Fungal Genetics and Biology 62 (2014) 1–10

Contents lists available at ScienceDirect

Fungal Genetics and Biology

journal homepage: www.elsevier .com/locate /yfgbi

Tools and Techniques

Establishing a versatile Golden Gate cloning system for geneticengineering in fungi

1087-1845/$ - see front matter � 2013 Elsevier Inc. All rights reserved.http://dx.doi.org/10.1016/j.fgb.2013.10.012

⇑ Corresponding author. Address: Institute for Microbiology, Build. 26.12.01,Heinrich-Heine University Düsseldorf, 40204 Düsseldorf, Germany.Fax: +49 (0)211 81 15370.

E-mail address: feldbrue@hhu.de (M. Feldbrügge).

Marius Terfrüchte a, Bastian Joehnk b, Rosa Fajardo-Somera c, Gerhard H. Braus b, Meritxell Riquelme c,Kerstin Schipper a, Michael Feldbrügge a,⇑a Institute for Microbiology, Center of Excellence on Plant Sciences (CEPLAS), Department of Biology, Heinrich-Heine University Düsseldorf, 40204 Düsseldorf, Germanyb Institute of Microbiology and Genetics, Department of Molecular Microbiology and Genetics, Georg-August-University, 37077 Göttingen, Germanyc Departamento de Microbiología, Centro de Investigación Científica y de Educación Superior de Ensenada CICESE, 22860 Baja California, Mexico

a r t i c l e i n f o

Article history:Received 26 September 2013Accepted 27 October 2013Available online 8 November 2013

Keywords:Golden Gate cloningHomologous recombinationResistance-marker recyclingFilamentous fungi

a b s t r a c t

The corn pathogen Ustilago maydis is a well-studied fungal model organism. Along with a broad set ofexperimental tools, versatile strategies for the generation of gene replacement mutants by homologousrecombination in U. maydis have been developed. Nevertheless, the production of corresponding linearDNA constructs still constitutes a time-limiting step. To overcome this bottleneck, various resistance cas-sette modules were adopted for use with the so-called Golden Gate cloning strategy. These modulesallow not only simple gene deletions but also more sophisticated genetic manipulations like insertingsequences for C-terminal protein tagging. The type IIs restriction enzyme BsaI was selected for this novelapproach as its recognition sites are comparatively rare in the U. maydis genome. To test the efficiency ofthe new strategy it was used to test the influence of varying flank lengths as well as the effect of non-homologous flank ends on homologous recombination. Importantly, to proof a broad applicability inother fungi the same strategy was used to generate mutants in the filamentous ascomycete Aspergillusnidulans. Hence, we present a highly efficient and economic cloning strategy that speeds up reversegenetic approaches in fungi.

� 2013 Elsevier Inc. All rights reserved.

1. Introduction

Ascomycete and basidiomycete fungi serve as importanteukaryotic models in fundamental and applied research. Withinthe Ascomycota well-known examples include the filamentousfungi Aspergillus nidulans and Neurospora crassa (Osmani andMirabito, 2004; Davis and Perkins, 2002; Borkovich et al., 2004;Galagan et al., 2005). Within the Basidiomycota, Ustilago maydis,a plant pathogen causing corn smut disease, is currently gainingincreasing momentum as a simple eukaryotic model organism(Bölker, 2001; Brefort et al., 2009; Dean et al., 2012; Vollmeisteret al., 2012; Feldbrügge et al., 2013). This pathogen shows a typicaldimorphism and is thus able to grow either in a yeast or in afilamentous form. The yeast form is genetically tractable by homol-ogous recombination yielding mitotically stable mutants. A largeset of genetic tools is available, including integrative and self-replicating plasmids, targeted insertion at defined genomic lociand constitutive and inducible promoters (Spellig et al., 1996;Stock et al., 2012; Brachmann et al., 2001; Zarnack et al., 2006).

The manually refined genome sequence of U. maydis was releasedin 2006 (Kämper et al., 2006; accessible at http://mips.helmholtz-muenchen.de/genre/proj/ustilago) facilitating reverse geneticapproaches.

Several years ago, a straightforward gene replacement strategybased on a vector library containing resistance cassette modulesthat allow different types of genetic manipulations was established(Brachmann et al., 2004; Kämper, 2004; Fig. 1A). At that time thecollection contained 32 plasmids harbouring four different types(I–IV) of gene replacement modules (Brachmann et al., 2004; see be-low). Each plasmid is available with four different selectable mark-ers, HygR, CbxR, NatR and PhleoR, mediating resistance againsthygromycin, carboxin, nourseothricin and phleomycin, respectively(Brachmann et al., 2004). As the most basic application, type Imodules allow for simple gene replacements by one of the fourresistance cassettes (Fig. 1B and C; Brachmann et al., 2004). As a sec-ond option, type II modules effect the expression of the gene ofinterest rather than deleting it (e.g., to render it conditional) by tran-scriptional fusion to different heterologous promoters. To this end,respective promoter cassettes, mediating inducible or constitutiveexpression of the target gene, were introduced upstream of the dif-ferent resistance cassettes (Fig. 1). Furthermore, the collection pro-vides type III modules to examine the expression of target genes by

Fig. 1. Gene deletion in U. maydis and rationale of the new library. (A) Schematic representation of the gene deletion strategy applied in U. maydis. Flanking regions areamplified by PCR and ligated to a resistance cassette (herein hygR mediating hygromycin B resistance). The resulting strain carries the resistance cassette instead of the geneof interest (goi). (B) Architecture of the gene replacement modules that are stored in storage vectors carrying the ampicillin resistance gene for selection in E. coli (AmpR). Upto three components can be contained in the module: The presence of the resistance cassette is crucial for selection. Optionally, a transcriptional reporter and a promotercassette can be added. BsaI sites (2) and (3) were introduced at the ends of the modules while Sfi(u/uC) and Sfi(d) as well as internal unique cloning sites from the traditionalsystem were kept. (C) Lists of available variants. Black font, cassettes are available for both the traditional and the novel Golden Gate based system. Grey font, cassettes areonly available in the traditional system but could easily be adopted.

2 M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10

replacing the ORF with a reporter gene such as gfp or mcherry. Forthis purpose vectors were generated, in which a reporter gene cas-sette was placed downstream of the resistance cassettes. In themost complex variants, the type IV modules, all three cassettes werecombined, e.g., to study expression of an essential gene (Fig. 1B andC; Brachmann et al., 2004).

Recently, additional compatible modules were developed thatadded increasing value to the existing library: the reporter cassettecan now be inserted in a way that leads to translational fusionswith the C-terminus of the encoded protein (type V modules;Fig. 1B and C; Becht et al., 2006). Different reporter genes encodingmCherry, eGFP, pGFP or RFP and tags like the Tap Tag, HA or Mycepitope have been established within the set of type V plasmids(Becht et al., 2006; König et al., 2009; Baumann et al., 2012). In or-der to delete several genes in one mutant, a HygR-cassette module,which can be removed using the FLP recombinase to allow markerrecycling, was designed as a variant for simple gene deletions(HygR-FRT; Khrunyk et al., 2010). Moreover, geneticin was estab-lished as a fifth dominant selectable marker for U. maydis(G418R; Kojic and Holloman, 2000; Baumann et al., 2012;Fig. 1C). The latter innovations now allow i.a. analysing multi-genefamilies with members of redundant function or introduction ofseveral reporter genes to co-localise multiple proteins (Khrunyket al., 2010). In summary, the actual library includes five differenttypes of replacement cassettes available with up to five selectablemarkers.

To generate a gene replacement construct, the module of choiceis fused to PCR-amplified flanks of about 1 kb using the restrictionenzyme SfiI (Brachmann et al., 2004; Kämper, 2004; Fig. 1A and B).The advantage of this enzyme is that the five core nucleotides be-tween its two identical recognition sequences can be chosen freely(50-GGCCNNNN;NGGCC-30). Thus, distinct core sequences (here re-ferred to as SfiI(u/uC) and SfiI(d); (Brachmann et al., 2004)) weredesigned to enable directional cloning with a single enzyme (Käm-per, 2004). Generation of the gene replacement construct is prefer-entially achieved by subcloning PCR-generated flanks intoappropriate vectors. In a second step these are combined withthe cassette of choice to result in a vector containing the final con-struct (for details see Brachmann et al., 2004). For the subsequentgeneration of the linear transformation constructs, the respectivepart was usually excised from the plasmid using restriction sitesthat are located close to the insertion sites. This led to short non-homologous sequences at each end of the two flanks. Alternatively,the deletion construct could be amplified from this vector or theligation reaction by PCR.

‘Golden Gate’ cloning is a recently described, very efficient high-throughput cloning procedure (Engler et al., 2008). The strategy isbased on type IIs restriction enzymes that do not cleave withintheir recognition sequence but a few basepairs downstream.(Szybalski et al., 1991; Engler et al., 2008). This allows seamlessjoining of DNA fragments with inverse oriented sites withoutintroducing foreign nucleotides (see below). Thus, a one-pot

M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10 3

reaction in which the restriction and the ligation can take place atthe same time can be conducted. This leads to very high cloningefficiency capable of combining even dozens of fragments in a sin-gle step (Weber et al., 2011).

In this study, we improved the standard U. maydis gene replace-ment system by including the newly developed tools and combin-ing it with the Golden Gate cloning strategy. To prove broadapplicability of the method, we also generated gene replacementmutants in the ascomycete fungus A. nidulans.

2. Materials and methods

2.1. Plasmids and plasmid constructions

Escherichia coli K12-derivative Top10 (Invitrogen/Life Technolo-gies) was used for cloning purposes (Sambrook et al., 1989). Stan-dard molecular techniques were followed. For PCR, genomic DNAof wild type strain UM521 (a1b1) served as a template (Kämperet al., 2006). All oligonucleotides used in this study are listed inSupplementary Table S2. All plasmids used for Golden Gate cloningwere depleted of BsaI recognition sites. If sites were present theywere mutagenised by PCR-mediated site-directed mutagenesis(Carter, 1986): To gain pUMa1788 (see Table 1), an intermediatevector (pUMa1508) was generated by inserting the 2.4 kb SfiI mod-ule of pUMa263 (Brachmann et al., 2004) into pUMa1466 using di-rected SfiI sites. pUMa1508 was subsequently mutagenised toremove the BsaI site within the PhleoR cassette using primersoRL752 and oRL753. The parental plasmid was eliminated by incu-bating the reaction with DpnI.

To generate pUMa1466 (pStor I), a linker obtained by annealingprimers oRL78 and oRL79 was introduced into pUMa1389, a pUC57derivative harbouring a gentamicin resistance gene, after restric-tion with SacI and XbaI. pStor II (pUMa1543) was generated simi-larly using the annealed primers oRL205 and oRL206. The

Table 1Storage vector I (pStorI) derivatives generated in this study.

Storage vector type I(SfiI(u)/SfiI(d))

pUManumber

Description/Version/Type

pStorI_1n pUMa1506 Type I: simple knockout with NatRpStorI_1 h pUMa1507 Type I: simple knockout with HygRpStorI_1rh pUMa1522 Type I: simple knockout with HygR/FRTpStorI_1 g pUMa1545 Type I: simple knockout with G418R

pStorI_1c pUMa1777 Type I: simple knockout with CbxRpStorI_1p pUMa1788 Type I: simple knockout with PhleoRpStorI_2–3 h pUMa1778 Type II: transcriptional fusion to constitutivepStorI_3 h pUMa1779 Type III: transcriptional fusion with Gfp (HygpStorI_4–3 h pUMa1780 Type IV: transcriptional fusion with constitu

transcriptional fusion with Gfp (HygR)

a BsaI site removed by site-directed mutagenesis, see Materials and methods section.

Table 2Storage vector II (pStorII) derivatives generated in this study.

Storage vector type II (SfiI(uC)/SfiI(d))

pUManumber

Description/Version/Type

pStorII_5–1 h pUMa1546 Type V: transcriptional fusion wpStorII_5–2 h pUMa1547 Type V: transcriptional fusion w

pStorII_5–13 h pUMa1559 Type V: transcriptional fusion w(HygR)

pStorII_5–15 h pUMa1557 Type V: transcriptional fusion w(HygR)

destination vector pUMa1467 is a pUC57 derivative containingan ampicillin resistance gene. It was generated from pUMa1390using SalI and SacI, introducing a linker of the annealed primersoRL76 and oRL77.

All pStorI derivatives were generated by shifting the respectiveSfiI modules from pre-existing library plasmids (Brachmann et al.,2004) to pUMa1466. Similarly, all pStorII derivatives to be used inU. maydis were generated by introducing the respective SfiI mod-ules to pUMa1543. A precise description of storage vectors I andII can be found in Tables 1 and 2, respectively. pUMa1056 has beengenerated by a four-fragment ligation and combines three copies ofthe mCherry-encoding gene with a HygR cassette, enabling trans-lational fusions.

Plasmids pUMa1976 to pUMa1985 were generated by conduct-ing BsaI-mediated Golden Gate reactions with the plasmidspUMa1467 (destination vector), pUMa1507 (storage vector I forsimple knockout with HygR) and two or three PCR products asindicated in Table S1. pUMa1565 was generated by BsaI-mediatedGolden Gate cloning using the destination vector pUMa1467, thestorage vector II derivative pUMa1559 and three PCR productsyielded by the primer combinations oRL274 and oRL275 (upstreamflank, 1.0 kb), oRL276 and oRL263 (downstream flank part a,0.52 kb) and oRL261 and oRL262 (downstream flank part b,0.53 kb) using UMa521 gDNA as a template.

For generation of pUMa1664 containing the deletion constructfor veA, a storage vector I derivative (pUMa1669) was assembledwhich contained a nourseothricin resistance cassette that is func-tional in A. nidulans (gpdA::natR; natR). To this end, oRL555 andoRL556 were used to amplify the 1.04 kb cassette frompME41722. Previously, to gain pME4172, the NatR cassette wasamplified with BJ3 and OZG192 from pME3167 (Bayram et al.,2008) and inserted into pJET1.2 (Fermentas). Subsequently, restric-tion hydrolysis with SfiI was used to insert the cassette intopUMa1466. pUMa1664 was then generated in a BsaI-mediatedGolden Gate reaction containing pUMa1669, pUMa1467 and three

Origin of SfiI module

pMF1n (pUMa262) Brachmann et al. (2004)pMF1h (pUMa194) Brachmann et al. (2004)pMF1rh (pUMa1441) Khrunyk et al. (2010)pMF1g (pUMa1057) Kojic and Holloman (2000)and Baumann et al. (2012)pMF1c (pUMa260) Brachmann et al. (2004)pMF1p (pUMa263a) Brachmann et al. (2004)

promoter Potef (HygR) pMF2–3 h (pUMa319) Brachmann et al. (2004)R) pMF4–3 h (pUMa229) Brachmann et al. (2004)

tive promoter Potef and pMF4–3 h (pUMa232) Brachmann et al. (2004)

Origin of SfiI module

ith sequence encoding Gfp (HygR) pMF5–1 h (pUMa317) Becht et al. (2006)ith sequence encoding Rfp (HygR) pMF5–2 h (pUMa890) Baumann et al. (in

press)ith sequence encoding 3xmCherry pMF5–13 h (pUMa1056) (This study)

ith sequence encoding mCherry pMF5–15 h (pUMa1409) Baumann et al.(in press)

4 M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10

PCR products gained with the primer combinations oRL557 andoRL561 (upstream flank part a, 0.8 kb), oRL562 and oRL558 (up-stream flank part b, 0.76 kb) and oRL559 and oRL560 (downstreamflank, 1.3 kb) on gDNA of A. nidulans AGB551 (Table S4).

All DNA modifications were confirmed by sequencing. Westored the basic vectors at the Fungal Genetics Stock Center(http://www.fgsc.net/) for distribution. All plasmids and sequencesare available upon request and will be published on our homepage(http://www.mikrobiologie.hhu.de/ustilago-community.html).

2.2. Golden Gate reactions

All plasmids and PCR products used in the reaction were puri-fied by standard procedures (Jetquick, Genomed; SureClean, Bio-line; according to the manufacturers’ protocols). PCR productswere excised from agarose gels and purified by gel extraction (Jet-Sorb, Genomed). Standard Golden Gate reactions for generation ofplasmids for gene replacement mutants were set up in 15 ll one-pot reactions containing 75 ng of the respective storage vectorderivative, 75 ng destination vector, 40 ng PCR product of eachflank, 1� T4 DNA ligase buffer (Roche), 0.5 ll BsaI-HF (NEB; 10U) and 0.75 ll T4 DNA ligase (Roche; 0.75 U). The reaction was per-formed in a PCR cycler using the following program: [37 �C 2 min,16 �C 5 min] 50 cycles, 37 �C 5 min, 50 �C 5 min, 80 �C 5 min. Fornegative controls, one PCR flank was excluded from the reaction.2 ll of each Golden Gate reaction was transformed into chemicallycompetent E. coli Top10 (Invitrogen) using dYT plates containing100 lg/ml ampicillin. For blue/white selection, plates were supple-mented with 60 ll X-Gal (2% (w/v) solved in DMSO) prior toplating.

2.3. Strains and growth conditions

Growth conditions for U. maydis strains and source of antibiot-ics were described previously (Brachmann et al., 2004; Baumannet al., 2012; Khrunyk et al., 2010). U. maydis strains were generatedby transformation of AB33 (Brachmann et al., 2001) derived strainswith SspI- or SwaI-linearised plasmids (Supplementary Table S1).Cells were incubated at 28 �C with 200 rpm shaking for liquid cul-tures. The A. nidulans strain AGB551 (Bayram et al., 2012) used fordeletion of veA was cultivated on minimal medium containing 1%(w/v) glucose, AspA (70 mM NaNO3, 7 mM KCl, 11,2 mM KH2PO4,pH 5.5), 2 mM MgSO4 and trace elements (76 lM ZnSO4, 178 lMH3BO3, 25 lM MnCl2, 18 lM FeSO4, 7.1 lM CoCl2, 6.4 lM CuSO4,6.2 lM Na2MoO4, 174 lM EDTA; Alic et al., 1991), supplementedwith 100 lM pyridoxine–HCl, 10 mM uracil and 1 mM uridine at37 �C. 2% (w/v) agar was added to solid media. For selective med-ium 120 lg/ml nourseothricin-dihydrogen sulphate was used.Vegetative cultures were grown in liquid submerged medium.

2.4. Gene disruption mutants

For U. maydis, gene disruption mutants were generated as pre-viously described (Brachmann et al., 2004). For generation offlanks, genomic DNA of wild type strain UM521 (a1b1) was usedas template (Kämper et al., 2006). For selection of hygromycinresistant mutants the antibiotic was used at a concentration of200 lg/ml. Correct gene replacement mutants were identifiedeither by means of counter-selection (strains were replica-platedon CM plates containing hygromycin or nourseothricin) or by char-coal bioassays (see below). For each experiment 22 independenttransformants were analysed. In addition to counter selection, mu-tants harbouring Rrm4 fused to fluorescent proteins were testedmicroscopically for the respective fluorescence (see below).

For A. nidulans, deletion of the veA gene was conducted accord-ing to published protocols. For transformation of the wild type

strain AGB551, plasmid pUMa1664 was linearised using SspI togenerate a 3.85 kb linear deletion construct. The cassette wastransformed into strain AGB551 by polyethylene glycol mediatedprotoplast fusion (Punt and van den Hondel, 1992). Transformantswere selected on nourseothricin-containing plates (120 lg/ml). Toisolate genomic DNA, vegetatively grown mycelium of AGB551 andeight transformants were ground in liquid nitrogen and subse-quently lysed in lysis-buffer (50 mM Tris–HCl (pH 7.2), 50 mMEDTA, 3% (w/v) sodium dodecyl sulphate, 1% (v/v) b-mercap-toethanol) at 65 �C. Genomic DNA was extracted with phenol andprecipitated with isopropanol containing 50 mM Na-acetate. Geno-mic DNA was dissolved in EB-buffer (Qiagen) containing RNase Aand subjected to Southern blot analysis (Southern, 1975). Alkalinephosphatase labelled probes were created with Amersham GeneImages AlkPhos Direct Labelling and Detection System (GE Health-care). The 50 flanking region of veA used for probe preparation wasamplified from genomic DNA with PCR using primers BJ194 andBJ195.

2.5. Microscopic analyses and image processing

A Zeiss Axio Observer.Z1 microscope equipped with a CCDcamera (Photometrics CoolSNAP HQ2) and objective lenses PlanNeofluar (40�, NA 1.3), Plan Apochromat (63� and 100�, NA1.4) and a-Plan Apochromat (100�, NA 1.46) was used for allmicroscopic analyses in this study (Baumann et al., 2012). Excita-tion of fluorescently-labelled proteins was carried out using anHXP metal halide lamp (LEj, Jena, Germany) in combination withfilter sets for Gfp (ET470/40BP, ET495LP, ET525/50BP) and mCher-ry (ET560/40BP, ET585LP, ET630/75BP; Chroma, Bellow Falls, VT,USA). All parts of the microscope system were controlled by thesoftware package MetaMorph (Molecular Devices, version 6 and7, Sunnyvale, CA, USA), which was also used for image processingincluding the adjustment of brightness and contrast as well asthe generation of kymographs (Baumann et al., 2012).

2.6. Bioassay to test filamentous growth in U. maydis

To assay filamentous growth of AB33 derivates, the respectivesporidial cells were grown in liquid complete medium (CM) sup-plemented with 1% (w/v) glucose to the exponential growth phase(OD600 = 0.8–1.0). Subsequently, cells were spotted on solid nitrateminimal medium (NM) supplemented with 1% (w/v) glucose, 10 g/lactivated charcoal and 20 g/l Bacto™ agar. The plates were sealedwith parafilm and incubated at 28 �C overnight.

2.7. Phenotyping of A. nidulans veA deletion strains

For sexual induction 104 conidia were point inoculated on solidmedium and incubated in the dark under oxygen limiting condi-tions for six days. The formation of cleistothecia was observed un-der an SZX-stereomicroscope (OLYMPUS) equipped with a SC30digital camera (OLYMPUS) and cellSens Dimension Software(OLYMPUS).

3. Results and discussion

3.1. Combining Golden Gate cloning with the classical genereplacement approach

The objective of this study was to combine the highly efficientGolden Gate cloning strategy (Engler et al., 2008) with traditionalprocedures to generate gene deletion or insertion mutants in U.maydis (Brachmann et al., 2004; Kämper, 2004). To identify themost suitable type IIs restriction endonucleases for use in our

M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10 5

experimental system, the elements of the pre-existing vector li-brary (Brachmann et al., 2004) as well as the genome sequenceof U. maydis (http://mips.helmholtz-muenchen.de/genre/proj/usti-lago) were analysed with respect to the presence of the corre-sponding candidate sites. Only those commercially availableenzymes were considered in the analysis that matched the follow-ing requirements: (i) the sticky end generated after hydrolysis is atleast 4 bp long to allow efficient ligation, (ii) the enzyme displaysfull activity in ligase buffer at temperatures of 6 37 �C and (iii)the restriction and the recognition sites are not separated by morethan ten bases to allow the incorporation into primers of suitablelength. The enzyme BsaI (isoschizomer: Eco31I) with the non-pal-indromic recognition/hydrolysis site occurring statistically onceevery 3592 bp turned out to match the most important criteriafor use in U. maydis. This enzyme is active at 37 �C, has a recogni-tion site of 6 bps separated by one random nucleotide from the siteof hydrolysis and generates a 4 bases overhang (50-GGTCTC(N)1-30/30-CCAGAG(N)5-50; recognition site underlined; Fig. 2A).

The rationale for our novel cloning system is depicted in Fig. 2B.In brief, flanking regions mediating homologous recombination are

Fig. 2. Scheme of the Golden Gate based gene replacement strategy. (A) Type IIs restrictioa few basepair downstream (dark blue). Thus, two fragments with inverse oriented sites cone-pot reaction in which the restriction and the ligation can take place at the same timewith the upstream flank using the specific site BsaI(1). Furthermore, an SspI site (red font)the linear gene replacement construct can be excised without including any non-homologa one-pot reaction including DNA ligase and the typeIIs enzyme BsaI. An appropriate storaselection can be chosen from the collection. Two flanking regions (orange) homologous tBsaI sites with unique overhangs called (1) to (4). In the end, the different fragments assresistance and allowing blue/white selection (lacZ). For excision of the linear transformainto the flanks. (For interpretation of the references to colour in this figure legend, the r

amplified from genomic DNA by PCR. Concurrently, two specific,inverse oriented BsaI sites are introduced at each end of the regionto be amplified by PCR, yielding products framed with unique BsaIsites (BsaI(1)-UF-BsaI(2) and BsaI(3)-DF-BsaI(4)). Using these sites,the flanks can be combined directionally with one of the resistancecassette modules stored in library vectors (designated as storagevector derivatives, pStor). Therefore, the existing vector collection(Brachmann et al., 2004) was re-designed by inserting the corre-sponding BsaI(2) and BsaI(3) sites up- and downstream of theexisting resistance cassette modules, respectively (Fig. 1B). Withthe specific BsaI sites on each flank end directional cloning into adestination vector (pDest) as well as directional fusion to the resis-tance cassette module of choice is possible (Figs. 1, 2B). Thus, thefour fragments (pDest, upstream flank, downstream flank andpStor with an appropriate resistance module) can be assembledin a highly efficient way conducting a so-called one-pot reactionin which restriction and ligation reactions are combined in onetube (Fig. 2B). Different bacterial resistance genes (pStor with gen-tamicin, GentR; pDest with ampicillin, AmpR) prevent the forma-tion of false-positive E. coli clones derived from the storage

n enzymes like BsaI do not cleave within their recognition sequence (light blue) butan be joined directionally while the recognition site is lost. This allows conducting a. The figures exemplifies this mechanism for the assembly of the destination vectoris added to the upstream flank PCR product. Thus, in the finally assembled plasmid,ous bases. (B) To generate a gene replacement vector, four components are mixed inge vector derivative carrying different modules including the resistance cassette for

o the target locus are generated by PCR. Both flanks and the modules are framed byemble in the destination vector pDest harbouring a gene that mediates gentamicintion construct from the gene replacement vector, SspI (or SwaI) sites are introducedeader is referred to the web version of this article.)

6 M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10

vector. Further accuracy is achieved by lacZ-mediated blue/whiteselection that allows screening for insertions into the destinationvector.

To increase the rate of homologous recombination (see below),recognition sites of either SwaI (50-ATTT; AAAT-30) or SspI(50-AAT;ATT-30) blunt end cutters were introduced to each flankbetween the BsaI sites (1)/(4) and the very upstream or down-stream end of the homologous region (Fig. 1B, Fig. 2). Flank endswere subsequently positioned in a way that the genomic regionconstitutes half of the respective SwaI or SspI recognition site. Toachieve this, half of the site needs to be present at the very flankend (e.g. 50-AAAT-30 for a SwaI site in the upstream flank), whereasthe second half is provided by the primer overhang (for example ifthe genomic sequence reads 50-GCCGTGAAATCGGCTAGCT-30, theresulting primer includes 50-atttAAATCGGCTAGCT-30). In parallel,these sites allow the excision of the construct for transformation.This assures that the SwaI or SspI linearised construct contains100% homologous flanking sequences and prevents insertion offalse nucleotides during homologous recombination due to thepresence of foreign non-homologous DNA.

Importantly, seamless cloning was used to stay fully compatiblewith the traditional SfiI system (Brachmann et al., 2004). Thus, theresulting plasmids can also be used for traditional cloning ap-proaches to exchange, for example, the resistance cassette in a sim-ple two-fragment ligation using SfiI. Further features contained inthe cassettes, as e.g. resistance cassette framing by NotI sites, werealso kept (see Brachmann et al., 2004; Fig. 1B).

In total, nine storage I and four storage II library vectors con-taining exemplary modules for all 5 types of gene replacementswere generated (Fig. 2B; Tables 1 and 2 with the details of eachvector; also compare to Table 1 in Brachmann et al., 2004) by shift-ing the SfiI framed modules of the original plasmids to pStorI or

Fig. 3. Influence of flank length and homologous flank ends on homologous recombinathave been generated. The 1 kb downstream flanks have been generated in two parts ahomologous ends. (C) Flanks used to generate rrm4 deletion mutants. Flank length varicompletely homologous to the genomic DNA and asterisks mark 10 bp regions leading tothe correlation between flank length and flank homology and the rate of homologous recC. For each flank combination, 22 transformants were analysed. (E) Filamentation bioaswith non-homologous ends. Strains AB33 (wt) and AB33rrm4D were used as filamenmorphology whereas strains lacking rrm4 grow less fuzzy. 22 transformants were teste

pStorII. In order to render them suitable for Golden Gate cloning,BsaI recognition sites present in some vector elements (e.g., thePhleoR cassette) were eliminated by site-directed PCR mutagenesis(Carter, 1986). Thus, all five types of gene replacement modules arenow available in the Golden Gate adapted system (Tables 1 and 2).

3.2. Influence of flank length and homologous ends on homologousrecombination

Currently, there are two major limitations in generating genereplacement constructs: The long duration (mostly about1–2 weeks) of generating plasmids containing the constructsneeded for U. maydis transformation, which is cumbersome espe-cially for inexperienced researchers, and the error-proneness ofswiftly generated PCR-based constructs (Brachmann et al., 2004;Kämper, 2004). To proof that our new strategy can be easily ap-plied to generate multiple gene replacement constructs in a shorttime, we set out to test the implication of different flank lengths(ranging from about 230 bp to 1090 bp) as well as the effect ofnon-homologous flank ends on homologous recombination in U.maydis (Table S2). According to previous publications, a flanklength of 0.8–1 kb was considered to be optimal whereas a shortlength of 80 bp, as is successfully used in Saccharomyces cerevisiae,seemed insufficient for U. maydis and most filamentous fungi(Brachmann et al., 2004; Ninomiya et al., 2004). However, more de-tailed studies regarding this question had not been performed todate. Thus, we aimed to test five different combinations of flanklength using homologous and non-homologous flank ends(Fig. 3A–C; Table S2). For generating non-homologous flank endswe added a random linker of 10 bp (Fig. 3C, Table S1). As proof-of-principle we chose the gene rrm4. This gene encodes the keymRNA-binding protein for endosomal mRNA transport in U. maydis

ion. (A and B) Agarose gels depicting the different up- and downstream flanks thats it contained a BsaI site. (A) Flanks with homologous ends. (B) Flanks with non-ed between about 250 bp, 500 bp and 1 kb. Orange bars represent regions that arenon-homologous flank ends. For details see Table S2. (D) Graphical representation ofombination in the rrm4 locus. The corresponding flank combinations are depicted insay on solid plates containing charcoal for the strain AB33rrm4D using 1 kb flankstation controls, respectively. Strains containing rrm4 show a fuzzy white colony

d for each flank combination.

M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10 7

and is essential for efficient unipolar filament formation (Bechtet al., 2005; Becht et al., 2006; Baumann et al., 2012). Thus, rrm4deletion mutants can be quickly identified in a bioassay due totheir failure to form wild type-like filaments.

Ten plasmids containing the deletion constructs with differentflank length combinations were generated (Fig. 3A and B;Table S2). Due to the presence of a single BsaI recognition siteabout 500 bp downstream of the rrm4 gene, the 1 kb downstreamflanks were amplified in two parts, UF1 and UF2, following thestrategy of enzymatic inverse PCR (Fig. 3A and B downstream;Table S1). This protocol involves two PCR products with primersoverlapping the interfering BsaI site. Combining the two productsintroduces a single nucleotide mismatch that eliminates the BsaIrecognition sequence (Stemmer and Morris, 1992; Engler et al.,2008). After amplification of all flanks (Fig. 3A), Golden Gate reac-tions including the PCR products of the two flanks of choice (corre-sponding to two or three DNA fragments) and the storage vectorpStorI_1 h (Table 1) containing a simple HygR deletion cassetteand the destination vector pDest (pUMa1467) were performed.Following E. coli transformation, three white clones were analysedfor each plasmid, yielding at least one correctly assembled plasmideach (Table S2). To sum up, these experiments proof the high effi-ciency and accuracy of this method in line with published proto-cols (Engler et al., 2008). Importantly, compared to formerprotocols, the time to generate the plasmids was reduced to fourdays while omitting the error-prone PCR amplification of the con-struct. This demonstrates the power of the novel strategy for use ingenerating U. maydis gene replacement vectors.

Following transformation of U. maydis AB33 with each of theten corresponding linearised plasmids (Fig. 3C), 22 transformantsof each strain (see Table S3) were cultured in liquid medium andsubsequently assayed for filamentous growth on charcoal-contain-ing plates (exemplified in Fig. 3D). AB33 is a haploid lab strain thathas been engineered to exhibit filamentous growth upon switch ofthe nitrogen source (Brachmann et al., 2001). Wild type-like unipo-lar filamentous growth can be monitored on nitrate-containingcharcoal plates where it leads to a white and fuzzy colony mor-phology. In contrast, bipolar growing rrm4 deletion mutants grow

Fig. 4. Combining strain generation and counter-selection to switch fluorescence prote(C-terminal Gfp fusion) to Rrm4C3 (C-terminal triple mCherry fusion). A NatR strain harbowith a construct that replaces gfp gene with mcherry and introduces a hygR cassette insteahygromycin but not on nourseothricin. (B) Counter-selection of 22 transformants of AB3were used as growth control. Lab strain AB33 (wt) grows neither on nourseothricin- norNat only and the rrm4 deletion strain AB33rrm4D-HygR growing on hygromycin only weshown to be positive in counter-selection (HygR, NatS). Besides DIC pictures Gfp and mCthe bidirectional movement of mCherry signals throughout the fungal cell.

less fuzzy (Banuett and Herskowitz, 1989; Fig. 3D). We observedthat a flank length of around 1 kb with 100% identity results inthe highest rates of homologous recombination (59%; Fig. 3C).The utilisation of shorter flanks or flanks with non-homologousends led to an increased number of ectopic insertions. Respectivetransformants seem to contain the hygromycin resistance cassettebut nevertheless do not show the expected rrm4 deletion pheno-type. In our study, rates between 0% and 32% homologous recom-bination were determined (Fig. 3C). Thus, we recommend usinghomologous flanks with a length of about 1 kb for efficient geneticengineering. In conclusion, the newly developed strategy allows forthe fast generation of multiple U. maydis mutants. The majoradvantage compared to previously applied strategies is the veryreliable and quick generation of the corresponding plasmids.Applying the novel strategy, we demonstrated for the first timethat homologous ends are important for efficient homologousrecombination in U. maydis.

3.3. Generation of C-terminal protein fusions by counter-selection

As second proof-of-principle we aimed to generate plasmidsthat allow the switch of reporter genes translationally fused toRrm4. To this end, a nourseothricin resistant strain that encodesa fusion of Rrm4 with eGfp (Rrm4G; enhanced version of the greenfluorescent protein, Clontech) at the endogenous locus (Fig. 4A)was transformed with a linear construct that mediates the switchto a fusion protein with triple mCherry (Rrm4C3). For this purpose,the upstream flank was placed such that it ends at the 30-nd of therrm4 open reading frame, thereby deleting the stop codon. Usingthe Golden Gate strategy, we generated a vector containing 100%homologous flanks of about 1 kb for homologous recombinationat the rrm4 locus as well as resistance cassette modules containingHygR and the reporter gene mCherry in triple copy (pUMa1565).This cassette for translational fusion had been introduced intothe storage vector pStorII in advance (pUMa1559). This storagevector differs from pStorI in containing SfiI(uC) instead of SfiI(u)(Table 2). As described for pStorI, BsaI sites with the overhangs(2) and (3) are flanking the cassettes introduced into this vector

in fusions. (A) Scheme of the genetic basis of switching the fusion protein Rrm4Guring a translational fusion of rrm4 to gfp in the endogenous locus was transformedd of natR. This allows screening mutants by counter-selection with cells growing on

3rrm4C3. Positive strains grow on hygromycin but not on nourseothricin. CM plateson hygromycin-containing plates. As controls, strain AB33rrm4G-NatR growing on

re used. (C) Microscopic validation of the colour switch for AB33rrm4C3#15 that washerry fluorescence was detected. Scale bar, 10 lm. The kymograph below visualises

8 M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10

allowing the cassette to assemble with the flanks in the GoldenGate reaction.

Due to the selected counter-selection strategy, transformantscould be pre-screened for growth on hygromycin but not onnourseothricin. A total of 22 transformants were tested by coun-ter-selection (Fig. 4B) with twelve showing the expected growthbehaviour (Table S3). We then further investigated all 22 transfor-mants microscopically to check whether the counter-selectionresults were indicative for a switch from green to red fluorescence.We indeed observed that all cells that were positive in counter-selection showed mCherry but no eGfp fluorescence. Fig. 4C shows

Fig. 5. Applying the Golden Gate cloning strategy to generate an A. nidulans veA deletrestriction enzymes NcoI and PvuI used for Southern analysis. Corresponding bands foreplacement of the veA locus by the nourseothricin resistance cassette (lanes 1–3 and 5Phenotypic comparison between wildtype and the DveA mutants 1 and 2. As previously scl) after sexual induction. Instead they continued producing asexual structures (conidio

microscopic pictures of one positive candidate of AB33rrm4C3

detecting signals in both the Gfp and mCherry channel. While onlybackground fluorescence is visible in the Gfp channel, red fluores-cent particles can be visualised, indicating functional mCherry. Asfunctional Rrm4 shuttles on endosomes along microtubules, werecorded movies of cells from the same strain with mCherry fluo-rescence and visualised particle movement by kymographs(Fig. 4C). These confirm that the observed particles show a similarmovement as described for Rrm4 in other studies (Becht et al.,2005; Baumann et al., 2012). These results demonstrate that theGolden Gate strategy can be used to quickly generate plasmids

ion strain. (A) Map of wildtype (wt) veA and DveA deletion loci with sites of ther these enzymes are shown in Southern hybridisations, which verify the correct). Mutants that were used for further characterisation are marked by asterisks. (B)

hown the DveA mutants were not able to form sexual fruiting bodies (cleisthothecia;phores; co).

M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10 9

for C-terminal protein fusions. Furthermore, counter-selection is avalid method to pre-screen transformants – however, it does noteliminate the need for a more careful verification of the geneticbackground, e.g. by Southern blot analysis (Brachmann et al.,2004; Southern, 1975).

3.4. Application of the Golden Gate cloning strategy in A. nidulans

To proof that our novel system is also applicable in other fungalmodel organisms, the same strategy was adopted for a gene dele-tion in A. nidulans. To this end, we chose a gene that leads to a dis-tinct morphologic phenotype upon deletion. We selected the veAgene which encodes a light response factor that activates sexualdevelopment in the dark (Bayram et al., 2008; Bayram et al.,2010; Kim et al., 2002). As a consequence veA deletion strains arestrongly reduced in the formation of sexual structures, the so-called cleistothecia, when grown under sexual inducing conditionslike in the dark or at low oxygen.

We introduced a nourseothricin resistance cassette specific forA. nidulans into pStorI using SfiI. This enabled the conduction of aGolden Gate reaction including the respective storage vectors,pDest as well as PCR-amplified flanks specific for the gene to bedeleted. This resulted in the generation of plasmid pUMa1664 fordeletion of veA. The linear transformation construct wassubsequently excised from the plasmids and used to transform A.nidulans AGB551, a strain that harbours a DnkuA-mutation andtherefore provides high rates of homologous recombination (Nayaket al., 2006). The mutant candidates obtained after transformationwere screened by Southern blot analysis (Fig. 5A). As expected, theveA deletion strains displayed a defect in the formation ofcleistothecia after six days of sexual development while theparental strain was able to form these structures (Fig. 5B).

Thus, clear deletion mutants could be identified, suggesting thatour newly developed strategy can be transferred to other modelfungi.

4. Conclusions

Fungal model organisms are important not only for basicresearch but also in applied sciences and synthetic biology(Feldbrügge et al., 2013). The long duration of mutant generationusually constitutes a time delay factor, which e.g. limits high-throughput approaches. This bottleneck can be tackled at the levelof molecular cloning as well as at the level of genetic manipulationof the fungi. The latter includes efficient transformation, homolo-gous recombination and convenient mutant analysis and is wellestablished for most fungal models (Park et al., 2011). As oneexample, for N. crassa, a highly efficient method to disrupt, modifyor replace a target gene by homologous recombination of its flank-ing sites was introduced in 2004 (Ninomiya et al., 2004). Themethod involved disruption of KU70 and KU80, which functionin non-homologous end-joining of double-stranded DNA breaksand provided highly efficient recipient N. crassa strains for genetargeting when 2 kb flanking regions were used for transformation.

We here concentrated on optimising the cloning methodologyand describe an improved strategy for the nucleic acid manipula-tion steps involved in the generation of gene replacement mutants.Besides demonstrating its applicability in U. maydis and A. nidulans,we also aimed on generating a N. crassa deletion mutant of sec-5,an exocyst component for which a UV mutant with seriously im-paired growth had been previously obtained (Seiler and Plamann,2003). We were successful in obtaining heterokaryons carryingwild type and gene deletions. However, no homokaryons couldbe identified because sec-5 appears to be essential in the geneticbackground of the transformed strain N. crassa FGSC#9718. Thus,

this further demonstrates that our method can be widely appliedin smuts and other model fungi. To our knowledge, this is the firstapproach in which a Golden Gate strategy has been applied in fil-amentous fungi.

Some recently developed high-throughput cloning strategiesare based on yeast recombinational cloning (Collopy et al., 2010).These strategies are efficient but time-consuming due to the neces-sity to transform S. cerevisiae (Oldenburg et al., 1997). This methodhas for example been used in N. crassa to generate knockout mu-tants (Colot et al., 2006). By 2010, knockout cassettes for all anno-tated ORFs in N. crassa were generated (Collopy et al., 2010).

In Aspergillus species, the so-called double-joint PCR was estab-lished in 2004 for the generation of deletion cassettes and gene fu-sions (Yu et al., 2004). This method was further improved with thefusion PCR strategy (Szewczyk et al., 2006). Both methods arebased on the fusion of DNA fragments within a single PCR reactionbased on overlapping DNA tails. Similarly, a PCR-based methodalso exists for gene-tagging to produce chromosomally encoded fu-sion proteins in N. crassa (Lai et al., 2010). This method is based onPCR synthesis of fusion cassettes. Although these methods are veryrobust, they harbour a major drawback when it comes to the fusionof long DNA constructs due to the error-proneness of PCR. This dis-advantage can be limited with the use of specialised but expensiveDNA polymerases like the Pfu-DNA polymerase. Notably, the hereestablished Golden Gate cloning system can completely bypass thisproblem because the single modules are conserved in plasmid vec-tors. Especially with a growing library of different modules thissystem would increase cloning efficiency compared to the fusionPCR-based system, requiring generation of new PCR products forevery construct.

In terms of efficiency, precision and simplicity the Golden Gatecloning based method is comparable to strategies that rely oneither site-specific recombination (e.g., Gateway cloning; Invitro-gen; Hartley et al., 2000) or isothermal Gibson assembly (Gibsonet al., 2009). However, in comparison to strategies based on site-specific recombination this approach provides the advantage ofnot leaving foreign recombination sites in the genome. Gibsonassembly could act as a good alternative method to the approachwe used because it allows seamless cloning based on overlappinghomologous regions (Gibson et al., 2009; Gibson, 2011). However,linear fragments are needed for the procedure resulting in an addi-tional experimental step before starting the actual cloning proce-dure. It is furthermore important to note, that the here presentedstrategy is also very inexpensive compared to many commerciallyavailable systems, as it only relies on low amounts of a singlerestriction enzyme and T4 ligase.

One drawback of the presented strategy, however, is that someloci are not suitable for the use of the specific enzyme BsaI becausethe flanking regions might contain this site too frequently. Thus, inthe future, we plan to expand our plasmid collection for the usewith alternative type IIs enzymes. The choice between at leasttwo enzymes would strongly increase the chance that the GoldenGate strategy can be applied to generate gene replacement con-structs for most genomic loci.

Conflicts of interest

The authors declare that the research was conducted in the ab-sence of any commercial or financial relationships that could beconstructed as a potential conflict of interest.

Acknowledgments

We thank Dr. A. Brachmann, LMU Munich, and all lab membersfor valuable discussion and critical reading of the manuscript. We

10 M. Terfrüchte et al. / Fungal Genetics and Biology 62 (2014) 1–10

greatly acknowledge B. Axler for excellent technical assistance, J.Stock and Dr. E. Vollmeister for providing ideas with respect tothe experimental design and E. Stratmann for proofreading of themanuscript. Special thanks to Drs. A. Brachmann, Y. Khrunyk andR. Kahmann for sharing plasmids and to G. Mannhaupt for deter-mining the number of Type IIs recognition sites of the U. maydisgenome. Genomic DNA of N. crassa was kindly provided by Drs.K. Kopke and U. Kück, Ruhr-University Bochum. We further thankT. Pohlmann and Dr. S. Baumann for supplying pUMa1056.

Our applied work was supported by the Ministry of Innovation,Science and Research of North Rhine-Westphalia and the HeinrichHeine University Düsseldorf (HHUD) through funding within theCLIB-Graduate Cluster Industrial Biotechnology and by a grantfrom the Strategic Research Fund of the HHUD to KS. Basic researchin the laboratory was funded by the Deutsche Forschungsgemeins-chaft (DFG) as part of the German/Mexican research groupFOR1334 (FE 448/5-1) and EXL1028 to M.F., as well as the HHUDgraduate schools iGRAD-MOI and iGRAD-Plant. Research at theUniversity of Göttingen was funded by the DFG as another partof the German/Mexican research group FOR1334 (BR 1502/13-2).Research in the Riquelme lab was supported by Mexican NationalCouncil for Science and Technology (CONACYT) grants U-45818Q,B0C022 and CONACYT-DFG 75306. RFS was funded by a CONACYTfellowship (176643).

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.2013.10.012.

References

Alic, M. et al., 1991. More gene manipulations in fungi. In: Bennette, J.W., Lasure, L.L.(Eds.). Academic Press Inc., San Diego, California.

Banuett, F., Herskowitz, I., 1989. Different a alleles of Ustilago maydis are necessaryfor maintenance of filamentous growth but not for meiosis. Proc Natl Acad SciUSA 86, 5878–5882.

Baumann, S. et al., 2012. Kinesin-3 and dynein mediate microtubule-dependent co-transport of mRNPs and endosomes. J. Cell Sci. 125, 2740–2752.

Baumann, S., et al., in press. Endosomal transport of septin mRNA and proteinindicates local translation on endosomes and is required for correct septinfilamentation. EMBO Rep. in press.

Bayram, O. et al., 2012. The Aspergillus nidulans MAPK module AnSte11-Ste50-Ste7-Fus3 controls development and secondary metabolism. PLoS Genet. 8,e1002816.

Bayram, O. et al., 2010. Spotlight on Aspergillus nidulans photosensory systems.Fungal Genet. Biol. 47, 900–908.

Bayram, O. et al., 2008. VelB/VeA/LaeA complex coordinates light signal with fungaldevelopment and secondary metabolism. Science 320, 1504–1506.

Becht, P. et al., 2006. The RNA-binding protein Rrm4 is essential for polarity inUstilago maydis and shuttles along microtubules. J. Cell Sci. 119, 4964–4973.

Becht, P. et al., 2005. Role for RNA-binding proteins implicated in pathogenicdevelopment of Ustilago maydis. Eukaryot. Cell 4, 121–133.

Bölker, M., 2001. Ustilago maydis – a valuable model system for the study of fungaldimorphism and virulence. Microbiology 147, 1395–1401.

Borkovich, K.A. et al., 2004. Lessons from the genome sequence of Neurosporacrassa: tracing the path from genomic blueprint to multicellular organism.Microbiol. Mol. Biol. Rev. 68, 1–108.

Brachmann, A. et al., 2004. A reverse genetic approach for generating genereplacement mutants in Ustilago maydis. Mol. Genet. Genomics 272, 216–226.

Brachmann, A. et al., 2001. Identification of genes in the bW/bE regulatory cascadein Ustilago maydis. Mol. Microbiol. 42, 1047–1063.

Brefort, T. et al., 2009. Ustilago maydis as a Pathogen. Annu. Rev. Phytopathol. 47,423–445.

Carter, P., 1986. Site-directed mutagenesis. Biochem. J. 237, 1–7.

Collopy, P.D. et al., 2010. High-throughput construction of gene deletion cassettesfor generation of Neurospora crassa knockout strains. Methods Mol. Biol. 638,33–40.

Colot, H.V. et al., 2006. A high-throughput gene knockout procedure for Neurosporareveals functions for multiple transcription factors. Proc Natl Acad Sci USA 103,10352–10357.

Davis, R.H., Perkins, D.D., 2002. Timeline: Neurospora: a model of model microbes.Nat. Rev. Genet. 3, 397–403.

Dean, R. et al., 2012. The Top 10 fungal pathogens in molecular plant pathology. MolPlant Pathol. 13, 414–430.

Engler, C. et al., 2008. A one pot, one step, precision cloning method with highthroughput capability. PLoS One 3, e3647.

Feldbrügge, M. et al., 2013. The biotechnological use and potential of plantpathogenic smut fungi. Appl. Microbiol. Biotechnol. 97, 3253–3265.

Galagan, J.E. et al., 2005. Sequencing of Aspergillus nidulans and comparativeanalysis with A. fumigatus and A. oryzae. Nature 438, 1105–1115.

Gibson, D.G., 2011. Enzymatic assembly of overlapping DNA fragments. MethodsEnzymol. 498, 349–361.

Gibson, D.G. et al., 2009. Enzymatic assembly of DNA molecules up to severalhundred kilobases. Nat. Methods 6, 343–345.

Hartley, J.L. et al., 2000. DNA cloning using in vitro site-specific recombination.Genome Res. 10, 1788–1795.

Kämper, J., 2004. A PCR-based system for highly efficient generation of genereplacement mutants in Ustilago maydis. Mol. Genet. Genomics 271, 103–110.

Kämper, J. et al., 2006. Insights from the genome of the biotrophic fungal plantpathogen Ustilago maydis. Nature 444, 97–101.

Khrunyk, Y. et al., 2010. The use of FLP-mediated recombination for the functionalanalysis of an effector gene family in the biotrophic smut fungus Ustilagomaydis. New Phytol. 187, 957–968.

Kim, H. et al., 2002. The veA gene activates sexual development in Aspergillusnidulans. Fungal Genet. Biol. 37, 72–80.

Kojic, M., Holloman, W.K., 2000. Shuttle vectors for genetic manipulations inUstilago maydis. Can. J. Microbiol. 46, 333–338.

König, J. et al., 2009. The fungal RNA-binding protein Rrm4 mediates long-distancetransport of ubi1 and rho3 mRNAs. EMBO J. 28, 1855–1866.

Lai, J. et al., 2010. Marker fusion tagging, a new method for production ofchromosomally encoded fusion proteins. Eukaryot. Cell 9, 827–830.

Nayak, T. et al., 2006. A versatile and efficient gene-targeting system for Aspergillusnidulans. Genetics 172, 1557–1566.

Ninomiya, Y. et al., 2004. Highly efficient gene replacements in Neurospora strainsdeficient for nonhomologous end-joining. Proc. Natl. Acad. Sci. USA 101, 12248–12253.

Oldenburg, K.R. et al., 1997. Recombination-mediated PCR-directed plasmidconstruction in vivo in yeast. Nucleic Acids Res. 25, 451–452.

Osmani, S.A., Mirabito, P.M., 2004. The early impact of genetics on ourunderstanding of cell cycle regulation in Aspergillus nidulans. Fungal Genet.Biol. 41, 401–410.

Park, G. et al., 2011. High-throughput production of gene replacement mutants inNeurospora crassa. Methods Mol. Biol. 722, 179–189.

Punt, P.J., van den Hondel, C.A., 1992. Transformation of filamentous fungi based onhygromycin B and phleomycin resistance markers. Methods Enzymol. 216,447–457.

Sambrook, J., Frisch, E.F., Maniatis, T., 1989. Molecular Cloning: A laboratorymanual. Cold Spring Harbour, New York.

Seiler, S., Plamann, M., 2003. The genetic basis of cellular morphogenesis in thefilamentous fungus Neurospora crassa. Mol. Biol. Cell 14, 4352–4364.

Southern, E.M., 1975. Detection of specific sequences among DNA fragmentsseparated by gel electrophoresis. J. Mol. Biol. 98, 503–517.

Spellig, T. et al., 1996. Green fluorescent protein (GFP) as a new vital marker in thephytopathogenic fungus Ustilago maydis. Mol. Gen. Genet. 252, 503–509.

Stemmer, W.P., Morris, S.K., 1992. Enzymatic inverse PCR: a restriction siteindependent, single-fragment method for high-efficiency, site-directedmutagenesis. Biotechniques 13, 214–220.

Stock, J. et al., 2012. Applying unconventional secretion of the endochitinase Cts1 toexport heterologous proteins in Ustilago maydis. J. Biotechnol. 161, 80–91.

Szewczyk, E. et al., 2006. Fusion PCR and gene targeting in Aspergillus nidulans. Nat.Protoc. 1, 3111–3120.

Szybalski, W. et al., 1991. Class-IIS restriction enzymes – a review. Gene 100, 13–26.Vollmeister, E. et al., 2012. Microtubule-dependent mRNA transport in the model

microorganism Ustilago maydis. RNA Biol. 9, 261–268.Weber, E. et al., 2011. A modular cloning system for standardized assembly of

multigene constructs. PLoS One 6, e16765.Yu, J.H. et al., 2004. Double-joint PCR: a PCR-based molecular tool for gene

manipulations in filamentous fungi. Fungal Genet. Biol. 41, 973–981.Zarnack, K. et al., 2006. Tetracycline-regulated gene expression in the pathogen

Ustilago maydis. Fungal Genet. Biol. 43, 727–738.