The sirodesmin biosynthetic gene cluster of the plant pathogenic fungus Leptosphaeria maculans

Authors


Summary

Sirodesmin PL is a phytotoxin produced by the fungus Leptosphaeria maculans, which causes blackleg disease of canola (Brassica napus). This phytotoxin belongs to the epipolythiodioxopiperazine (ETP) class of toxins produced by fungi including mammalian and plant pathogens. We report the cloning of a cluster of genes with predicted roles in the biosynthesis of sirodesmin PL and show via gene disruption that one of these genes (encoding a two-module non-ribosomal peptide synthetase) is essential for sirodesmin PL biosynthesis. Of the nine genes in the cluster tested, all are co-regulated with the production of sirodesmin PL in culture. A similar cluster is present in the genome of the opportunistic human pathogen Aspergillus fumigatus and is most likely responsible for the production of gliotoxin, which is also an ETP. Homologues of the genes in the cluster were also identified in expressed sequence tags of the ETP producing fungus Chaetomium globosum. Two other fungi with publicly available genome sequences, Magnaporthe grisea and Fusarium graminearum, had similar gene clusters. A comparative analysis of all four clusters is presented. This is the first report of the genes responsible for the biosynthesis of an ETP.

Introduction

Sirodesmin PL is a non-host specific phytotoxin produced by the dothideomycete Leptosphaeria maculans‘brassicae’ (Desm.) Ces. et de Not. [anamorph = Phoma lingam (Tode: Fr.) Desm.]. In addition to causing chlorotic (yellow) lesions on plant leaves, sirodesmin PL has antibacterial and antiviral properties (Rouxel et al., 1988). Sirodesmin PL is a member of the epipolythiodioxopiperazine (ETP) class of fungal secondary metabolites and is characterized by a disulphide bridge across a diketopiperazine ring (Fig. 1).

Figure 1.

The related epipolythiodioxopiperazines (ETPs) sirodesmin PL (A) and gliotoxin (B). The core ETP moiety is the disulphide bridged diketopiperazine ring.

Leptosphaeria maculans causes blackleg of canola (Brassica napus), the most damaging disease of this crop worldwide (for review see Howlett et al., 2001). The role of sirodesmin PL in blackleg disease is unclear. Its detection in infected canola leaves depends on the plant growth conditions (Pedras and Seguin-Swartz, 1992; Sock and Hoppe, 1999). Ultraviolet light induced mutants that produce low levels of sirodesmin PL in culture have similar pathogenicity to that of wild type isolates in cotyledon assays (Sock and Hoppe, 1999). However, these mutants cause smaller stem lesions compared to those caused by the parent isolates (Sock and Hoppe, 1999). Sirodesmin PL appears to be involved in a complex interplay between the host and the fungus; for instance, its production is suppressed by brassinin, a phytoalexin (antimicrobial compound) of canola (Pedras and Taylor, 1993). In contrast, closely related Leptosphaeria species such as L. biglobosa (formerly B group isolates) do not produce sirodesmins (Pedras and Biesenthal, 2000).

Like sirodesmin PL, other ETPs have antibacterial and antiviral properties as well as being immunosuppressive (Waring and Beaver, 1996). The most studied ETP is gliotoxin (Fig. 1) which is involved in the mammalian mycotoxicosis invasive aspergillosis (Sutton et al., 1994; Sutton et al., 1996). Gliotoxin is produced by several fungi including Aspergillus fumigatus, an opportunistic human pathogen, and others such as Trichoderma spp (Wilhite et al., 2001), Penicillium spp (Macdonald and Slater, 1975) and some Candida spp (Shah et al., 1995). In vitro experiments have shown that gliotoxin causes both apoptotic and necrotic cell death (Hurne et al., 2002). The sporidesmin ETPs are produced by Pithomyces chartarum, and cause liver damage in livestock (Cheeke, 1995). Additionally, ETPs are concentrated within the target cell by reduction of the disulphide bond, which enhances their toxicity (Bernardo et al., 2003).

There are at least two reported mechanisms of toxicity of ETPs (Chai and Waring, 2000). First, ETPs cause the generation of reactive oxygen species via redox cycling between the oxidized (disulphide) and reduced (dithiol) forms. This activity has been well characterized for the sporidesmins (Munday, 1984; 1987). Second, the formation of mixed disulphides with free thiol groups on proteins leading to their inactivation has been demonstrated extensively with gliotoxin (Mullbacher et al., 1986; Hurne et al., 2000; Moerman et al., 2003).

As yet there are no reports of the genes responsible for the biosynthesis of an ETP. The early steps for biosynthesis of both sirodesmin PL and gliotoxin are predicted to include condensation of two amino acids by a non-ribosomal peptide synthetase; tyrosine and serine for sirodesmin PL (Ferezou et al., 1980), phenylalanine and serine for gliotoxin (Bose et al., 1968a,b). For the biosynthesis of sirodesmin, a prenyl transferase would add a dimethylallyl group to either the dipeptide cyclo-l-tyrosyl-l-serine or free tyrosine, producing the readily detectable intermediate phomamide (Ferezou et al., 1980). Subsequently, a series of oxidations, sulphurization, a claisen rearrangement reaction, methylation and acetylation are also predicted (Ferezou et al., 1980; Bu’Lock and Clough, 1992). A proposed pathway for biosynthesis of sirodesmin PL is shown in Fig. 2. Evidence for this pathway is provided by labelling experiments (Ferezou et al., 1980; Bu’Lock and Clough, 1992) and analysis of putative intermediates (Pedras et al., 1990; Pedras, 2001).

Figure 2.

Proposed biosynthetic pathway of sirodesmin PL by Leptosphaeria maculans. Known compounds are named in bold type. Postulated intermediates are bracketed and numbered. The pathway is a modified version of that presented by others (Bu’Lock and Clough, 1992; Ferezou et al., 1980). Dimethylallyl-PP, dimethylallyl pyrophosphate. Some steps are likely to be multistep reactions catalysed by more than one enzyme.

The genes responsible for the biosynthesis of fungal secondary metabolites are typically clustered in the genome (Keller and Hohn, 1997). For example, gene clusters for the production of aflatoxin and related compounds have been identified in a range of distantly related ascomycetes (Bradshaw et al., 2002). While the biological reason for this clustering is still the subject of debate (Rosewich and Kistler, 2000), this clustering makes the identification of many of the genes involved in secondary metabolite biosynthesis relatively straight forward. In this paper we report the cloning of the gene cluster in L. maculans responsible for the biosynthesis of sirodesmin PL and describe a similar gene cluster in the gliotoxin-producing organism A. fumigatus. The disruption of the non-ribosomal peptide synthetase present in the gene cluster will allow a direct assessment of the role of sirodesmin PL in blackleg disease of canola. The identification of a cluster of genes responsible for the biosynthesis of, and self-protection against, sirodesmin may lead to novel strategies to deal with ETP toxins in a diverse range of diseases.

Results and discussion

The sirodesmin biosynthetic gene cluster is predicted to contain 18 genes

A homologue of the dimethylallyl tryptophan synthase gene from the fungi Neotyphodium coenophialum, Claviceps pupurea and Penicillium paxilli and a protein containing a thioredoxin reductase domain (Marchler-Bauer et al., 2003) were identified in a set of expressed sequence tags (ESTs) generated from L. maculans mycelia grown in 10% Campbell's V8 juice (A. Idnurm, A.J. Cozijnsen and B.J. Howlett, unpubl. data). These genes were present on the same cosmid clone. Further sequencing of this and two linked cosmids revealed a group of genes reminiscent of fungal secondary metabolite production gene clusters. In total, 23 putative genes, 18 of which are thought to be part of the sirodesmin biosynthesis cluster (see below), were identified in a 68 kb region of L. maculans DNA (Fig. 3, GenBank AY553235). Their database matches and proposed function are shown in Table 1. These genes are denoted with the prefix sir followed by a single letter. Other genes that flank the gene cluster that are not predicted to play a role in sirodesmin PL biosynthesis are also listed in Table 1, and are denoted with the prefix Lm followed by an abbreviation of their proposed function.

Figure 3.

Putative sirodesmin biosynthesis gene cluster (shown in black) and surrounding genes (grey) in Leptosphaeria maculans. The best Blast match and predicted function of each gene is shown in Table 1. The diagram is approximately to scale.

Table 1. Best matches of genes present in 68 kb of contiguous sequence of Leptosphaeria maculans DNA containing the putative sirodesmin biosynthetic gene cluster. Genes with putative roles in the sirodesmin pathway are designated sir. Other flanking genes with no predicted function in sirodesmin biosynthesis have an Lm prefix.
Gene Blast matcha: Gene (function), SpeciesGenBank Accessione-valueProposed function/commentsb
  • a. 

    Match listed excludes hypothetical proteins. For most genes, matches with higher e-values were identified to hypothetical proteins from various fungi.

  • b. 

    Denoted as being involved in biosynthesis of core ETP moiety if there is a match in Aspergillus fumigatus (see text). Most comments and assignments of function are tentative, because of the predicted nature of many of the intermediates of the biosynthesis of sirodesmin PL.

LmHDX1 Salicylate hydroxylase, Pseudomonas putida BAA61829 1e-31Hydroxylase
LmPKS1 PksA (Aflatoxin biosynthesis polyketide
synthase), Aspergillus nidulans.
AAR32704 0Polyketide synthase
LmUV1I-h UVI-1 (Ultraviolet inducible gene 1), Bipolaris oryzae BAA96294 3e-69Transcript accumulates when exposed to UV light and during appressorium formation
sirH Tri7 (T-2 toxin C-4 acetylation), Gibberella zeae AAK53575 4e-16Deacetylsirodesmin PL acetyl transferase
sirQ Progesterone 5-β-reductase, Digitalis purpurea CAC80137 2e-14Multiple hypothetical matches in fungal
genomes with e-values as high as 8e-45
sirI 1-Aminocyclopropane-1-carboxylic acid synthase (ACCS), Penicillium citrinum JC7730 3e-46Core ETP moiety biosynthesis
sirJ Microsomal dipeptidase, Schizosaccharomyces pombeNP_5941931e-74Hypothetical matches with e-values as high
as 1e-90. Possibly involved in release of the
cyclic dipeptide from the peptide synthetase
or autoprotection. Core ETP moiety
biosynthesis.
sirR Progesterone 5-β-reductase, Digitalis purpurea CAC80137 3e-18Hypothetical matches with e-value as high
as 5e-35. Domain match to Nucleoside-
diphosphate-sugar epimerase (96% aligned
e-value 2e-11)
sirB Cytochrome P450 monooxygenase, Fusarium fujikuroi CAA76703 5e-48Oxidation
sirP Peptide synthetase NcpA, Nostoc sp. ATCC 53789 AAO23333 e-178Two-module peptide synthetase.
Condensation of (dimethylallyl) tyrosine and serine. Core ETP moiety biosynthesis
sirG Glutathione S-transferase, Alternaria alternata AAR98813 2e-49Autoprotection and/or removal of glutathione
conjugates. Core ETP moiety biosynthesis
sirC Cytochrome P450 monooxygenase (cholesterol desmolase), Mesocricetus auratusQ9EPT41e-12Oxidation (intermediates 1–2, Fig. 2). Core ETP moiety biosynthesis
sirM MmcR (O-methyl transferase), Streptomyces lavendulae AAD32742 e-21Methyl transferase. Core ETP moiety
biosynthesis
sirE Cytochrome P450 monooxygenase, Fusarium fujikuroi CAA76703 2e-47Oxidation
sirS Progesterone 5-β-reductase, Digitalis purpurea CAC80137 6e-11Hypothetical matches with e-values as high
as 5e-30. Domain match to Nucleoside-
diphosphate-sugar epimerase (81% aligned
e-value 3e-5)
sirZ PrnA (Zinc finger transcriptional regulator), A. nidulans CAA11374 6e-6Transcriptional regulation of sir genes
sirA MDR1 (ABC transporter), A. fumigatus AAB88657 0.0Sirodesmin efflux
sirN Putative methyltransferase (tricothecene), Fusarium sporotrichioides AAO27746 1e-10Unknown. Domain match to methyl transferase only partial. Core ETP moiety biosynthesis
sirO Aflatoxin β1 aldehyde reductase, Mus musculusNP_0796138e-45Reduction of carbonyl group of intermediate
4 to produce deacetyl sirodesmin PL
sirD l-Tryptophan dimethylallyl transferase,
Neotyphodium coenophialum
AAP81207 3e-30Transfer of dimethylallyl group to tyrosyl residue
sirT Thioredoxin reductase, Cytophaga hutchinsoniiZP_001179449e-25Disulphide bond formation (oxidation). Core ETP moiety biosynthesis
LmMP1 Maltose permease, Aspergillus oryzae BAB59002 e-100Maltose transport
LmZnF3 Trancription factor, Schizosaccharomyces pombe T37601 1e-04Transcriptional regulator

Based on homology to other genes, a number of genes present in the cluster could be assigned putative roles in the biosynthesis of sirodesmin. In addition to the prenyl transferase (sirD) and thioredoxin reductase (sirT), the presence of a two-module non-ribosomal peptide synthetase (sirP) supported the role of the gene cluster in the production of sirodesmin. An acetyl transferase also required for the biosynthesis of sirodesmin (final step in Fig. 2), is the function proposed for sirH, a homologue of the trichothecene acetyl transferase Tri7 (Brown et al., 2001). An N-methyl transferase was expected to be involved in the production of sirodesmin, to methylate the seryl residue nitrogen atom (intermediate 3 in Fig. 2). Two putative methyl transferases were identified in the gene cluster, an O-methyl transferase (sirM) and unassigned methyl transferase (sirN). The domain match for the sirN predicted protein was weak and only partially aligned (57%), possibly indicating that the matching region only represents a substrate binding site; for example, for the methyl donor S-adenosyl methionine (SAM). Whether either of these function as an N-methyl transferase remains to be determined. Transporter proteins are common components of fungal secondary metabolite clusters (Del Sorbo et al., 2000) and provide the dual role of toxin export and self protection. The L. maculans cluster contained an ATP-binding cassette (ABC) type transporter, sirA (formerly LmABCt4, Gardiner and Howlett, 2004). Many of the other genes present in the cluster, such as the cytochrome P450 monooxygenases (sirB, sirC and sirE) and oxidoreductase (sirO) are common components of other secondary metabolite clusters and would be expected to catalyse oxidation/reduction reactions in sirodesmin PL biosynthesis (Fig. 2).

Some other genes in the cluster have no immediately obvious function in sirodesmin biosynthesis, as their homologues have not been characterized in secondary metabolite production gene clusters previously. In particular glutathione S-transferase (sirG), dipeptidase (sirJ), aminocyclopropane-1-carboxylic acid synthase (ACCS) (sirI) and putative progesterone 5-β-reductase (sirQ, sirR and sirS) homologues, are to varying extents novel components of fungal secondary metabolite gene clusters and their possible roles in the biosynthesis of sirodesmin are discussed below.

Glutathione S-transferases (GST) are common detoxification enzymes (Hayes and Pulford, 1995; Hayes and Strange, 1995). The A. nidulans sterigmatocystin biosynthetic gene cluster contains a predicted protein with a GST domain although it is described as an elongation factor homologue (Brown et al., 1996), and remains uncharacterized with respect to metabolite production. Total cellular GST activity is also correlated with aflatoxin production in A. parasiticus (Allameh et al., 2002). The observations that gliotoxin-glutathione conjugates form non-enzymatically (Bernardo et al., 2001) and that glutathione is necessary for intracellular accumulation of gliotoxin in target cells (Bernardo et al., 2003), may indicate that ETP-glutathione conjugates could occur during the biosynthetic process and require degradation by the reverse reaction of a GST. While the sirG product may be involved in auto-detoxification, a role in the biosynthetic process cannot be excluded. Likewise, the dipeptidase gene (sirJ) which could be involved in autoprotection (by cleaving the peptide bond of excess toxin), may also have a biosynthetic role. Because the predicted sirP product lacks a recognizable thioesterase (cyclization) domain, the sirJ gene product may remove the dipeptide tethered to the peptide synthetase.

ACCS is a key step in the biosynthesis of ethylene in plants (Bleecker and Kende, 2000). A fungal homologue has been identified in P. citrinum and biochemically shown to catalyse the same reaction as in plants (Jia et al., 1999), but has not been assessed for a role in the production of secondary metabolites. The ACCS reaction uses SAM and releases methylthio-adenosine (CH3S-adenosine) (Bleecker and Kende, 2000). A number of CH3S-substituted dioxopiperazine compounds have been identified from ETP-producing organisms including Aspergillus spp, L. maculans and Fusarium clamydosporum, but while the existence of these compounds may suggest a role in ETP biosyntheses they have not been placed in the biochemical pathways as yet and may also be by-products (Kirby and Robins, 1980; Kawahara et al., 1987; Pedras et al., 1990; Usami et al., 2002). Whether the ACCS reaction is related to the presence of CH3S- groups is unknown. Labelling experiments have shown that methionine, cysteine and sodium sulphate can all act as sulphur donors in ETP biosyntheses, although cysteine is thought to be the direct donor and occurs without methyl groups attached (P. Waring, pers. comm.; Kirby and Robins, 1980).

Despite the weak homology to one putative reductase involved in the biosynthesis of the plant steroid cardenolide (Table 1), the sirQ, sirR and sirS genes are novel genes with respect to fungal secondary metabolite gene clusters. All other database matches were to hypothetical proteins. The three gene products had a high degree of sequence similarity to each other (minimum pairwise identity 34% and similarity 53%). None of them have identifiable oxidoreductase domains, but sirR and sirS are predicted to contain nucleoside diphosphate sugar epimerase domains (Marchler-Bauer et al., 2003). This type of epimerase catalyses the stereochemical shifting of a hydroxyl group around a carbon atom via a transient NAD+ dependent oxidation of the hydroxyl group (Tanner, 2002). The epimers of sirodesmin occur at the junction of the two terminal five member rings but the reactions that determine the final stereochemistry may occur before ring rearrangements (Bu’Lock and Clough, 1992). Determining whether these genes are also present in the related dothidiomycete Sirodesmium diversum which also produces sirodesmins (Curtis et al., 1977), but in differing abundances of chiral isomers (Bu’Lock and Clough, 1992), may provide some insight in the role of these genes in the biosynthesis of sirodesmin.

The peptide synthetase (sirP) is essential for sirodesmin PL biosynthesis

Disruption of sirP was carried out with a targeting construct containing 7 kb of homologous sequence and two copies of the toxic flanking marker thymidine kinase, which increases the efficiency of recovery of transformants with a targeted gene (Fig. 4) as previously described (Gardiner and Howlett, 2004). Fifty transformants that grew in the presence of trifluorothymidine were screened using PCR (data not shown) and one disruptant was identified, and its identity confirmed using Southern analysis (Fig. 4). The sirP mutant shows similar in vitro growth to the parent isolate (sporulation and growth rate) and causes lesions of similar sizes to that of the parent isolate on cotyledons (data not shown).

Figure 4.

Targeted disruption of Leptosphaeria maculans peptide synthetase gene, sirP.
A. A 7 kb genomic fragment was PCR-amplified using aatB recombination site tailed primers and cloned into a Gateway®-fitted thymidine kinase (tk) containing binary vector. The disruption cassette was introduced using an in vitro transposition reaction that introduces the region shaded in grey.
B. Southern analysis of sirP disruptant (Δ) and two ectopic integrants (e). DNA was digested with HindIII. The probe was generated by PCR using psc16 and psc21 primers and is indicated by a black box in (A).
RB, right border; LB, left border; CAT, chloramphenicol acetyl transferase gene; HygR., hygromycin phosphotransferase gene; aatB1 and aatB2, bacteriophage lambda (Gateway®) recombinase recognition sites; H, HindIII.

The production of sirodesmin PL in the parent isolate (IBCN 18) and sirP mutant after 6 days culture in 10% Campbells V8 juice was analysed using HPLC (Fig. 5). The HPLC profile of the sirP mutant lacked the peak corresponding to sirodesmin PL. Of three transformants tested in which ectopic integration of the targeting construct had occurred, all showed similar HPLC profiles to the parent isolate (data not shown).

Figure 5.

HPLC analysis of ethyl acetate extracts of 6 d old culture filtrates of the parent isolate (IBCN 18) (A) and sirP disruptant (B) of Leptosphaeria maculans. The major peak (2) eluting at 18.8 min has identical retention properties to a standard preparation of sirodesmin PL and is absent from the sirP mutant. The peak (1) eluting at 14 min is the internal control, gliotoxin, added to the culture filtrate immediately before extraction. Minor peaks eluting at 12.5 and 13.5 min in the wild type but not in the sirP mutant are most likely deacetyl sirodesmin PL and phomamide, respectively (Pedras and Biesenthal, 1998).

Genes in the sirodesmin biosynthetic gene cluster are co-regulated and their expression corresponds with sirodesmin production

Analysis of the regulation of genes within the cluster was carried out using quantitative reverse transcription polymerase chain reaction (qRT-PCR). Co-regulation of nine (sirP, sirD, sirA, sirG, sirH, sirI, sirJ, sirT, sirZ) of the 18 sir genes was assessed during a culture time course using isolate IBCN 18 grown in 10% V8 juice (Fig. 6). The expression of each gene peaked at 4 days after inoculation of the culture. The correlation coefficients for pairwise correlation between each gene was, with the exception of sirA and sirZ, greater than 0.95. Expression of sirA had a correlation with all other genes of the cluster of greater than 0.8, while sirZ showed a correlation of at least 0.75 with the other genes. Expression of the transporter gene (sirA) initially increased in a similar pattern to that of the other genes, but remained high after the expression level of the other genes decreased. This is consistent with the need for L. maculans to continually maintain a low intracellular level of sirodesmin PL. The strong correlation of the sirG and sirJ genes with other genes with predicted biosynthetic functions supports these genes having a biosynthetic roles rather than involvement in autoprotection. At the very least, expression of sirG and sirJ occurs at a level related to the rate of biosynthesis rather than the amount of end product (cf. sirA). The magnitude of expression of all genes (relative to actin) except sirH and sirZ was similar. While this would be expected for the transcriptional regulator sirZ, it was not expected for the transcript of a catalytic enzyme such as the predicted acetyl transferase (sirH). It is possible that the lower level of sirH expression is responsible for the relative abundance of the deacetyl derivative of sirodesmin PL (compared to other intermediates such as phomalirazine). The biological significance of this lower expression of sirH is unknown.

Figure 6.

Time course of expression of 12 genes from a 68 kb region of the Leptosphaeria maculans genome. Gene expression level is relative to actin. β-Tubulin is included to show that the calculated expression levels were not resulting from variations in the expression level of actin. All genes thought to be part of the cluster showed similar regulation during culture, peaking in expression at day 4. Other genes, LmMP1, LmPKS1 and LmUVI-1h were not co-regulated with the sir genes. Each data point is the average of three replicate PCR reactions with the standard error of the mean normalized expression level indicated by error bars. All scales are the same except for LmMP1.

Expression of genes not thought to be involved in the biosynthesis of sirodesmin, namely LmMP1, LmPKS1 and LmUVI-1h in addition to β-tubulin, did not show correlation with the sir genes. Interestingly expression of the adjacent genes LmPKS1 and LmUVI-1h were strongly correlated with each other (r = 0.98), albeit with different magnitudes.

Sirodesmin production was quantified using the culture filtrates of the mycelia for which the time course of gene expression was carried out. Sirodesmin PL was first detected in culture filtrates 4 days after inoculation at the same time that expression of the genes was at a maximum (Fig. 7). The close correspondence between the profiles of gene expression (Fig. 6) and rate of sirodesmin PL production is seen in Fig. 7.

Figure 7.

Sirodesmin PL production during in vitro growth of Leptosphaeria maculans. Culture filtrates of the mycelia used for quantitative RT-PCR experiments (Fig. 6) were analysed. Data points are expressed as the mean of two separate extractions.

A similar cluster is present in the gliotoxin-producing fungus Aspergillus fumigatus

Using sirT as a query in genome database searches, a homologue was identified in the A. fumigatus genome closely linked to a two-module non-ribosomal peptide synthetase gene similar to sirP. These A. fumigatus homologues also resided in a cluster of genes (Fig. 8), many of which were similar to those in the L. maculans sirodesmin biosynthetic gene cluster. Based on this homology, the A. fumigatus cluster is likely to be responsible for the production of gliotoxin. No other peptide synthetase genes of A. fumigatus were identified that had a thioredoxin reductase nearby. A similar cluster could not be identified in the genome of A. nidulans, which does not produce gliotoxin.

Figure 8.

Comparative analysis of the Leptosphaeria maculans sirodesmin biosynthetic gene cluster with the putative gliotoxin gene clusters in Aspergillus fumigatus, expressed sequence tags (ESTs) from Chaetomium globosum, and clusters of unknown function from Magnaporthe grisea and Fusarium graminearum. Nine homologous genes are shared between the sirodesmin biosynthetic gene cluster and putative gliotoxin cluster, and six of these were also identified in ESTs from C. globosum, which also produces ETP toxins. The M. grisea cluster is annotated with the dipeptidase and peptide synthetase, and O-methyl transferase and glutathione S-transferase as being fused. This prediction partially truncates the first module of the peptide synthetase, but alternative prediction did not appear plausible. The F. graminearum peptide synthetase contains seven complete modules. Both the L. maculans and F. graminearum clusters contain ATP-binding cassette type transporters, whereas the A. fumigatus cluster has a transporter belonging to the major facilitator superfamily of transporters. The sirC homologues encode cytochrome P450 monooxygenases that have higher similarity to each other than to other cytochrome P450 monooxygenases. ‘Oxygenase’ includes cytochrome P450 and NADH type monooxygenases. The A. fumigatus cluster was extracted from contig number 69, M. grisea cluster (GenBank accession AACU01000945), F. graminearum cluster (GenBank accession AACM01000002) and C. globosum ESTs (GenBank accessions, BP099008, sirI homologue; BP113141, sirG; BP099970, sirC; BP099783, sirM; BP113200, sirN; and BP113179sirT). The gene predictions for both M. grisea and F. graminearum have not been changed from those entered in GenBank.

Comparative analysis of the genes shared between the sirodesmin biosynthetic cluster and putative gliotoxin cluster will enable prediction of the genes involved in biosynthesis of the core ETP moiety and those involved in side group modifications. The mode of introduction of the sulphur residues to the diketopiperazine intermediates remains elusive, but identification of genes common to both L. maculans and A. fumigatus will allow directed analysis of the steps involved by gene disruption experiments. For example, an ACCS homologue is present in both the sirodesmin biosynthetic cluster and putative gliotoxin gene cluster suggesting a role in biosynthesis of the core ETP as proposed above. Comparative analyses will also provide useful information for the cytochrome P450 monooxygenases. The cytochrome P450 monooxygenase, proximal to the peptide synthetase in the A. fumigatus cluster is homologous to sirC (30% identical 50% similar, e-value 1e-69) compared with the next best pair sirE and the cytochrome P450 monooxygenase adjacent to the thioredoxin reductase (shown as ‘oxygenase’ in Fig. 8; 26% identical, 43% similar, e-value 4e-25). The sirN homologue in the A. fumigatus cluster (51% identical, 66% similar, e-value 6e-77) was not identified as having a methyl transferase domain. Homologues of sirD and sirH were absent from the A. fumigatus cluster. These genes would not be predicted to be required for gliotoxin biosynthesis. Interestingly a geranylgeranyl pyrophosphate (GGPP) synthase was identified 22 kb from the A. fumigatus sirT homologue but would not be expected to be involved in gliotoxin biosynthesis; a similar gene is present in the paxilline biosynthetic cluster from P. paxilli which also contains a sirD homologue (Young et al., 2001). The region separating these two genes contains a cytochrome P450 monooxygenase (proximal to the GGPP synthase) and the remnants of a retroelement (data not shown). The absence of homologues of the novel genes sirQ, sirR and sirS from the putative gliotoxin biosynthesis gene cluster would suggest a role for the gene products in the modifications made to the complicated ring structures of sirodesmin PL derived from the tyrosine side chain.

Other fungi also have similar genes to those in the sirodesmin biosynthetic gene cluster

A query of the EST database at NCBI using all 18 sir genes identified a number of Chaetomium globosum homologues. Chaetomium species produce the ETPs chetomin and chaetocin (Sekita et al., 1981). C. globosum ESTs were identified with matches to sirI, sirG, sirC, sirM, sirN and sirT, as shown in Fig. 8. These six genes also have homologues in the A. fumigatus cluster.

In addition to the A. fumigatus and C. globosum matches, sirT homologues linked to peptide synthetase genes were identified in the Magnaporthe grisea and Fusarium graminearum genomes (Whitehead Institute), and are shown in Fig. 8. The M. grisea cluster contained genes for a zinc finger protein, dipeptidase, peptide synthetase, ACCS homologue, thioredoxin reductase, O-methyl transferase, GST and SirC cytochrome P450 monooxygenase homologue. No ETP type toxins are reported to be produced by M. grisea but published details of the secondary metabolites of M. grisea are limited. However, given the small size of the M. grisea contig and that the prediction that some of the gene products are fused (Fig. 8), this may not represent a functional gene cluster. The F. graminearum cluster contained genes predicted to encode a seven-module peptide synthetase, thioredoxin reductase, ABC transporter, zinc finger protein and an ACCS homologue (Fig. 8). There are no published reports of a metabolite that would be produced by a seven-module peptide synthetase for F. graminearum. Interestingly the F. graminearum cluster was located close (23 kb) to a paxM (FAD dependant monooxygenase) homologue from the paxilline biosynthetic gene cluster (data not shown) (Young et al., 2001) and a maltose permease gene (data not shown), as is the case in L. maculans (Fig. 3). Homologues in the Candida albicans genome could not be identified suggesting that the isolate sequenced may not produce gliotoxin. Production of gliotoxin by Candida spp has only been reported in a limited number of clinical isolates (Shah and Larsen, 1991; Shah et al., 1995). Alternatively a different mode of biosynthesis could be possible as a bacterial diketopiperazine, albonoursin, is synthesized independent of a non-ribosomal peptide synthetase (Lautru et al., 2002). The only other available genome sequence of a filamentous ascomycetous fungus, Neurospora crassa, did not contain a similar gene cluster.

The presence of similar clusters in a number of distantly related ascomycetes raises interesting questions about the origins of a putative parent cluster. Whether this parent cluster was distributed by horizontal gene transfer or vertical transmission and subsequent gene loss in species that do not contain homologues of the sirodesmin biosynthesis cluster is unknown. The remains of a retroelement including a transposase flanking the A. fumigatus cluster (data not shown) are an interesting feature. These regions are abundant in stop codons and appear to have undergone a process whose end point resembles repeat induced point mutations, known to occur in the sexually outcrossing N. crassa and L. maculans (Cambareri et al., 1989; Idnurm and Howlett, 2003). Whether this is relevant to the evolution of the cluster remains to be determined. The sequencing of other similar clusters and flanking genes, including the gliotoxin-producing clusters from Trichoderma and Penicillium species as well as the sirodesmin production cluster from S. diversum, and genes for chetomin biosynthesis, will allow these questions to be addressed.

Concluding remarks

The generation of a sirP mutant and its inability to produce sirodesmin PL at a detectable level clearly demonstrates a role for this gene in the biosynthesis of sirodesmin PL. The presence of sirP in a cluster of genes, many with homology to genes involved in the biosynthesis of other fungal secondary metabolites, suggests a role for many of these in the biosynthesis of sirodesmin. Indeed a number of the predicted functions of the identified genes can be tentatively assigned to predicted steps in the biochemical pathway. The coregulation of the genes with the biosynthesis of sirodesmin is also convincing evidence for the role of the other genes involvement in sirodesmin production.

While targeted disruption of every gene in the cluster would provide a great deal of information regarding both the intermediates of the pathway and autoprotection mechanisms, the effort required to disrupt just one gene in L. maculans makes this a difficult task. Heterologous expression of the entire cluster, with defined mutations, in a sirodesmin non-producing host may prove a useful alternative to targeted disruption experiments in determining the role of each gene.

ETPs are an important class of fungal compounds. This is the first report of a gene cluster responsible for the biosynthesis of an ETP. The generation of a transformant via homologous recombination that does not produce sirodesmin PL provides a means to assess the role of this compound in the process of pathogenicity towards canola and/or competition with other microorganisms. The identification of other genes potentially involved in the biosynthesis of ETPs will make the cloning of further ETP gene clusters a relatively straight forward process. Comparative analyses between these clusters will provide insight into the genes necessary for the biosyntheses of the core ETP structure.

Experimental procedures

DNA sequencing

A cosmid library of L. maculans isolate IBCN 18 (constructed in the pWEB vector Epicentre, USA) was probed with an L. maculans EST with best match to dimethylallyl tryptophan synthase of N. coenophialum to isolate the first of three cosmids sequenced. Two other cosmids were isolated sequentially using the end sequence of the first then second cosmid, to give a total coverage of about 68 kb of L. maculans DNA (Fig. 3). Cosmids were sequenced using a combination of primer walking and shotgun cloning at the Australian Genome Research Facility (Brisbane, Australia) and Macrogen sequencing facility (Seoul, South Korea). Intron positions and transcriptional start and stop sites were determined using RT-PCR and the GeneRacer 5′- and 3′ rapid amplification of cDNA ends kit (Invitrogen, USA). Products were cloned (pCR2.1 TOPO-TA cloning kit, Invitrogen) and sequenced (GenBank AY553235).

Preliminary sequence data for A. fumigatus was obtained from The Institute for Genomic Research website at http://www.tigr.org. Putative genes were predicted using Fgenesh software at http://www.softberry.com

Fungal culture

The wild type isolate used for culture and transformation was ICBN 18 (M1). Isolates were grown on 10% Campbell's V8 juice agar for pycnidiospore production at 22°C with 12/12 h light dark cycle. For liquid cultures, pycnidiospores (106) were inoculated into 50 ml of 10% Campbell's V8 juice and grown at 22°C in the dark without agitation.

Fungal transformation and gene disruption

The disruption cassette for sirP was assembled in a Gateway® (Invitrogen) fitted version of pPZPtk8.10 (Gardiner and Howlett, 2004). A 7 kb genomic fragment containing the coding region of sirP was amplified with Platinum Taq DNA polymerase (Invitrogen) using aatB1 and aatB2 tailed primers (sirPKO1 and sirPKO2, Table 2) and cloned into pDONR207 using BP clonase (Invitrogen). LR clonase (Invitrogen) was used to move the fragment into the Gateway® fitted pPZPtk8.10 derivative. The fungal selectable marker was introduced using the in vitro transposition-based GPS-M mutagenesis system (NEB, USA) as previously described (Gardiner and Howlett, 2004). The final targeting T-DNA is shown in Fig. 4. Transformation of isolate IBCN 18 was carried out using Agrobacterium-mediated DNA delivery and transformants were initially selected on hygromycin and subsequently selected with hygromycin and trifluorothymidine (Gardiner and Howlett, 2004).

Table 2. PCR primers.
NameSequence (5′-3′)
LmMT-l1fTCAAACATCACCATGTCTCC
LmMT-l1rGCTTCAACTTCTGGCAAAC
sirPKO1GGGGACAAGTTTGTACAAAAAAGCAGGCTTGACCACAGCCCTGAATCAC
sirPKO2GGGGACCACTTTGTACAAGAAAGCTGGGTACCCTTGACCACATAAAGCAC
psc16CTTATCGATGGCCAAAGC
psc21CAAGTTACCAGGGGCTATC
actinfTTCCAGCTTGGAGAAGTCGT
actinrCTGACATCGACATCGCACTT
tubulinfGTCCTCCACCTTTGTCGGTA
tubulinrATTGCTGGTACTCGGACACC
sirGfACTGGTTGACCTTGCACACA
sirGrTCTGCAATTGTCGGTCTGTC
sirDfCCACACCTTGGATCTGACCT
sirDrAGTACGCCTTCAGACGAGGA
sirTfCTACTTGCCGGAGACTTTGC
sirTrCTGGAAGTTTACGCCAGAGC
sirAfCTCGAGGACGAGATGTGTGA
sirArTTGCTCGGTCGTGTACTCTG
sirHfGAAAGTAATGGCCCAACGAA
sirHrGAAATAGCGTCGTGGGTAGC
LmPKS1fACGAACTGACCGACGATACC
LmPKS1rGGGAGTCACCAACAGCATCT
LmMP1fTTTTGAACGGCACACAGGTA
LmMP1rGAGGATTGCTCCGGTCATAA
sirIfCTCATAATCCCCTGGGTCAA
sirIrTAAAAGCTGGCGTCACTCCT
sirJfCCGAGAAACATTGCAACAGA
sirJrTCTGGTGTAGGCCCTCAATC
sirPfTCACATGGTGAAATCGGCTA
sirPrAATTCCCAACGCATCAACTC
sirZfCCAAAAGGAAGCAGGAACAA
sirZrGCCGAGTCTGTATCCGAATG
LmUVI-1hfCAATGGTCCCCTGATCCTTA
LmUVI-1hrTGCCAATGAGGATGTTCAAA

Gene expression analysis

Expression of the genes in the cluster was examined in mycelia grown in 10% Campbell's V8 juice by qRT-PCR analysis. RNA was purified using the RNeasy Plant Mini kit (QIAgen, Germany) and DNaseI treated (Invitrogen) before oligo-dT primed reverse transcription with ThermoScript (Invitrogen). DNA digestion was confirmed for each sample by the absence of PCR amplification of the metallothionein like gene (LmMT-l1, GenBank AY541064, MT-l1f and MT-l1r primers, Table 2), from a RT reaction set up without reverse transcriptase.

Quantitative RT-PCR was performed using Rotor-Gene 3000 equipment (Corbett Research, Australia) and QuantiTectTM. SYBR® Green PCR kit (QIAgen). A standard curve of amplification efficiency of each gene was generated from purified RT-PCR products across a five orders of magnitude dilution series (10−4 to 10−8 dilution) in triplicate. Samples were analysed in triplicate from a five-fold dilution of the original RT products. Diluted RT product (1 µl) was added to 19 µl of PCR mix and subjected to 45 cycles of PCR (30 s at each of 94°C, 60–64°C and 72°C, with the annealing temperature optimized for particular primers). The amplified product was detected every cycle at the end of the 72°C step. Melt curve analysis after the cycling confirmed the absence of non-specific products in the reaction. The fluorescence threshold (Ct) values were determined for standards and samples using the Rotor-Gene 5 software. Ct values were exported to Microsoft Excel and analysed as described by Muller et al. (2002). All calculations were performed using actin (GenBank AY547274) as a reference gene.

High performance liquid chromatography

Filtrate of L. maculans grown in 10% Campbell's V8 juice (15 ml) was extracted twice with ethyl acetate (10 ml) and the organic phase dried under nitrogen. The residue was resuspended in 1 ml ethanol and 100 µl was analysed by HPLC using Beckman System Gold solvent module 126 and detector 168 apparatus fitted with a UV detection system. A Phenomenex Luna 5 µm particle size, 4.6 id × 250 mm C18(2) column was used. Mobile phase was as follows; 2 min with 25% CH3CN – 75% H2O, followed by a linear gradient from 25% to 100% CH3CN over 35 min, with 100% CH3CN maintained for a further 8 min Detection was at 210 and 240 nm.

Retention times of peaks in samples were compared to that of a standard preparation of sirodesmin PL (Pedras et al., 1990). For quantification of sirodesmin PL in culture filtrates, samples were spiked with 20 µg of gliotoxin (Sigma, USA) to follow extraction efficiencies. Extraction efficiencies were calculated on the recovery of gliotoxin from 48 and 72 h culture filtrates as after this time, the appearance of compounds that coeluted with gliotoxin made the use of gliotoxin as an internal standard impossible (recovery efficiency 91 ± 7%). Sirodesmin PL peak areas were compared to a standard curve (r2 = 0.9997).

Acknowledgements

This work was funded by the Australian Grains Research and Development Corporation. Preliminary sequence data were obtained from The Institute for Genomic Research website at http://www.tigr.org. Sequencing of A. fumigatus was funded by the National Institute of Allergy and Infectious Disease U01 AI 48830 to David Denning and William Nierman, the Wellcome Trust, and Fondo de Investicagiones Sanitarias. We thank Dr Candace Elliott for critical comments on the manuscript.

Ancillary