Fum3p, a 2-Ketoglutarate-Dependent Dioxygenase Required for C-5 Hydroxylation of Fumonisins in Fusarium verticillioides

Abstract

Fumonisins are polyketide-derived mycotoxins produced by several agriculturally important Fusarium species. The B series fumonisins, FB₁, FB₂, FB₃, and FB₄, are fumonisins produced by wild-type Fusarium verticillioides strains, differing in the number and location of hydroxyl groups attached to the carbon backbone. We characterized the protein encoded by FUM3, a gene in the fumonisin biosynthetic gene cluster. The 33-kDa FUM3 protein (Fum3p) was heterologously expressed and purified from Saccharomyces cerevisiae. Yeast cells expressing the Fum3p converted FB₃ to FB₁, indicating that Fum3p catalyzes the C-5 hydroxylation of fumonisins. This result was verified by assaying the activity of Fum3p purified from yeast cells. The C-5 hydroxylase activity of purified Fum3p required 2-ketoglutarate, Fe²⁺, ascorbic acid, and catalase, all of which are required for 2-ketoglutarate-dependent dioxygenases. The protein also contains two His motifs that are highly conserved in this family of dioxygenases. Thus, Fum3p is a 2-ketoglutarate-dependent dioxygenase required for the addition of the C-5 hydroxyl group of fumonisins.

Fumonisins are mycotoxins produced by the filamentous fungus Fusarium verticillioides (Sacc.) Nirenberg (synonym F. moniliforme, teleomorph Gibberella moniliformis, synonym Gibberella fujikuroi mating population A) (22, 25). The fungus is a widespread pathogen of maize. It causes ear and stalk rot and can colonize plant tissues without causing visible symptoms (4, 11). Fumonisins induce several animal diseases, including leukoencephalomalacia in horses (20, 23), pulmonary edema in swine (15), and cancer in rats (14). In some areas of China and South Africa, the consumption of fumonisin-contaminated maize is correlated with high incidences of human esophageal cancer (9, 34).

B-series fumonisins are typically the most abundant fumonisins in maize, with fumonisin B₁ (FB₁) making up to approximately 70% of the total content (25). Fumonisins consist of a linear 20-carbon backbone with hydroxyl, methyl, and tricarballylic acid moieties attached at various positions along the backbone (1). This structure is similar to sphinganine, an intermediate in the biosynthesis of sphingolipid. Fumonisins competitively inhibit sphinganine N-acyltransferase, thereby blocking the biosynthesis of sphingolipids, essential components of cell membranes (24, 30). This disruption may be the cause of the fumonisin-induced animal diseases.

The biosynthetic origins of fumonisins are partly established. Carbons 3 to 20 (C-3 to C-20) of the fumonisin backbone are derived from acetate and the amino group and C-1 and C-2 from alanine (2, 3). The methyl groups at C-12 and C-16 are derived from methionine (26), the C-3 hydroxyl group is from an acetate-derived carbonyl group, and the C-5 and C-10 hydroxyl groups are from molecular oxygen (7). The origin of vicinal diol at C-14 and C-15 is not totally clear, but it also could be from molecular oxygen (7). Recently, Proctor et al. (27-29, 32) cloned a 15-gene (FUM1 and FUM6 through FUM19) cluster responsible for the biosynthesis of fumonisins in F. verticillioides. The proteins deduced from the DNA sequence are consistent with the hypothesis that fumonisins are synthesized by a polyketide mechanism followed by modifications of the polyketide backbone. Among these genes, FUM13 is known to encode a C-3 ketoreductase (6), and FUM3 (previously called FUM9) is known to encode the C-5 hydroxylase (5). FUM3 deletion mutants produce fumonisins that lack the C-5 hydroxyl group (i.e., FB₃ and FB₄), the same phenotype exhibited by mutants of F. verticillioides with defective alleles at the meiotically defined Fum3 locus (12). So far, none of the FUM genes has been biochemically characterized. The data from FUM3 mutants suggest that Fum3p catalyzes the C-5 hydroxylation of fumonisins, but direct evidence for the function of Fum3p is still lacking. Our objective in this study was to heterologously express FUM3 and evaluate the Fum3p enzyme. We found that Fum3p is a 2-ketoglutarate-dependent dioxygenase required for C-5 hydroxylation of fumonisins.

MATERIALS AND METHODS

Materials, strains, and vectors.

High-performance liquid chromatography (HPLC)-grade acetonitrile was from EM Science (Darmstadt, Germany) or from Fisher Scientific (Pittsburgh, Pa.). Yeast nitrogen base without amino acids (but with ammonium sulfate), uracil, raffinose, and salmon sperm DNA were from Sigma (St. Louis, Mo.). Standard FB₁, FB₂, FB₃, and FB₄ were gifts from Ronald D. Plattner (U.S. Department of Agriculture [USDA], Peoria, Ill.) or were purchased from Promec Unit, Medical Research Council, Tygerberg, South Africa. F. verticillioides wild-type strain A0149 (FGSC number 7600) was provided by David Gilchrist (University of California, Davis). Plasmid preparation and DNA extraction were carried out with Qiagen kits (Valencia, Calif.), and all other manipulations were carried out according to standard methods (31). Escherichia coli strain DH5α was used as the host for general DNA propagation, and cloning vectors were the pGEM-zf series from Promega (Madison, Wis.). Saccharomyces cerevisiae strain INVSc1 was purchased from Invitrogen (Carlsbad, Calif.). The strain is auxotrophic for histidine, leucine, tryptophan, and uracil. Yeast expression vectors pYES2/NT A, B, and C also were from Invitrogen.

Fum3p expression in yeast.

The FUM3 coding region was amplified by PCR with primers fum3-F-BamHI, 5′-CCGCATGGATCCAACAAGGAAAAGGTTCCC-3′ (underlined letters indicate BamHI site), and fum3-R-EcoRI, 5′-CCCGCGGAATTCTCAATCATCTAACCCAAG-3′ (underlined letters indicate EcoRI site), with F. verticillioides genomic DNA as template and Pfu-Ultra polymerase (Stratagene, La Jolla, Calif.). The PCR product was digested with EcoRI and BamHI and ligated into pGEM-3zf (Promega) at the same sites to produce pDU1. After confirmation by DNA sequencing that no errors were introduced into the coding region during PCR, the cloned FUM3 was transferred to pYES2/NT C via the BamHI and EcoRI sites to produce pDU2. The construct was subsequently transformed into yeast following the manufacturer's instructions (Invitrogen). For protein expression and extraction from the yeast cells, the same procedure was followed as described in the manufacturer's manual. The lysis buffer contained 50 mM sodium phosphate (pH 7.4), 5% glycerol, 1 mM phenylmethylsulfonyl fluoride. To precipitate proteins from the cell extracts, 0.338 g of ammonium sulfate was added slowly to 1 ml of extract solution (50% saturation at 4°C). After gentle shaking for 2 to 3 h, the solution was centrifuged at 16,000 × g at 4°C for 10 min. The supernatant was transferred to a new tube, and 0.201 g of ammonium sulfate was added to the solution (85% saturation at 4°C). The solution was gently shaken at 4°C overnight, and the precipitated proteins were collected by centrifugation at 16,000 × g at 4°C for 10 min. The protein pellets were redissolved in 500 μl of phosphate-buffered saline (PBS) buffer (140 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄; pH 7.4) and dialyzed twice against 2,000 volumes of PBS buffer at 4°C. The proteins were loaded onto a Ni-nitrilotriacetic acid (NTA) column (Qiagen) preequilibrated with PBS buffer, and Fum3p was eluted in PBS buffer containing 250 mM imidazole. Finally, the eluents containing purified Fum3p were combined and desalted on a PD-10 column (Sephadex G-25; Pharmacia Biotech, Piscataway, N.J.) into 50 mM sodium phosphate buffer, pH 7.8, containing 1 mM dithiothreitol and 5% glycerol and stored at −80°C until used in in vitro assays.

Experiments in yeast cultures.

One microliter of FB₃ stock solution (10 μg/μl, in 50% acetonitrile) was added to 200 μl of freshly prepared yeast cells, which had grown in induction medium at 30°C, 250 rpm, for 17 h. After the culture was incubated an additional 30 to 60 min under the same conditions, a 100-μl aliquot was centrifuged at 16,000 × g at room temperature for 8 min. The supernatant was transferred to a clean tube, and the cell pellet was combined with 20 μl of acetonitrile. The mixture was placed at −80°C for 2 to 3 h and thawed under running tap water. After this freezing-thawing procedure was repeated two more times, the mixture was centrifuged at 16,000 × g at room temperature for 8 min. The resulting supernatant was recovered and dried in a SpeedVac concentrator (Savant, Albertville, Minn.). The residue was suspended in 50 μl of H₂O to dissolve the trace amount of fumonisins. An aliquot (50 μl) of the solution was injected into an HPLC apparatus for analysis. As a control, a culture of INVSc1 carrying pYES2/NT C without FUM3 was processed in the same manner.

Enzymatic assays for Fum3p.

The enzyme assay followed previously described methods for 2-ketoglutarate-dependent dioxygenases (13, 19). A typical reaction mixture contained 10 μM Fum3p, 142 μM FB₃ or 290 μM FB₄, 100 μM freshly made Fe(NH₄)₂(SO₄)₂, 200 μM 2-ketoglutarate, 200 μM ascorbic acid, and 100 μg of catalase/μl in a total volume of 50 μl. After incubation at 30°C for 30 min, 50 μl of methanol was added to stop the reaction. The mixture was centrifuged at 16,000 × g at room temperature for 5 min, and a fraction of the supernatant (typically 50 μl) was injected into the HPLC apparatus for analysis. Reactions with boiled Fum3p (10 min in a boiled water bath), without 2-ketoglutarate, Fe(NH₄)₂(SO₄)₂, ascorbic acid, or catalase, also were conducted in parallel. Each reaction had at least three replications.

HPLC-ELSD and LC-ESMS methods.

The HPLC system was a ProStar, model 210, from Varian (Walnut Creek, Calif.) with a column of Alltima C18LL, 5 μm, 250 by 4.6 mm inner diameter (Alltech, Deerfield, Ill.). The mobile phases were water-TFA (100:0.025 [vol/vol]) (A) and acetonitrile-TFA (100:0.025 [vol/vol]) (B), with a gradient of 0 to 40% B in A in the first 5 min, 40 to 60% B from 5 to 10 min, 60 to 80% B from 10 to 15 min, 80 to 80% B from 15 to 20 min, and 80 to 0% B from 20 to 25 min. The flow rate was 1.0 ml/min, and 50 μl was injected in analytic experiments. The conditions set for the evaporative laser scattering detector (ELSD 2000; Alltech) were 45°C of drift tube temperature, 2.0-liter/min nitrogen gas flow, and gain value of 1 in the impactor-on mode. For preparative HPLC, the respective fractions were collected directly from the column according to their retention times. A 50-μl aliquot of the concentrated fractions was reanalyzed by HPLC-ELSD to confirm the presence of the desired compounds. These salt-free fractions were used in liquid chromatography-electrospray mass spectrometry (LC-ESMS) analysis. LC-ESMS used a column of RFC₁₈ Vydac, 5 μm, 250 by 1 mm inner diameter (C₁₈ MassSpec; Vydac). The solvent system was the same as for HPLC-ELSD. The flow rate was 0.05 ml/min, and the injection volume was 50 μl. All positive electrospray spectra were acquired using a VG Platform II mass spectrometer.

RESULTS

Conversion of FB₃ to FB₁ by yeast cells expressing FUM3.

Cultures of the yeast strain INVSc1 carrying pDU2 were fed FB₃. HPLC-ELSD analysis of extracts from these cultures revealed a peak with a retention time of 12.6 min that was not produced in control cultures (INVSc1 transformed with pYES2/NT) that had been fed FB₃. The retention time of this new peak was the same as that of FB₁ (Fig. 1). FB₃ had a retention time of 13.2 min under the same conditions. Since the only difference between FB₃ and FB₁ is a C-5 hydroxyl group, the results indicated that Fum3p expressed in yeast catalyzes C-5 hydroxylation of fumonisins.

FIG. 1.

HPLC-ELSD analysis of extracts from yeast cells fed FB₃. (A) Standard FB₃; (B) acetonitrile-water (1:1 [vol/vol]) extracts of yeast cells expressing FUM3; (C) standard FB₁.

Enzymatic activity of Fum3p.

The Fum3p protein was expressed at a moderate level and was partially soluble under the experimental conditions used. Using a Ni-NTA affinity column, the His₆-tagged Fum3p was purified and yielded a 33-kDa band in sodium dodecyl sulfate-polyacrylamide gel electrophoresis (Fig. 2). Incubation of the purified protein with FB₃ in the presence of 2-ketoglutarate, iron(II), ascorbic acid, and catalase yielded a compound with the same retention time (12.3 min) as FB₁ in HPLC-ELSD (Fig. 3A). The fractions corresponding to all fumonisins (retention times from 12 to 15 min) on HPLC-ELSD were pooled and dried in vacuo. After redissolving in water, the fractions were analyzed by LC-ESMS. Two peaks appeared on the LC. The first had a retention time of 14.8 min and a molecular weight of 722.6, which are identical to the retention time and [M+H]⁺ ion of the FB₁ standard (data not shown). The second peak had a retention time of 15.2 min and a molecular weight of 706.3, which are identical to the retention time and [M+H]⁺ ion of the FB₃ standard (data not shown). The results showed that Fum3p catalyzes the conversion of FB₃ to FB₁. In a reaction mixture containing 5 μg (142 μM) of FB₃, the amount of FB₁ produced by Fum3p was 1,292 ± 91 ng, which represents a 24 to 28% conversion of FB₃ to FB₁. When boiled Fum3p was used in the reaction, no FB₁ was detected by HPLC-ELSD (Fig. 3B). A similar result was obtained when the reactions were run without iron(II) (Fig. 3C), 2-ketoglutarate (Fig. 3D), ascorbic acid (data not shown), or catalase (data not shown). Thus, Fum3p has the characteristics of a typical 2-ketoglutarate-dependent dioxygenase.

FIG. 2.

Sodium dodecyl sulfate-polyacrylamide gel analysis of Fum3p expressed in S. cerevisiae. Lane 1, molecular mass markers; lane 2, total proteins after induction by isopropyl β-d-thiogalactopyranoside; lane 3, unbound proteins of Ni-NTA column; lane 4, 250 mM imidazole eluent of Ni-NTA column.

FIG. 3.

HPLC-ELSD analysis of Fum3p reactions using FB₃ as substrate. (A) Full reaction mixture containing 10 μM Fum3p, 142 μM FB₃, 100 μM Fe(NH₄)₂(SO₄)₂, 200 μM 2-ketoglutarate, 200 μM ascorbic acid, and 100 μg of catalase/μl in a total volume of 50 μl. (B) Full reaction mixture but with boiled Fum3p. (C) Full reaction without Fe(NH₄)₂(SO₄)₂. (D) Full reaction mixture without 2-ketoglutarate.

Parallel to the reactions using FB₃ as substrate, a series of reactions with FB₄ as the substrate also were conducted. Under the same conditions, the standard FB₄ had a retention time of 14.4 min, and the standard FB₂ had a retention time of 13.4 min on HPLC-ELSD (data not shown). However, no observable FB₂ was produced in any of the Fum3p reaction mixtures with FB₄ as the substrate. Varying the concentration of Fum3p or FB₄, with or without the cosubstrate or cofactors, in the reactions did not change these results.

DISCUSSION

So far, eight (FUM1, FUM3, FUM6, FUM8, FUM13, FUM17, FUM18, and FUM19) of the 15 FUM genes have been disrupted or deleted by homologous recombination (5, 6, 27, 28, 32), but only two (FUM3 and FUM13) of these disruptions resulted in the accumulation of detectable intermediates (5, 6). FUM13 encodes a ketoreductase required for the reduction of the C-3 keto group of the carbon backbone (6), and FUM3 encodes a hydroxylase required for the hydroxylation of the C-5 position (5). We biochemically confirmed that the function of the FUM3-encoded protein was to hydroxylate the C-5 position of fumonisins. This conclusion is based on both in vivo and in vitro conversion of FB₃ to FB₁ by the heterologously expressed Fum3p enzyme.

FUM3 is a 903-bp gene located in the middle of the FUM gene cluster (27, 32). Fum3p shares 30 to 40% identity at the amino acid level to several hypothetical proteins, including a hypothetical protein (AAK01519) from Pseudomonas aeruginosa, a conserved hypothetical protein (AAN68422) from Pseudomonas putida KT2440, and hypothetical protein Rv3633 (NP_218150) from Mycobacterium tuberculosis H37Rv, but it did not exhibit any significant similarity to proteins with known functions. In several short regions (50 to 60 amino acid residues), Fum3p has some similarity (approximately 40%) to epoxidase subunit A from Penicillium decumbens (BAA75924) (35), a probable phytanoyl-coenzyme A 2-hydroxylase from Homo sapiens (NP_006205) (8), Rattus norvegicus (P57093) (18), and Mus musculus (NP_034856) (17), l-proline 4-hydroxylase from Dactylosporangium sp. (BAA20094) (33), and MmcH (a putative 2-ketoglutarate-dependent dioxygenase) from Streptomyces lavendulae (AF127374) (21). Thus, the biochemical characterization of Fum3p has implications for the function of the hypothetical proteins identified in other organisms.

Fum3p belongs to the family of 2-ketoglutarate-dependent dioxygenases. These enzymes catalyze a wide range of substrate conversions, including hydroxylation, desaturation, and epoxidation, using O₂ and 2-ketoglutarate as cosubstrates and Fe²⁺ as a cofactor (10, 16). One atom of O₂ is inserted into 2-ketoglutarate, which is subsequently decomposed to give succinate and CO₂, and the other is inserted in the second organic product (e.g., FB₁). Most of these oxygenases require, but not absolutely, ascorbate and catalase for optimal substrate conversion in vitro (10, 16). The role of ascorbate probably is indirect and unrelated to the reaction mechanism, whereas the role of catalase is presumably to protect the dioxygenase from the H₂O₂ generated in the reaction medium (10, 16). The activity of Fum3p was clearly dependent on 2-ketoglutarate, Fe²⁺, ascorbate, and catalase. The conclusion that Fum3p is a 2-ketoglutarate-dependent dioxygenase also is supported by a more detailed sequence analysis. Although an initial database search yielded no clear matches, a careful analysis of the deduced amino acid sequence of the FUM3 product identified His-1 and His-2 motifs, which are highly conserved in this group of dioxygenases (13, 19, 33). The two motifs usually are arranged in the general pattern of H-X-D-X_53-57-H (X being any amino acid residue), in which the two histidines and one aspartate constitute the endogenous ligands of the Fe²⁺ active site. Thus, the residues H₁₄₆, D₁₄₈, and H₂₂₁ of Fum3p could be the facial triad for its active site common to all 2-ketoglutarate-dependent dioxygenases.

Fum3p was expected to convert FB₄ to FB₂. Like FB₁, FB₂ contains the C-5 hydroxyl group, but not the C-10 hydroxyl. FB₃ and FB₄ have the same structure except that FB₃ has the C-10 hydroxyl group. The failure of Fum3p to convert FB₄ to FB₂ suggests that the presence of a C-10 hydroxyl group may be important for the C-5 hydroxylation activity of Fum3p. The enzyme responsible for the addition of the C-5 hydroxyl group of FB₂ has not been identified. One possibility is that FB₄ may not be the direct precursor of FB₂. Further biochemical characterization of other redox enzymes in the FUM cluster, e.g., FUM6, FUM12, FUM 13, and FUM15, should shed further light on the origins of the hydroxyl groups of fumonisins.