FUM9 Is Required for C-5 Hydroxylation of Fumonisins and Complements the Meitotically Defined Fum3 Locus in Gibberella moniliformis

Abstract

Deletion of the Gibberella moniliformis FUM9 gene resulted in mutants that produce only fumonisins that lack a C-5 hydroxyl group. This phenotype is identical to that of previously described mutants with defective alleles at the meiotically defined Fum3 locus. Transformation with a wild-type FUM9 gene into a Fum3-defective mutant restored wild-type fumonisin production. These results indicate that the FUM9 protein catalyzes the C-5 hydroxylation of fumonisins and that FUM9 and the Fum3 locus are the same gene.

Gibberella moniliformis is one of the fungi most commonly associated with maize worldwide. The fungus can cause severe ear and stalk rots, but also is present in healthy maize tissue (9) and can produce high concentrations of the carcinogenic fumonisin mycotoxins. In naturally infected maize, wild-type strains of the fungus produce fumonisins B₁ (FB₁), B₂ (FB₂), B₃ (FB₃), and B₄ (FB₄) (13). These toxins are synthesized via polyketide metabolism and consist of a linear 20-carbon backbone with an amine, one to three hydroxy moieties, and two methyl and two tricarboxylate functions at various positions along the backbone.

Genetic analysis of naturally occurring variants (10) identified four loci required for fumonisin biosynthesis (5-7, 11): Fum1 confers the ability to produce fumonisin, Fum2 confers the ability to hydroxylate carbon atom 10 (C-10) of the fumonisin backbone, Fum3 confers the ability to hydroxylate carbon atom 5 (C-5) of the backbone, and Fum4 controls the amount of fumonisins produced. For example, strains with a defective Fum3 allele produce only fumonisins that lack a C-5 hydroxyl (i.e., FB₃ and FB₄) (6, 15).

The fumonisin biosynthetic cluster in G. moniliformis contains 15 coregulated genes, designated FUM1 and FUM6 through FUM19 (14). FUM9 is one of seven genes in the G. moniliformis fumonisin biosynthetic gene cluster that has not been functionally characterized. Amino acid comparison of the predicted FUM9 protein yielded only low levels of similarity to dioxygenases. However, the predicted FUM9 protein does share significant homology to actinomycete sequences, which in turn are similar to oxoglutarate-dependent dioxygenases (18). Here, we report the results of deletion of FUM9, sequence analysis of a naturally occurring mutant with a fumonisin phenotype identical to the FUM9 deletion mutant, and complementation of the naturally occurring mutation with the wild-type FUM9 gene.

To study the role of FUM9 in fumonisin biosynthesis, we first deleted the gene. A FUM9 deletion vector, pFUM9KOH, was constructed as previously described (2, 3). Briefly, the 1-kb regions immediately upstream and downstream of the FUM9 coding region were amplified by PCR and subcloned into the same vector so they were separated by an AscI restriction site. The hygromycin B resistance gene (HygB) was then inserted between the two fragments, utilizing the AscI site to yield vector pFUM9KOH. Primers 9-1, 9-2, 9-4, and 9-4 were used to amplify these regions (Fig. 1 [primer sequences: 9-1, GACGGATCCGCGGCCTATTGGGACGTACTA; 9-2, GACGGCGCGCCTGCATTGGCGTTGGCAAA; 9-3, GACGGCGCGCCGACGTTTGAATTGTCTTGGCGT; 9-4, GACCTCGAGGGCAACAAACTCCCTGCAAT; 9-5, TCAAGTTCCTCGTAATCGC; 9-6, CACAAGTGGGAGTTCAACC; 9-7, GAAGGTGATGAAGTGTCGG; 10-1, GACACGCGTCAAGGAAATTGGCGCACATAG; rp250, CTGCTGCATTCCCATTCCCATCGT; 1098, ACCAAGCCTATGCCTACAGCATCC]). All PCR products were generated with Pfu polymerase and sequenced to confirm the absence of errors.

FIG. 1.

Southern analysis of FUM9 deletion mutants and 575-R-5 complemented with FUM9. (A) Genomic region of the wild-type (top) and the deleted (bottom) FUM9 coding region (large arrow). H, HindIII; B, BglII. Small arrows indicate positions and orientations of PCR primers. (B) BglII-digested genomic DNA from FUM9 deletion mutants probed with the 773-bp fragment shown in panel A. Note in the mutants GMT-9-206 and GMT-9-211 the loss of the wild-type 1.2-kb FUM9 fragment and the gain of the 3.6-kb fragment resulting from integration of vector sequences by double-homologous recombination and the resulting replacement of the 0.9-kb FUM9 coding region with HygB. (C) ApaI-digested genomic DNA from Fum3-complemented mutants, GMT-9-5-4 and GMT-9-10-8, hybridized to a 2.0-kb fragment carrying the entire FUM9 coding region, including 453 bp upstream of the start site and 672 bp downstream of the stop site. Note additional hybridizing bands in DNA from complemented strains.

To delete FUM9, wild-type G. moniliformis strain M-3125 (8) was transformed with the FUM9 deletion vector via the protoplast method, and transformants were selected on hygromycin B-amended regeneration medium as previously described (3, 16). Eighteen hygromycin B-resistant transformants were recovered and screened by PCR to identify those in which the FUM9 coding region was deleted. The PCR strategy was designed to amplify unique combinations of sequence elements that were formed from homologous recombination between the deletion vector and the FUM9 locus. The PCR primers used for this screen were 9-5, rp250, 1098, and 9-7 (Fig. 1). Transformants also were screened by PCR for the absence of the FUM9 coding region by using primers 9-5, 9-6, 10-1, and 9-7 (Fig. 1). Selected transformants were regenerated from single conidia to ensure genetic homogeneity and then subjected to Southern analysis to confirm deletion of the FUM9 coding region. The Southern analysis employed standard protocols (17) and probes labeled with ³²P by the RediPrime kit (Amersham Biosciences, Piscataway, N.J.). The PCR screens and Southern analysis identified two transformants, GMT-9-206 and GMT-9-211, in which the FUM9 coding region was deleted and replaced with the HygB gene (Fig. 1).

Fumonisin production in transformants was assessed by liquid chromatography-mass spectroscopy of acetonitrile-water (1:1) extracts of 3-week-old cracked corn cultures as previously described (12, 16). The only fumonisins produced by the FUM9 deletion mutants were FB₃ and FB₄. In contrast, the wild-type complement of FB₁, FB₂, FB₃, and FB₄ was produced by transformants in which FUM9 remained intact. Because the C-5 hydroxyl is the only structural feature that is absent in both FB₃ and FB₄ but present in FB₁ and FB₂, these data indicate that FUM9 is required for the C-5 hydroxylation of the fumonisin backbone.

The FUM9 deletion mutants had the same phenotype as previously described mutants with defective alleles of the meiotically defined Fum3 locus. To determine if Fum3-defective mutants carry mutations in FUM9, we amplified and sequenced a 1,591-bp fragment spanning the FUM9 coding region from strain 575-R-5, which has the mutant Fum3-3 allele and therefore cannot hydroxylate the C-5 position of fumonisins. 575-R-5 is a progeny from a sexual cross of wild-type strain M-3125 and a UV-induced mutant derived from wild-type strain M-3120 (15). The sequence of the FUM9 coding region in strain 575-R-5 was identical to that of M-3125, except for a C-to-T transition at nucleotide 94 that is predicted to introduce a stop codon (ochre mutation) in the coding region and a G-to-C transversion at nucleotide 495 that is predicted to result in no amino acid change. Sequence analysis of the same region of DNA in M-3120, which produces the wild-type complement of FB₁, FB₂, FB₃, and FB₄ (6), showed that it also carries the G-to-C transversion.

To determine if FUM9 can complement a Fum3-defective mutant, we transformed 575-R-5 with a wild-type copy of FUM9. A 2.0-kb genomic region containing the wild-type FUM9 coding region was amplified via PCR with Pfu DNA polymerase from cosmid clone 4-5 (14). Nucleotide sequence analysis indicated that the amplified FUM9 did not have any errors. The 2.0-kb amplification product was subcloned into the hygromycin B-containing vector pUCH2-8 (1) to yield the complementation vector pUCH2-8F9. 575-R-5 was transformed with circular pUCH2-8F9, and hygromycin B-resistant putative transformants were recovered and demonstrated by PCR to carry the vector. Southern analysis of two selected transformants revealed that one, GMT-9-5-4, had multiple copies of pUCH2-8F9, while the second, GMT-9-10-8, had only one copy of the vector (Fig. 1). Fumonisin production assays, as described above, revealed that both GMT-9-5-4 and GMT-9-10-8 produced the wild-type complement of B-series fumonisins (Table 1). These results indicate that the wild-type copy of FUM9 complemented the Fum3-defective mutant 575-R-5 and therefore provide further evidence that the ochre mutation in the FUM9 coding region of this mutant results in its altered fumonisin production phenotype.

TABLE 1.

Fumonisin production by wild-type G. moniliformis strain M-3125, Fum3-defective mutant 575-R-5, and transformants GMT-9-5-4 and GMT-9-10-8

Straina	% Fumonisin recovered
	FB1	FB2	FB3
M-3125	77	16	7
575-R-5	0	0	100
GMT-9-5-4	71	14	15
GMT-9-10-8	37	33	30

^aTransformants GMT-9-5-4 and GMT-9-10-8 were generated by transforming 575-R-5 with a vector (pUCH2-8F9) carrying a wild-type copy of FUM9.

In this study, we have identified a direct link between the FUM9 gene and hydroxylation of the C-5 position of the fumonisin backbone. FUM9 deletion mutants produce only FB₃ and FB₄, which lack the C-5 hydroxyl, but not FB₁ and FB₂, which have the hydroxyl. These results suggest that the FUM9 protein is a fumonisin C-5 hydroxylase. This conclusion is consistent with the prediction that FUM9 encodes an oxoglutarate-dependent dioxygenase (18). Such dioxygenases frequently catalyze hydroxylation reactions (4).

Our results also reconcile classical and molecular genetic analyses of fumonisin biosynthesis in G. moniliformis by demonstrating that the meiotically defined Fum3 locus and the molecularly defined FUM9 are the same gene. The first evidence for this identity was that the FUM9 deletion mutants had the same phenotype as Fum3 mutants. Further evidence was that the FUM9 coding region in Fum3-defective mutant 575-R-5 had an ochre mutation that should result in a truncated FUM9 protein. The final evidence was that a wild-type copy of FUM9 could complement the Fum3-defective mutant. Based on these results, we propose that hereafter FUM9/Fum3 be designated FUM3. This designation is consistent with the conventional designation of fumonisin biosynthetic genes (i.e., FUM) and with the precedent set by Fum3 being described before FUM9.