Differential Retention of Gene Functions in a Secondary Metabolite Cluster Hannah T

Home , Alternaria, Alternaria brassicicola, Fusarium langsethiae

Differential Retention of Gene Functions in a Secondary Metabolite Cluster Hannah T. Reynolds,‡,1 Jason C. Slot,*,1 Hege H. Divon,2 Erik Lysøe,3 Robert H. Proctor,4 and Daren W. Brown*,4 1Department of Plant Pathology, The Ohio State University, Columbus, OH 2Norwegian Veterinary Institute, Oslo, Norway 3Department of Biotechnology and Plant Health, NIBIO – Norwegian Institute of Bioeconomy Research, A˚s, Norway 4Mycotoxin Prevention and Applied Microbiology Research Unit, U.S. Department of Agriculture†, Agricultural Research Service, Peoria, IL †Mention of trade names or commercial products in this article is solely for the purpose of providing speciﬁc information and does not imply recommendation or endorsement by the US Department of Agriculture. USDA is an equal opportunity provider and employer. ‡Present address: Department of Biological and Environmental Sciences, Western Connecticut State University, Danbury, CT. Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 *Corresponding authors: E-mails: [email protected]; [email protected]. Associate editor: Jeffrey Townsend

Abstract In fungi, distribution of secondary metabolite (SM) gene clusters is often associated with host- or environment-specific benefits provided by SMs. In the plant pathogen Alternaria brassicicola (Dothideomycetes), the DEP cluster confers an ability to synthesize the SM depudecin, a histone deacetylase inhibitor that contributes weakly to virulence. The DEP cluster includes genes encoding enzymes, a transporter, and a transcription regulator. We investigated the distribution and evolution of the DEP cluster in 585 fungal genomes and found a wide but sporadic distribution among Dothideomycetes, Sordariomycetes, and Eurotiomycetes. We confirmed DEP gene expression and depudecin production in one fungus, Fusarium langsethiae. Phylogenetic analyses suggested 6–10 horizontal gene transfers (HGTs) of the cluster, including a transfer that led to the presence of closely related cluster homologs in Alternaria and Fusarium. The analyses also indicated that HGTs were frequently followed by loss/pseudogenization of one or more DEP genes. Independent cluster inactivation was inferred in at least four fungal classes. Analyses of transitions among functional, pseudogenized, and absent states of DEP genes among Fusarium species suggest enzyme-encoding genes are lost at higher rates than the transporter (DEP3)andregulatory(DEP6) genes. The phenotype of an experimentally-induced DEP3 mutant of Fusarium did not support the hypothesis that selective retention of DEP3 and DEP6 protects fungi from exogenous depudecin. Together, the results suggest that HGT and gene loss have contributed significantly to DEP cluster distribution, and that some DEP genes provide a greater fitness benefit possibly due to a differential tendency to form

Article network connections. Key words: secondary metabolites, gene cluster, phylogenetic analysis, pseudogenization, horizontal gene transfer, depudecin.

Introduction proteins that export the SM out of the cells in which it is synthesized, and a transcription factor that activates expres- Metabolites that provide a selective advantage in certain en- sion of the cluster genes. vironments but otherwise are not essential for growth, devel- Recent studies in fungal genome evolution have provided opment or reproduction are often referred to as secondary evidence that horizontal gene transfer (HGT) of SM biosyn- metabolites (SMs). The evolutionary history of SMs can be thetic gene clusters plays a significant role in fungal ecological dynamic because of shifting selection pressures that result adaptation (Walton 2000; Jargeatetal.2003; Bain et al. 2007; from environmental changes and co-evolution of species Fedorova et al. 2008; Khaldi et al. 2008; Slot and Rokas 2011; that interact via SMs (Stone and Williams 1992; Challis and Campbell et al. 2013; Chen et al. 2013; McGary et al. 2013; Hopwood 2003; Wink 2003). In bacteria and fungi, genes Proctor et al. 2013; Schumacher et al. 2013; Greene et al. 2014; encoding related metabolic functions are frequently found Brown et al. 2015). These observations are consistent with the in clusters (Brown et al. 1996, 2015; Bangera and “selfish cluster/operon” model of gene cluster evolution Thomashow 1999; Fischbach et al. 2008; Slot and Rokas (Lawrence and Roth 1996; Lawrence 1999). Under this model, 2011; Takos and Rook 2012; Campbell et al. 2013; Staehlin genes sharing a common function (e.g., a biosynthetic path- et al. 2016). Such clusters typically include genes coding for way) have a higher probability of persisting when they are enzymes that catalyze synthesis of the SM(s), transport horizontally transferred together in a cluster compared with

Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution 2017. This work is written by US Government employees and is in the public domain in the US.

2002 Mol. Biol. Evol. 34(8):2002–2015 doi:10.1093/molbev/msx145 Advance Access publication April 28, 2017 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE when they are transferred individually, because individual Fusarium genome dataset. Hypotheses for the role of selec- cluster genes cannot confer a selectable function alone. tion in the evolutionary mode of the DEP cluster, including Thus, organization of genes into a cluster benefits the genes the preservation of DEP function following HGT, and the themselves. Selfish clustering is expected to preferentially ben- differential retention of DEP3 and DEP6, were tested using efit genes in clusters responsible for synthesis of SMs under experimental approaches in Fusarium. weak or transient selection, because if the genes were not clustered they would be less likely to be retained in a genome Results through vertical inheritance. A gene cluster thought to have a weakly selected function Distribution and Clustering of DEP Genes is responsible for synthesis of the SM depudecin, a small linear We surveyed the presence of DEP genes in phylogenetically polyketide originally isolated from the fungus Alternaria bras- diverse fungi by BLAST analysis using nucleotide sequences or sicicola (Dothideomycetes, Ascomycota) that acts as a his- predicted amino acid sequences of the six A. brassicicola DEP tone deacetylase (HDAC) inhibitor (Matsumoto et al. 1992). genes as queries against 585 genomes and the Non- Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 In plants, HDACs and histone acetyltransferases play an inte- redundant sequence database at NCBI. These analyses gral role in modifying regulation of gene expression and affect led to the identification of DEP gene homologs in 120 a variety of cellular and developmental processes (Hollender genomes of species in five classes of the Ascomycota: and Liu 2008). Depudecin makes a small contribution to The Dothideomycetes, Eurotiomycetes, Leotiomycetes, pathogenesis; depudecin-non-producing mutants of A. bras- Sordariomycetes, and Lecanoromycetes (see supplemen- sicicola exhibit a 10% reduction in virulence on cabbage tary figs. S1–S6, Supplementary Material online, table 1 (Wight et al. 2009). In contrast, the HDAC inhibitor HC toxin and see supplementary table S1, Supplementary Material makes a more substantial contribution to virulence of online). In preliminary phylogenetic analyses, DEP1 and Cochliobolus carbonum (Dothideomycetes) on maize DEP3–DEP5 were found in large, multigene families. (Walton 2006). Depudecin also has moderate antiprotozoan Sequences in these families were manually removed at activity against parasites such as Plasmodium (Isaka et al. strongly supported nodes in order to improve the evalu- 2000)andNeospora (Kwon et al. 2003). However, it requires ation of homologs most closely related to the DEP genes. a markedly higher concentration (500 ng/ml; 0.47 mM) to in- Sequences were removed from the DEP3 gene family, hibit Neospora cell replication to the same degree as apicidin which had the highest number of homologs, so that it (15 ng/ml; 0.02 mM), another HDAC inhibitor. Thus depude- included all taxa containing five or more DEP homologs cin is expected to provide a weak and transient benefit to based on the other DEP gene trees (see supplementary fungi compared with more potent HDAC inhibitors, and to figs. S1–S6, Supplementary Material online). At least one be under inconsistent selection. DEP gene sequence was found in the 120 genomes, while 95 The depudecin biosynthetic gene (DEP) cluster consists of DEP3 homologs were distributed among 15 Dothideomycete, six adjacent and coregulated genes, DEP1–DEP6 (Wightetal. 15 Eurotiomycete, 1 Lecanoromycete, 1 Leotiomycete, and 63 2009). Three genes encode enzymes: DEP5 encodes a polyke- Sordariomycete genomes (see supplementary table S1, tide synthase (PKS), and DEP2 and DEP4 encode monooxy- Supplementary Material online, figs. 1 and 2). genases. The PKS is predicted to synthesize the carbon Multiple DEP cluster homologs exhibited gene rearrange- backbone of depudecin, while the monooxygenases likely ox- ment and loss (fig. 1, table 2). The gene order in A. brassicicola ygenate the backbone. DEP3 encodes a Major Facilitator (DEP1, DEP2, DEP3, DEP4, DEP5, DEP6) was present only in a Superfamily (MFS) transporter, DEP6 encodes a Zn2Cys6 group of Dothideomycete and Sordariomycete species that DNA-binding transcription factor (C6) and DEP1 encodes a were resolved into one clade in the DEP3 gene tree. In this protein of unknown function. Alternaria brassicicola mutant clade, the full DEP cluster was found in A. brassicicola and strains in which DEP5 or DEP6 has been inactivated do not Dothideomycetes species 5, Sordariomycete species produce depudecin, or any obvious biosynthetic intermedi- Colletotrichum fioriniae and C. higginsianum,aswellaseleven ates or shunt products, whereas mutants in which either Fusarium species (fig. 2). Seven other gene arrangements DEP2 or DEP4 has been inactivated do not produce depude- within the DEP cluster occurred in genomes with all six DEP cin, but accumulate epoxide-containing metabolites that genes on a single scaffold (fig. 1, see supplementary fig. S7, could be biosynthetic intermediates and/or shunt products Supplementary Material online and table 2). In most fungi, (Wight et al. 2009). scaffolds with the DEP cluster included more than six open Since its identification in A. brassicicola, depudecin pro- reading frames (ORFs) flanking the DEP genes, and some fungi duction has been reported in two other Ascomycete fungi: had additional enzyme-coding genes between DEP homologs Nimbya scirpicola (Dothideomycetes; Tanaka et al. 2000)and (fig. 1, see supplementary fig. S7, Supplementary Material on- Xylaria sp. BCC 1067 (Sordariomycetes; Amnuaykanjanasin line, table 2). Some of the predicted functions of the proteins et al. 2005). These reports of production in only a few diver- encoded by these additional ORFs within the DEP cluster gent fungi suggest a discontinuous distribution of the DEP were evident based on sequence homology to genes of cluster, which could be a result of HGT. Here, we conducted a known function. For example, in the Stachybotrys clusters, survey of depudecin genes in 585 fungal genomes, and mod- additional ORFs were predicted to encode two dehydroge- eled the birth and degradation of the cluster using phyloge- nases and an oxidase, and in the cluster of Sordariomycetes netic approaches, with specific attention to a well-sampled species 1, an additional ORF was predicted to encode a

2003 Reynolds et al. . doi:10.1093/molbev/msx145 MBE Table 1. DEP Clusters from Diverse Fungi (At Least Two Functional DEP Genes or Pseudogenes at the Same Locus).

Species Accession (DEP3a) Genesb Cluster Functionc Dothideomycete Alternaria brassicicola ACZ57546 6, 6 Depudecin, secrete Alternaria alternata OAG22887 2, 2 Secrete Dothideomycetes species 5d n/a 6, 6 Depudecin, secrete Setosphaeria turcica XP_008026365 6, 6 Depudecin, secrete Settur1_699796 Parastagonospora nodorum XP_001797040 2, 2 Secrete Cochliobolus lunatus Coclu2_47787 5, 4 Secrete Dothideomycetes species 12e n/a 4, 4; 1, 1 Depudecin, secrete Sordariomycete Fusarium langsethiae KPA37943 6, 6 Depudecin, secrete Fusarium verticillioidesf EWG38548 6, 1 Depudecin, depudecin-like, secrete g g

Fusarium oxysporum Fo Variable Depudecin, depudecin-like, secrete Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 Fusarium fujikuroi KLO83786 4, 2 Secrete Fusarium solani XP_003043274 2, 2 Secrete 19 other Fusarium sp. — Variable Depudecin, depudecin-like, secrete Chaetomium globosum XP_001226016 3, 3 Depudecin-like, secrete Colletotrichum higginsianum CCF33228 6, 6 Depudecin, secrete Colletotrichum fioriniae XP_007600323 6, 6 Depudecin, secrete Colletotrichum graminicola XP_008096245 5, 2 Secrete Sordariomycetes species 6 n/a 6, 6 Depudecin, secrete Sordariomycetes species 2 n/a 6, 6 Depudecin, secrete Sordariomycetes species 4h n/a 6, 4i Depudecin-like Stachybotrys chartarum KEY75120 6, 6 Depudecin, secrete Stachybotrys chlorohalonata KFA66941 6, 6 Depudecin, secrete Sordariomycetes species 1 n/a 6, 6 Depudecin, secrete Thozetella sp. PMI 491 ThoPMI4911_721523 4, 4 Depudecin-like, secrete Sodiomyces alkalinus Sodal1_32352 1, 1; 4, 4 Depudecin-like, secrete Coniochaeta ligniaria Conli1_11707 4, 4 Depudecin-like, secrete Conlig1_299260 (Jimenez et al. 2017) Clonostachys roseaj Cloro1_528243 and Cloro1_533975 2, 2; 6, 6 Depudecin, secrete Corollospora maritima Corma2_669553 2, 2 Secrete Dothideomycetes species 6 n/a 3, 2 Secrete Eurotiomycetes Arthroderma gypseum XP_003170085 6, 6 Depudecin, secrete Arthroderma otae XP_002843078 4, 2 Secrete Uncinocarpus reesii XP_002541852 5, 5 Depudecin, secrete Coccidioides posadasii EFW22347 4, 4 Depudecin-like, secrete Coccidioides immitis XP_001248625 4, 4 Depudecin-like, secrete Penicillium expansum Penex1_395006 2, 2 Secrete Penicillium chrysogenume Pench1_75739 4, 4; 2, 1 Depudecin, secrete PenchWisc11_140429 4, 4; 3,2 Aspergillus flavus XP_002378444 2, 2 Secrete Aspergillus terreus XP_001217025 2, 2 Secrete Aspergillus oryzae Aspor1_31221 2, 2 Secrete Talaromyces stipitatus XP_002482124 2, 2 Secrete Arthroderma otae Micca1_8368 3, 3 Secrete Leotiomycetes Cadophora sp. DSE1049j Cadsp1_557367 and Cadsp1_576979 2, 2; 4, 4 Depudecin, secrete Lecanoromycetes Cladonia grayi CLAGR_001305k 3, 3 Depudecin-like Xanthoria parietina Xanpa1_56935 4,1 Secrete aThe NCBI Accession or JGI Protein ID corresponds to the DEP3 homolog. bThe first number refers to the number of DEP genes and the second number refers to number of functional DEP genes. cCluster function refers to whether the DEP cluster present have the potential to synthesize depudecin or depudecin-like chemical. Secrete refers to the potential to transport a chemical, possibly depudecin or depudecin-like chemical, from the fungus and is based on the presence of a putative functional DEP3 homolog. dAs of February 17, 2017, genomes sequenced by JGI that have not yet been described in a primary publication are indicated by class and a unique number. Principle investigator and current status of genomes can be obtained from the authors. eDEP genes are located at two different loci. fThree F. verticillioides strains genomes were examined. One genome has a complete, functional DEP cluster (FGSC 7603), whereas two have partial DEP clusters: FGSC 7600 (6, 5) and MRC 826 (4, 2). gTwelve F. oxysporum strains genomes were examined. Three genomes have complete DEP clusters (Fo5176, NRRL 25433 and PHW808) whereas nine have partial DEP clusters: II5 (6, 1), Fol4287 (5, 3), NRRL 26405 (5, 3), CL57 (6, 3), Fo47 (5, 3), MN25 (5, 3), FoSC3a (4, 2), HDV247 (3, 1), and PHW815 (3, 1). hSordariomycetes species 4 has six DEP genes located on four small contigs. DEP1 and 5 are on one contig, DEP4 and 6 are on another contig and all appear functional. DEP2 and 3 are on two other contigs and may not be non-functional. iThe functionality of DEP3 could not be predicted because it is located at the edge of a contig such that 231 nucleotides of coding sequence relative to Alternaria brassicicola DEP3 is missing. Because DEP3 is not intact, it was not included in figure 2. jClonostachys rosea and Cadophora sp. have two DEP clusters containing DEP3 homologs. kJGI Protein ID for DEP4 homolog. 2004 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE

85 Fusarium * 100 Alternaria alternata 93 Alternaria brassicicola 94 89 Cochliobolus lunatus Key to Classes 97 1 71 Colletotrichum graminicola Dothideomycetes Colletotrichum higginsianum Eurotiomycetes 81 89 Pyrenophora tritici-repentis 80 Dothideomycetes species 5 Lecanoromycetes Colletotrichum fioriniae Leotiomycetes 2 Clonostachys rosea Corollospora maritima Sordariomycetes 98 Cadophora sp. DSE1049 100 Fusarium sp. CS3069 92 Fusarium solani 100 Coccidioides immitis 90 Coccidioides posadisii 99 100 Uncinocarpus reesii 3 Dothideomycetes species 2 4 100 Arthroderma gypseum Arthroderma otae 99 94 Aspergillus terreus 99 Aspergillus flavus 100 Aspergillus oryzae 97 Penicillium expansum 99 Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 Penicillium chrysogenum (rubens) Wisc54-1255 92 Penicillium chrysogenum Pench1 Clonostachys rosea Seteosphaeria turcica Et28A 5 Seteosphaeria turcica NY001 100 Sordariomycetes species 2 100 Sordariomycetes species 6 Stachybotrys chartrarum Stachybotrys chlorohalonata Sordariomycetes species 1 100 Thozetella sp. PMI491 100 Coniochaeta ligniaria CBS111746 11707 C. ligniaria NRRL30616 299260 100 Trichoderma atroviride Melanconium sp. NRRL54901 100 Thielava terrestris 97 96 Coniochaeta sp. PMI546_889453 100 100 C. ligniaria CBS11746 1_629912 100 C. ligniaria NRRL30616 10775 100 Sporotrichum thermophile Mycellophthora heterothallica 100 Exophiala dermatitidis UT8656 Penicillium glabrum 418343 6 Lollipopaia minuta 100 Cadophora sp. DSE1049 7 Chaetomium globosum 99 100 Dothideomycetes species 7 75 Dothideomycetes species 11 99 Dothideomycetes species 12 100 Sodiomyces alkalinus 8a 100 Acremonium alcalophilum Daldinia eschscholzii 8b 100 Didymella exigua 100 Stagonospora 8c nodorum Dothideomycetes species 8 Talaromyces stipitatus

0.2

FIG.1.DEP3 phylogeny inferred from maximum likelihood analysis (left) and variation in gene content and arrangement in DEP cluster homologs (right). Numbered circles indicate putative horizontal gene transfer events, where the color of the number indicates the class of the recipient. Arrow colors indicate gene identity: red for DEP1, orange for DEP2, yellow for DEP3, green for DEP4, blue for DEP5, violet for DEP6. Grey arrows represent genes that do not have a homolog in the Alternaria brassicicola DEP cluster. Genes located on different scaffolds of the same genome are represented by arrows on different sides of a backward slash. Solid lined boxes indicate cluster homologs with six putatively functional DEP genes; whereas dashed lined boxes indicate cluster homologs in which the three enzyme-encoding genes (DEP2, DEP4, and DEP5) are putatively functional but one or more of the other DEP genes is/are pseudogenized or absent. Unfilled arrows represent pseudogenes. Arrows point in the direction of transcription, and arrow length corresponds to coding region length. The triangle adjacent to Fusarium represents genes from 39 different Fusarium species. The asterisk reflects multiple gene cluster organizations as indicated in figure 2. decarboxylase (see supplementary fig. S7, Supplementary from sequences of individual DEP genessuggestsmultiple Material online). Multiple partial clusters that lacked a HGTs and subsequent gene losses (fig. 1, see supplementary DEP6 transcriptional regulator were adjacent to an ORF pre- figs. S1–S6, Supplementary Material online). The trees suggest dicted to encode a different transcription factor. Such clusters that Sordariomycetes have been the most frequent donors of included the DEP3–DEP5 cluster in Chaetomium globosum, DEP genes.InthecaseofextanttaxawithpartialDEP clusters, the DEP1–DEP5 cluster in Uncinocarpus reesii,andtheDEP1 it is unknown whether the HGT included only the partial and DEP3–DEP5 cluster in Coccidioides posadasii. Additionally, cluster or the entire cluster, which subsequently decayed to Aspergillus campestris contains a DEP6 and a pseudogenized the partial cluster. On the basis of the DEP3 tree (fig. 1), we DEP5–DEP6 cluster at a different locus. identified at least ten possible HGT events of either intact or partial cluster homologs from the Sordariomycetes to the HGT Dothideomycetes (HGT events 1, 5, 8a, and 8c), from the We identified possible horizontal transfer events based on the Sordariomycetes to the Eurotiomycetes (HGT events 4 presence of closely related DEP gene homologs in distantly and 6), from the Sordariomycetes to the Leotiomycetes related fungi (i.e., fungi from different classes). The gene con- (HGT events 2 and 7), from the Eurotiomycetes to the tent of DEP cluster homologs combined with trees inferred Leotiomycetes (HGT event 3), and from the

2005 Reynolds et al. . doi:10.1093/molbev/msx145 MBE

1 1 1 1 1 1 A F. babinda B F. redolens Fusarium sp. 25184 0 Ψ 0 Ψ 0 Ψ 0 Ψ 0 Ψ 0 XΨ F.foetens F. oxysporum II5 1 1 1 1 1 1 F. oxysporum 25433 F. oxysporum PHW808 F. oxysporum Fo5176 0 Ψ 0 Ψ 0 Ψ 0 Ψ 0 Ψ 0 XΨ F. oxysporum HDV247 F. oxysporum PHW815 F. oxysporum FoSC3a 1 1 1 1 1 1 F. oxysporum Fo47 F. oxysporum MN25 0 Ψ 0 Ψ 0 Ψ 0 Ψ 0 Ψ 0 XΨ F. oxysporum CL57 . F oxysporum 4287_ann1 dep1 dep2 dep3 dep4 dep5 dep6 F. oxysporum 4287_ann2 F. oxysporum 26406 F. commune C F . gaditjirri 15 dep1 F . miscanthi . 10 Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 F dlaminii 5 Fusarium sp. 52700 0 F . subglutinans . 15 dep2 F anthophilum 10 F . succisae F. bulbicola 5 0 F. circinatum F. sacchari 15 dep3 10 F. proliferatum . 5 F fujikuroi 0 F. thapsinum F. verticill oides 7603 15 dep4 10 F. verticill oides MRC 826 . 5 F verticillioides 7600 0 F. avenaceum F. acuminatum 15 dep5 Fusarium sp. CS3069 10 5 . (1000s) Trees Number of F scirpi 0 F. ay thiae 15 dep6 F. culmorum 10 F. 5 graminearum 0 F. pseudograminearum F. poae 20 22 24 26 28 210 F. eu Transition Rate (H2) F. solani

dep1dep2dep3dep4dep5dep6 q(1 0) q(1 Ψ) q(Ψ 0) 0.05

FIG.2.(A) Fusarium species phylogeny (left) and character grid showing patterns of DEP gene loss/retention (right). The species phylogeny was inferred by maximum likelihood analysis of concatenated sequences of 20 housekeeping genes. Supported nodes (> 95% bootstrap from 500 replicates) are in black, and unsupported nodes are in grey. Species names/strains presented in green text indicate taxa with six functional DEP genes and red text indicate taxa that have no DEP genes. Character grid indicates state of each DEP gene: functional (green), pseudogenized (yellow), or absent (white). (B) Estimated transition rates between functional, pseudogenized and absent states for each of the six DEP genes. BayesTraits was run under three hypotheses using 20,000 trees generated for individual HK genes, with 1,000 trees used per gene. The hypotheses were: H0—all transitions are permitted; H1—function can be regained only from an absent state, representing horizontal gene transfer; H2—only losses permitted. Arrow color indicates the estimated transition rate, with darker arrows representing high rates of transition, and state intensity is proportional to the number of states observed. DEP genes are colored according to function: DEP1—protein with unknown function; DEP2 and DEP4—monooxygenases; DEP3—MFS transporter; DEP5—polyketide synthase; DEP6—transcription factor. (C) Transition rates for each of the depudecin genes under the hypothesis that lost or pseudogenized genes are not regained (H1). Histogram indicates the number of Bayesian trees from a BayesTraits analysis of 20,000 trees supporting a given transition rate.

Table 2. Gene Order in Single-scaffold Clusters Containing DEP1– Dothideomycetes to the Sordariomycetes (HGT event DEP6. 8b). Among HGT recipients of the DEP cluster, the Gene order Class Species Eurotiomycetes exhibited the broadest phylogenetic diversity, with species from three orders (Chaetothyriales, 1,2,3,4,5,6 Dothideomycete Alternaria brassicicola D. species 5 Eurotiales, and Onygenales) receiving DEP genes. In contrast, Sordariomycetes Fusarium langsethiae current putative Dothideomycete recipients were restricted Colletotrichum ﬁoriniae to the order Pleosporales, and the sole current Leotiomycete 1,2,3,4,5,6r Sordariomycetes Colletotrichum higginsianum recipient of DEP genes was Cadophora sp. DSE1049 in the 6,4,5,1,2,3 Sordariomycetes S. species 6 order Helotiales. This Cadophora strain contained two DEP 1,2,3,4,5,þ,6 Sordariomycetes Eight Fusarium spp. cluster homologs derived from two independent HGT 6,þ,2,þ,4,1,5,þ,þ,3 Sordariomycete Stachybotrys chartarum 1,2,þ,þ,3,6,4,5 Sordariomycetes S. species 1 events: One cluster consisting of DEP3 and DEP6 homologs 6,1,2,3,4,5 Sordariomycetes Clonostachys rosea that were closely related to Fusarium DEP homologs (fig. 1, 4,6,þ,3,2,5,1 Dothideomycete Setosphaeria turcica see supplementary figs. S3 and S6, Supplementary Material 1,2,6,3,4,5 Eurotiomycetes Arthroderma gypseum online); and the other cluster consisting of DEP2–DEP5 ho- r ¼ Direction of transcription is reversed; þ¼additional ORF. mologs that were closely related to C. globosum DEP

2006 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE homologs (fig. 1, see supplementary figs. S3–S5, loss of all DEP genes except DEP3; 2) HGT of another segment Supplementary Material online). A second Leotiomycete, of DNA that include only DEP3 flanked by other possible SM Oidiodendron maius, contained a DEP4 homolog only (see genes; or 3) HGT of DEP3 by itself. The pattern of genes supplementary fig. S4, Supplementary Material online), but flanking DEP3 in the two non-DEP gene clusters in multiple the donor could not be ascertained given that closely related taxa (table 3) is consistent with the hypothesis that some of DEP4 homologs were present in a diversity of related taxa. the clusters underwent rearrangement or decay following DEP3 homologs were also present in two putative gene HGT. clusters that did not include other DEP genes and that had a In addition to complete gene loss, DEP pseudogenes were sporadic distributions among five or more species of the also present in phylogenetically diverse taxa. DEP pseudo- Dothideomycetes, Eurotiomycetes and Sordariomycetes, pre- geneswerecommonintheFusarium genomes in our sample, sumably as a result of HGT (table 3). One cluster was present with 25 of the 39 genomes examined having at least one DEP in the Eurotiomycetes and Sordariomycetes and encoded a pseudogene (fig. 2). Pseudogenes were also found in three transcription factor in addition to DEP3; whereas the other Dothideomycete species (Cochliobolus lunatus, Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 cluster was present in Dothideomycetes and Sordariomycetes Parastagonospora nodorum,andParaconiothyrium sporulo- and encoded a cytochrome P450 monooxygenase, two dehy- sum), six Eurotiomycete species (Arthroderma otae, A. gyp- drogenases, a FAD-binding domain protein, and a zinc-finger seum, T. marneffei, P. chrysogenum, P. rubens,andA. transcription factor in addition to DEP3. In these non-DEP campestris), one Lecanoromycete species (Xanthoria parie- clusters, the transcription factor gene occurred most fre- tina), and one other non-Fusarium Sordariomycete species quently across taxa. (C. graminicola; fig. 1 and see supplementary fig. S3, Supplementary Material online). Physical linkage of DEP ho- Loss of DEP Genes and Clusters mologs with each other varied from gene to gene (table 4). Although DEP genes were detected in 120 genomes repre- The PKS gene (DEP5) exhibited the highest frequency of link- senting 105 fungal species and 53 genera, most genomes age to at least one other functional DEP gene (97.7%), whereas lacked the full-length, six-gene DEP cluster (figs. 1 and 2,see the MFS transporter gene (DEP3) exhibited the lowest fre- supplementary tables S1 and S2, Supplementary Material on- quency (74.7%). The two monooxygenase genes (DEP2 and line). In addition to A. brassicicola, DEP genes were detected in DEP4) differed in their frequency of linkage to other DEP 21 Dothideomycete species, but only two of these, genes (93.0 and 89.6%, respectively). Setosphaeria turcica and Curcurbitaria berberidis,hadthe full-length cluster. Likewise, DEP genes were detected in 21 Patterns of DEP Cluster Degeneration in Fusarium Eurotiomycete species, but only one, Arthroderma gypseum, In the 39 Fusarium genomes with DEP genes, 11 had a full- had the full-length cluster. Two Eurotiomycete species, length cluster consisting of apparently functional homologs Penicillium rubens and P. chrysogenum,hadhomologsofall of DEP1–DEP6, 22 had a cluster consisting of a functional six DEP genes, but the genes were dispersed between two loci: DEP3 and DEP6 homolog and one or more pseudogenized DEP1, DEP2, DEP4,andDEP5 were at one locus, and DEP3 and DEP genes, two had a cluster with a functional DEP6 homolog DEP6 were at the other. Two other Eurotiomycetes species, and three or four pseudogenized DEP genes and four had a Talaromyces stipitatus and T. marneffei, also had homologs of cluster consisting solely of functional DEP3 and DEP6 homo- all six DEP genes, but the genes were dispersed between three logs. The DEP3 and DEP6 homologs in Fusarium babinda and four loci respectively. Among the Sordariomycetes, DEP (KY582470) were located on different contigs and both genes were detected in 58 species, but a full-length DEP clus- were adjacent to pseudogenized DEP genes. DEP cluster de- ter was present in only 19: Sordariomycetes species 1, generation via complete deletion or pseudogenization of one Sordariomycetes species 2, Sordariomycetes species 6, or more genes is hypothesizedtohaveoccurredindepen- Clonostachys rosea, C. fioriniae, C. higginsianum, Stachybotrys dently on multiple occasions and likely most recently once chartarum, S. chlorohalonata,and11Fusarium species. In S. in Fusarium verticillioides and once in F. oxysporum or recent chlorohalonata, DEP3 was found on a different contig than ancestors of these species. We hypothesize that this loss has the other DEP genes, but the DEP genes were found at the led to an inability to synthesize depudecin. In F. oxysporum, ends of the contigs, and the separation may be an artifact of theninepartialDEP clusters represent seven distinct patterns thegenomesequenceassembly.Colletotrichum graminicola of degeneration (fig. 3). Although the genetic events leading had DEP genes at two loci: One locus with an apparently to pseudogenization in some taxa are clearly related (e.g., functional DEP3 flanked by pseudogenized homologs of DEP4 and DEP5 indels in 26406 and Fol4287 are very similar), DEP2, DEP4, DEP5,andDEP6; and a second locus with an some new indels have accumulated after these strains di- apparently functional DEP6 but no other DEP genes. verged from their common ancestor (e.g., regions within The widespread distribution of partial DEP clusters and DEP2 are present in Fol4287 but absent in 26406). single DEP genes among Pezizomycotina indicates that com- Interestingly, and as noted above, each partial cluster in F. plete DEP gene loss has occurred frequently in this group of oxysporum contains intact, and presumably functional, ho- fungi. Some taxa that are likely recipients in HGT events have mologs of DEP3 and DEP6. Striking examples of this apparent DEP3 but no other DEP genes(HGTlabeled6and8ainfig. 1). retention of functional homologs of DEP3 and DEP6 during The presence of only DEP3 in these taxa could have resulted cluster degeneration are in isolates FoSC3a and HDV247, in from: 1) HGT of a partial or complete DEP cluster followed by which all DEP genes except DEP3 and DEP6 have been either

2007 Reynolds et al. . doi:10.1093/molbev/msx145 MBE

Table 3. Frequent KOGs &/or Functions Associated with DEP3 Not Clustered with Other DEP Genes. Only Genes Found in at Least 5 Taxa Are Displayed. Taxa Are Ordered According to the DEP3 Phylogeny.

Taxon Class KOG 0157 KOG1208 KOG1262 KOG1611 MCLgrp6 MCLgrp9 MCLgrp10 Cloro1 Sordariomycetes þ –––––– S. species 4 Sordariomycetes þ –––þ –– Triat2 Sordariomycetes – – – – þ –– Melsp1 Sordariomycetes – – – – þ –– Thite2 Sordariomycetes – þ ––þþ – ConPMI546 Sordariomycetes – – – – þ –– Conli1 Sordariomycetes – – – – þ –– Conlig1 Sordariomycetes – – – – þ –– Myche1 Sordariomycetes – – – – þ –– Eder Eurotiomycetes – – – – þ ––

Cadsp1 Leotiomycetes – – – – þ ––Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 Chagl1 Sordariomycetes – – – – þ –– D. species 7 Dothideomycetes þþþþ––þ D. species 11 Dothideomycetes þþþþ– þþ D. species 12 Dothideomycetes þþþþ– þþ DalEC121 Sordariomycetes þþþþ– þþ Acral2 Sordariomycetes þ – þþ– þþ Sodal1 Sordariomycetes þþþþ– þþ Didex1 Dothideomycetes þ –––––– Stano2 Dothideomycetes – – þþ––þ D. species 8 Dothideomycetes – – þþ––þ

Functions: KOG0157 cytochrome P450; KOG1208 dehydrogenases; KOG1262 FAD-binding protein; KOG1611 dehydrogenase; MCL group 6 transcription factor; MCL group 9 cupin-binding domain protein; MCL group 10 transcription factor.

Table 4. Frequency of DEP Homologs Clustering with at Least One cluster consisting of solely pseudogenized genes. To test how Other Functional DEP Homolog. gene decay rates varied among DEP genes, we analyzed tran- DEP Homolog Proposed Function Clustered Solo % Clustered sition rates among the functional, pseudogenized, and absent states under three models of BayesTraits: H —all transitions 1 Unknown 40 1 97.6 0 2 Monooxygenase 40 3 93.0 allowed; H1—all loss types allowed, regain of function only 3 MFS transporter 71 24 74.7 possible from the absent state; and H2—only losses through 4 Monooxygenase 43 5 89.6 pseudogenization and deletion allowed. This analysis con- 5 PKS 42 1 97.7 firmed that DEP3 and DEP6 had lower rates of gene decay 6 C6 transcription factor 60 6 90.9 (deletion and pseudogenization) than other genes in the DEP cluster, with DEP6 showing the lowest decay rates overall (fig. pseudogenized or deleted. DEP cluster degeneration was also 2B, see supplementary tables S4 and S5, Supplementary apparent in other genera. The pattern of full-length clusters Material online). DEP3 had the lowest rates of gene deletion versus degenerated clusters suggests that similar degenera- in the DEP cluster, but high rates of pseudogene loss. Under tion has occurred in Colletotrichum (Sordariomycetes) after the model permitting all transition types (H0), with regain of divergence of C. graminicola and C. fioriniae,inArthroderma function or pseudogenes, each of the six DEP genes showed a (Eurotiomycetes) after divergence of A. otae and A. gypseum, different transition rate pattern. The H2 model, which per- and in Alternaria (Dothideomycetes) after divergence of A. mitted only decay transitions, predicted the lowest transition brassicicola and A. alternata (fig. 3). rates of the three models. Under this model, decay rates from Bayesian and maximum likelihood trees inferred from con- a functional state to either an absent or a pseudogenized state catenated sequences of 20 housekeeping genes from 45 were similar (fig. 3B), but for DEP1, DEP2, DEP3,andDEP4,loss Fusarium strains (see supplementary table S3, of pseudogenes was the most common transition, whereas Supplementary Material online), including 32 species, showed for DEP5, pseudogenization of functional genes was most concordant topologies (fig. 2A). Comparisons of the relation- common (fig. 2B and C, see supplementary tables S4 and ships of species inferred in these trees to different states (func- S5, Supplementary Material online). tional, pseudogenized and absent) of DEP genes indicated a highly complex pattern of DEP retention and decay (fig. 2A), Confirmation of Functionality of DEP Cluster in with fully functional (all DEP genes present) and fully absent Fusarium (all DEP genes absent) clusters being polyphyletic. Fully absent To determine the functionality of a Fusarium DEP cluster, we clusters were found only in the species Fusarium euwallacea, examined expression of DEP genes and depudecin production F. avenaceum, F. culmorum, F. graminearum, F. poae and F. in Fusarium langsethiae. The deduced amino acid sequences pseudograminearum.AllotherstrainsofFusarium contained of the F. langsethiae and A. brassicicola DEP genes shared at least a functional DEP3 and DEP6, except Fusarium circina- 64.1–79.7% identity (table 5). We observed moderate to tum and F. dlaminii, and no Fusarium genomes had a DEP high levels of expression of DEP1–DEP5 on oat grain medium

2008 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE

Sordariomycetes Strain Fusarium 25433, PHW808, oxysporum and Fo5176 1 243 5 6 II5

Fol4287

26405

CL57

Fo47 and MN25 Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021

FoSC3a

HDV247 and PHW815

Colletotrichum ﬁoriniae

Colletotrichum graminicola

Dothideomycetes

Alternaria brassicicola Alternaria alternata Euroomycetes

Arthroderma gypseum

Arthroderma canis

FIG.3.Comparison of the patterns of DEP cluster degeneration in multiple fungal genera. Three strains of Fusarium oxysporum have DEP1–DEP6 intact, while the DEP gene clusters in nine strains represent seven distinct patterns of degeneration. Cluster degeneration is apparent in a second Sordariomycete, Colletotrichum, the Dothideomycete Alternaria, and the Eurotiomycete Arthroderma, where in each case at least one species has DEP1–DEP6, intact, and a second species has a DEP gene cluster that has degenerated. The direction of transcription for each gene is indicated by an arrow, and the length of each arrow corresponds to the length of each gene, respectively. Arrow colors indicate gene identity: red for DEP1, orange for DEP2, yellow for DEP3, green for DEP4, blue for DEP5, violet for DEP6. Grey arrows represent genes that do not have a homolog in the A. brassicicola DEP cluster. Unﬁlled arrows represent pseudogenes.

(OG) and minimal levels of expression on oat flower medium verticillioides and their wild-type progenitor strain for growth (OF; fig. 4A). Failure to detect expression of the transcription in the presence and absence of exogenous depudecin in liquid factor gene, DEP6, was not unexpected because DEP6 was and on solid growth media. Conidial germination, mycelial significantly less expressed than the other DEP genes in A. growth, and sporulation were not significantly affected by the brassicicola (Wight et al. 2009). Low levels of expression of the presence of the exogenous depudecin in any of the three monooxygenase genes (DEP2 and DEP4) but no other DEP strains examined. genes were observed on the complete medium (CM), which lacked oat material (fig. 4A). In HPLC analysis, solvent extracts Discussion of OG cultures of F. langsethiae yielded a peak with the same Prior to the era of genome sequencing, the distribution of retention time and UV spectrum as a depudecin standard most fungal SM biosynthetic gene clusters was unknown or (fig. 4B). This peak was not detected in extracts of F. langse- thought to be narrow based on reports of production of the thiae cultures grown in OF or CM. corresponding SMs in a limited number of species. Analysis of genome sequences from diverse fungi has altered this percep- DEP3 and F. verticillioides Resistance to Depudecin tion. For example, early genomic studies revealed a discon- In order to test the role of the transporter gene DEP3 in tinuous distribution of orthologous PKS genes among five resistance to depudecin, we examined two independent fungal genera from three classes (Kroken et al. 2003). The DEP3 deletion strains (DDEP3.2 and DDEP3.9) of F. discovery that widely divergent fungi synthesize the PKS-

2009 Reynolds et al. . doi:10.1093/molbev/msx145 MBE

Table 5. Percent Identities for the Predicted Dep Proteins of Alternaria brassicicola and Fusarium langsethiae.

Gene A. brassicicola protein Accession F. langsethiae protein Accession % ID Proposed Function DEP1 369 ACZ57550 365 KPA37945 69.6 Unknown DEP2 528 ACZ57545 538 KPA37944 72.9 Monooxygenase DEP3 564 ACZ57546 533 KPA37943 79.7 Transporter (MFS) DEP4 581 ACZ57547 580 KPA37942 77.1 Monooxygenase DEP5 2,376 ACZ57548 2,371 KPA37941 70.5 PKS DEP6 646 ACZ57549 608 KPA37940 64.1 Transcriptional regulator (C6) Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021

FIG.4.DEP gene expression and depudecin production in Fusarium langsethiae.(A) RNA-Seq analysis of Fusarium langsethiae after growth on complete (CM), oat ﬂower (OF) and oat grain (OG) media. Levels of gene expression are presented as RPKM (Reads Per Kilobase of exon model per Million mapped reads). (B) HPLC-detection of depudecin in extracts of a Fusarium langsethiae oat grain culture. derived SM depudecin (Matsumoto et al. 1992; Tanaka et al. recipients, two to Eurotiomycete recipients, and two to 2000; Amnuaykanjanasin et al. 2005) and the characterization Leotiomycete recipients. We also found evidence that multi- of the depudecin gene cluster in A. brassicicola (Wightetal. ple, independent gene loss events contributed significantly to 2009) led to our identification of closely related cluster in the the presence of partial DEP clusters. And finally, we found distantly related fungus F. langsethiae and evidence that its evidence that DEP gene loss was biased; that is, when gene function in depudecin production was conserved. loss has occurred, the transporter (DEP3) and transcription These findings indicated a potential for widespread pro- factor (DEP6) genes have been retained more frequently than duction of depudecin among Pezizomycotina. We surveyed other DEP genes. 585 fungal genomes for the DEP cluster, and found that the Loss of DEP genes was evident from the occurrence of cluster has a wide but discontinuous distribution. That is, DEP partial DEP clusters, including partial clusters with pseudogenes occur in distantly related genomes, but their occur- genized DEP genes. In genera other than Fusarium,anintact, rence varies among more closely related genomes. We de- six-gene DEP cluster at a single locus is rare, occurring in only tected homologs of the full-length, six-gene DEP cluster or 12 of 540 non-Fusarium genomes surveyed. The discovery of partial DEP cluster in multiple species of Dothideomycetes, more intact DEP clusters may require sequencing additional Sordariomycetes and Eurotiomycetes, as well as two species species within a genus or additional genera. In Fusarium,in- of Lecanoromycetes and one species of Leotiomycetes. HGT traspecies variation in DEP cluster content was observed in F. of DEP cluster homologs was evident from the close phylo- oxysporum (fig. 3)andF. verticillioides. How such variation has genetic relationships of DEP homologs in distantly related arisen and become distributed across F. oxysporum is not fungi, that is, fungi from different classes. For example, in trees clear. However, because the DEP cluster is located within inferred from each of the six DEP genes, a clade that includes thecoregenomeofF. oxysporum,itisunlikelythatthevar- closely related DEP homologs from Sordariomycete and iation has become distributed from the system of HGT of Dothideomycete species provides evidence for HGT of a regions of the accessory genome that contributes to genetic full-length homolog of the cluster from Sordariomycetes to variationinthefungus(Ma et al. 2010). We found a single Dothideomycetes (HGT event 1 in fig. 1). This event likely pseudogene in the sparsely sampled Fusarium clades, but resulted in the presence of the DEP cluster in A. brassicicola numerous pseudogenes with a complex history in the well- (Wight et al. 2009). Elsewhere in the DEP3 gene tree (fig. 1), we sampled oxysporum/verticillioides/babinda clade. This sug- identified seven additional possible HGT events from a gests that fine-scale sampling of fungal genomes can reveal Sordariomycete donor: Three more to Dothideomycete broader evolutionary patterns. Sampling of additional

2010 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE

Fusarium species may reveal similar pseudogenization pat- thereisnoevidenceforthepresenceofthisgenepairin terns in other species complexes, or may provide evidence biosynthetic clusters other than the DEP cluster. Rather, there that the relatively high rate of pseudogenization is limited to is evidence for the presence of DEP3 homologs in other clus- the oxysporum/verticillioides/babinda clade. We also found ters that include a different transcription factor. Overall, DEP3 intragenus variation in DEP gene content in Colletotrichum, exhibited low fidelity to the DEP cluster, and among the ge- Alternaria,andArthroderma (figs. 1, 3). However, it is possible nomes examined there are at least two other transcription that an intact DEP cluster is not essential for depudecin pro- factor genes physically linked to it (tables 3 and 4), indicating duction. The presence of functional monooxygenase (DEP2 that the DEP3-encoded transporter has an especially non- and DEP4)andPKS(DEP5) genes could be sufficient for dep- specific or modular function. The presence of DEP3 adjacent udecin biosynthesis if other transporter and transcription fac- to putative SM biosynthetic genes in genomes that do not tor genes are adapted to compensate for the absence of DEP3 include other DEP genes suggests that in some fungi DEP3 has and DEP6. If this were the case, depudecin production could relocated to other gene clusters, which may have allowed it to be slightly more frequent than thepresenceofanintactsix- persist in organisms that do not produce depudecin. Fourth, Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 gene DEP cluster given that apparently functional homologs selective gene retention may have resulted from physiological of DEP2, DEP4 and DEP5 are present in three genomes that constraints imposed on the metabolic pathway. In this sce- lack DEP3 and DEP6. Furthermore, depudecin-like com- nario, toxic biosynthetic intermediates accumulate due to pounds could be synthesized by the 22 clusters containing deletion or mutation of an enzyme-encoding gene, regulator, afunctionalDEP5 along with either a functional DEP2 (1 or transporter, favoring the repeated loss of enzymes that cluster) or DEP4 (21 clusters). Fungi that contain non-DEP catalyze formation of the intermediates (Slot and Rokas enzyme-coding genes between or adjacent to DEP genes, as is 2011; Campbell et al. 2013; McGary et al. 2013; Greene et al. the case in Stachybotrys, Sordariomycetes species 1, and P. 2014; Wisecaver et al. 2014; Gluck-Thaler and Slot 2015). chrysogenum, for example, may also produce metabolites that However, deletion of DEP2 or DEP4 in A. brassicicola led to are structurally similar to depudecin. the accumulation of two different unidentified epoxide- The relatively high frequency of DEP pseudogenes and containing compounds but caused no other reported phe- partial DEP clusters among the fungal genomes examined notypic change (Wightetal.2009). Additionally, the deletion provides evidence that frequent degeneration of the cluster analysis performed here revealed that F. verticillioides DEP3 has occurred, which in turn suggests sporadic selection for deletion mutants were not sensitive to exogenous depudecin, depudecin production. The clustering of genes encoding non- indicating that F. verticillioides does not rely on DEP3 for re- essential functions under weak selection is explained by the sistance to depudecin, or possibly that resistance to exoge- “selfish clustering” model: That is, the full complement of nous depudecin is not required. This result is similar to the genes required for a weakly selected metabolic pathway are finding that the ABC transporter gene FUM19 in the gene more likely to be transferred via HGT, and therefore propa- cluster responsible for synthesis of fumonisin mycotoxins is gated, if the genes are clustered (Lawrence and Roth 1996; not required for resistance to fumonisins (Proctor et al. 2003). Lawrence 1999; Gluck-Thaler and Slot 2015). In recent years, However, these tests were likely performed on haploid strains, there have been a number of reports of HGT of SM gene and it is possible that retention of DEP3 and DEP6 is driven by clusters between distantly related fungi (Khaldi et al. selection imposed on a heterokaryon, or that DEP3 aids in 2008; Slot and Rokas 2011; Campbell et al. 2012; Proctor intracellular depudecin resistance, but that there is redun- et al. 2013;Wisecaver et al. 2014). By utilizing evidence dancy in depudecin (i.e., HDAC) resistance mechanism from 585 fungal genomes, the current study provides a (Baidyaroy et al. 2002). Thus, there is no evidence that gene more detailed understanding of the tempo or frequency loss in the DEP cluster leads to a decrease in fitness resulting of this birth and death process of a gene cluster that is from accumulation of toxic metabolites. Further experimen- weakly selected. tation on phenotypic changes induced by depudecin in var- Here, we observed the selective retention of DEP3 and ious genetic backgrounds is needed to better understand how DEP6 genes across the diversity of fungi examined is not im- this SM affects fungal fitness. mediately explained by “selfish clustering.” We propose four Depudecin was originally isolated over 25 years ago as a hypothetical mechanisms to explain this biased pattern of potential anticancer agent produced by A. brassicicola DEP gene retention and loss. First, DEP3 and DEP6 might (Matsumoto et al. 1992). Here we show that the gene cluster confer resistance to depudecin or depudecin-like metabolites responsible for depudecin synthesis occurs broadly but dis- produced by other fungi that occupy the same habitat. continuously among species of Pezizomycotina, and that Second, DEP3 and DEP6 may interact with functional net- HGT and gene loss have contributed to this pattern of dis- works outside the DEP cluster. For example, the DEP3- tribution, as predicted by the selfish cluster model. We also encoded transporter may export additional metabolic prod- provide evidence that the transporter and regulatory genes in ucts, while the DEP6-encoded transcription factor may inter- the DEP cluster are differentially retained, possibly as a result act with other regulatory proteins thereby exposing of their integration into other metabolic networks and/or deleterious mutations in these regulators for remediation, their contributions to fitness beyond that provided by depu- thus impacting broader regulatory networks. Third, DEP3 decin production. These observations provide broader insight and DEP6 homologs may function in additional biosynthetic into evolution of SM gene clusters. That is, the observations pathways.However,whiletheDEP3/DEP6 pairing is frequent, suggest that differential retention of transporters encoded by

2011 Reynolds et al. . doi:10.1093/molbev/msx145 MBE one SM gene cluster may benefit fungi exposed to structurally and maximum likelihood phylogenetic analysis was per- similar SMs produced by competing fungi and/or facilitate formed using RAxML v.8.1.20 (Stamatakis 2006). export of other SMs or xenobiotics. Both of these benefits A BayesTraits v.2.0 (Pagel et al. 2004; available from http:// could increase the likelihood that cluster-encoded transpor- www.evolution.rdg.ac.uk/; last accessed October 21, 2016) ters and proteins that regulate transporter expression are analysis of gene transition rates across the Fusarium species preferentially repurposed for other functions. Such repurpos- examined was conducted to model the rates of gene gain, ing could lead to incorporation of the corresponding genes degradation, and loss in a taxonomically well-sampled data- into other gene clusters and subsequent HGT of the genes as set. A species phylogeny for the 46 Fusarium genomes in- part of the remodeled clusters. cluded in this analysis was constructed from 20 housekeeping genes (see supplementary table S3, Materials and Methods Supplementary Material online) with Fusarium solani f. sp. pisi as an outgroup. First, we created a concatenated align-

DEP Gene Cluster Identification ment of the 20 housekeeping genes for a RAxML analysis Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 Putative DEP genes and pseudogenes were identified by using the GTRGAMMA model and 100 bootstrap replicates BLAST searches of multiple databases with the A. brassicicola (Stamatakis 2006).Second,foreachgene,agenetreewas DEP genes (Accessions ACZ57544–ACZ57549) as query se- constructed using MrBayes v.3.2.5 (Huelsenbeck and quences. A local database of 556 fungal proteomes (see sup- Ronquist 2001; Ronquist and Huelsenbeck 2003)usingthe plementary table S1, Supplementary Material online) was mixed amino acid prior, the autoclose setting, 1,000,000 gen- searched using ublast (Edgar 2010)withane-value of 5 erations and two chains, thus producing 40 Bayesian *.tre files. 1 10 and an “accel” of 0.8. An extended set of 45 The individual gene trees were used to construct a supertree Fusarium proteomes, which includes 16 from the set of 556, using Bucky v.1.4.3 (alpha ¼ 1, burnin ¼ 450, 2 chains, 100,000 was searched using BLAST in CLC Genomics Workbench generations; Ane´ et al. 2007; Larget, et al. 2010). The final 500 (version 8.0; see supplementary table S2, Supplementary trees from each file were also used as the basis of a BayesTraits Material online). About 53% of the genomes used in these analysis of transition rates for functional, pseudogenized, and analyses (308 at the time of manuscript preparation) are absent genes using the ML setting, which was performed currently unpublished and not freely available for publication separately for each of the six DEP genes. Three models of here. When observations were noted for these genomes, we rate transitions were analyzed: H ) all transitions among func- refer to them only by taxonomic class. Nucleotide sequence 0 100 tional, pseudogenized, and absent genes are possible; H1)all of genes with e-values < 1 10 to a query sequence, and loss transition types are possible, but the only decay reversal 20 kb flanking sequences, were retrieved, assembled in permitted is from the absent to functional state, representing Sequencher (version 5.2.4; Gene Codes Corp.) and subjected possible HGT; and H2) only decay is permitted, with no regain to BLASTX analysis against the NCBI-NR databases to identify of function. putative functional DEP genes and DEP pseudogenes as compared with A. brassicicola DEP genes. Homologs (predicted DEP Gene Expression Analysis functional or pseudogene) to the query genes in the local In vitro and transcriptomic assays in this study used F. lang- proteome database or assemblies were considered clustered if sethiae wild-type strain Fl201059 (IBT9951 or NRRL 54940; separated by no >6 intervening gene models (Wisecaver et al. Thrane et al. 2004). The F. langsethiae genome was sequenced 2014). Non-DEP clusters containing DEP3 homologs were and assembled, using a combination of 454 pyrosequencing identified by searching for shared synteny of predicted genes and Illumina HiSeq (Lysøe et al. 2016). RNA-Seq analysis of F. within five positions of DEP3 sequencesacrossthefungal langsethiae wasconductedaftergrowthonOGagarmedium, phylogeny (Slot and Rokas 2011). In the C. higginsianum as- OF agar medium, and complete liquid medium (CM; Harris sembly, DEP1 was found on a different scaffold than DEP2– et al. 1994).OGwasmadefromlatemilkstagewholeOG DEP6; however, further analysis indicated these two scaffolds from cultivar Belinda grown in a farmer’s field located near overlap by 332 bp, and therefore are marked in figure 1 as a the Norwegian University of Life Sciences, Norway. OF was single, complete cluster. made of a mix of Norwegian cultivars Belinda, Aragon, Ringsaker, and 595-7, grown in a greenhouse and kindly pro- Phylogenetic Analyses and Estimation of DEP Gene vided by Helge Skinnes, Norwegian University of Life Sciences, Decay Rates Norway. Floral organs were separated from the rachis, rinsed Protein translations for all complete homologs of each DEP with sterile distilled water (SDW), suspended in 0.5% NaOCl gene that lacked in-frame stop codons were combined into a and agitated for 90 s, suspended in 70% ethanol and agitated single file, aligned using mafft v.7.2.2.1 (Katoh et al. 2002)with for 60 s, and rinsed twice with SDW. The surface sterilized default parameters, and trimmed using trimAl v.1.4 with the OGs and OFs were then embedded in 1.5% water agar con- “automated1” algorithm (Capella-Gutierrez et al. 2009). An taining 0.05 mg/ml chloramphenicol and covered with sterile initial estimation of gene evolution was made using FastTree cellophane. (Price et al. 2009), and the dataset was pruned to the earliest The inoculum for the media were obtained from five day node that included all clusters of at least three DEP genes. The old potato dextrose broth (PDB) cultures grown at 24 Cand best model of protein evolution (JTT, WAG, LG) according to from a two week old Spezieller N€ahrstoffarmerAgar(SNA; AICc was inferred using ProtTest v.3.4 (Abascal et al. 2005) Leslie and Summerell 2006). The OF and OG plates were

2012 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE inoculated with 100 ml spore suspension at 3.3 106 spores/ on a Waters XBridge C18-column (100 4.6 mm, 3.5 mmpar- ml and incubated for 84 h in sealed, transparent plastic bags ticle diameter) using a 1 ml/min gradient flow of acetonitrile/ at 22–24 C. Mycelia were harvested by scraping the cello- water (10–23% acetonitrile for 5 min, followed by 90% aceto- phane with a scalpel, flash frozen in liquid nitrogen, and nitrile for 2 min). Depudecin was detected at 4.4 min, using stored at -80 C prior to RNA extraction. The CM culture UV 210 nm and a bandwidth of 4 nm. Quantification was was prepared as described by Guldener et al. (2006):100ml done by external standard calibration with depudecin stan- of CM in a 250 ml flask was inoculated with 1 108 spores, dard dissolved in 12% acetonitrile to concentrations of 0.1– cultivated at 25 C, and shaken at 150 rpm for 28 h. The my- 100 mg/ml. celia were harvested, washed with SDW, and re-inoculated into 100 ml CM. After 16 h growth at 25 C and 150 rpm, the DEP3 Deletion Experiments in F. verticillioides mycelia were harvested by vacuum filtration through The wild-type strain of F. verticillioides used in this study was Whatman paper, flash frozen in liquid nitrogen, and stored FGSC 7600 (FRC M-3125 or NRRL 13447; Leslie et al. 1992). The

Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 at -80 C prior to RNA extraction. coding region of the DEP3 (FVEG_01738) homolog in F. verti- Total RNA was isolated from 100 mg of mycelia using the cillioides was deleted via split-marker and protoplast transfor- RNeasy midi kit (Qiagen) according to the manufacturer’s mation methods with HygB as the selectable marker as protocol. Poly Aþ mRNA was isolated with an Oligotex previously described (Brown et al. 2015; Catlett et al. 2002; Direct mRNA Mini Kit (Qiagen) from 20 mgtotalRNAwith Proctor et al. 1999). The wild-type progenitor strain (FGSC RIN values >8. cDNA was synthesized using a RevertAid H 7600) and two dep3 deletion mutants (DDEP3.2, and Minus First Strand cDNA Synthesis Kit (Fermentas) as di- DDEP3.9) were tested for sensitivity to depudecin as follows. rected by the manufacturer. cDNA was further purified using Conidia from nine-day-old cultures (28 C) on V-8 juice agar theMinElutePCRPurificationKit(Qiagen),andcDNAcon- (Tuite 1969) were harvested and suspended in sterile water at centration and fragment size range were measured using an 1 105 conidia per ml. To assess spore germination and growth Agilent Bioanalyzer with an Agilent DNA 7500 chip. in liquid media, 10,000 spores were inoculated into 1 ml of GYP Preparation of the three libraries for sequencing was car- (2% glucose, 1% peptone, 0.3% yeast extract) or 10% GYP con- ried out by the Norwegian High-Throughput Sequencing taining either 0, 1, 10 or 150 ng of depudecin per well in a 24-well Centre at the University of Oslo, Norway. Briefly, each library plate. The plate was shaken at 150 rpm at 25 C. To assess spore was prepared from 2 mg cDNA using the 454 DNA Library germination and growth on solid media, 10,000 spores were Preparation Kit (Roche, Life Sciences) and standard methods spread across the surface of a V-8 juice agar medium and a to generate single-stranded DNA for emulsion PCR (emPCR). 1% water agar. Whatman assay discs (5 mm), containing 0, 10 or The cDNA was size-fractionated with the cut-off of 250–300 155 ng depudecin, were placed, evenly distributed, on the sur- base pairs (bp). The libraries were nebulized, separately MID- face of the agar plates. For both liquid and solid media, spore tagged, and sequenced, according to standard protocols for germination and growth were monitored by light microscopy the GS-FLX technology using the 454 Life Sciences/Roche and by eye every 2 h starting 10 h after inoculation. sequencing platform. A total of 1,335 million reads, averaging 278 bp, were de novo assembled using default parameters in Supplementary Material CLC Genomics Workbench (version 4.03), resulting in 18,742 contigs and 15,264 singletons for a total of 34,006 expressed Supplementary data are available at Molecular Biology and sequence tags (ESTs). RNA-seq mapping to de novo contigs Evolution online. was performed to determine the relative gene expression (RPKM; Reads Per Kilobase of exon model per Million Acknowledgments mapped reads; Mortazavi et al. 2008) in the three libraries. Analyses in this study were performed on the Ohio Supercomputer (OSC) and the Vanderbilt University Analysis of Depudecin Production in F. langsethiae Advanced Computing Center for Research & Education Depudecin (4,5:8,9-dianhydro-1,2,6,7,11-pentadeoxy-D-threo- (ACCRE), c/o A. Rokas. This research was supported by the D-ido-undeca-1,6-dienitol) standard was purchased from Ohio Agricultural Research and Development Center (JCS) Sigma–Aldrich Co. For chemical detection of depudecin, and The Norwegian Research Council and Foundation for 100–150 mg mycelia from F. langsethiae OG agar cultures Research Levy on Agricultural Products (H.H.D. and E.L.). were ground in liquid nitrogen and extracted with 500 ml The GenBank accession numbers for the 19 Fusarium DEP dichloromethane. The resulting mixture was vortexed lightly clusters described are KY582465–KY582483. Thanks to Marit and then sonicated for 30 min. Cell debris was removed by Almvik at NIBIO for chemical analysis of depudecin in F. centrifugation at 14,000 rpm for 5 min and the solvent was langsethiae. Chris McGovern, Amy McGovern, Crystal transferred to a glass tube and evaporated to dryness under a Probyn, and Nathane Orwig of NCAUR provided much ap- stream of nitrogen gas. The extract was dissolved in 500 mlof preciated technical assistance. We thank Scott Baker, Petr 12% acetonitrile (v/v) and insoluble material was pelleted by Baldrian, Gregory Bonito, Kathryn Bushley, Irina Druzhinina, centrifugation at 14,000 rpm for 15 min. The cleared extracts Paul Dyer, David Ezra, Dave Greenshields, Alexey Grum- were analyzed by high performance liquid chromatography Grzhimaylo, Patrik Inderbitzin, Nada Krasevec, Christian with photodiode-array detection (Agilent 1100 HPLC-DAD). Kubicek, Jon Karl Magnuson, Francis Martin, Byoungnam 20 ml extract was injected and extract components separated Min, Donald Natvig, Kerry O’Donnell, Amy Jo Powell,

2013 Reynolds et al. . doi:10.1093/molbev/msx145 MBE

Steven Singer, Joseph Spatafora, Adrian Tsang, and B. Gillian islands in the pathogenic filamentous fungus Aspergillus fumigatus. Turgeon for providing access to unpublished genome data PLoS Genet. 4:e1000046. produced by the U.S. Department of Energy Joint Genome FischbachMA,WalshCT,ClardyJ.2008. The evolution of gene collec- tives: how natural selection drives chemical innovation. Proc Nat Institute, a DOE Office of Science User Facility, supported by Acad Sci U S A. 105:4601–4608. the Office of Science of the U.S. Department of Energy under Gluck-Thaler E, Slot JC. 2015. Dimensions of horizontal gene trans- Contract No. DE-AC02-05CH11231. Genomes of Corollospora fer in eukaryotic microbial pathogens. PLoS Pathogens. maritima, Didymella exigua, Clonostachys rosea,and 11:e1005156. have been sequenced under the Greene GH, McGary KL, Rokas A, Slot JC. 2014. Ecology drives the dis- Botryosphaeria dothidea tribution of specialized tyrosine metabolism modules in fungi. JGI 1000 Fungal Genomes project, and Coniochaeta ligniaria Genome Biol Evol. 6:121–132. CBS111746 was sequenced as part of the Assembling the Guldener U, Seong KY, Boddu J, Cho S, Trail F, Xu JR, Adam G, Mewes Fungal Tree of Life project (AFTOL). HW, Muehlbauer GJ, Kistler HC. 2006. Development of a Fusarium graminearum Affymetrix GeneChip for profiling fungal gene expres-

sion in vitro and in planta. Fungal Genet Biol. 43:316–325. Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 Harris SD, Morrell JL, Hamer JE. 1994. Identification and characterization References of Aspergillus nidulans mutants defective in cytokinesis. Genetics Abascal F, Zardoya R, Posada D. 2005. ProtTest: selection of best-fit 136:517–532. models of protein evolution. Bioinformatics 21:2104–2105. Hollender C, Liu Z. 2008. Histone deacetylase genes in Arabidopsis de- Amnuaykanjanasin A, Punya J, Paungmoung P, Rungrod A, velopment. J Integr Plant Biol. 50:875–885. Tachaleat A, Pongpattanakitshote S, Cheevadhanarak S, Huelsenbeck JP, Ronquist F. 2001. MRBAYES: Bayesian inference of phy- Tanticharoen M. 2005. Diversity of type I polyketide synthase logenetic trees. Bioinformatics 17: genes in the wood-decay fungus Xylaria sp. BCC 1067. FEMS Isaka M, Jaturapat A, Kladwang W, Punya J, Lertwerawat Y, Tanticharoen Microbiol Lett. 251:125–136. M, Thebtaranonth Y. 2000. Antiplasmodial compounds from the Ane´ C, Larget B, Baum DA, Smith SD, Rokas A. 2007. Bayesian estimation wood-decayed fungus Xylaria sp. BCC 1067. Planta Med. 66:473–475. of concordance among gene trees. MolBiolEvol. 24:412–426. Jargeat P, Rekangalt D, Verner M-C, Gay G, Debaud J-C, Marmeisse R, Baidyaroy D, Brosch G, Graessle S, Trojer P, Walton JD. 2002. Fraissinet-Tachet L. 2003. Characterisation and expression analysis of Characterization of inhibitor-resistant histone deacetylase activity a nitrate transporter and nitrite reductase genes, two members of a in plant-pathogenic fungi. Eukaryot Cell. 1:538–547. gene cluster for nitrate assimilation from the symbiotic basidiomy- Bain JM, Tavanti A, Davidson AD, Jacobsen MD, Shaw D, Gow NA, Odds cete Hebeloma cylindrosporum. Curr Genet. 43:199–205. FC. 2007. Multilocus sequence typing of the pathogenic fungus Jimenez DJ, Hector RE, Riley R, Lipzen A, Kuo RC, Amirebrahimi M, Barry Aspergillus fumigatus. J Clin Microbiol. 45:1469–1477. KW, Grigoriev IV, van Elsas JD, Nichols NN. 2017. Draft genome Bangera MG, Thomashow LS. 1999. Identification and characterization sequence of Coniochaeta ligniaria NRRL 30616, a lignocellulolytic of a gene cluster for synthesis of the polyketide antibiotic 2,4-diac- fungus for bioabatement of inhibitors in plant biomass hydrolysates. etylphloroglucinol from Pseudomonas fluorescens Q2-87. JBacteriol. Genome Announc.5:1–2. 181:3155–3163. Katoh K, Misawa K, Kuma K-I, Miyata T. 2002. MAFFT: a novel method Brown DW, Lee SH, Kim LH, Ryu JG, Lee S, Seo Y, Kim YH, Busman M, for rapid multiple sequence alignment based on fast Fourier trans- Yun SH, Proctor RH, et al. 2015. Identification of a 12-gene fusaric form. Nucleic Acids Res. 30:3059–3066. acid biosynthetic gene cluster in Fusarium species through compar- Khaldi N, Collemare J, Lebrun MH, Wolfe KH. 2008. Evidence for hori- ative and functional genomics. Mol Plant-Microbe Interact. zontal transfer of a secondary metabolite gene cluster between fungi. 28:319–332. Genome Biol. 9:R18. Brown DW, Yu J-H, Kelkar HS, Fernandes M, Nesbitt TC, Keller NP, Kroken S, Glass NL, Taylor JW, Yoder OC, Turgeon BG. 2003. Adams TH, Leonard TJ. 1996. Twenty-five co-regulated transcripts Phylogenomic analysis of type I polyketide synthase gene in patho- define a sterigmatocystin gene cluster in Aspergillus nidulans. Proc genic and saprobic ascomycetes. Proc Nat Acad Sci U S A. Nat Acad Sci U S A. 93:1418–1422. 100:15670–15675. Campbell MA, Rokas A, Slot JC. 2012. Horizontal transfer and death of a Kwon HJ, Kim JH, Kim M, Lee JK, Hwang WS, Kim DY. 2003. Anti-par- fungal secondary metabolic gene cluster. Genome Biol Evol. asitic activity of depudecin on Neospora caninum via the inhibition 4:289–293. of histone deacetylase. Vet Parasitol. 112:269–276. Campbell MA, Staats M, van Kan JA, Rokas A, Slot JC. 2013. Repeated loss Larget BR, Kotha SK, Dewey CN, Ane´ C. 2010. BUCKy: gene tree/species of an anciently horizontally transferred gene cluster in Botrytis. tree reconciliation with Bayesian concordance analysis. Mycologia 105:1126–1134. Bioinformatics 26:2910–2911. Capella-Gutierrez S, Silla-Martinez JM, Gabaldon T. 2009. trimAl: a tool Lawrence J. 1999. Selfish operons: the evolutionary impact of gene clus- for automated alignment trimming in large-scale phylogenetic anal- tering in prokaryotes and eukaryotes. Curr Opin Genet Dev. yses. Bioinformatics 25:1972–1973. 9:642–648. Catlett NL, Lee BN, Yoder OC, Turgeon BG. 2002. Split-marker recom- Lawrence JG, Roth JR. 1996. Selfish operons: horizontal transfer may drive bination for efficient targeted deletion of fungal genes. Fun Genet the evolution of gene clusters. Genetics 143:1843–1860. News. 49:9–11. Leslie JF, Plattner RD, Desjardins AE, Klittich CJR. 1992. Fumonisin B1 Challis GL, Hopwood DA. 2003. Synergy and contingency as driving production by strains from different mating populations of forces for the evolution of multiple secondary metabolite produc- Gibberella fujikuroi (Fusarium section Liseola). Phytopathology tion by Streptomyces species. Proc Nat Acad Sci U S A. 100(Suppl 82:341–345. 2):14555–14561. Leslie JF, Summerell BA. 2006. The Fusarium laboratory manual. Ames ChenL,YueQ,ZhangX,XiangM,WangC,LiS,CheY,Ortiz-Lopez FJ, (IA): Blackwell Publishing. Bills GF, Liu X, et al. 2013. Genomics-driven discovery of the pneu- LysøeE,FrandsenRJN,DivonHH,TerziV,Orru L, Lamontanara A, mocandin biosynthetic gene cluster in the fungus Glarea lozoyensis. Kolseth A-K, Nielsen KF, Thrane U. 2016. Draft genome se- BMC Bioinformatics. 14:339. quence and chemical profiling of Fusarium langsethiae,an Edgar RC. 2010. Search and clustering orders of magnitude faster than emerging producer of type A trichothecenes. Int J Food BLAST. Bioinformatics 26:2460–2461. Microbiol. 221:29–36. Fedorova ND, Khaldi N, Joardar VS, Maiti R, Amedeo P, Anderson MJ, Ma LJ, van der Does HC, Borkovich KA, Coleman JJ, Daboussi MJ, Di Crabtree J, Silva JC, Badger JH, Albarraq A, et al. 2008. Genomic Pietro A, Dufresne M, Freitag M, Grabherr M, Henrissat B, et al. 2010.

2014 Differential Retention of Gene Functions . doi:10.1093/molbev/msx145 MBE

Comparative genomics reveals mobile pathogenicity chromosomes rare strains of Botrytis cinerea is positively controlled by in Fusarium. Nature 464:367–373. VELVET. PLoS ONE. 8:e53729. Matsumoto M, Matsutani S, Sugita K, Yoshida H, Hayashi F, Terui Y, Slot JC, Rokas A. 2011. Horizontal transfer of a large and highly toxic Nakai H, Uotani N, Kawamura Y, Matsumoto K, et al. 1992. secondary metabolic gene cluster between fungi. Curr Biol. Depudecin: a novel compound inducing the flat phenotype of 21:134–139. NIH3T3 cells doubly transformed by Ras- and Src-oncogene, pro- Staehlin BM, Gibbons JG, Rokas A, O’Halloran TV, Slot JC. 2016. duced by Alternaria brassicicola. JAntibiot. 45:879–885. Evolution of a heavy metal homeostasis/resistance island reflects McGary KL, Slot JC, Rokas A. 2013. Physical linkage of metabolic increasing copper stress in enterobacteria. Genome Biol Evol. genes in fungi is an adaptation against the accumulation of 8:811–826. toxic intermediate compounds. Proc Nat Acad Sci U S A. Stamatakis A. 2006. RAxML-VI-HPC: maximum likelihood-based phylo- 110:11481–11486. genetic analyses with thousands of taxa and mixed models. Mortazavi A, Williams BA, McCue K, Schaeffer L, Wold B. 2008. Mapping Bioinformatics 22:2688–2690. and quantifying mammalian transcriptomes by RNA-Seq. Nat Stone MJ, Williams DH. 1992. On the evolution of functional secondary Methods. 5:621–628. metabolites (natural products). Mol Microbiol.6:29–34.

Pagel M, Meade A, Barker D. 2004. Bayesian estimation of ancestral Takos AM, Rook F. 2012. Why biosynthetic genes for chemical defense Downloaded from https://academic.oup.com/mbe/article/34/8/2002/3778301 by guest on 28 September 2021 character states on phylogenies. Syst Biol. 53:673–684. compounds cluster. Trends Plant Sci. 17:383–388. Price MN, Dehal PS, Arkin AP. 2009. FastTree: computing large minimum Tanaka M, Fujimori T, Nabeta K. 2000. Biosynthesis of depudecin, a evolution trees with profiles instead of a distance matrix. Mol Biol metabolite of Nimbya scirpicola. Biosci Biotechnol Biochem. Evol. 26:1641–1650. 64:244–247. Proctor RH, Brown DW, Plattner RD, Desjardins AE. 2003. Co-expression Thrane U, Adler A, Clasen PE, Galvano F, Langseth W, Lew H, Logrieco A, of 15 contiguous genes delineates a fumonisin biosynthetic gene Nielsen KF, Ritieni A. 2004. Diversity in metabolite production by cluster in Gibberella moniliformis. Fungal Genet Biol. 38:237–249. Fusarium langsethiae, Fusarium poae,andFusarium sporotrichioides. Proctor RH, Desjardins AE, Plattner RD, Hohn TM. 1999. A polyketide IntJFoodMicrobiol. 95:257–266. synthase gene required for biosynthesis of fumonisin mycotoxins in Tuite J. 1969. Plant pathological methods: fungi and bacteria. Minnesota, Gibberella fujikuroi mating population A. Fungal Genet Biol. Minneapolis: Burgess Publishing Company. 27:100–112. Walton JD. 2000. Horizontal gene transfer and the evolution of second- Proctor RH, Van Hove F, Susca A, Stea G, Busman M, van der Lee T, ary metabolite gene clusters in fungi: an hypothesis. Fungal Genet Waalwijk C, Moretti A, Ward TJ. 2013. Birth, death and hori- Biol. 30:167–171. zontal transfer of the fumonisin biosynthetic gene cluster dur- Walton JD. 2006. HC-toxin. Phytochemistry 67:1406–14013. ing the evolutionary diversification of Fusarium. Mol Microbiol. Wight WD, Kim KH, Lawrence CB, Walton JD. 2009. Biosynthesis and 90:290–306. role in virulence of the histone deacetylase inhibitor depudecin from Ronquist F, Huelsenbeck JP. 2003. MrBayes 3: Bayesian phylogenetic Alternaria brassicicola. Mol Plant-Microbe Interact. 22:1258–1267. inference under mixed models. Bioinformatics 19:1572–1574. Wink M. 2003. Evolution of secondary metabolites from an ecological SchumacherJ,GautierA,MorgantG,StudtL,DucrotP-H,Le and molecular phylogenetic perspective. Phytochemistry 64:3–19. P^echeur P, Azeddine S, Fillinger S, Leroux P, Tudzynski B, Wisecaver JH, Slot JC, Rokas A. 2014. The evolution of fungal metabolic et al. 2013. A functional bikaverin biosynthesis gene cluster in pathways. PLoS Genet. 10:e1004816.

2015