatgg
MC842	gtgggccatctaggcctcgtatagcaacggatgaaattcgtg	Amplification of the 3′ region of uhd1 for deletion construct
MC843	cacagcggccgccaagaatcgatgcttgccgg
RB89	gagccatggtaaaaatgacaaaggaagccg	Amplification of fhd1 ORF
RB90	gaggcggccgcgcccgcccgacgatagttat
RB125a	ccatggccacccagactctatc	Amplification of ahd1 ORF
RB125b	gcggccgctagccgtctggcccgaatac
RB18	agaggagtgaggcggtttgg	Amplification of internal probe for gene fat2 (RT-PCR)
RB19	caagagcccaacgctgaacg
RB27	cctcctgctgctgctgctgc	Internal control primers for fhd1
RB28	gccattcgaagtgtacatgg
RB72	ccaacgtcttcttcgacatc	Amplification of internal probe for gene ppi (RT-PCR)
RB73	gcgccgtagatcgacttgcc
RB112	gcctcctcgccaccgaggtc	Amplification of internal probe for gene ahd1 (RT-PCR)
RB113	ggtgccgatgaggaggccgg

^aUnderlined letters indicate introduced restriction sites.

Generation of U. maydis strains carrying P. flocculosa genes.

For expression of P. flocculosa fhd1 and ahd1 (GenBank accession no. HQ292619 [Fhd1] and JN039370 [Ahd1]) in U. maydis, plasmid p123, carrying an enhanced green fluorescent protein (eGFP) reporter under the control of the strong constitutive P_otef promoter and a carboxin resistance (ip) cassette as a selectable marker in U. maydis, was used (25). Genes fhd1 and ahd1 were amplified with primers RB89 and RB90 (fhd1) and primers RB125a and RB125b (ahd1) digested with NcoI-NotI and introduced into p123 under the control of the P_otef promoter. Prior to integrative transformation into the ip locus of U. maydis strain FB1, FB1Δuhd1, or FB1Δahd1, plasmids were linearized with SspI. For selection of transformants, PDA plates containing 2 μg/ml carboxin were used. Mutants were confirmed by PCR amplification using primers RB27 and RB28 (fhd1) and primers RB112 and RB113 (ahd1) (Table 1).

Isolation of glycolipids.

Extracellular glycolipids were extracted from supernatants of suspension cultures with ethyl acetate and analyzed by thin-layer chromatography (TLC) as described previously (9).

Mass spectrometry of glycolipids.

Extracellular glycolipids were extracted from supernatants of suspension cultures with ethyl acetate. High-performance liquid chromatography (HPLC) separation of extracted cellobiose lipids (100 ml) was performed with an 1100-HPLC system (Agilent, Germany) equipped with a 3-mm-diameter Nucleosil 250/3 C8 column (Macherey-Nagel, Germany). The gradient applied at a flow rate of 0.4 ml min⁻¹ and a column temperature of 45°C was as follows (buffer A was water with 0.05% formic acid; buffer B was methanol with 0.045% formic acid): a linear gradient from 60% buffer B to 95% buffer B within 30 min and then holding of 95% buffer B for 10 min. Electrospray ionization-mass spectrometry (ESI-MSⁿ) for structural elucidation of the compounds was performed with a Finnigan LTQ FT mass spectrometer (Thermo Electron, Germany) equipped with a static nanospray source. The ionization voltage was optimized in the range of 800 to 1,800 V to obtain a stable signal. The capillary temperature was set to 200°C, capillary voltage to 37 V, and lens tube voltage to 120 V. Accurate values for masses (and therefore chemical formulas) were obtained by using an FT mass analyzer, which was operated with a resolution of 100,000. For structural elucidations, compounds of interest were subjected to MSⁿ fragmentation using the LTQ mass analyzer and an isolation width of 3 m/z at a normalized collision energy of 35 V, an activation Q of 0.25, and an activation time of 30 ms. Fragment ions were alternatively analyzed in the LTQ or FT mass analyzer.

Reverse transcriptase PCR (RT-PCR).

Total RNA from P. flocculosa cultures was isolated with TRIzol reagent (Invitrogen, Burlington, Ontario, Canada) following the manufacturer's protocol. P. flocculosa was grown at 28°C to the logarithmic phase in YMPD medium (6 g/liter yeast extract, 3 g/liter malt extract, 5 g/liter peptone water, 10 g/liter dextrose) on a rotary shaker. Cells were transferred to either a flocculosin-inducing medium (MOD) (7) or a flocculosin-repressing medium (YMPD) and grown for 24 h at 28°C. RNA was prepared after 24 h.

Total RNA was transcribed into cDNA with SuperScript II reverse transcriptase (Invitrogen, Burlington, Ontario, Canada) following the manufacturer's protocol. A standard PCR using primers RB112 and RB113 (ahd1), primers RB18 and RB19 (fat2), and primers RB72 and RB73 (ppi) was performed (Table 1).

P. flocculosa genome.

The complete genome sequence of P. flocculosa, assembled into 1,281 scaffolds, was used for protein identification based on homology with U. maydis sequences (26). The protein sequences of U. maydis were used to identify the predicted open reading frame (ORF) in the database of the sequenced genome of P. flocculosa by the use of CLC Genomics Workbench version 3.6.1 software (CLC bio USA, Cambridge, MA).

Nucleotide sequence accession number.

The sequence for the ahd1 gene has been deposited in GenBank under accession number JN039370.

RESULTS

The uhd1 and fhd1 genes show homology to the superfamily of extended SDR.

The UA and flocculosin biosynthesis gene clusters contain two homologous genes, uhd1 (UA hydroxylase) and fhd1 (flocculosin hydroxylase), that have not been characterized so far. The deduced nucleic acid sequence of uhd1 encodes a protein of 300 amino acid residues and contains one intron, as confirmed by a query of the MUMDB. The coding region of fhd1 is composed of three exons interrupted by two introns encoding a protein of 323 amino acids. The introns have been confirmed by cDNA sequencing and alignment with the genome sequence of P. flocculosa. To elucidate the function of the two proteins, the deduced amino acid sequences were analyzed with the GenBank BLAST algorithm. This analysis revealed that Uhd1 and Fhd1 are members of the superfamily of the extended short-chain dehydrogenase-reductases (SDR) and show homology to proteins of different plants (Arabidopsis thaliana and Ricinus communis) and fungi (Aspergillus flavus and Aspergillus oryzae) (Fig. 3). All proteins shared the typical TGxxGxxG and HxAS patterns of extended SDR family proteins necessary for cofactor binding, as well as the active site pattern YxxxK (11, 12) (Fig. 3).

Fig. 3.

The genes uhd1 and fhd1 show homology to genes encoding the extended short-chain dehydrogenase/reductase (SDR) superfamily. Sequence alignment of the short signature sequences (TGxxGxxG, HxAS, and YxxxK) typical of members of the extended SDR superfamily is shown. The abbreviations representing species and GenBank accession numbers are as follows: Pf Fhd1, P. flocculosa ADN97212.1; Um Uhd1, U. maydis XP_762613.1; At CAD, Arabidopsis thaliana NP_197445.1 At CCR, A. thaliana NP_178345.1; Rc CCR, Ricinus communis XP_002519377.1; Ao 1820731, Aspergillus oryzae XP_001820731.1; Af 2376456, Aspergillus flavus XP_002376456.1; Pm 2148433, Penicillium marneffei XP_002148433.1. Sequences were aligned using ClustalW. The common motifs are shaded in black. Highly conserved amino acids are shaded in gray.

The U. maydis protein Uhd1 hydroxylates the short-chain fatty acid.

To determine the function of Uhd1 in U. maydis, deletion mutants were generated and the secreted UAs of the mutant strains were analyzed by mass spectrometry. U. maydis FB1 wild-type strains produce four different UA derivatives with measured masses of 791.4078 m/z, 807.4028 m/z, 819.4392 m/z, and 835.4345 m/z [M + Na]⁺ (Fig. 4A) (27). The FB1Δuhd1 deletion strains secreted four derivatives with measured masses of m/z 775.4140, m/z 791.4092, m/z 803.4455, and m/z 819.4383, each of which showed an m/z mass difference of 15.994 compared to the masses of the UAs produced by the wild-type strain (Fig. 4C). In each case, this indicated that a hydroxyl group was missing from the altered UAs produced by Δuhd1 mutants, since the calculated mass for one oxygen atom is 15.999 m/z. It has already been shown that during UA biosynthesis, the two cytochrome P450 monooxygenases Cyp1 and Cyp2 are necessary for the terminal and subterminal hydroxylation of the long-chain fatty acid while the α-hydroxylase Ahd1 hydroxylates its α position (9, 27). This led to the assumption that Uhd1 might be required for the hydroxylation of the remaining hydroxyl group on the short-chain fatty acid. To confirm this hypothesis, the secreted glycolipids were analyzed by tandem mass spectrometry (MS/MS). Fragmentation of the wild-type UA with the mass of m/z 791.4078 showed that the molecule was cleaved into the two glucose molecules, from which the part containing the long-chain fatty acid (m/z 515 [M + Na]⁺) was captured (Fig. 4B) (27). Figure 4D shows the fragmentation spectrum of the smallest compound (m/z 775.4140) secreted by uhd1 deletion strains. The measured fragment showed the same m/z value as the wild-type fragment, indicating that the long-chain fatty acid was not affected by the mutation. This confirms the assumption that Uhd1 hydroxylates the β position of the short-chain fatty acid (Fig. 5C).

Fig. 4.

Mass spectrometry of uhd1 deletion strains. Glycolipid extracts were subjected to high-resolution FTMS analysis coupled with tandem MSⁿ. (A) U. maydis wild-type strains secrete four UA derivatives with masses of m/z 791.4078, m/z 807.4028, m/z 819.4392, and m/z 835.4345 [M + Na]⁺. WT, wild type. (B) Mass fragmentation of the smallest compound (m/z 791.4078) showed that the molecule was situated between the two glucose molecules from which the part containing the long-chain fatty acid (m/z 515 [M + Na]⁺) was captured. (C) Uhd1 deletion strains also synthesize four UA derivatives with masses of m/z 775.4140, m/z 791.4092, m/z 803.4455, and m/z 819.4383 [M + Na]⁺, each differing by exactly 15.994 Da from its corresponding wild-type compound (calculated mass of an oxygen atom, 15.999), suggesting a missing hydroxyl group on the molecule. (D) Mass fragmentation of the smallest compound (m/z 775.4140) secreted by uhd1 deletion strains. The captured fragment showed the same mass as the wild-type fragment, indicating that the long-chain fatty acid was not affected. This suggests that Uhd1 hydroxylates the short-chain fatty acid.

Fig. 5.

Functional analysis of Fhd1. (A) TLC analysis of secreted glycolipids produced by the U. maydis FB1 wild-type strain (WT) and mutant strains FB1Δuhd1 (Δuhd1), FB1Δuhd1-P_otef::fhd1 (Δuhd1+fhd1), FB1Δahd1 (Δahd1), FB1Δahd1-P_otef::fhd1 (Δahd1+fhd1), and FB1Δahd1-P_otef::P.f.ahd1 (Δahd1+ahd1). Deletion of uhd1 in U. maydis resulted in strains producing UAs that showed different UA patterns when analyzed by TLC (lane 2). Expressing the P. flocculosa protein Fhd1 in FB1Δuhd1 strains rescued the phenotype (lane 3). Introducing fhd1 in U. maydis ahd1 deletion strains did not change the phenotype of the mutant (lane 6). P. flocculosa Ahd1, when expressed in U. maydis ahd1 deletion strains, was able to hydroxylate UA, but the reaction was very weak compared to that seen with the wild type (marked with a circle in lane 7). Lanes WT show the typical glycolipid pattern produced by U. maydis wild-type strains. “MEL” indicates the mannosylerythritol lipids also produced by U. maydis. (B) Reverse transcriptase PCR of gene ahd1. cDNA was prepared from strains under inducing (+) and noninducing (−) conditions. Amplification was performed with primers specific for P. flocculosa ahd1 (lanes 1 and 2), fat2 (lanes 3 and 4), and ppi (lanes 5 and 6). While the fat2 control gene was upregulated under flocculosin-inducing conditions, ahd1 showed the same expression characteristics under inducing and noninducing conditions. The ppi gene, which encodes the peptidyl-prolyl cis-trans isomerases, is constitutively expressed. (C) Hypothetical reaction mechanism for Uhd1 and Fhd1. (D) While the α-hydroxylase Ahd1 hydroxylates UA (27), it remains unclear which enzyme hydroxylates flocculosin.

Functional analysis of Fhd1 from P. flocculosa.

To analyze whether the Fhd1 homologous protein from P. flocculosa catalyzed the same reaction as Uhd1, we attempted to complement the Δuhd1 mutants with fhd1, since we were unable to generate P. flocculosa deletion strains (26). For this purpose, fhd1 was constitutively expressed in the FB1Δuhd1 mutant strain. Glycolipids produced by U. maydis wild-type strains and transformants were analyzed by TLC (Fig. 5A); in each case, all four wild-type UA derivatives could be detected (Fig. 5A, lane 1). However, UA secreted by Δuhd1 mutants showed a different glycolipid pattern (lane 2). Figure 5A shows that fhd1 was able to complement the phenotype of a Δuhd1 mutant (lane 3), suggesting that Fhd1 had the same function during flocculosin biosynthesis as Uhd1 during UA biosynthesis (Fig. 5C).

Hydroxylation of the β position of the long-chain fatty acid.

One of the major differences between UA and flocculosin is the position of the hydroxyl groups on the long-chain fatty acid: whereas UA is hydroxylated at the α position, a reaction catalyzed by the α-hydroxylase Ahd1 (27), flocculosin carries a hydroxyl group at the β position. Since the flocculosin gene cluster does not contain an ahd1 homolog, we wondered whether Fhd1 might also be necessary for this reaction. We therefore expressed the corresponding gene in FB1Δahd1 strains, which are mutant strains producing UAs lacking the α-hydroxyl group on the long-chain fatty acid (27). TLC analysis of the UAs produced by these mutants showed that Fhd1 was not able to hydroxylate the β position of the long-chain fatty acid (Fig. 5A, lanes 5 and 6), indicating the specificity of Fhd1 for hydroxylation of shorter fatty acids.

To detect a gene that would catalyze hydroxylation of the β position of the long-chain fatty acid, we inspected the genome sequence for candidate genes and found one with 63% homology to ahd1 (E-value, 2 × 10⁻¹⁰¹). TLC analysis of Δahd1 mutant strains expressing ahd1 of P. flocculosa showed that the latter was able to hydroxylate UA but that the reaction was very weak compared to that seen with the wild type (Fig. 5A, lane 7). This led us to question whether ahd1 was upregulated under flocculosin-producing conditions. Therefore, we analyzed expression of ahd1 under flocculosin-repressing (−) and -inducing (+) conditions by RT-PCR. Figure 5B shows that the fat2 control gene, which is part of the cluster, is strongly upregulated under flocculosin-inducing conditions. In contrast, Ahd1 shows the same expression characteristics under either set of conditions, indicating that ahd1 is not involved in flocculosin biosynthesis (Fig. 5D).

DISCUSSION

In this work, we have identified two β-hydroxylase genes, uhd1 from U. maydis and fhd1 from P. flocculosa, that play a role in the biosynthesis of UA and flocculosin, respectively. Both genes encode proteins belong to the extended SDR superfamily involved in different oxidoreductase reactions. By mutational analysis, mass spectrometry, and complementation assays, we have showed that Uhd1 and Fhd1 are necessary for the hydroxylation of the β position of the short-chain fatty acid of their corresponding glycolipids. Deletion of uhd1 in U. maydis resulted in mutant strains producing UA that lacked the β-hydroxyl group on the short-chain fatty acid. Complementation of Δuhd1 strains with fhd1 from P. flocculosa restored the Δuhd1 phenotype, demonstrating the same function for Fhd1 during flocculosin biosynthesis.

The exact reaction mechanism(s) of Uhd1 and Fhd1 is still unknown. Both proteins show homology to proteins of the extended SDR superfamily. Extended SDRs form a diverse collection of proteins, including isomerases, epimerases, oxidoreductases, and lyases (5, 12). They are similar to the “classical” SDRs but have a less conserved C-terminal extension of approximately 100 amino acids (11). All SDRs are characterized by a typical α/β folding pattern, the Rossmann fold, consisting of a central β sheet flanked by α helices required for NAD(P) (H)-binding (21). Uhd1 and Fhd1in particular showed sequence homology to plant proteins of the cinnamoyl-coenzyme A (cinnamoyl-CoA) reductase (CCR) family involved in lignin biosynthesis and to the dihydroflavonol-4-reductase (DFR) necessary during anthocyanin biosynthesis (10, 14, 24). In both cases, the enzyme catalyzes the reduction of its appropriate substrate, thereby yielding a hydroxylated product. The DFR reaction mechanism has been previously described for Gerbera hybrids. DFR catalyzes the stereospecific reduction of dihydroflavonols to the respective leucoanthocyanidins, which are precursorss for the synthesis of anthocyanins, the major water-soluble pigments in flowers and fruits (15). Through this reaction, a ketone is reduced to an alcohol. It has not been shown that the shorter fatty acid carries a ketone group in the β position during flocculosin or UA biosynthesis. If Uhd1 and Fhd1 were necessary for the reduction of a ketone group, deletion of uhd1 should result in the production of cellobiose lipids carrying a β-ketone group on the short-chain fatty acid. This indicates that Uhd1 and Fhd1 do not reduce a ketone group but directly transfer a hydroxyl group to the substrate (see Fig. 5C). They might have a reaction mechanism similar to that of cytochrome P450 monooxygenases, which are able to transfer one oxygen atom from molecular oxygen directly to various substrates by the use of NAD(P)H as an electron donor.

Interestingly, Uhd1 and Fhd1 seem to exhibit substrate specificity for short-chain fatty acids. Expressing fhd1 in U. maydis FB1Δahd1 deletion strains did not change the ahd1 deletion phenotype, indicating that Fhd1 was not able to hydroxylate long fatty acid chains. This specificity might therefore make the enzyme suitable for practical applications in metabolic engineering where one could, for instance, use the enzyme to hydroxylate short-chain fatty acids specifically. Hydroxylated fatty acids have been reported to be valuable compounds in the chemical and medical industry (3).

Which enzyme catalyzes the transfer of the β-hydroxyl group to the long-chain fatty acid of flocculosin is still an open question. We demonstrated that neither Fhd1 nor Ahd1 was responsible for this reaction. For U. maydis, it has been shown that hydroxylation of the α position is done by Ahd1 (27). Expression of a P. flocculosa homolog (located outside the flocculosin biosynthesis cluster) in U. maydis strains lacking ahd1 showed that this protein had very low specificity with respect to hydroxylation of UA. Based on this result, the fact that it is not expressed under flocculosin-producing conditions, and its exclusion from the cluster, we conclude that P. flocculosa Ahd1 is not significantly involved in flocculosin biosynthesis. As a matter of fact, one could even question the necessity of ahd1 in the U. maydis cluster, given that only half of the molecules are hydroxylated. Preliminary data show that UA produced by Δahd1 strains still show antibiotic activities (B. Teichmann and M. Bölker, unpublished data), but we do not know whether they are lacking another characteristic. The fact that P. flocculosa found another mechanism to systematically hydroxylate the long-chain fatty acid seems to indicate that this hydroxyl group is essential for the activity of the molecule or, at the very least, that it confers a fitness advantage to the organism.

In conclusion, we have found that the production of unusual glycolipids by two related yet disparate organisms is the result of an intricate and well-conserved enzymatic process exclusive to the two studied fungi. From a biological or evolutionary point of view, one has to assume that conservation of this cluster serves a distinct purpose, even though evidence to that effect is still lacking. Two closely related plant pathogens, Sporisorium reilianum and Ustilago hordei, lack such a cluster (23; J. Schirawski, R. Kahmann, and G. Bakkeren, personal communication) but are much less prevalent than U. maydis (16). This could indicate that production of glycolipids with antibiotic activity does extend the ecological niche and enhance the ability to reproduce.