Identication of a cis-acting factor modulafumonisin-biosynthetic gene in the fungal

y c

, ViaGabersi

Aspergillus nigerFusarium fujikuroi

ryheafum

identied a degenerated, over-represented motif potentially involved in the cis-regulation of FUM genes,

duce apropenot es

the same (Proctor et al., 2003, 2008). Two other Fusarium specieshost a full version of the FUM cluster, only missing the FUM17orthologue (encoding one of the two putative ceramide synthasesin the cluster): F. proliferatum and Fusarium fujikuroi (B. Tudzynski,personal communication). Both fumonisin-producing and non-producing strains were identied in these species (Munkvold,2003; Proctor et al., 2004). In A. niger instead, the cluster containsonly 11 homologues of the 17 FUM genes identied (FUM1, FUM610, FUM1315, FUM19 and FUM21) (Pel et al., 2007). Here too, bothproducing and non-producing strains were identied (Frisvadet al., 2007). The difference in gene composition and order of the

factor binding site; TSS, transcription start-site.q F.C. conceived of the study, designed the experiments with M.P., P.K. M.P.

performed bioinformatic analyses, drafted the corresponding manuscript sections.V.M. performed the experiments, drafted the rest of the manuscript. I.V. generatedthe Fv-PFUM1-1, -3 control transformants, performed the experiment in Fig. 2. V.M.,I.V., M.P., F.C. analyzed the data. F.C., P.K. funded the work, provided logistic help.I.V., M.P., P.K., F.C. revised the paper. All authors read, approved the nalmanuscript. Corresponding author. Fax: +39 011 2368875.

E-mail addresses: valeria.montis@unito.it (V. Montis), pasquali@lippmann.lu (M.Pasquali), ivan.visentin@unito.it (I. Visentin), pkarlov@gwdg.de (P. Karlovsky),francesca.cardinale@unito.it (F. Cardinale).

Fungal Genetics and Biology 51 (2013) 4249

Contents lists available at

Fungal Genetic

journal homepage: www.e1 These authors contributed equally to this work.ferentiation or reproduction of their producer but increase itstness in certain ecological contexts (Fox and Howlett, 2008). Infungi, most genes needed for the synthesis of secondary metabo-lites are organized in clusters. Physical clustering may allow thegenes to be co-regulated by chromatin modication (Keller et al.,2005), and it may also facilitate horizontal gene transfer betweenspecies (Rosewich and Kistler, 2000).

strains of Fusarium oxysporum (Rheeder et al., 2002) and by somespecies of Aspergillus, including A. niger and related species (Frisvadet al., 2007; Varga et al., 2010). Fumonisin B1 (FB1) is the majorfumonisin produced by F. verticillioides and F. proliferatum (Bartket al., 2006; Nelson et al., 1993). Since fumonisins are natural con-taminants of food and feed and a serious threat to animal and hu-man health (reviewed by Voss et al. (2007) and Wan Norhasimaet al. (2009)), a considerable effort is being devoted to a betterunderstanding of the regulation of their synthesis.

FUM genes are clustered in F. oxysporum as they are in F. verti-cillioideswith the number, order and orientation of the genes being

Abbreviations: dw, dry weight; FB1, B-series fumonisin 1; GFP, green uorescentprotein; PFUM1, FUM1 promoter; PFUM1mod, mutated FUM1 promoter; RT-qPCR,quantitative reverse-transcriptase PCR; TF, transcription factor; TFBS, transcription-Fusarium oxysporumFusarium proliferatumGene-cluster regulationZinc-nger protein

1. Introduction

Filamentous fungi are able to prometabolites endowed with bioactiveolites are organic molecules that are1087-1845/$ - see front matter 2012 Elsevier Inc. Ahttp://dx.doi.org/10.1016/j.fgb.2012.11.009and of fumonisin biosynthesis. The samemotif was not found in various FUM homologues of fungi that donot produce fumonisins. Comparison of the transcriptional strength of the intact FUM1 promoter with asynthetic version, where the motif had been mutated, was carried out in vivo and in planta for F. verticil-lioides. The results showed that the motif is important for efcient transcription of the FUM1 gene.

2012 Elsevier Inc. All rights reserved.

multitude of secondaryrties. Secondary metab-sential for growth, dif-

The fumonisin (FUM) gene cluster contains genes involved inthe synthesis of fumonisins. Fumonisins are polyketide-derivedmycotoxins produced mainly by the maize pathogens Fusariumverticillioides and Fusarium proliferatum (teleomorphs Gibberellamoniliformis and Gibberella intermedia) (Munkvold, 2003), by someAvailable online 3 December 2012

Keywords:

In eukaryotes, coordination of transcription can be attained through shared transcription factors,whose specicity relies on the recognition of cis-regulatory elements on target promoters. A bioinfor-matic analysis on FUM promoters in the maize pathogens Fusarium verticillioides and Aspergillus nigerV. Montis a,1, M. Pasquali b,1, I. Visentin a, P. KarlovskaDipartimento di Scienze Agrarie, Forestali e Alimentari, Universit degli Studi di Torinob Environment and Agro-biotechnologies Department (EVA), Centre de Recherche Public-cDepartment of Crop Sciences, Molecular Phytopathology and Mycotoxin Research, Univ

a r t i c l e i n f o

Article history:Received 7 August 2012Accepted 26 November 2012

a b s t r a c t

Fumonisins, toxic secondastrong agro-economic andtered and co-expressed inll rights reserved.ting the transcription of FUM1, a keymaize pathogen Fusarium verticillioidesq

, F. Cardinale a,Leonardo da Vinci, 44, 10095 Grugliasco (TO), Italyriel Lippmann, rue du Brill, 41, L-4422 Belvaux, Luxembourgty of Gttingen, Grisebachstrasse 6, D-37077 Gttingen, Germany

metabolites produced by some Fusarium spp. and Aspergillus niger, havelth impacts. The genes needed for their biosynthesis, named FUM, are clus-onisin producers.

SciVerse ScienceDirect

s and Biology

lsev ier .com/ locate/yfgbi

ics aFusarium vs. Aspergillus FUM clusters likely reects a long history ofindependent evolution from a common ancestor. Indeed, the FUMcluster in A. niger probably originated by horizontal gene transferfrom a common ancestor prior of the divergence between F. verti-cillioides and F. oxysporum (Khaldi and Wolfe, 2011). Interestingly,a gene-by-gene phylogenetic analysis revealed a total of 13 puta-tive orthologues of FUM genes (FUM1, FUM67, FUM1019, herebycalled FUM-like), scattered in the genomes of ve other Euascomy-cetes: Cochliobolus heterostrophus (Kroken et al., 2003), Fusariumgraminearum, Neurospora crassa, Magnaporthe grisea, and Aspergil-lus nidulans (Khaldi and Wolfe, 2011). Physical clustering of genesmay also inuence the evolution of their regulatory sequences,likely due to the possibility of tight co-regulation of expressionby localized events of chromatin remodelling, and to the activityof pathway-specic regulators embedded in the cluster (Bayramand Braus, 2012; Yu and Keller, 2005). One proposed mechanismof regulation of fumonisin synthesis is indeed at the epigenetic le-vel (Visentin et al., 2012), as for other secondary metabolic geneclusters (Palmer and Keller, 2010; Reyes-Dominguez et al., 2010).However, chromatin-controlled regulation normally coexists witha shared set of transcription factors (TFs), which bind to the pro-moters of co-regulated genes by recognizing common cis-regula-tory elements acting as TF-binding sites (TFBSs). Knowledge ofthe interplay between trans-factors and their cognate cis-se-quences is crucial to understanding the molecular underpinningsof the coordinated regulation of gene expression.

A number of proteins with a demonstrated or postulated globalfunction in signaling and chromatin remodeling were connected tofumonisin biosynthesis, together with broad-domain TFs withgenes outside of the FUM cluster (reviewed in Picot et al. (2010)).At present, only one narrow-domain TF, encoded by FUM21, hasbeen characterized as required for fumonisin biosynthesis. Fum21is a predicted Zn(II)2Cys6 zinc-nger protein found in all ve FUMgene clusters sequenced so far (Brown et al., 2007; Pel et al.,2007; Proctor et al., 2008; B. Tudzynski, personal communication).No cis-regulatory elements with TFBS function have been reportedin FUM gene promoters yet. One plausibly successful approach toidentify them would be based on their conservation among co-regulated FUM promoters within the same genome as well asamong orthologous FUM clusters. Extension of sequence conserva-tion to regions adjacent to TFBSs by functional constraints mayfacilitate identication (Doniger et al., 2005). A motif-discoveryalgorithm has been already successfully used for the identicationof cis-regulatory elements in the promoters of F. graminearum genesinduced during trichothecene synthesis (Seong et al., 2009).

In this work, our aim was to identify putative cis-regulatory ele-ments involved in the coordinated transcription of FUMgenes and tovalidate the prediction experimentally, with the broader goal ofdeepeningour understandingof themolecularmechanisms control-ling fumonisin synthesis. We detected two similar DNA sequences,over-represented within the promoters of the clustered FUM genesof the fumonisin-producing fungi F. verticillioides and A. niger, whilenot in the scattered FUM-like genes of C. heterostrophus, F. graminea-rum, A. nidulans, M. grisea and N. crassa, which do not producefumonisins. These sequences are also present in the regulatory se-quences of the clustered FUM genes in F. oxysporum, F. proliferatumand F. fujikuroi. The involvement of one of them in the transcrip-tional regulation of FUM1, encoding a key polyketide synthase(Proctor et al., 1999), was demonstrated in vivo in F. verticillioides.

2. Materials and methods

2.1. In silico analyses

V. Montis et al. / Fungal GenetTo investigate motif over-representation in the promoters ofFUM genes, 1 kb upstream of the TSS of all FUM and FUM-likegenes were used as query in RSAT (http://rsat.ulb.ac.be/) (Tho-mas-Chollier et al., 2008; van Helden, 2003). FUM20 was excludedfrom the analysis since its function is unknown and the geneshares no signicant similarity with any previously described genein the NCBI database, although it was proven to be transcribed(Brown et al., 2005). Sequences were downloaded from the BroadInstitute for all fungi (http://www.broadinstitute.org/scientic-community/data) with the exception of sequences of the A. nigerstrain ATCC1015 (Andersen et al., 2011), which were obtained fromthe JGI database (http://genome.jgi-psf.org/Aspni1/Aspni1.ho-me.html), and of the upstream sequence of the FUM1-like from C.heterostrophus (Kroken et al., 2003), from the JGI database(http://genomeportal.jgi-psf.org/CocheC5_3/CocheC5_3.home.html).

F. verticillioides and F. oxysporum sequences (NCBI accessionsAF155773.5 and EU449979.1, respectively) are from strains 7600and O-1890, respectively. A. niger sequences presented in the man-uscript are from strain CBS 513.88 (Pel et al., 2007).

F. graminearum, A. nidulans, M. grisea and N. crassa sequenceswere retrieved according to a recent article (Khaldi and Wolfe,2011). F. fujikuroi (IMI58289) and F. proliferatum (strain ET1)FUM cluster sequences were retrieved from a recent genomesequencing program (B. Tudzynski and co-workers, unpublished).For pattern discovery, the oligo-analysis (words) and the dyad-analysis (spaced pairs) tools of RSAT were used. For patternmatching, DNA-pattern (string description) and genome-scaleDNA-pattern, tools of the same software were used, and featuremaps generated. Background frequencies for all these analyseswere calculated genome-wide on 1000 bp upstream of the TSSsfor all the analyzed fungi, and for the range 100600 or 6001000 bp. As negative controls, three sets of 20 random sequences(1 kb upstream of TSSs) from the genomes of F. verticillioides, F.graminearum and Saccharomyces cerevisiae were created by therandom gene-selection tool (RSAT), and a subsequent pattern-matching computation for the motif found was performed onthem. To prevent methodological biases (Tompa et al., 2005), thequery sequences from F. verticillioides and A. niger were alsosearched for conserved regulatory sequences using the stand-aloneversion of Weeder (among the MoD Tools; http://159.149.109.9/modtools/) (Pavesi et al., 2004, 2006). Background frequencies forthis analysis were calculated on genome-wide sequences upstreamof TSSs of both F. verticillioides and A. niger by using the FrequencyMaker program of Weeder. The Motif Locator and the Motif p-valueCalculator tools were used to compute signicance P-values and tolocate instances of a given motif in a set of sequences, respectively.

The databases JASPAR Core Fungi (http://jaspar.genereg.net/),TRANSFAC (http://www.gene-regulation.com/cgi-bin/pub/dat-abases/transfac/search.cgi), and YEASTRACT (http://www.yea-stract.com/) were employed for a rst characterization of the6-mers found.

2.2. Nucleic acids methods

A mutated version (PFUM1mod) of FUM1 putative promoter se-quence (PFUM1) was created by replacing the main 6-mer sequencein PFUM1 (CGGATA) with the XbaI target sequence (TCTAGA). Wereplaced PFUM1 with this modied version within the constructpCAM-PFUM1::GFP (green uorescent protein) (Visentin et al., 2012)in several cloning steps, using a set of three modied primer pairs(Xba_seq3_for and seq3_Sma_rev; Aat_seq1_for and seq1_Xba_rev;Xba_seq2_for and seq2_Xba_rev, see Table S1) to create the nalconstruct pCAM-PFUM1mod::GFP, which was fully sequenced (Fig.S1). For cloning purposes, PCR reactions were set up in 25 ll (nal

nd Biology 51 (2013) 4249 43volume) of Pfu Reaction Buffer containing 1.5 mM MgSO4 (Prome-ga), dNTPs 0.2 mM each, primers 0.5 lM each, DNA template 1 ng,Pfu DNA polymerase 0.375 U and GoTaq DNA polymerase 0.875 U

ics a(Promega). PCR conditionswere the following: 94 C for 2 min; then35 cycles at 94 C for 45 s, appropriate annealing temperature(Table S1) for 45 s, 72 C for 1 min; and nal extension at 72 C for5 min. PCR products were puried from agarose gels by NucleoSpinExtract II kit (MachereyNagel). All restriction enzymes were fromFermentas, while T4 DNA ligase was from Promega and DNA phos-phatase was from Amersham Pharmacia Biotech.

For the extraction of genomic DNA to be used in Southern blotanalyses, we used a variant of the CTAB method (Brandfass andKarlovsky, 2008) with slight modications, and analyzed 5 lg foreach strain as described previously (Visentin et al., 2012). Forinduction kinetics and plant infection experiments, total RNAwas extracted using the NucleoSpin RNA Plant kit (MachereyNa-gel) according to the manufacturers instructions. For all othersamples, RNA was obtained from 15 mg of fungal mycelium by aslightly modied Changs method (Chang et al., 1993). Qualityand quantity of RNA were checked with the RNA 6000 Nano Lab-Chip kit in the Agilent 2100 Bioanalyzer (Agilent Technologies).Only samples with spectrophotometric ratios of A260/A280P 1.9,A260/A230P 2, and RNA integrity number valuesP 7 were used.After extraction, RNA samples were stored at 80 C. cDNA wassynthesized starting from 10 lg of total RNA with the HighCapacity cDNA Reverse Transcription Kit (Applied Biosystems),following the manufacturers instructions. 1 ll of each samplewas analyzed by NanoDrop ND 1000 equipped with the ND-1000V3.7 software (Thermo Fisher Scientic). cDNA samples werestored at 20 C.

For quantitative RT-PCR experiments (RT-qPCR) analyses, abso-lute quantication (copy number) of target DNA was performed onthe StepOnePlus Real-Time PCR System (Applied Biosystems) withthe FAM dye lter set in combination with SYBR Green dye (SYBRGreen PCR Master Mix, Applied Biosystems). RT-qPCR reactionswere set up in 10 ll (nal volume) of SYBR Green PCR MasterMix, with primers (500 nM each) and cDNA (10 ng). Transcriptsfrom each of the two targets and one reference gene (FUM1, GFPand TUB2) were quantied in each of three independent biologicalreplicates per experimental condition, in analytical triplicates. Forstandard curves, target sequences from each of the three geneswere amplied from genomic DNA of F. verticillioides (primer pairsFUM1_for/_rev, GFP_for/_rev and TUB2_for/_rev in Table S1, thesame used for RT-qPCR) (Flaherty et al., 2003) and cloned inpGEM-T Easy (Promega). Five-point curves were obtained by seri-ally diluting (1:5) the obtained constructs; concentrations rangedfrom 25 pg to 40 fg of each plasmid for the three target sequences,corresponding to 7.30 1061.17 104 copies/reaction for FUM1and to 7.04 1061.13 104 copies/reaction for GFP and TUB2.Differences in copy number with respect to quantity are due tothe difference in length of the PCR amplication products (TableS1). All standard samples were run in analytical triplicates. PCRconditions were the following: 95 C for 10 min; 41 cycles at95 C for 15 s and 60 C for 1 min; and 95 C for 15 s. Differencesin the amplication efciencies of the reference (TUB2) and the tar-get gene transcripts (FUM1 and GFP) in RT-qPCR experiments wereconsidered acceptable if 615% within each run and between inde-pendent experiments, with slopes of the standard curves rangingfrom 3.5 to 3.2, R2 values larger than 0.98, and CT valuesbetween 18 and 30 for target genes and about 23 for TUB2 inall experiments. Transcripts of FUM1 and GFP were normalizedusing TUB2 values and ratios of GFP to FUM1 transcripts werecalculated.

2.3. Agrobacterium-mediated transformation of F. verticillioides andcharacterization of fungal transformants

44 V. Montis et al. / Fungal GenetF. verticillioides strain VP2 (Visentin et al., 2009) was used to cre-ate transgenic strains named Fv-PFUM1mod via Agrobacterium tum-efaciens-mediated transformation (Takken et al., 2004), withminor modications (Visentin et al., 2012). Briey, competent A.tumefaciens strain EHA 105 was prepared by CaCl2 treatment (Yonget al., 2006) and transformed by heat shock with pCAM-PFUM1-mod::GFP plasmid DNA (Fig. S1). Rifampicin (50 lg/ml) and Kana-mycin (50 lg/ml) were used to select positive transformants.Vector acquisition by Agrobacterium was checked by PCR. Fungaltransformation was carried out as described before (Visentin et al.,2012). Hygromycin-resistant fungal colonies were tested for thepresence of the Hygromycin-resistance cassette by PCR, using FTAcards according to the manufacturers instructions (WhatmanSchleicher and Schuell). For each positive Fv-PFUM1mod transform-ant, a single conidiumwas pickedmanually, transferred onto a plateof Czapek medium supplemented with Hygromycin (100 lg ml1)and Cefotaxime (200 lM), and incubated for 4 d at 25 C. Monocon-idial stock colonies were stored at 4 C onto slants of Malt ExtractAgar (Sigma). Southern blot and hybridization analysis wereperformed as follows. 5 lg of genomic DNA from wild-type F.verticillioides strain VP2 and 13 independent Fv-PFUM1mod transfor-mants were analyzed as in Visentin et al. (2012). TransformantFv-PFUM1mod-C was excluded from analyses because of its alteredcolony morphology. Single-copy, independent Fv-PFUM1modtransformants (indicated by capital letters B, D E and F in Fig. S2)were selected for further analyses. Their toxigenicity was comparedto the one of the wild-type parental strain by quantifying FB1production on cracked maize kernels after 18 d of culture at 25 Cin the light (106 spores in 3 ml sterile water were used to inoculate8 g of autoclaved cracked kernels, in triplicate). Concentration ofFB1 was determined by HPLC/MS as described previously (Visentinet al., 2012).

2.4. Fungal cultures and media, plant infection experiments

For Southern blot analysis, F. verticillioides strains were grownfor 7 d at 25 C on Czapek medium (fumonisin non-inducing. In1 l: 2 g NaNO3, 1 g KH2PO4, 0.5 g MgSO47H2O, 0.5 g KCl, 0.01 gFeSO47H2O, 30 g sucrose). For quantitative analysis of PFUM1 vs.PFUM1mod transcriptional activity, the transgenic control strainsFv-PFUM1-1 and -3 (Visentin et al., 2012) and the Fv-PFUM1mod-B-D -E and -F transformants were grown in triplicate (18 d at25 C in the dark and in stationary conditions) on Czapekmedium or on two different inductive substrates for fumonisinproduction: GYAM (0.12 M glucose, 50 mM malic acid, 8 mML-Asp, 1.7 mM NaCl, 4.4 mM K2HPO4, 2 mM MgSO4, 8.8 mMCaCl2, 0.05% w/v yeast extract, pH 3) and fructose medium (in1 l: 0.5 g malt extract, 1 g yeast extract, 1 g peptone, 1 g KH2PO4,0.3 g MgSO47H2O, 0.3 g KCl, 0.05 g ZnSO47H2O, 0.01 g CuSO45H2O, 20 g fructose). For induction kinetics, wild-type, Fv-PFUM1-1and Fv-PFUM1mod-E were grown for 7, 14 or 21 d at 25 C inthe dark and in stationary conditions in liquid fructose medium.All liquid media (40 ml per replicate, in 100 ml asks) wereinoculated with a starting concentration of 105 conidia ml1.Hyphal mats were harvested by ltration (0.8 lm pores,Millipore), frozen in liquid nitrogen and stored at 80 C. Forseedling infection experiments, maize kernels (Silver Queenhybrid, Syngenta Seeds) were incubated overnight in steriledistilled water containing, per ml, 106 conidia of transformantsFv-PFUM1-1, Fv-PFUM1mod-E or Fv-PFUM1mod-F. Infected andcontrol, non-infected seeds were germinated on water-agarplates (1 week at RT, 12 h dark/light cycle), then sown in steril-ized soil and grown in the greenhouse for one additional week(1025 C and about 12 h dark/light cycle). At harvest, rootletswere quickly washed under tap water, frozen in liquid nitrogen

nd Biology 51 (2013) 4249and stored at 80 C until RNA extraction. Three independentpools of three rootlets each were analyzed for each sample.The experiment was conducted thrice.

2.5. Statistical analysis

Fishers tests (http://www.langsrud.com/sher.htm) for assess-ing hexamers representation within the two analyzed fungalgenomes were calculated on occurrences in the FUM clusters ascompared to the full genome, both on 1000 bp and in 100600 bp regions upstream of TSSs. Bonferroni correction was ap-plied and corrected P values 6 0.05 were considered signicant.Analysis of variance between the TUB2-normalized GFP/FUM1 ra-tios was performed using the LSD test, with 95% or 99% P-values.Data were analyzed by the SGwin software (StatGraphics Plus forWindows version 2.1, Statistical Graphics Corp.).

3. Results

3.1. In silico analysis reveals similar over-represented sequences

Most signicant 6-mers in the FUM cluster as identied by oligo analysis (RSAT) on1 kb sequences upstream of TSSs in FUM genes of the fumonisins producers F.

V. Montis et al. / Fungal Genetics averticillioides and A. niger, and in the promoters of FUM-like genes scattered in thegenome of four other fungal species, which do not produce fumonisins (F. graminea-rum, A. nidulans, N. crassa, M. grisea). Full results of RSAT analysis are provided inTable S2. The second column shows the over-represented sequences identied, in bothorientations. P-value is the probability to observe the same result by chance (falsepositive). E-value is the number of patterns with the same level of over-representa-tion which would be expected by chance alone. Signicance index (SI) is theLog10(E-value), and the higher values are associated to the more signicant patterns.

Organism 6-mer P E SI

Most highly-scoring sequences by oligo-analysis (RSAT)F. verticillioides CGGATA|TATCCG 1.5e

08 3.2e05 4.49

A. niger CGGCTA|TAGCCG 1.9e04 3.8e01 0.42

F. graminearum CTCCAC|GTGGAG 1.7e06 3.4e03 2.47A. nidulans AAAAAA|TTTTTT 5.1e12 1.0e08 7.99within promoters of clustered FUM genes in fumonisin producers F.verticillioides and A. niger

In silico pattern-discovery (by RSAT software) detected two sim-ilar 6-mers signicantly over-represented in the 1000 bp upstreamof transcription start-sites (TSSs) of clustered FUM genes of bothfungi, as compared to background values obtained on the corre-sponding regions of all protein-coding genes (Tables 1 and S2).All sequences will be reported hereafter in a simplied way, i.e.only in their forward orientation, but they are meant to indicateboth forward and reverse orientation. The sequences CGGATA,present in 87.5% of FUM promoters in F. verticillioides, and CGGCTA,present in 63.6% of FUM promoters in A. niger, were suggested asmost signicant by RSAT oligo-analysis and were the only onesin common with the Weeder analysis output on the two FUM clus-ters (Table S2). They both contain the CGG triplet. Instead, runningthe pattern-discovery tool (RSAT) on FUM-like genes scattered inthe genomes of the fumonisin non-producers F. graminearum, M.grisea, A. nidulans and N. crassa identied different over-repre-sented hexamers than in the promoters of clustered FUM genesin F. verticillioides and A. niger (Tables 1 and S2). Notably, theCGG triplet was not present anymore within the suggested hexa-mers. The motif was absent also from the putative regulatory se-quence of the FUM1-like gene of C. heterostrophus (Kroken et al.,2003). When the dyad-search algorithm was performed as imple-mented in RSAT, the most signicant output was the CGGN(0)ATAdyad, which is equivalent to the sequence identied in F. verticillio-ides. In A. niger no signicant dyads were found (Table S2).

The sequences identied in F. verticillioides (CGGATA) and in A.niger (CGGCTA) were analyzed for their distribution and abundance

Table 1N. crassa TCTTCA|TGAAGA 4.43e04 4.0e01 0.1M. grisea AAAAAA|TTTTTT 2.6e15 5.2e12 11.28within both fungal genomes; Fishers Exact Test with Bonferronicorrection was used to test over-representation of each of the twosequences in the genome of the other fungal species by comparingsequence occurrences in the regulatory regions of the cluster andof the full corresponding genomes. Results conrmed the signicantover-representation (P 6 0.05) of each sequence in both genomeswhen the analysis was limited to the region 0.10.6 kb upstreamof the TSSs, where TFBSs are most probably located (Table S3). It isnoteworthy that both sequences display the best signicance inthe species where theywere initially discovered, while being signif-icantly over-represented also in the FUMcluster of the other species.This latter nding is compatible with the hypothesis that both se-quences contribute to cis-regulation of the FUM genes, possibly asTFBSs, being under selective pressure for conservation in phyloge-netically distant species. Combining themain sequences as obtainedfrom the above analyses, a degenerated motif CGGMTA can be de-ned (with M = A or C). This is common to the two species and isover-represented as such in the FUM clusters of the two species(P < 0.05, Tables S2 and S3). Fig. S3 localizes visually theCGGMTAse-quence and its orientationwithin the FUMpromoters of F. verticillio-ides and A. niger, and shows a similar distribution of occurrences pergene in the two species. Motif distribution was also determined inthe A. niger strain ATCC1015, which does not produce fumonisins(Andersen et al., 2011). No differences in position and number ofoccurrences of themotifwere foundbetween the twoA.niger strains(data not shown). Localization of themotif in the regulatory regionsof the clustered FUM genes of fumonisin-producing F. oxysporumand F. proliferatum strains, and of a non-producing F. fujikuroiisolate (the latter two sets of sequences were kindly providedby B. Tudzynski, Westflische Wilhelms-Universitt Mnster,Germany) was also carried out (Fig. S3). The lack of genomesequences for these fungi did not make it possible to evaluate thesignicance of the motif over-representation, but it is clear thatalmost all FUM genes, if clustered, have at least one motif in theregion 0.10.6 kb upstream of the TSS, both in fumonisin-producers(F. verticillioides, F. oxysporum, F. proliferatum, A. niger) and non-producers (F. fujikuroi).

When the 6-mers identied were searched as such within pub-lic databases specic for TFBSs (TRANSFAC, JASPAR and YEA-STRACT), no matches were found. However, when bindingstringency was lowered to 80% or less in the JASPAR database,the main sequence of F. verticillioides appeared as recognizable byseveral S. cerevisiae TFs belonging to the group of the zinc (orbinuclear) cluster proteins (Table S4). This protein family, uniqueto the fungal kingdom, is characterized by a zinc-nger domainof the Zn(II)2Cys6 type, where two Zn atoms are coordinated bysix Cys residues in the DNA-recognition domain. Binding sites forthis family of TFs invariably contain a CGG triplet (MacPhersonet al., 2006).

Overall, our in silico results suggest that the CGGMTA motifidentied on the basis of its abundance within the FUM promotersmight be the conserved core of a cis-regulatory element. Possibly,this motif is part of a TFBS specic for a zinc-cluster TF involvedin fumonisin biosynthesis.

3.2. A version of F. verticillioides PFUM1 mutated in the main sequencedrives signicantly less transcription than native PFUM1

The main sequence CGGATA is present twice in the FUM1 pro-moter of F. verticillioides, PFUM1 (at 643 and390 bp from TSS; Ta-ble 2). To validate the results obtained in silico, we replaced bothinstances with XbaI restriction sites (TCTAGA) to generate a modi-ed version of PFUM1 (PFUM1mod). This was inserted upstream of a

nd Biology 51 (2013) 4249 45GFP-encoding reporter gene (Fig. S1) in a construct then trans-ferred into F. verticillioides. A set of 14 transformants (Fv-PFUM1-mod-A to -O) was generated by A. tumefaciens-mediated

mai

ics atransformation in two independent experiments. Single-copy,independent transformants Fv-PFUM1mod-B, -D, -E and -F (Fig. S2)with wild-type colony morphology were retained for further inves-tigation. Toxigenicity assays on cracked maize kernels revealedthat these transformants produced fumonisins at levels compara-ble to the parental strain (300500 lg FB1 per g of substrate; datanot shown).

GFP transcripts were quantied in one non-inducing (Czapek)and two fumonisin-inducing media (GYAM and fructose med-ium). To evaluate if the rate of expression driven by PFUM1was changed by the mutation of the two putative TFBSs, Fv-PFUM1mod transformants were compared to strains Fv-PFUM1-1and -3. These bear a cognate expression cassette with the intactPFUM1 sequence coupled to a GFP reporter gene (Visentin et al.,2012). After 18 d of growth in the fumonisin non-inducing med-ium, all transformants showed detectable amounts of GFP tran-script, while the transcript of the endogenous FUM1 gene wasalmost undetectable (Fig. 1A). GFP expression on this mediumwas higher to a statistically signicant degree (P < 0.01) in thecontrol transformants Fv-PFUM1-1 and -3 with respect to thetransformants containing the PFUM1mod::GFP expression cassette(Fig. 1A). The trend was conrmed in the fumonisin-inducingmedia: GFP expression was lower in all the Fv-PFUM1mod com-pared to the Fv-PFUM1 transformants (data not shown). The re-sults are better visualized by GFP/FUM1 transcript ratios,

Table 2Summary of the number of occurrences (OCCs) and positions relative to the TSSs of the

Gene Putative function

FUM21 Zn2Cys6 transcription factorFUM1 (FUM5) Polyketide synthaseFUM6 Cytochrome P450 monooxygenase/reductaseFUM7 Alcohol dehydrogenaseFUM8 Oxoamine synthaseFUM3 (FUM9) DioxygenaseFUM10 Fatty acyl-CoA synthetaseFUM11 Tricarboxylate transporterFUM2 (FUM12) Cytochrome P450 monooxygenaseFUM13 Short-chain dehydrogenase/reductaseFUM14 Peptide synthetase/condensation domainsFUM15 Cytochrome P450 monooxygenaseFUM16 Fatty acyl-CoA synthetaseFUM17 Ceramide synthaseFUM18 Ceramide synthaseFUM19 ABC transporter

46 V. Montis et al. / Fungal Genetwhich are signicantly lower in the latter set of transformantscompared to the former (P < 0.01) (Fig. 1B and C). The timecourse of GFP and FUM1 transcript levels in Fv-PFUM1mod-Ewas assessed on fructose medium at 7, 14 and 21 d post-inoc-ulation and compared to strain Fv-PFUM1-1. Impairment of PFUM1activity by the mutation of the putative TFBSs approximately bya factor of 3 was consistent over the whole incubation period(Fig. 1D). These differences were also observed for strains colo-nizing maize rootlets, conrming that the mutation has an ef-fect also under physiologically relevant conditions (Fig. 2).Comparison of Fv-PFUM1 vs. a Fv-PFUM1mod strain in fructosemedium revealed that GFP transcript was less abundant thantranscript of the endogenous FUM1 in the transformant contain-ing the mutated PFUM1, but more abundant than endogenousFUM1 transcript in the transformant containing native, unmod-ied PFUM1 (Fig. S4). The difference was consistent and signi-cant at all three time points. Summarizing, on all syntheticsubstrates tested as well as during plant infection, targetedmutation of the two putative cis-regulatory elements CGGATAfrom the promoter of FUM1 signicantly reduced the level oftranscript generated by the promoter ectopically inserted in F.verticillioides genome.4. Discussion

4.1. A candidate cis-regulatory element is found in promoters ofclustered FUM genes, and a regulatory role for the main sequence isconrmed experimentally in F. verticillioides

In silico analyses revealed a set of two 6 bp-long sequences com-mon to clustered FUM genes in the fumonisin-producing fungi F.verticillioides and A. niger, which can be described by the commonmotif CGGMTA. Depending on the fungus, one of the two se-quences, the most abundant, can be regarded as the main se-quence; while the other is less represented and can be dened asan auxiliary or sister sequence. Neither of these sequences wasfound to be signicantly present in the promoters of the FUM-likegenes scattered in the genomes of fungi that do not producefumonisins (F. graminearum, M. grisea, N. crassa, A. nidulans, C.heterostrophus).

To test the involvement of the putative cis-regulatory elementin the regulation of the FUM cluster, we focused on the contribu-tion of the main sequence found in F. verticillioides to the expres-sion of the key biosynthetic gene FUM1. Comparison of the intactFUM1 promoter with a synthetic version in which the sequencewas mutated revealed that under all conditions, in vitro as wellas in planta, the transcriptional strength of the mutated promoterwas signicantly reduced. The transcript of the GFP gene driven

n sequence discovered by oligo analysis in the clustered FUM genes of F. verticillioides.

CGGATA|TATCGG OCC Start position (relative to TSS)

2 978, 1312 643, 3903 815, 116, 852 244, 2132 217, 1504 828, 647, 228, 1102 127, 9 3 906, 827, 4993 367, 316, 299 1 8181 1521 1831 1151 161

nd Biology 51 (2013) 4249by the intact PFUM1 promoter was more abundant than the tran-script of the endogenous FUM1 gene; as commented elsewhere,this might be due to lack of epigenetic control on the ectopic PFUM1(Visentin et al., 2012) However, the transcript level of the GFP dri-ven by FUM1 promoter deprived of the putative cis-regulatory ele-ment was signicantly lower, in vivo, than that of the endogeneousFUM1, suggesting that positive control of FUM1 via the main puta-tive cis-regulatory element plays a crucial role together with epige-netically mediated control. Since the main sequence is present innearly all FUM gene promoters, it is likely involved in the controlof the FUM gene cluster as a whole in F. verticillioides. As in tricho-thecene biosynthesis (Seong et al., 2009), comparative genomicsproved here successful in identifying putative TFBSs, conrmingthat statistical approaches for canonical motif nding perform beston sets of genes where co-regulation is strictly required (as in tox-in-biosynthetic clusters).

4.2. FUM clusters in closely and distantly-related fungal species showlocalized conservation for the putative cis-regulatory element

The signicant local over-representation of the degeneratedmotif CGGMTA with respect to the whole genome was conrmed

ics aV. Montis et al. / Fungal Genetin the 100 to 600 regions of clustered FUM genes for the twofumonisin producers for which whole reference genomes are avail-able. This positional conservation suggests functional constraints

norm

aliz

ed tr

ansc

ript a

mou

nts

0

4

2

6

8

10

Fv-PFUM1 Fv-PFUM1mod

1 F3 B D E

(A)GFPFUM1

A A

B B B B

GFP

/FUM1

trans

crip

t rat

ios

0

2

4

3

1

A

B

Fv-PFUM1 Fv-PFUM1mod

1 F3 B D E

((C)

Fig. 1. Mutations in the main sequence CGGATA of F. verticillioides impairs FUM1-prverticillioides transformants Fv-PFUM1-1 and -3 and Fv-PFUM1mod-B, -D, -E, -F grown for 1inducing) and (C) fructose medium (fumonisin-inducing). Results for GFP and FUM1 werebefore being divided by each other. For statistical analysis, Fv-PFUM1 transformants wereDifferent letters indicate a statistically signicant difference (LSD test, P < 0.01). (D) GFP/FE during a 3-point kinetics in fructose medium (fumonisin-inducing). For each time-pdpi = days post-inoculation.

GFP

/FUM1

trans

crip

t rat

ios

0

2

4

3

1

A

B

Fv-PFUM1-1 Fv-PFUM1mod

FE

B

Fig. 2. Mutations in the main sequence CGGATA of F. verticillioides impairs FUM1-promoter (PFUM1) activity during plant infection. GFP/FUM1 transcript ratios in F.verticillioides transformants Fv-PFUM1-1, Fv-PFUM1mod-E and PFUM1mod-F wereassessed in infected maize rootlets 1 week after sowing. Results for GFP andFUM1 were obtained by absolute RT-qPCR and then normalized over TUB2transcript copy number, before being divided by each other. Signicant differencesare indicated by different letters on top of bars (LSD test, P < 0.05).GFP

/ FUM1

trans

crip

t rat

ios

0

2

1

3

4

5

6

A

B

Fv-PFUM1 Fv-PFUM1mod

1 F3 B D E

** *

*Fv-PFUM1mod-EFv-PFUM1-1

2.5

2

1.5

trans

crip

t rat

ios

D)

(B)

nd Biology 51 (2013) 4249 47on the putative TFBSs (Hare and Palumbi, 2003) and is perfectlycompatible with the hypothesis of a remote horizontal gene trans-fer event (Khaldi and Wolfe, 2011). The sequence was also found tobe similarly distributed in the regulatory regions of clustered FUMgenes in a fumonisin-producing F. oxysporum strain, and in F. pro-liferatum and F. fujikuroi. The lack of whole reference genome se-quences for these species and strains did not allow estimation ofthe statistical signicance of this nding. However, the numberand distribution of the instances of the putative cis-regulatory ele-ment is similar to those observed on F. verticillioides and A. niger,further suggesting its association with the regulatory regions ofclustered FUM genes. We expect that when whole backgroundgenomes become available, it will be possible to understand ifthe core motif identied here should be further degenerated orchanged for these fungi. Finally, the overall comparable numberand distribution of the occurrences in the FUM promoters of theF. fujikuroi and A. niger ATCC1015 strains analyzed, which bearthe cluster but do not produce fumonisins, may imply that an epi-static level of repression on the cluster must exist and be regulateddifferently than in fumonisin-producing strains.

The fact that the same motif is signicantly over-represented inF. verticillioides and A. niger is surprising, since cis-regulatory ele-ments are typically divergent in sequence between fungi belongingto different classes (Gasch et al., 2004). This nding is even moreintriguing since the conditions inducing fumonisin synthesis inthe two fungi are completely different (Frisvad et al., 2007);

0

1

0.5

dpi7 14 21

GFP

/FUM1

omoter (PFUM1) activity in vivo. Transcript abundance for FUM1 and GFP from F.8 days in (A) Czapek (non-inducing medium for fumonisins); (B) GYAM (fumonisin-obtained by absolute RT-qPCR and then normalized over TUB2 transcript quantitieseither individually (A) or collectively (B and C) compared with Fv-PFUM1mod strains.UM1 transcript ratios in F. verticillioides transformants Fv-PFUM1-1 and Fv-PFUM1mod-oint, signicant differences are indicated by stars. H P < 0:05; HH P < 0:01.

ics afumonisin production may indeed serve as a model for the diversi-cation of metabolic regulation in spite of the retention of sharedcis-regulatory elements and of the key TF. However, our observa-tion also supports the hypothesis that orthologous TFs (most prob-ably, but not necessarily, a narrow-domain TF located within theFUM cluster) may recognize these conserved cis-regulatory ele-ments in the two species. Binding sites of TFs belonging to the Zincbinuclear cluster and possessing a Zn(II)2Cys6 motif, which are onlyfound in fungi, share the CGG core contained in the motif identiedin this study (Campbell et al., 2008; MacPherson et al., 2006). Wetherefore hypothesize that the TF binding to the putative cis-regu-latory elements is a Zinc binuclear TF with a Zn(II)2Cys6 motif. In-deed, several Zinc binuclear TFs from S. cerevisiae may recognizethe primary sequence of F. verticillioides, as seen when the strin-gency for the binding constraints was lowered to 80% in searchingthe JASPAR database. Fum21 belongs to this same family (Brownet al., 2007), as many other narrow-domain TFs specic for geneclusters in secondary metabolism (Bok et al., 2006; Fox et al.,2008; Wiemann et al., 2009; Woloshuk et al., 1994). FUM21 is pres-ent in all FUM clusters analyzed (Khaldi and Wolfe, 2011; Pel et al.,2007). The most parsimonious hypothesis is therefore that Fum21is the TF binding the identied motif, assuming that this is a TFBS.fum21 Knock-out mutants of F. verticillioides accumulate only verylimited amounts of FUM1 transcripts under fumonisin-inducingconditions (Brown et al., 2007). In our experimental system, we ob-served residual activity by PFUM1mod (although signicantly im-paired with respect to PFUM1). Together, these results suggest thatthe effect of mutating the putative TFBSs of PFUM1 is weaker thanthat of deleting FUM21. We reasoned that this may be due to aincomplete inactivation of the binding site, if (as likely) the hexa-mer is part of a larger binding domain. If this is the case, residualbinding activity may account for a certain extent of activity byPFUM1mod. As an obvious alternative hypothesis, another TF maybe involved, which is not absolutely required for FUM1 transcrip-tion. However, of course, more experiments are needed to proveor disprove this hypothesis.

Acknowledgments

Wewish to thank Prof. G. Tamietti for nancial and D. Valentinofor technical support, respectively; M. Beinhoff and K. Dll for helpin Southern blot analysis and FB1 quantication, respectively; B.Amato for assistance in statistical analysis; U. Gldener and B.Tudzynski for sharing unpublished FUM regulatory sequences fromF. proliferatum and F. fujikuroi; G. Munkvold for critical reading ofthe manuscript.

Appendix A. Supplementary material

Supplementary data associated with this article can be found, inthe online version, at http://dx.doi.org/10.1016/j.fgb.2012.11.009.

References

Andersen, M.R. et al., 2011. Comparative genomics of citric-acid-producingAspergillus niger ATCC 1015 versus enzyme-producing CBS 513.88. GenomeRes. 21, 885897.

Bartk, T. et al., 2006. Detection of new fumonisin mycotoxins and fumonisin-likecompounds by reversed-phase high-performance liquid chromatography/electrospray ionization ion trap mass spectrometry. Rapid Commun. MassSpectrom. 20, 24472462.

Bayram, O., Braus, G.H., 2012. Coordination of secondary metabolism anddevelopment in fungi: the velvet family of regulatory proteins. FEMSMicrobiol. Rev. 36, 124.

Bok, J.W. et al., 2006. Secondary metabolic gene cluster silencing in Aspergillusnidulans. Mol. Microbiol. 61, 16361645.

48 V. Montis et al. / Fungal GenetBrandfass, C., Karlovsky, P., 2008. Upscaled CTAB-based DNA extraction and real-time PCR assays for Fusarium culmorum and F. graminearum DNA in plantmaterial with reduced sampling error. Int. J. Mol. Sci. 9, 23062321.Brown, D.W. et al., 2007. The Fusarium verticillioides FUM gene cluster encodes aZn(II)2Cys6 protein that affects FUM gene expression and fumonisin production.Eukaryot. Cell 6, 12101218.

Brown, D.W. et al., 2005. Comparative analysis of 87,000 expressed sequence tagsfrom the fumonisin-producing fungus Fusarium verticillioides. Fungal Genet.Biol. 42, 848861.

Campbell, R.N. et al., 2008. Metabolic control of transcription: paradigms andlessons from Saccharomyces cerevisiae. Biochem. J. 414, 177187.

Chang, S. et al., 1993. A simple and efcient method for isolating RNA from pinetrees. Plant Mol. Biol. Rep. 11, 113116.

Doniger, S.W. et al., 2005. Identication of functional transcription factor bindingsites using closely related Saccharomyces species. Genome Res. 15, 701709.

Flaherty, J.E. et al., 2003. PAC1, a pH-regulatory gene from Fusarium verticillioides.Appl. Environ. Microbiol. 69, 52225227.

Fox, E. et al., 2008. A Zn(II)2Cys6 DNA binding protein regulates the sirodesmin PLbiosynthetic gene cluster in Leptosphaeria maculans. Fungal Genet. Biol. 45,671682.

Fox, E.M., Howlett, B.J., 2008. Secondary metabolism: regulation and role in fungalbiology. Curr. Opin. Microbiol. 11, 481487.

Frisvad, J.C. et al., 2007. Fumonisin B production by Aspergillus niger. J. Agric. FoodChem. 55, 97279732.

Gasch, A.P. et al., 2004. Conservation and evolution of cis-regulatory systems inascomycete fungi. PLoS Biol. 2, e398.

Hare, M.P., Palumbi, S.R., 2003. High intron sequence conservation across threemammalian orders suggests functional constraints. Mol. Biol. Evol. 20, 969978.

Keller, N.P. et al., 2005. Fungal secondary metabolism from biochemistry togenomics. Nat. Rev. Microbiol. 3, 937947.

Khaldi, N., Wolfe, K.H., 2011. Evolutionary origins of the fumonisin secondarymetabolite gene cluster in Fusarium verticillioides and Aspergillus niger. J. Evol.Biol. 2011, 17.

Kroken, S. et al., 2003. Phylogenomic analysis of type I polyketide synthase genes inpathogenic and saprobic ascomycetes. Proc. Natl. Acad. Sci. USA 100, 1567015675.

MacPherson, S. et al., 2006. A fungal family of transcriptional regulators: the zinccluster proteins. Microbiol. Mol. Biol. Rev. 70, 583604.

Munkvold, G., 2003. Epidemiology of Fusarium diseases and their mycotoxins inmaize ears. Eur. J. Plant Pathol. 109, 705713.

Nelson, P.E. et al., 1993. Fumonisins, mycotoxins produced by Fusarium species:biology, chemistry, and signicance. Annu. Rev. Phytopathol. 31, 233252.

Palmer, J.M., Keller, N.P., 2010. Secondary metabolism in fungi: does chromosomallocation matter? Curr. Opin. Microbiol. 13, 431436.

Pavesi, G. et al., 2004. Weeder web: discovery of transcription factor binding sites ina set of sequences from co-regulated genes. Nucleic Acids Res. 32, W119.

Pavesi, G. et al., 2006. MoD tools: regulatory motif discovery in nucleotidesequences from co-regulated or homologous genes. Nucleic Acids Res. 34,W566W570.

Pel, H.J. et al., 2007. Genome sequencing and analysis of the versatile cell factoryAspergillus niger CBS 513.88. Nat. Biotechnol. 25, 221231.

Picot, A. et al., 2010. Factors of the Fusarium verticillioides-maize environmentmodulating fumonisin production. Crit. Rev. Microbiol. 36, 221231.

Proctor, R.H. et al., 2003. Co-expression of 15 contiguous genes delineates afumonisin biosynthetic gene cluster in Gibberella moniliformis. Fungal Genet.Biol. 38, 237249.

Proctor, R.H. et al., 2008. A fumonisin biosynthetic gene cluster in Fusariumoxysporum strain O-1890 and the genetic basis for B versus C fumonisinproduction. Fungal Genet. Biol. 45, 10161026.

Proctor, R.H. et al., 1999. A polyketide synthase gene required for biosynthesis offumonisin mycotoxins in Gibberella fujikuroimating population A. Fungal Genet.Biol. 27, 100112.

Proctor, R.H. et al., 2004. Discontinuous distribution of fumonisin biosyntheticgenes in the Gibberella fujikuroi species complex. Mycol. Res. 108, 815822.

Reyes-Dominguez, Y. et al., 2010. Heterochromatic marks are associated with therepression of secondary metabolism clusters in Aspergillus nidulans. Mol.Microbiol. 76, 13761386.

Rheeder, J.P. et al., 2002. Production of fumonisin analogs by Fusarium species. Appl.Environ. Microbiol. 68, 21012105.

Rosewich, U.L., Kistler, H.C., 2000. Role of horizontal gene transfer in the evolutionof fungi. Annu. Rev. Phytopathol. 38, 325363.

Seong, K.-Y. et al., 2009. Global gene regulation by Fusarium transcription factorsTri6 and Tri10 reveals adaptations for toxin biosynthesis. Mol. Microbiol. 72,354367.

Takken, F.L. et al., 2004. A one-step method to convert vectors into binaryvectors suited for Agrobacterium-mediated transformation. Curr. Genet. 45,242248.

Thomas-Chollier, M. et al., 2008. RSAT: regulatory sequence analysis tools. NucleicAcids Res. 1, W119W127.

Tompa, M. et al., 2005. Assessing computational tools for the discovery oftranscription factor binding sites. Nat. Biotechnol. 23, 137144.

van Helden, J., 2003. Regulatory sequence analysis tools. Nucleic Acids Res. 31,35933596.

Varga, J. et al., 2010. Fumonisin contamination and fumonisin producing blackAspergilli in dried vine fruits of different origin. Int. J. Food Microbiol. 143, 143149.

nd Biology 51 (2013) 4249Visentin, I. et al., 2012. Transcription of genes in the biosynthetic pathway forFumonisin mycotoxins is epigenetically and differentially regulated in thefungal maize pathogen Fusarium verticillioides. Eukaryot. Cell 11, 252259.

Visentin, I. et al., 2009. The ITS region as a taxonomic discriminator betweenFusarium verticillioides and Fusarium proliferatum. Mycol. Res. 113, 11371145.

Voss, K.A. et al., 2007. Fumonisins: toxicokinetics, mechanism of action and toxicity.Anim. Feed Sci. Technol. 137, 299325.

Wan Norhasima, W.M. et al., 2009. The health and toxic adverse effects of Fusariumfungal mycotoxin, fumonisins, on human population. Am. J. Infect. Dis. 5, 273281.

Wiemann, P. et al., 2009. Biosynthesis of the red pigment bikaverin in Fusariumfujikuroi: genes, their function and regulation. Mol. Microbiol. 72, 931946.

Woloshuk, C.P. et al., 1994. Molecular characterization of AR, a regulatory locus foraatoxin biosynthesis. Appl. Environ. Microbiol. 60, 24082414.

Yong, W.T.L. et al., 2006. Optimization of Agrobacterium-mediated transformationparameters for Melastomataceae spp. using green uorescent protein (GFP) as areporter. Sci. Hortic. 109, 7885.

Yu, J.H., Keller, N.P., 2005. Regulation of secondary metabolism in lamentous fungi.Annu. Rev. Phytopathol. 43, 437458.

V. Montis et al. / Fungal Genetics and Biology 51 (2013) 4249 49

Identification of a cis-acting factor modulating the transcription of FUM1, a key fumonisin-biosynthetic gene in the fungal maize pathogen Fusarium verticillioides1 Introduction2 Materials and methods2.1 In silico analyses2.2 Nucleic acids methods2.3 Agrobacterium-mediated transformation of F. verticillioides and characterization of fungal transformants2.4 Fungal cultures and media, plant infection experiments2.5 Statistical analysis

3 Results3.1 In silico analysis reveals similar over-represented sequences within promoters of clustered FUM genes in fumonisin producers F. verticillioides and A. niger3.2 A version of F. verticillioides PFUM1 mutated in the main sequence drives significantly less transcription than native PFUM1

4 Discussion4.1 A candidate cis-regulatory element is found in promoters of clustered FUM genes, and a regulatory role for the main sequence is confirmed experimentally in F. verticillioides4.2 FUM clusters in closely and distantly-related fungal species show localized conservation for the putative cis-regulatory element

AcknowledgmentsAppendix A Supplementary materialReferences