Evidence that a secondary metabolic biosynthetic gene cluster has grown by gene relocation during evolution of the filamentous fungus Fusarium

Evidence that a secondary metabolic biosynthetic genecluster has grown by gene relocation during evolutionof the filamentous fungus Fusariummmi_6927 1128..1142

Robert H. Proctor,* Susan P. McCormick,Nancy J. Alexander and Anne E. DesjardinsU. S. Department of Agriculture, Agricultural ResearchService, National Center for Agricultural UtilizationResearch, Peoria, IL, USA.

Summary

Trichothecenes are terpene-derived secondarymetabolites produced by multiple genera of filamen-tous fungi, including many plant pathogenic speciesof Fusarium. These metabolites are of interestbecause they are toxic to animals and plants andcan contribute to pathogenesis of Fusarium onsome crop species. Fusarium graminearum andF. sporotrichioides have trichothecene biosyntheticgenes (TRI ) at three loci: a 12-gene TRI cluster andtwo smaller TRI loci that consist of one or two genes.Here, comparisons of additional Fusarium specieshave provided evidence that TRI loci have a complexevolutionary history that has included loss, non-functionalization and rearrangement of genes as wellas trans-species polymorphism. The results also indi-cate that the TRI cluster has expanded in somespecies by relocation of two genes into it from thesmaller loci. Thus, evolutionary forces have drivenconsolidation of TRI genes into fewer loci in somefusaria but have maintained three distinct TRI loci inothers.

Introduction

Filamentous fungi can produce numerous secondarymetabolites that include pigments, compounds that aretoxic to plants and/or animals, plant growth regulators andantibiotics or other pharmaceuticals. The majority offungal secondary metabolites described to date are

derived from the activities of one of three classes ofenzymes: terpene synthases, polyketide synthases andnon-ribosomal peptide synthetases (Keller et al., 2005).Each enzyme class utilizes relatively simple primarymetabolites as substrates and rearranges or condensesthem into structurally more complex molecules, i.e. terpe-nes, polyketides or non-ribosomal peptides. These morecomplex molecules can undergo enzyme-catalysed modi-fications, such as oxygenation, cyclization, isomerizationand/or condensation with other metabolites to form themyriad of fungal secondary metabolites that, collectively,are highly variable in chemical structure and biologicalactivity. Secondary metabolite biosynthetic genes areoften clustered in filamentous fungi (Keller et al., 2005).The clusters typically include a gene encoding a synthaseas well as genes encoding structure-modifying enzymes.The clusters can also include genes encoding metabolitetransport proteins and proteins that regulate transcriptionof cluster genes.

The fungal genus Fusarium (teleomorph Gibberella)consists of over 70 species, many of which are plantpathogens (Desjardins, 2006; Leslie and Summerell,2006). Some species can produce secondary metabo-lites, known as mycotoxins, that are toxic to animals.Trichothecenes are a family of terpene-derived mycotox-ins produced by some species of Fusarium and severalother fungal genera in the order Hypocreales. Tricho-thecenes are among the most economically significantmycotoxins worldwide because of their widespreadoccurrence in important grain crops such as barley, maizeand wheat (Council for Agricultural Science and Tech-nology, 2003). Trichothecenes are also an agriculturalconcern because they are toxic to plants and can con-tribute to pathogenesis of Fusarium on some crops(Desjardins et al., 1996; Maier et al., 2006).

The trichothecene biosynthetic gene cluster has beenwell characterized in Fusarium graminearum sensu strictoand F. sporotrichioides, two species that have served asmodels to elucidate the genetics and chemistry of tri-chothecene biosynthesis (Brown et al., 2001; 2002; Leeet al., 2002). In both species, the core cluster consists of12 genes that are responsible for synthesis of the coretrichothecene molecule and several modifications to it.

Accepted 8 October, 2009. *For correspondence. E-mail [email protected]; Tel. (+1) 309 681 6830; Fax (+1) 309 6816686. Mention of trade names or commercial products in this article issolely for the purpose of providing specific information and does notimply recommendation or endorsement by the U.S. Department ofAgriculture.

Molecular Microbiology (2009) 74(5), 1128–1142 � doi:10.1111/j.1365-2958.2009.06927.xFirst published online 13 November 2009

Journal compilation © 2009 Blackwell Publishing LtdNo claim to original US government works

The cluster genes are: the terpene synthase gene TRI5,the cytochrome P450 monooxygenase genes TRI4,TRI11 and TRI13, the acyl transferase genes TRI3 andTRI7, the esterase gene TRI8, the regulatory genes TRI6and TRI10, and the transporter gene TRI12. TRI9 andTRI14 are also located in the core TRI cluster, but theylack similarity to genes with known functions and theirspecific functions in trichothecene biosynthesis are notknown.

Fusarium graminearum and F. sporotrichioides alsohave two smaller loci that encode trichothecene biosyn-thetic enzymes. Linkage analysis and genome sequencedata indicate that these two loci and the core TRI clusterare all on different chromosomes in F. graminearum(Cuomo et al., 2007; Lee et al., 2008). The first of thesesmaller loci consists of a single acyl transferase gene,TRI101, that is responsible for esterification of acetate tothe hydroxyl function at carbon atom 3 (C-3) of trichoth-ecenes (Fig. 1) (Kimura et al., 1998a; McCormick et al.,1999). The second locus consists of two genes: TRI1,which encodes a cytochrome P450 monooxygenase,and TRI16, which encodes an acyl transferase. InF. sporotrichioides, the TRI1 enzyme catalyses hydroxy-lation of trichothecenes at C-8 (Fig. 1), and the TRI16enzyme catalyses esterification of a five-carbon carboxy-lic acid, isovalerate, to the C-8 oxygen (Brown et al.,2003; Meek et al., 2003; Peplow et al., 2003). InF. graminearum, the TRI1 enzyme catalyses hydroxyla-tion of trichothecenes at both C-7 and C-8 (Fig. 1), butTRI16 is non-functional because of the presence of frameshifts and stop codons in its coding region (Brown et al.,2003; McCormick et al., 2004; 2006). As a result, inF. graminearum-produced trichothecenes the C-8 oxygenis not esterified but instead undergoes oxidation to form acarbonyl function by an as yet unknown mechanism. Thefunctional differences of the TRI1 and TRI16 enzymes in

F. graminearum and F. sporotrichioides are responsiblefor important structural differences of trichothecenes pro-duced by the two species. F. graminearum can producetrichothecenes (e.g. nivalenol and deoxynivalenol) witha C-7 hydroxyl and a C-8 carbonyl, whereas F. sporo-trichioides can produce trichothecenes (e.g. T-2 toxin)that have an isovalerate ester at C-8 and lack the C-7hydroxyl.

Comparisons of nucleotide sequences from multiplespecies have provided insight into evolution of fungalsecondary metabolite gene clusters. Some clustersappear to have moved into fungal genomes by horizontalgene transfer from either prokaryotes or other fungi(Brakhage et al., 2005; Khaldi et al., 2008). In addition,vertical inheritance has resulted in distribution of clustersacross multiple genera of fungi, and differential inherit-ance and/or deletion has resulted in discontinuous distri-bution of clusters among groups of related fungi (Krokenet al., 2003; Proctor et al., 2004; Patron et al., 2007;Glenn et al., 2008). High levels of sequence identityamong some cluster genes indicate that gene duplicationhas contributed to formation of plant secondary metabo-lite biosynthetic clusters and clusters that regulate devel-opment in animals (Gierl and Frey, 2001; Benderoth et al.,2006; Lemons and McGinnis, 2006). In addition, interge-nus comparisons indicate that gene relocation has con-tributed to formation of an allantoin utilization cluster inyeast (Wong and Wolfe, 2005) and some secondarymetabolite gene clusters in plants (Field and Osbourn,2008).

Gene duplication has been proposed to contribute toformation and growth of secondary metabolite gene clus-ters in filamentous fungi (Cary and Ehrlich, 2006; Saikiaet al., 2008). However, there is limited evidence for thecontribution of either gene duplication or relocation in theformation of such clusters (Cary and Ehrlich, 2006;Carbone et al., 2007). In this study, we compared theorganization of TRI loci in fungi that represent a widerange of the phylogenetic and chemical diversity oftrichothecene-producing species of Fusarium. We foundthat TRI1 and TRI101 are located in the core TRI clusterin some species but not in others. This difference amongspecies provided an ideal system to address two compet-ing hypotheses: (i) the core TRI cluster has grown byrelocation of TRI1 and TRI101 into it from elsewhere inthe genome versus (ii) the core cluster has grown by geneduplication, which was followed by relocation of TRI1 andTRI101 to other loci in the genome.

Results

Fusarium equiseti TRI cluster

In F. graminearum and F. sporotrichioides, the core TRIcluster is identical with respect to order and orientation of

Fig. 1. A basic chemical structure of Fusarium trichothecenesshowing the positions affected by the TRI1 and TRI101 enzymes.R indicates atoms and functional groups that vary among differenttrichothecenes.

Fungal gene cluster growth by gene relocation 1129

Journal compilation © 2009 Blackwell Publishing Ltd, Molecular Microbiology, 74, 1128–1142No claim to original US government works

genes in and immediately flanking it (Brown et al., 2002;2004). In the current study, phylogenetic analysis with fiveprimary metabolic genes indicated that F. graminearumand F. sporotrichioides are more closely related to oneanother than to some other trichothecene-producingspecies, such as F. longipes, F. equiseti and F. semi-tectum (Fig. 2A). Therefore, we sequenced the corecluster in a strain (NRRL 13405) of the more distantlyrelated, morphological species, F. equiseti. The sequence

analysis revealed significant structural differencescompared with the cluster in F. graminearum andF. sporotrichioides (Fig. 3). The most striking difference isthe presence of orthologues of TRI1 and TRI101 in theF. equiseti cluster. As noted above, TRI1 and TRI101 areat other loci in F. graminearum and F. sporotrichioides.The F. equiseti cluster also differs from the F. grami-nearum and F. sporotrichioides core cluster by the follow-ing: (i) it lacks the transporter gene TRI12, (ii) the region

Fig. 2. A. One of two most parsimonious trees of trichothecene-producing and non-producing species of Fusarium examined in this study.The tree was generated by maximum parsimony analysis of combined DNA sequences (4506 characters) of CPR1, HIS3, RPB2, TEF1 andTUB2 fragments. Intron sequences were removed to improve alignments. Numbers near branches indicate bootstrap values based on 1000pseudoreplicates.B. Organization of genes near TRI101 and YTRI101 homologues in fungi shown in the phylogenetic tree in Fig. 2A. Arrows indicate genesand their orientation. An X through an arrow indicates a pseudogene. Arrow colours correspond to the following genes: orange – TRI101; blue– from left to right, PHO5 and URA7; green – from left to right TRI14 and TRI3; grey – a putative aldehyde dehyrdrogenase gene (protein ID64833 in the N. haematococca genome sequence database). Diagrams of the TRI101 region in F. graminearum and F. fujikuroi are basedon previously published data (Brown et al., 2001; 2002; Kimura et al., 2003b; McCormick et al., 2004). Not included in the figure are twoadditional genes that are located between PHO5 and URA7 in F. fujikuroi, F. oxysporum (FOXG_03788 and FOXG_03787) and F. verticillioides(FVEG_12070 and FVEG_12071).C. Organization of genes in the TRI1 region in trichothecene-producing species of Fusarium shown in Fig. 2A. Arrows indicate genes and theirorientation. Arrow colours correspond to the following genes: orange – TRI1; blue – TRI16; purple – from left to right, homologues ofFG00070, FG00072 and FG00073; pink – from left to right, PDB1 and ORF1; yellow – putative WD repeat protein-encoding gene; brown –from left to right, genes encoding a putative transcription factor, cytochrome P450 monooxygenase and b-mannosidase; green – from left toright TRI10, TRI9, TRI11 and TRI13, except for F. camptoceras where the arrow to the far right corresponds to TRI14. The vertical lines tothe far right, labelled as GE1 through GE4, delimit the four genetic environments in which TRI1 is located. GenBank accession numbers:CPR1, GQ915450–GQ915466; HIS3, GQ915467–GQ915483; RPB2, GQ915484–GQ915499; TEF1, GQ915500–GQ915515; TUB2,GQ915434–GQ915449; TRI1 region, GQ915516–GQ915528.

1130 R. H. Proctor, S. P. McCormick, N. J. Alexander and A. E. Desjardins �


spanning TRI3, TRI7 and TRI8 is in the opposite orien-tation and located at the opposite end of the cluster and(iii) it includes a putative Cysteine 6 (Cys6) transcrip-tion factor gene (gene F in Fig. 3) between TRI5 andTRI6.

We also obtained limited sequence data for regionsflanking the TRI cluster in F. equiseti NRRL 13405: 3.8 kbflanking the TRI4 end of the cluster and 4.6 kb flanking theTRI8 end. These data indicate that the genes immediatelyflanking the F. equiseti cluster are not the same as thoseflanking the core cluster in F. graminearum andF. sporotrichioides (Fig. 3) (Brown et al., 2004). However,among the genes flanking the TRI4 end of the F. equiseticluster is an orthologue of a gene (gene D in Fig. 3)designated as FGSG_00072 in the F. graminearumgenome sequence in the Fusarium ComparativeDatabase. In F. graminearum, FGSG_00072 is located inthe 3′-flanking region of TRI1 (Fig. 3). Thus, bothFGSG_00072 and TRI1 have been transferred betweenthe TRI1 region of F. graminearum and the core TRIcluster regions of F. equiseti. The deduced amino acid

sequences of the F. graminearum and F. equisetiFGSG_00072 orthologues are 62% identical across thecarboxy-terminal halves of the proteins but exhibit only35% identity across their amino-terminal halves. In addi-tion, there are multiple insertions and deletions in theamino-terminal halves of the protein orthologues relativeto each other. In contrast, the FGSG_00072 orthologuesin F. oxysporum (FOXG_05799) and F. verticillioides(FVEG_03669) are 58 and 64% identical, respectively, tothe F. graminearum orthologue over the entire lengths ofthe proteins. This indicates that the amino-terminal half ofthe F. equiseti orthologue is a product of rapid divergencefrom the orthologues in the other three fusaria or aproduct of recombination with another locus.

We also amplified and sequenced a fragment of theF. equiseti TRI16 orthologue with primers 1472 and 1477(Table 1) and used GenomeWalker analysis to obtain theentire TRI16 coding region as well as 3.3 and 1.8 kb ofsequence for the 5′- and 3′-flanking regions respectively.Unlike in F. graminearum, the TRI16 coding region inF. equiseti did not include any frame shifts or stop codons

Fig. 3. Comparison of the two trichothecene biosynthetic loci in F. equiseti strain NRRL 13405 with the three previously described TRI loci inF. graminearum (Brown et al., 2001; 2002; Kimura et al., 2003b; McCormick et al., 2004). Arrows indicate genes and direction of transcription.Green arrows labelled with numbers indicate known TRI genes (e.g. 4 and 5 indicate TRI4 and TRI5 respectively). TRI1, TRI16 and TRI101are spelled out and depicted with orange or blue arrows for emphasis. YTRI16 indicates TRI16 pseudogene. Other arrows labelled with lettersindicate genes flanking TRI loci, with the exception of gene F, which is located between TRI5 and TRI6 in F. equiseti NRRL 13405. Flankinggenes D, I, J, K, L, M, N and O correspond to FGSG_00072, FGSG_00070, FGSG_03529, FGSG_03530, FGSG_03531, FGSG_03544¸FGSG_03545 and FGSG_03546 respectively, in the Fusarium Comparative Database. The TRI flanking genes in F. equiseti NRRL 13405labelled as B, C, E and H exhibit high levels of identity (> 50% in blastx analysis) to the Fusarium Comparative Database genesFGSG_02263, FGSG_04750, FGSG_04692 and FGSG_04624 respectively. Genes A, F and G exhibit only low levels of identity (< 35% inblastx analysis) to the Fusarium Comparative Database genes FOXG_17386, FGSG_10458 and FGSG_12447 respectively. Dashed linesindicate genes that have been transferred between regions. GenBank accessions numbers for F. equiseti NRRL 13405 sequences: core TRIcluster region, GQ865563; TRI16 region, GQ865564.



relative to the F. sporotrichioides TRI16 coding region.Therefore, the F. equiseti TRI16 may be functional. Thegenes (A and B in Fig. 3) immediately flanking theF. equiseti TRI16 do not exhibit significant similarity/identity to the genes that flank the TRI1-TRI16 locus orthe other TRI loci in F. graminearum or F. sporotrichioides.Together, our sequence analyses of the core TRI clusterand TRI16 regions indicate that TRI1 and TRI16 are aminimum of 12 kb apart in F. equiseti NRRL 13405.However, the exact distance between these two genes orwhether they are located on different chromosomesremains to be determined.

TRI101 relocation

The presence of TRI1 and TRI101 in the core TRIcluster in F. equiseti but not in F. graminearum andF. sporotrichioides raises the question, did the two genesoriginate in the cluster and subsequently move out of itduring the evolution of some Fusarium species, or didTRI1 and TRI101 originate outside the cluster and subse-quently move into it during the evolution of some species?To address this question with respect to TRI101, wecompared the location of TRI101 orthologues in 16trichothecene-producing species and three non-

Table 1. Oligonucleotide primers used for PCR amplification and sequence analysis of target sequences indicated.

Experiment Target gene Primer designation Nucleotide sequencea (5′→3′)

TRI1 fragment TRI1 1285 GCGTCTCAGCTTCATCAAGGCAKCKAMTGAWTCGTRI1 1292 CTTGACTTSMTTGGCKGCAAAGAARCGACCA

TRI3 fragment TRI3 1912 TGTGTMGGYGCWGAGGCVATYGTTGGTRI3 1914 ACRGCAGCRGTCTGRCACATGGCGTA

TRI4 fragment TRI4 1450 ACCTTGAGTTCTACCATGAAGTCATCTRI4 1455 GCACTGTCTAGARCCCTGAGAGAAGT

TRI5 fragment TRI5 1558 GGCATGGTCGTGTACTCTTGGGTCAAGGTTRI5 1559 GCCTGMYCAWAGAAYTTGCRGAACTT

TRI11 fragment TRI11 1482 CACACYCTCCTSATGCTYTGTGGACTTRI11 1483 TCCCAMACTGTYCTYGCCAGCATCAT

TRI16 fragment TRI16 1472b CCTCTCTCCCCTTGAYCAATTRAACTCTTRI16 1473b CTTCCCGATCCCRAYGAGCCTCTTACACTRI16 1474b GCCTTATMTKGGTAATGTCGTGCTKACATRI16 1475b AAGAGGCTCRTYGGGATCGGGAAGGTTCTRI16 1476b CARCCGACGATGTMAGCACGACATTACCTRI16 1477b CAATATACGGATACCGCACAAAGACTGG

TRI101 fragment TRI101 109 CCATGGGTCGCRGGCCARGTSAATRI101 178 AACTCSCCRTCIGGYTTYTTNGGCAT

TRI101-PHO5 TRI101 2069 GATGTACTTTATGCCCAAGAAGCintergenic region PHO5 2071 GGCGACTATTGATACTCTTGGACTRI101-URA7 TRI101 2070 GTTCCTGTGTTTCCYTCGCTRATGintergenic region URA7 2074 CACTCAGCATGCCATTCGACAAGTTRI1-TRI9 TRI9 1787 CCCAGAGTARGCGCAGACTTCTAGCintergenic region TRI1 1789 GCATGAGAGTCRCTRACCTGCTTTCTTRI1-TRI11 TRI1 1790 ARAARAAGCAGAYGCCGCCAGGGAGGATintergenic region TRI11 1794 CCKCGRAATTGYATTGGHATGAAGTAAGCPR1 fragment CPR1 1825 CTCTCACAACCCCTACATTGCCCCTAT

CPR1 1826 CTTACCTGCCAYTCCTTYTSGTACATGAHIS3 fragment HIS3 1701 CAATGGCTCGCACTAAGCAGAC

HIS3 1702 CTCRTGAACGCCTTGATTATCCRPB2 fragment A RPB2 5F2 GGGGWGAYCAGAAGAAGGC

RPB2 7cR CCCATRGCTTGYTTRCCCATRPB2 fragment B RPB2 7cF ATGGGYAARCAAGCYATGGG

RPB2 11aR GCRTGGATCTTRTCRTCSACCTEF1 fragment TEF1 ef1b ATGGGTAAGGARGACAAGAC

TEF1 ef2b GGARGTACCAGTSATCATGTTTEF1 ef22 AGGAACCCTTACCGAGCTC

TUB2 fragment TUB2 T1b AACATGCGTGAGATTGTAAGTTUB2 T11 AATTGGTGCTGCTTTCTGGCATUB2 T21 GGTTTGCCAGAAAGCAGCACCTUB2 T22b TCTGGATGTTGTTGGGAATCCTUB2 T121 CCACCTGTCTCCGTTTCCCCGTUB2 T222 GACCGGGGAAACGGAGACAGG

a. In nucleotide sequences, K = G or T, M = A or C, S = C or G, R = A or G, W = A or T, Y = C or T, and I = inosine.b. Primers used for PCR amplification.All sequences are shown in the 5′ to 3′ orientation.



producing species of Fusarium as well as a more distantlyrelated, trichothecene non-producing fungus, Nectria hae-matococca (synonym Haematonectria haematococca,anamorph F. solani ) (Fig. 2A).

In F. graminearum and F. sporotrichioides, TRI101 islocated between orthologues of a phosphate permeasegene, PHO5, and a UTP-ammonia ligase gene, URA7(Kimura et al., 1998b). We used a PCR approach withprimers shown in Table 1 to amplify PHO5-TRI101 andTRI101-URA7 intergenic regions in 13 trichothecene-producing fusaria in addition to F. graminearum,F. sporotrichioides and F. equiseti. The ends of ampliconswere sequenced to confirm that the expected productswere amplified. With this approach, we obtained evidencethat PHO5 and TRI101 are in the same orientation as inF. graminearum and are located within 2.5 kb of eachother in 10 of the species examined (Fig. 2B). We alsoobtained evidence that TRI101 and URA7 were in thesame orientation and within 2.5 kb of each other in sevenof the species examined (Fig. 2B).

The PCR approach did not provide evidence forlinkage of TRI101, PHO5 and URA7 in F. scirpi, F. semi-tectum or F. camptoceras. However, GenomeWalkeranalysis of these species yielded sequence data dem-onstrating that, in all three species, TRI101 is locatedbetween TRI3 and TRI14, the same position as in F.equiseti (Fig. 2B).

To examine the location of TRI101 in the trichoth-ecene non-producing species F. oxysporum and F. verti-cillioides, we utilized the genome sequence databasesof these fungi in the Fusarium Comparative Database.The databases indicate that these two species have anon-functional remnant or pseudogene of TRI101(YTRI101) located between orthologues of PHO5 andURA7. These findings are consistent with previousreports of a YTRI101 between orthologues of PHO5 andURA7 in F. oxysporum and another trichothecene non-producing species F. fujikuroi (Kimura et al., 2003a;Tokai et al., 2005). We also examined the N. haemato-cocca genome sequence database at the Joint GenomeInstitute and identified an orthologue of TRI101 (proteinID 97980) and PHO5 (protein ID 64827) 3.5 kb apart onchromosome 13. We found no evidence of a URA7orthologue in this region of the N. haematococcagenome sequence.

Our phylogenetic analysis with five primary metabolicgenes resolved the 16 trichothecene-producing species ofFusarium into a well-supported clade that is distinct fromthe clade formed by the three trichothecene non-producing fusaria (Fig. 2A). The analysis also resolvedF. equiseti, F. scirpi, F. semitectum and F. camptocerasinto a well-supported clade, which, for the purposes of thisstudy, we will hereafter refer to as the F. equiseti clade.The F. equiseti clade was part of the larger clade of tri-

chothecene producers but formed a sister group to all theother species within it. Comparison of the phylogenetictree and the locations of TRI101 suggests that TRI101was located between PHO5 and URA7 prior to diver-gence of trichothecene-producing and non-producingspecies of Fusarium, and that TRI101 was located next toPHO5 prior to divergence of fusaria with the Gibberellateleomorph and those with the Nectria teleomorph(Fig. 2A and B). Because the F. equiseti clade lies withinthe trichothecene-producing clade, it follows that the pres-ence of TRI101 in the core TRI cluster in the F. equiseticlade is a more recently derived condition than its pres-ence in the PHO5-URA7 region.

TRI1-flanking regions

The F. oxysporum, F. verticillioides and N. haematococcagenome sequence databases were uninformative withrespect to TRI1 movement, because a TRI1 orthologuewas not detected in them. Therefore, we compared thegenetic environment in which TRI1 occurs in the 16trichothecene-producing species shown in Fig. 2. First,PCR with degenerate primers 1285 and 1292 (Table 1)was employed to amplify a ~1200 bp fragment of the TRI1coding region from a representative strain of each speciesfor which we lacked data. Sequence data from the ampli-fication products were then used in GenomeWalker analy-sis to obtain 3–6 kb of flanking sequence on each side ofTRI1.

The resulting sequence data revealed that TRI1 can belocated within one of four distinct genetic environments,which we have designated GE1, GE2, GE3 and GE4(Fig. 2C). GE1 occurs in eight species, includingF. graminearum (Brown et al., 2003; McCormick et al.,2004). In these species, an orthologue of FGSG_00070and TRI16 (or a TRI16 pseudogene, YTRI16) are locatedin the 5′-flanking region of TRI1, and an orthologue ofFGSG_00072 and sometimes FGSG_00073 are locatedin the 3′-flanking region (Fig. 2C). GE2 occurs inF. sporotrichioides (Brown et al., 2003; Meek et al., 2003;Peplow et al., 2003), F. armeniacum and Fusarium sp.NRRL 36351. In GE2, the gene PDB1 is consistentlylocated in the 5′-flanking region of TRI1, but the3′-flanking region varies. In F. armeniacum andF. sporotrichioides, an orthologue of TRI16 and an ortho-logue of a gene previously designated as Orf1/orf1(Brown et al., 2003; McCormick et al., 2004) are locatedin the TRI1 3′-flanking region. However, in Fusarium sp.NRRL 36351, a gene encoding a putative WD repeatprotein is located in the 3′-flanking region, and there is noevidence for either TRI16 or Orf1/orf1 in 3 kb region 3′ tothe TRI1 stop codon. GE3 occurs only in F. longipesamong the species examined here. In GE3, genes encod-ing a putative transcription factor and a cytochrome P450



monooxygenase are located in the TRI1 5′-flankingregion, and a gene encoding a b-mannosidase-likeprotein is located in the 3′-flanking region. GE4 occurs inthe four species that form the F. equiseti clade, i.e.F. equiseti, F. scirpi, F. semitectum and F. camptoceras(Fig. 2C). In GE4, TRI1 is located in the core TRI cluster,with TRI10 and TRI9 in the 5′-flanking region of TRI1, andTRI11 and TRI13 in the 3′-flanking region. The exceptionto this is F. camptoceras, where TRI11 and TRI14 arelocated in the 3′-flanking region.

The analysis of TRI1-flanking regions indicated thatan apparently functional TRI16 is located adjacent toTRI1 in five species: F. sambucinum, F. venenatum,F. kyushuense, F. sporotrichioides and F. armeniacum(Fig. 2C). That the gene is functional is evident by theabsence of insertions, deletions or stop codons inits coding region. The analysis also indicated that aYTRI16 is adjacent to TRI1 in five species: F. boothii,F. graminearum, F. crookwellense, F. culmorum andF. poae (Fig. 2C). PCR analysis with primers 1472 and1477 (Table 1) followed by GenomeWalker analysisrevealed that F. scirpi and F. semitectum have an appar-ently functional TRI16. However, the presence of TRI1 inthe core cluster of F. scirpi and F. semitectum indicatesthat TRI16 is not adjacent to TRI1 in these two species.

The PCR analysis with multiple pairs of degenerateTRI16 primers (Table 1) failed to yield a TRI16 am-plicon for F. camptoceras, F. longipes or Fusarium sp.NRRL 36351. In addition, TRI16 was not detected inF. camptoceras, F. longipes or Fusarium sp. NRRL 36351by low-stringency Southern blot analysis (data notshown). Thus, TRI16 may be absent from the genomes ofthese three species or may be highly diverged comparedwith TRI16 orthologues in the other fusaria included in thisstudy.

TRI1 location and species phylogenies

Comparison of the genetic environment in which TRI1 islocated and the phylogenetic relationship of the 16trichothecene-producing fusaria revealed that TRI1 loca-tion tends to be the same in more closely related speciesand different in more distantly related species (Fig. 2Aand C). As noted above, TRI1 is located in the core cluster(GE4) in the F. equiseti clade. The three species in whichTRI1 is located in GE2 formed another strongly supportedclade, but this clade was resolved within a larger cladethat included species with TRI1 in the GE1 context. Theanalysis also indicated that F. longipes, with its uniqueTRI1 location (GE3), represents a distinct lineage fromspecies with TRI1 located in GE1 and GE2 (Fig. 2A andC). These findings are consistent with movement of TRI1into or out of the core TRI cluster during the evolution oftrichothecene-producing fusaria.

TRI1 and TRI16 trans-species polymorphism

We conducted a maximum parsimony analysis of a~1200 bp fragment of the TRI1 coding region in 51 strainsthat represent 16 trichothecene-producing species ofFusarium. We also analysed the entire ~ 1800 bp codingregion in one representative strain of each species. Theseanalyses resolved TRI1 orthologues into four majorclades, each of which was supported by bootstrap valuesof 97–100 (Fig. 4). A comparison of the TRI1 phylogenywith the species phylogeny, inferred from combinedsequences of five primary metabolic genes, revealedincongruencies between the two phylogenies (Fig. 5).For example, F. kyushuense and F. venenatum weremost closely related to F. poae and F. sambucinum inthe species phylogeny, but more closely related toF. camptoceras, F. equiseti, F. longipes, F. scirpi, andF. semitectum in the TRI1 phylogeny (Fig. 5). The lack ofcorrelation between TRI1 and species phylogenies is con-sistent with trans-species polymorphism, a phenomenonthat has been described in animals, plants and fungi(Muirhead et al., 2002; Ward et al., 2002; Klein et al.,2007; Powell et al., 2007).Trans-species polymorphism

Fig. 4. Phylogenetic analysis of a ~1200 bp region of theTRI1 coding region from 51 strains representing 16 of thetrichothecene-producing species of Fusarium. The coloured boxesdemarcate the four major clades resolved by the analysis. Thephylogram was generated by maximum parsimony analysis.Numbers near branches are bootstrap values based on 1000pseudoreplications. Strain designations are the same as in Table 2,except that the letters NRRL have been removed from designationsfor strains from the NRRL collection.



can arise when an ancestral species carries multiplealleles of a gene and when the alleles are inherited differ-entially during subsequent speciation events (Klein et al.,2007). This can result in closely related species inheritingless similar alleles and distantly related species inheritingmore similar alleles.

Phylogenies inferred from maximum parsimony analy-sis of DNA sequences of TRI101 and the representativeTRI cluster genes TRI4, TRI5 and TRI11 were highlycorrelated with each other and with the species phylogenyderived from primary metabolic gene sequences.Because of the similarity of the TRI4 and TRI5 trees andbecause they employed an outgroup from the sameorganism (i.e. Myrothecium roridum), a phylogenetic treewas generated with the combined sequence data for TRI4

and TRI5 (Fig. 5). Sequence data for TRI11 and Tri101were not combined with each other or with othersequences, and the phylogenetic relationships for thesegenes are presented as separate trees (Fig. S1). In thespecies phylogeny and the phylogenies based on TRI4,TRI5, TRI11 and TRI101, the 16 fusaria analysed wereconsistently resolved into five major clades, and thespecies within each of the five clades were the same in allof the phylogenies (Fig. 5, Fig. S1). These five clades aredesignated Clades 1 through 5 in the combined TRI4/TRI5 tree in Fig. 5 and in the TRI11 tree in Fig. S1. Giventheir similarities to the species phylogeny, it is not surpris-ing that the phylogenies based on TRI4, TRI5, TRI11 andTRI101 exhibited marked incongruencies with the TRI1-based phylogeny (Fig. 5, Fig. S1).

Fig. 5. Comparison of the TRI1 phylogeny with phylogenies based on the combined sequences for five primary metabolic genes(CPR1, HIS3, RPB2, TEF1 and TUB2) and the combined sequences for two selected TRI cluster genes (TRI4 and TRI5). Phylogenies weregenerated by maximum parsimony analysis, and numbers near branches are bootstrap values based on 1000 pseudoreplications. Colouredboxes around species names correspond to the different types of TRI1 orthologues presented in Fig. 4. GenBank accession numbers for TRI4are GQ915529–GQ915541, and for TRI5 are GQ915542–GQ915554.



We also conducted a maximum parsimony analysis withthe DNA sequence of TRI16 and compared the resultingphylogeny with the TRI1 phylogeny (Fig. 6). The analysisof TRI16 was more limited than those described above,because TRI16 is either a pseudogene or not detectablein half of the trichothecene-producing species included inthis study. Therefore, the TRI16 phylogeny was inferredfrom sequence data of eight species that had an appar-ently functional TRI16. In contrast to the comparisonsnoted above, the TRI16 phylogeny was correlated withthat of TRI1. For example, F. sambucinum was resolvedinto a clade distinct from the clade that includedF. kyushuense and F. venenatum in both the TRI1 andTRI16 phylogenies. In the species, TRI4, TRI5, TRI11and TRI101 phylogenies, however, F. sambucinum,F. kyushuense and F. venenatum always resolved into thesame clade (e.g. Clade 1, Fig. 5).

Discussion

Previous molecular genetic analyses indicated that TRI1,TRI101 and the core TRI cluster are located at threedistinct loci in F. graminearum and F. sporotrichioides

(Brown et al., 2002; 2003; Kimura et al., 2003b; McCor-mick et al., 2004). Moreover, linkage analysis (Lee et al.,2008) and genome sequence data (Cuomo et al., 2007)indicate that the three loci are on different chromosomesin F. graminearum. Therefore, it is intriguing that TRI1 andTRI101 are located in the core TRI cluster in theF. equiseti clade. These findings raise the question, didTRI1 and TRI101 originate in the core cluster and subse-quently move out of it, or did the two genes originateoutside the cluster and move into it? The location ofTRI101/YTRI101 in the PHO5-URA7 region in both thetrichothecene-producing and non-producing clades ofFusarium indicates that TRI101 was located in this regionprior to divergence of the two clades. Because theF. equiseti clade is within the trichothecene-producingclade, it follows that TRI101 most likely moved directly orindirectly from the PHO5-URA7 region to the core TRIcluster during evolution of the F. equiseti clade. Criticalevidence for the directionality of TRI1 movement is thesimilar pattern of trans-species polymorphism observedfor orthologues of TRI1 and TRI16 but not for orthologuesof the TRI cluster genes TRI4, TRI5 and TRI11. Thepresence/absence of this pattern of trans-species poly-

Fig. 6. Comparison of TRI1 and TRI16 phylogenies in Fusarium species with apparently functional TRI16 orthologues. Phylogenies weregenerated by maximum parsimony analysis, and numbers near branches are bootstrap values based on 1000 pseudoreplicates. MP indicatesmost parsimonious. GenBank accession numbers for the F. semitectum and F. scirpi TRI16 are GQ915568 and GQ915569 respectively.For TRI16 from other species see TRI1 region GenBank accessions GQ915519, GQ915521, GQ915522 and GQ915524.



morphism among the various TRI genes would havearisen more readily if TRI1 and TRI16 had been closelylinked to one another but not to the core TRI cluster in theancestral Fusarium. If this were the case, it follows thatTRI1 moved into the cluster during the evolution of theF. equiseti clade. On the other hand, if TRI1 had origi-nated in the cluster, orthologues of other TRI clustergenes would be expected to exhibit a pattern of trans-species polymorphism similar to that exhibited by TRI1.But, they do not. Thus, the most parsimonious explanationfor the observed data is that TRI1 originated outside thecluster and subsequently relocated into it during the evo-lution of the F. equiseti clade. This explanation is consis-tent with relocation of TRI101 into the cluster.

The mechanism by which TRI1 and TRI101 relocated tothe core TRI cluster is not known. However, analysis ofTRI locus-flanking genes may provide insight into themechanism. For example, the presence of TRI1 in thecluster and the FGSG_00072 orthologue flanking thecluster in F. equiseti NRRL 13405 suggests that recombi-nation occurred between an ancestral core cluster and anancestral locus that included TRI1 in a GE1-like context.The lack of identity of the 5′ half of the F. equisetiFGSG_00072 orthologue relative to orthologues in otherfusaria suggests a possible link between relocation of thegene and the marked changes in its sequence inF. equiseti.

The presence of TRI1 and TRI101 in the core TRIcluster in the F. equiseti clade indicates that relocation ofthe genes into the cluster occurred relatively early in theevolution of this clade. It remains to be determinedwhether the relocation of the two genes resulted fromindependent or related events. Likewise, it remains to bedetermined whether TRI1 and/or TRI101 relocation arerelated to the absence of TRI12, the rearrangement of theTRI3-TRI7-TRI8 region, and the presence of the Cys6transcription factor gene in the core cluster in F. equisetiNRRL 13405.

It is not clear what evolutionary forces drove relocationof TRI1 and TRI101 in the F. equiseti clade but maintainedthese genes at separate loci distinct from the cluster inother species. One hypothesis to explain the existence ofsecondary metabolite biosynthetic gene clusters is thatclustering facilitates co-ordinated regulation of genesresponsible for the same biosynthetic pathway (Kelleret al., 2005). Although we have no direct evidence for itsrole in trichothecene biosynthesis, the presence of a Cys6transcription factor gene between TRI5 and TRI6 in theF. equiseti cluster suggests that the gene may regulateexpression of the other TRI cluster genes. This and theabsence of a closely related orthologue of the Cys6 tran-scription factor gene in the F. graminearum genomesuggest that there may be fundamental differences inregulation of TRI gene expression in F. equiseti compared

with F. graminearum. Functional characterization of theCys6 gene should provide insight into whether such adifference in TRI gene regulation exists in the twospecies.

Another question raised by the results of the currentstudy is: which of the TRI1 loci has the more ancestralorganization? As discussed above, the presence of TRI1in the core cluster is more likely a derived ratherthan ancestral condition. Given the TRI1 and TRI16trans-species polymorphism, it is likely that withintrichothecene-producing fusaria the ancestral TRI1 locusincluded both TRI1 and TRI16. Moreover, TRI16 wasprobably functional, because it is more likely for a func-tional gene to degenerate into a non-functional gene thanfor a non-functional gene with multiple deletions andinsertions to become functional. Among the genetic envi-ronments in which TRI1 occurs, both GE1 and GE2include a functional TRI16. The four major types of TRI1orthologues resolved by phylogenetic analysis of the gene(Fig. 4) likely correspond to different alleles in the ances-tral, trichothecene-producing Fusarium. Thus, there weremost likely at least four alternate alleles for the TRI1coding sequence in the ancestral TRI1-TRI16 locus.Among the species examined these four types of TRI1orthologues (ancestral alleles) are represented in GE1by F. poae, F. graminearum, F. sambucinum and F. kyu-shuense. Likewise, two of the TRI1 orthologue typesoccur among species with GE2. Together, these findingssuggest that the GE1 and GE2 contexts of TRI1 representmore ancestral organizations compared with the GE3 andGE4 contexts.

Fusarium is a member of the ascomycetous orderHypocreales, which includes several other genera oftrichothecene-producing fungi. Three of these, Myroth-ecium, Stachybotrys and Trichoderma, produce trichoth-ecenes that lack an oxygen atom at the C-8 position(Jarvis, 1991; Nielsen et al., 2005), whereas species ofFusarium, Spicellum and Trichothecium can produce tri-chothecenes with an oxygen atom at C-8 (Machida andNozoe, 1972; Kralj et al., 2007). These differences in tri-chothecene production among genera suggest that theability to oxygenate trichothecenes at C-8 arose afterdivergence of Fusarium from a common ancestor that itshared with Myrothecium, Stachybotrys and Trichoderma.This hypothesis is consistent with the origin of TRI1outside the TRI cluster, because it is responsible for C-8oxygenation in Fusarium. It is not known whether C-8oxygenation of trichothecenes produced by Spicellum andTrichothecium is catalysed by an orthologue of the TRI1enzyme or an unrelated hydroxylase.

Trichothecene C-3 oxygenation and the TRI101enzyme-catalysed acetylation of the C-3 oxygen mayalso be relatively recent innovations in trichothecenebiosynthesis, because trichothecenes produced by



Myrothecium, Spicellum, Stachybotrys, Trichoderma andTrichothecium lack a C-3 oxygen. Analysis of the phylo-genetic relationships of TRI genes in these speciesshould provide further insight into the evolution of thetrichothecene C-3, C-7 and C-8 oxygenations and otherbiochemical reactions in the trichothecene biosyntheticpathway.

Although trans-species polymorphism similar to thatobserved among orthologues of TRI1 and TRI16 was notobserved among core TRI cluster genes, other patterns oftrans-species polymorphism have been observed previ-ously for TRI cluster genes in the F. graminearum speciescomplex and F. culmorum (Ward et al., 2002; Chandleret al., 2003). The F. graminearum species complex con-sists of at least 11 closely related lineages, which haverecently been elevated to species rank (O’Donnell et al.,2000; 2006). The complex includes the phylogeneticspecies F. graminearum sensu stricto (F. graminearumlineage 7) and F. boothii (F. graminearum Lineage 3).Although the function(s) of trans-species polymorphismamong orthologues of TRI cluster genes has not beendemonstrated, the polymorphisms are associated with dif-ferent trichothecene production profiles. For example,deletions within TRI13 are trans-species in that the samedeletions can occur in F. graminearum and F. culmorum(Chandler et al., 2003). Strains of these fungi with a func-tional TRI13 (i.e. without the deletions) produce NIV,whereas strains with a non-functional TRI13 (i.e. withdeletions) produce DON rather than NIV (Lee et al., 2002;Chandler et al., 2003). Similarly, among the trans-speciesorthologues of TRI1, the F. graminearum TRI1 orthologueleads to DON or NIV production, whereas theF. sporotrichioides TRI1 leads to T-2 toxin production(Meek et al., 2003; McCormick et al., 2004; 2006). Thus, itis possible that trans-species polymorphism of TRI genescontributes to production of structurally diverse tricho-thecenes within and among species.

Fusarium is a large and complex genus that includesmultiple species that are sometimes difficult to distinguishfrom each other by morphology and some that are ill-defined (Geiser et al., 2004; Leslie and Summerell, 2006).Some species, such as F. equiseti and F. semitectum(synonym F. pallidoroseum), are genetically diverse andlikely represent species complexes (Geiser et al., 2004;Leslie and Summerell, 2006) similar to previouslydescribed species complexes of Fusarium (O’Donnellet al., 1998; 2000, 2006). DNA sequence-based phyloge-netic analyses have helped to resolve relationshipsamong and within species and to define new species(O’Donnell et al., 2000; 2006; Yli-Mattila et al., 2004;Leslie et al., 2007). In the current study, the relationshipsinferred from sequences of primary metabolic genesare by and large consistent with relationships inferredfrom previous studies focused on broad ranges of

trichothecene-producing species (Mulè et al., 1997; Kris-tensen et al., 2005).

Relocation of TRI1 and TRI101 into the core TRI clusterof the F. equiseti clade provides evidence for growth of afungal secondary metabolite gene cluster by gene reloca-tion rather than by gene duplication. Phylogenetic analy-sis of 376 cytochrome P450 monooxygenase genes fromfour species of filamentous fungi provides independentsupport for this conclusion (Deng et al., 2007). The analy-sis revealed that the four F. graminearum monooxyge-nase genes (TRI1, TRI4, TRI11 and TRI13) involved intrichothecene biosynthesis are more closely related toother monooxygenase genes than they are to each other.Thus, even though the four trichothecene biosyntheticmonooxygenases have chemically similar substrates andshare a similar enzymatic mechanism, it is unlikely thatthe genes encoding them evolved directly from the sameancestral TRI gene. This contrasts proposals that fungalcluster genes with similar biochemical functions canevolve by duplication of a preexisting cluster gene (Caryand Ehrlich, 2006; Carbone et al., 2007; Saikia et al.,2008). The TRI gene data also contrast data showing thatgene duplication has contributed to growth of some sec-ondary metabolite gene clusters in plants (Gierl and Frey,2001; Benderoth et al., 2006).

The results of the current and previous studies (e.g.Brown et al., 2001; Lee et al., 2002; Ward et al., 2002;Chandler et al., 2003) provide evidence for a complexevolutionary history of TRI loci that has included loss,non-functionalization and rearrangement of genes as wellas trans-species polymorphism. Together, the studiesdemonstrate that multispecies comparisons of TRI genescan provide important insights into the evolution of sec-ondary metabolism in filamentous fungi.

Experimental procedures

Strains and media

Strains of Fusarium used in this study are shown in Table 2.Specific epithets used for these species conform to namesused by Leslie and Summerell (2006) except for F. boothii(O’Donnell et al., 2006). With the exception of F. avenaceum,all species listed in Table 2 have been previously reported toproduce trichothecenes and were confirmed to produce tri-chothecenes during the course of this study (S.P. McCormickand R.H. Proctor, unpublished). Strains were stored as 15%glycerol stocks at -80°C and grown on V-8 juice agar medium(Tuite, 1969) and in liquid GYEP medium (2% glucose, 0.1%yeast extract, 0.1% peptone) (Seo et al., 2001).

Nucleic acid manipulations

To prepare fungal genomic DNA, strains were grown in liquidGYEP medium for 2–4 days depending on the growth rate.Fungal growth was harvested by vacuum filtration, lyophilized



and ground to a powder. The ground material was suspendedin extraction buffer (200 mM Tris-Cl, pH 8, 250 mM NaCl,25 mM EDTA pH 8 and 0.5% SDS) at 50 mg per 250 mlbuffer. Subsequently, genomic DNA used for GenomeWalkerlibraries was purified with the DNeasy Plant Mini Kit asdescribed by the manufacturer (Qiagen). Genomic DNA usedfor PCR only was sometimes purified by extraction with anequal volume of a 1:1 (v/v) mixture of TRIS-equilibratedphenol and chloroform : isoamyl alcohol (24:1). The resultingaqueous phase was mixed with 2 vols of NaI solution and 5 mlof UltraBind solution, and then further purified by the Ultra-Clean DNA Purification kit (Mo Bio Laboratories) as specifiedby the manufacturer.

Fragments of TRI1, TRI4, TRI5, TRI11 and TRI101 wereamplified from multiple Fusarium species and sequencedwith primers shown in Table 1. PCR and sequencing primerswere synthesized by Integrated DNA Technologies or bySigma Life Science-Genosys. Standard PCR methodsemployed Taq DNA polymerase (Qiagen) or Platinum PCRSuperMix polymerase (Invitrogen Life Technologies) usingconditions recommended by the manufacturers. PCR prod-ucts were purified by agarose gel electrophoresis followed byband purification with the UltraClean DNA Purification kit(Mo Bio Laboratories).

The GenomeWalker protocol (Clontech) was used toamplify regions of DNA flanking TRI gene fragments that hadbeen amplified with primers (Table 1) that were designedbased on F. graminearum and F. sporotrichioides sequences.GenomeWalker libraries were prepared as specified by themanufacturer. Briefly, genomic DNA was digested separatelywith each of the restriction endonucleases DraI, EcoRV, PvuIIand StuI. Each DNA digestion was then ligated separately toa DNA-adapter fragment supplied in the GenomeWalker kit.

Dilutions of the resulting ligation products were employed astemplates in nested PCR with Advantage II DNA polymerase(Clontech) and with the cycling conditions specified by themanufacturer. In both the primary and secondary reactionsof the nested PCR, primer pairs consisted of one primercomplementary to Fusarium DNA and the other complemen-tary to the adapter DNA. The nucleotide sequence of thesecondary PCR products was used to design additionalprimers specific to Fusarium DNA in order to extend thesequence data by the GenomeWalker procedure. Fusarium-specific primers were also designed to sequence the entirelength of GenomeWalker PCR products that were longer than~2 kb and to obtain second-strand sequence data.

For Southern blot analysis of TRI16, hybridization probeswere prepared with the Ready-to-Go DNA labelling kit asdescribed by the manufacturer (Amersham Biosciences).Probe templates consisted of fragments of the TRI16 codingregion that were amplified by PCR with primers 1472 and1477 from genomic DNA of F. equiseti NRRL 13405 andF. sporotrichioides NRRL 3299. Hybridization conditions wereas previously described (Proctor et al., 2004).

Sequence analyses

Nucleotide sequence analysis employed BigDye Terminatorversion 3.1 (Applied Biosystems) reagents and UltraClean-purified PCR products (Mo Bio Laboratories) as DNAtemplates. Sequence reactions were purified with the BigDyeXterminator Purification protocol and analysed with a 3730DNA analyser at the USDA-ARS-NCAUR DNA SequenceFacility. Sequence data were viewed and edited withSequencher version 4.5 (Gene Codes).

Table 2. Fusarium strains used in this study.

Species Straina

F. armeniacum FRC R-09335b, NRRL 26908, NRRL 29133F. avenaceum FRC R-9495b

F. boothii (synonym F. graminearum lineage 3) NRRL 26916b, NRRL 29011, NRRL 29105F. camptoceras NRRL 13381b

F. crookwellense (synonym F. cerealis) FRC R-09624b, ITEM 664, ITEM 665, ITEM 4334, NRRL 25805Fusarium sp. (received as F. chlamydosporum) NRRL 36351b, NRRL 36542F. culmorum FRC R-09618b, NRRL 25475F. equiseti NRRL 13405b, ITEM 3708, ITEM 4743, ITEM 6453, NRRL 25795, NRRL 29128F. graminearum (synonym F. graminearum lineage 7) NRRL 5883, NRRL 34097, PH-1 (= NRRL 31084), Z-3639b

F. kyushuense NRRL 3509, NRRL 25348b

F. longipes NRRL 20695b

F. poae FRC T-0962b, ITEM 6700, ITEM 6703, NRRL 13714F. sambucinum FRC R-04995, FRC R-06380, FRC R-07843b, FRC R-09198, FRC R-09200F. scirpi FRC R-06979b, NRRL 29134F. semitectum (synonym F. pallidoroseum) NRRL 31160b

F. sporotrichioides FRC T-0972, ITEM 3902, ITEM 4594, 6813, ITEM 7365, NRRL 3299b

F. venenatum FRC R-09186b, ITEM 831, ITEM 833, ITEM 834, ITEM 835

a. Strains with FRC designation are from the Fusarium Research Center culture collection at Pennsylvania State University, State College,Pennsylvania. Strains with ITEM designation are from the culture collection at the Institute of Sciences of Food Production, National ResearchCouncil, Bari, Italy. Strains with NRRL designation are from the Northern Regional Research Center culture collection at USDA ARS NCAUR,Peoria, Illinois. Strain PH-1 was provided by F. Trail, Michigan State University and Strain Z-3639 was provided by Robert L. Bowden USDA/ARSKansas State University.b. Strains used for sequence analysis of TRI1 region and as representative strains of each species in the phylogenetic analyses shown in Figs 2,4 and 5 and Fig. S1.



DNA sequence of the core TRI cluster in F. equiseti wasdetermined by the GenomeWalker analysis as describedabove. Gene-specific primers for this protocol were designedbased on the DNA sequences of fragments of TRI1, TRI4,TRI5 and TRI11 amplified with the primers shown in Table 1.The resulting sequence data were then used to designF. equiseti-specific primers for the GenomeWalker protocol toexpand and connect sequences for the TRI genes. With thisstrategy, we obtained 37 kb of contiguous, double-strandedDNA sequence for the TRI cluster region in F. equiseti. Forother Fusarium species, primers 1285 and 1292 were used toamplify and sequence a TRI1 fragment from a representativestrain of each species. The resulting sequence data were usedto design Fusarium-specific primers for the GenomeWalkerprotocol in order to amplify and sequence the regions flankingTRI1. For most species, multiple steps with the Genom-eWalker protocol were necessary to obtain sequence data for3–6 kb of DNA on both sides of TRI1. In a few instances, theGenomeWalker protocol did not yield a secondary PCRproduct. In these cases, sequence data were obtained bydesigning primers based on sequence data from one or moreclosely related species for which the GenomeWalker protocoldid yield secondary PCR products. The primers were thenused in standard PCR protocols and for sequence analysis toobtain sequence data of the desired region.

Phylogenetic analysis

All phylogenetic relationships were determined with DNAsequences. DNA sequence alignments were done withthe ClustalW+ programme in GCG version 11.1.3Unix(Accerlrys). When necessary, the resulting alignments wereadjusted manually. Phylogenetic relationships were inferredby maximum parsimony analysis with PAUP version 4.0b10 forUnix. This programme was also used to determine statisticalsupport for branches within phylogenetic trees by bootstrapanalysis. Alignment of sequences revealed the presenceof a 62-nucleotide intron and 39-nucleotide partial intronsequence that were present in the M. roridum TRI4 sequencebut not in the Fusarium sequences. These M. roridum intronsequences were deleted from the alignment prior to phyloge-netic analysis.

To determine phylogenetic relationships between species,we employed DNA sequences of five primary metabolicgenes that have been used previously for phylogenetic analy-ses of Fusarium. The genes were: (i) CPR1, an NADPH-dependent cytochrome P450 reductase gene; (ii) HIS3, theHistone H3 gene; (iii) RPB2, the gene encoding the secondlargest subunit of RNA polymerase II; (iv) TEF1 (also ef-1a),the translation elongation factor 1a gene; and (v) TUB2, theb-tubulin gene (Steenkamp et al., 2002; Malonek et al., 2005;O’Donnell et al., 2006; 2007). Primers used to amplify andsequence the gene fragments are shown in Table 1. To deter-mine phylogenetic relationships of TRI orthologues from dif-ferent Fusarium species, we employed sequence dataobtained from PCR-amplified fragments of TRI1, TRI4, TRI5,TRI11 and TRI101.

Some analyses employed previously generated sequencedata. Data for the TRI1 and TRI101 regions inF. graminearum and F. sporotrichioides and for the TRI101region in F. fujikuroi were obtained from the National Center

for Biotechnology Information GenBank database (http://www.ncbi.nlm.nih.gov). Sequence data for the Neurosporacrassa monooxygenase gene NCU05376 and the M. roridumTRI genes were also obtained from the GenBank database.Sequence data for the YTRI101 region, CPR1, HIS3, RPB2,TEF1 and TUB2 in F. oxysporum and F. verticillioides wereobtained from the Fusarium Comparative Database website(http://www.broadinstitute.org/). Gene designations thatinclude FGSG_, FOXG_ and FVEG_ are for genes identifiedin the F. graminearum, F. oxysporum and F. verticillioidesgenome databases, respectively, with the ComparativeDatabase. DNA sequences from N. haematococca (F. solani )were obtained from the N. haematococca v2.0 genomicsequence database website (http://genome.jgi-psf.org).

Other than the trichothecene-producing species that werethe subject of this study, we have not yet identified an organ-ism with closely related homologues of all the TRI genesused in our phylogenetic analyses. Therefore, sequencesused as outgroups in the different phylogentic trees were notalways from the same organism. For example, the outgroupfor the combined TRI4 and TRI5 tree was the combinedsequence of the M. roridum orthologues of these genes.However, because we were unable to detect a M. roridumTRI1 orthologue, the outgroup for the TRI1 tree was theN. crassa gene NCU05376, which is the gene in theGenBank database with the highest identity to TRI1.

Acknowledgements

We are grateful to Marcie L. Moore, April Stanley and Kim-berly M. MacDonald for technical assistance, to Jennifer N.Steel and Nathane Orwig at the USDA-ARS-NCAUR DNASequence Facility for clean-up and electrophoretic analysisof sequence reactions, and to Stephen W. Peterson for assis-tance with phylogenetic analyses.

References

Benderoth, M., Textor, S., Windsor, A.J., Mitchell-Olds, T.,Gershenzon, J., and Kroymann, J. (2006) Positive selec-tion driving diversification in plant secondary metabolism.Proc Natl Acad Sci USA 103: 9118–9123.

Brakhage, A.A., Al-Abdallah, Q., Tüncher, A., and Spröte, P.(2005) Evolution of b-lactam biosynthesis and recruitmentof trans-acting factors. Phytochemistry 66: 1200–1210.

Brown, D.W., McCormick, S.P., Alexander, N.J., Proctor,R.H., and Desjardins, A.E. (2001) A genetic and biochemi-cal approach to study trichothecene diversity in Fusariumsporotrichioides and Fusarium graminearum. FungalGenet Biol 32: 121–133.

Brown, D.W., McCormick, S.P., Alexander, N.J., Proctor,R.H., and Desjardins, A.E. (2002) Inactivation of a cyto-chrome P-450 is a determinant of trichothecene diversity inFusarium species. Fungal Genet Biol 36: 224–233.

Brown, D.W., Proctor, R.H., Dyer, R.B., and Plattner, R.D.(2003) Characterization of a Fusarium 2-gene clusterinvolved in trichothecene C-8 modification. J Agric FoodChem 51: 7936–7944.

Brown, D.W., Dyer, J.M., McCormick, S.P., Kendra, D.F.,and Plattner, R.D. (2004) Functional demarcation of the



Fusarium core trichothecene gene cluster. Fungal GenetBiol 41: 462.

Carbone, I., Ramirez-Prado, J.H., Jakobek, J.L., and Horn,B.W. (2007) Gene duplication, modularity and adaptationin the evolution of the aflatoxin gene cluster. BMC Evol Biol7: 111.

Cary, J.W., and Ehrlich, K.C. (2006) Aflatoxigenicity inAspergillus: molecular genetics, phylogenetic relationshipsand evolutionary implications. Mycopathologia 162: 167–177.

Chandler, E.A., Simpson, D.R., Thomsett, M.A., and Nichol-son, P. (2003) Development of PCR assays to Tri7 andTri13 trichothecene biosynthetic genes, and characteriza-tion of chemotypes of Fusarium graminearum, Fusariumculmorum and Fusarium cerealis. Physiol Mol Plant Pathol62: 355–367.

Council for Agricultural Science and Technology (CAST)(2003) Mycotoxins: Risks in Plant, Animal, and HumanSystems. Task Force Report No. 139. Ames, IA: Council forAgricultural Science and Technology.

Cuomo, C.A., Güldener, U., Xu, J.-R., Trail, F., Turgeon,B.G., Pietro, A.D., et al. (2007) The Fusarium graminearumgenome reveals a link between localized polymorphismand pathogen specialization. Science 317: 1400–1402.

Deng, J., Carbone, I., and Dean, R.A. (2007) The evolution-ary history of cytochrome P450 genes in four filamentousascomycetes. BMC Evol Biol 7: 30.

Desjardins, A.E. (2006) Fusarium Mycotoxins Chemistry,Genetics and Biology. St. Paul, MN: APS Press.

Desjardins, A.E., Proctor, R.H., Bai, G., McCormick, S.P.,Shaner, G., Buechley, G., and Hohn, T.M. (1996) Reducedvirulence of trichothecene-nonproducing mutants of Gib-berella zeae in wheat field tests. Mol Plant Microbe Interact9: 775–781.

Field, B., and Osbourn, A.E. (2008) Metabolic diversification– independent assembly of operon-like gene clusters indifferent plants. Science 320: 543–547.

Geiser, D.M., del Mar Jiménez-Gasco, M., Kang, S.,Makalowska, I., Veeraraghavan, N., Ward, T.J.Z.N., et al.(2004) FUSARIUM-ID v. 1.0: a DNA sequence database foridentifying Fusarium. Eur J Plant Pathol 110: 473–479.

Gierl, A., and Frey, M. (2001) Evolution of benzoxazinonebiosynthesis and indole production in maize. Planta 213:493–498.

Glenn, A.E., Zitomer, N.C., Zimeri, A.M., Williams, L.D.,Riley, R.T., and Proctor, R.H. (2008) Transformation-mediated complementation of a FUM gene cluster deletionin Fusarium verticillioides restores both fumonisin produc-tion and pathogenicity on maize seedlings. Mol PlantMicrobe Interact 21: 87–97.

Jarvis, B.B. (1991) Macrocyclic trichothecenes. In Mycotox-ins and Phytoalexins. Sharma, R.P., and Salunkhe, D.K.(eds). Boca Raton: CRC Press, pp. 361–421.

Keller, N.P., Turner, G., and Bennett, J.W. (2005) Fungalsecondary metabolism – from biochemistry to genomics.Nat Rev Microbiol 3: 937–947.

Khaldi, N., Collemare, J., Lebrun, M.-H., and Wolf, K.H.(2008) Evidence for horizontal transfer of a secondarymetabolite gene cluster between fungi. Genome Biol 9:R18.1–R18.10.

Kimura, M., Kaneko, I., Komiyama, M., Takatsuki, A.,

Koshino, H., Yoneyama, K., and Yamaguchi, I. (1998a)Trichothecene 3-O-acetyltransferase protects both the pro-ducing organism and transformed yeast from relatedmycotoxins. J Biol Chem 273: 1654–1661.

Kimura, M., Matsumoto, G., Shingu, Y., Yoneyama, K., andYamaguchi, I. (1998b) The mystery of the trichothecene3-O-acetyltransferase gene analysis of the region aroundTri101 and charactrization of its homologue from Fusariumsporotrichioides. FEBS Lett 435: 163–168.

Kimura, M., Tokai, T., O’Donnell, K., Ward, T.J., Fujimura, M.,Hamamoto, H., et al. (2003a) The trichothecene biosynthe-sis gene cluster of Fusarium graminearum F15 contains alimited number of essential pathway genes and expressednon-essential genes. FEBS Lett 359: 105–110.

Kimura, M., Tokai, T., Matsumoto, G., Fujimura, M., Hama-moto, H., Yoneyama, K., et al. (2003b) Trichothecene non-producer Gibberella species have both functional andnonfunctional 3-O-actyltransferase genes. Genetics 163:677–684.

Klein, J., Sato, A., and Nikolaidis, N. (2007) MHC, TSP, andthe origin of species: from immunogenetics to evolutionarygenetics. Annu Rev Genet 41: 281–304.

Kralj, A., Gurgui, M., Konig, G.M., and van Echten-Deckert, G.(2007) Trichothecenes induce accumulation of glucosylce-ramide in neural cells by interfering with lactosylceramidesynthase activity. Toxicol Appl Pharmacol 225: 113–122.

Kristensen, R., Torp, M., Kosiak, B., and Holst-Jensen, A.(2005) Phylogeny and toxigenic potential is correlated inFusarium species as revealed by partial translation elon-gation factor 1 alpha gene sequences. Mycol Res 109:173–186.

Kroken, S., Glass, N.L., Taylor, J.W., Yoder, O.C., andTurgeon, B.G. (2003) Phylogenomic analysis of type Ipolyketide synthase genes in pathogenic and saprobicascomyctes. Proc Natl Acad Sci USA 100: 15670–15675.

Lee, J., Jurgenson, J.E., Leslie, J.F., and Bowden, R.L.(2008) Alignment of Genetic and Physical Maps of Gibber-ella zeae. Appl Environ Microbiol 74: 2349–2359.

Lee, T., Han, Y.-K., Kim, K.-H., Yun, S.-H., and Lee, Y.-W.(2002) Tri13 and Tri7 determine deoxynivalenol- andnivalenol-producing chemotypes of Gibberella zeae. ApplEnviron Microbiol 68: 2148–2154.

Lemons, D., and McGinnis, W. (2006) Genomic evolution ofHox gene clusters. Science 313: 1918–1922.

Leslie, J.F., and Summerell, B.A. (2006) The Fusarium Labo-ratory Manual. Ames: Blackwell Publishing.

Leslie, J.F., Anderson, L.L., Bowden, R.L., and Lee, Y.-W.(2007) Inter- and intra-specific genetic variation inFusarium. Int J Food Microbiol 119: 25–32.

McCormick, S.P., Alexander, N.J., Trapp, S.C., and Hohn,T.M. (1999) Disruption of TRI101, the gene encodingtrichothecene 3-O-acetyltransferase, from Fusariumsporotrichioides. Appl Environ Microbiol 65: 5252–5256.

McCormick, S.P., Harris, L.J., Alexander, N.J., Ouellet, T.,Saparno, A., Allard, S., and Desjardins, A.E. (2004) Tri1 inFusarium graminearum encodes a P450 oxygenase. ApplEnviron Microbiol 70: 2044–2051.

McCormick, S.P., Alexander, N.J., and Proctor, R.H. (2006)Heterologous expression of two trichothecene P450genes in Fusarium verticillioides. Can J Microbiol 52: 220–226.



Machida, Y., and Nozoe, S. (1972) Biosynthesis of trichoth-ecin and related compounds. Tetrahedron 28: 5113–5117.

Maier, F.J., Miedaner, T., Hadeler, B., Felk, A., Salomon, S.,Lemmens, M., et al. (2006) Involvement of trichothecenesin fusarioses of wheat, barley and maize evaluated by genedisruption of the trichodiene synthase (Tri5) gene in threefield isolates of different chemotype and virulence. MolecPlant Pathol 7: 449–461.

Malonek, S., Bömke, C., Bornberg-Bauer, E., Rojas, M.C.,Hedden, P., Hopkins, P., and Tudzynski, B. (2005) Distri-bution of gibberellin biosynthetic genes and gibberellinproduction in the Gibberella fujikuroi species complex.Phytochemistry 66: 1296–1311.

Meek, I.B., Peplow, A.W., Ake, C., Phillips, T.D., and Ber-emand, M.N. (2003) Tri1 encodes the cytochrome P450monooxygenase for C-8 hydroxylation during trichoth-ecene biosynthesis in Fusarium sporotrichioides andresides upstream of another new Tri gene. Appl EnvironMicrobiol 69: 1607–1613.

Muirhead, C.A., Glass, N.L., and Slatkin, M. (2002) Multilocusself-recognition systems in fungi as a cause of trans-species polymorphism. Genetics 161: 633–641.

Mulè, G., Logrieco, A., Stea, G., and Bottalico, A. (1997)Clustering of trichothecene-producing Fusarium strainsdetermined from 28S ribosomal DNA sequences. ApplEnviron Microbiol 63: 1843–1846.

Nielsen, K.F., Gräfenhan, T., Zafari, D., and Thrane, U.(2005) Trichothecene production by Trichodermabrevicompactum. J Agric Food Chem 53: 8190–8196.

O’Donnell, K., Cigelnik, E., and Nirenberg, H.I. (1998)Molecular systematics and phylogeography of the Gibber-ella fujikuroi species complex. Mycologia 90: 465–493.

O’Donnell, K., Kistler, H.C., Tacke, B.K., and Casper, H.H.(2000) Gene genealogies reveal global phylogeographicstructure and reproductive isolation among lineages ofFusarium graminearum, the fungus causing wheat scab.Proc Natl Acad Sci USA 97: 7905–7910.

O’Donnell, K., Ward, T.J., Geiser, D.M., Kistler, H.C., andAoki, T. (2006) Genealogical concordance between themating type locus and seven other nuclear genes supportsformal recognition of nine phylogenetically distinct specieswithin the Fusarium graminearum clade. Fungal Genet Biol41: 600–623.

O’Donnell, K., Sarver, B.A.J., Brandt, M., Chang, D.C.,Noble-Wang, J., Park, B.J., et al. (2007) Phylogeneticdiversity and microsphere array-based genotyping ofhuman pathogenic Fusaria, including from the multistatecontact lens-associated U.S. keratitis outbreaks of 2005and 2006. J Clin Microbiol 45: 2235–2248.

Patron, N.J., Waller, R.F., Cozijnsen, A.J., Straney, D.C.,Gardiner, D.M., Nierman, W.C., and Howlett, B.J. (2007)Origin and distribution of epipolythiodioxopiperazine (ETP)gene clusters in filamentous ascomycetes. BMC Evol Biol7: 174.

Peplow, A.W., Meek, I.B., Wiles, M.C., Phillips, T.D., and

Beremand, M.N. (2003) Tri16 is required for esterificationof position C-8 during trichothecene mycotoxin productionby Fusarium sporotrichioides. Appl Environ Microbiol 69:5935–5940.

Powell, A.J., Jacobson, D.J., and Natvig, D.O. (2007) Ances-tral polymorphism and linkage disequilibrium at the het-6region in pseudohomothallic Neurospora tetrasperma.Fungal Genet Biol 44: 896–904.

Proctor, R.H., Plattner, R.D., Brown, D.W., Seo, J.-A., andLee, Y.-W. (2004) Discontinuous distribution of fumonisinbiosynthetic genes in the Gibberella fujikuroi speciescomplex. Mycol Res 108: 815–822.

Saikia, S., Nicholson, M.J., Young, C., Parker, E.J., andScott, B. (2008) The genetic basis for indole-diterpenechemical diversity in filamentous fungi. Mycol Res 112:184–199.

Seo, J.-A., Proctor, R.H., and Plattner, R.D. (2001) Charac-terization of four clustered and coregulated genes associ-ated with fumonisin biosynthesis in Fusarium verticillioides.Fungal Genet Biol 34: 155–165.

Steenkamp, E.T., Wingfield, B.D., Desjardins, A.E., Marasas,W.F.O., and Wingfield, M.J. (2002) Cryptic speciation inFusarium subglutinans. Mycologia 94: 1032–1043.

Tokai, T., Fujimura, M., Inoue, H., Aoki, T., Ohta, K., Shibata,T., et al. (2005) Concordant evolution of trichothecene3-O-acetyltransferase and rDNA species phylogeny oftrichothecene-producing and nonproducing fusaria andother ascomycetous fungi. Microbiology 151: 509–519.

Tuite, J. (1969) Plant Pathological Methods: Fungi andBacteria. Minneapolis: Burgess Publishing, pp. 1–239.

Ward, T.J., Bielawski, J.P., Kistler, H.C., Sullivan, E., andO’Donnell, K. (2002) Ancestral polymorphism and adaptiveevolution in the trichothecene mycotoxin gene cluster ofphytopathogenic Fusarium. Proc Natl Acad Sci USA 99:9278–9283.

Wong, S., and Wolfe, K.H. (2005) Birth of a metabolic genecluster in yeast by adaptive gene relocation. Nat Genet 37:777–782.

Yli-Mattila, T., Mach, R.L., Alekhina, I.A., Bulat, S.A., Koski-nen, S., Fullnig-Gradinger, C.M., et al. (2004) Phylogeneticrelationship of Fusarium langsethiae to Fusarium poaeand Fusarium sporotrichioides as inferred by IGS, ITS,b-tubulin sequences and UP-PCR hybridization analysis.Int J Food Microbiol 95: 267–285.

Supporting information

Additional supporting information may be found in the onlineversion of this article.

Please note: Wiley-Blackwell are not responsible for thecontent or functionality of any supporting materials suppliedby the authors. Any queries (other than missing material)should be directed to the corresponding author for the article.



Evidence that a secondary metabolic biosynthetic gene cluster has grown by gene relocation during evolution of the filamentous fungus Fusarium

Documents