Fungal Genetics and Biology 45 (2008) S15–S21

Contents lists available at ScienceDirect

Fungal Genetics and Biology

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

Review Sex in smut fungi: Structure, function and evolution of mating-type complexes

Guus Bakkeren a, Jörg Kämper b, Jan Schirawski c,* a Agriculture & Agri-Food Canada, Pacific Agri-Food Research Centre, Summerland, BC, Canada V0H 1Z0 b University of Karlsruhe, Institute for Applied Biosciences, Department of Genetics, 76187 Karlsruhe, Germany c Max-Planck-Institute for Terrestrial Microbiology, Karl-von-Frisch-Strasse, 35043 Marburg, Germany article info abstract

Article history: Smut fungi are basidiomycete plant pathogens that pose a threat to many important cereal crops. In order Received 4 February 2008 to be pathogenic on plants, smut fungal cells of compatible mating-type need to fuse. Fusion and path- Accepted 11 April 2008 ogenicity are regulated by two loci, a and b, which harbor conserved genes. The functions of the encoded Available online 22 May 2008 mating-type complexes have been well-studied in the model fungus Ustilago maydis and will be briefly reviewed here. Sequence comparison of the mating-type loci of different smut and related fungi has Keywords: revealed that these loci differ substantially in structure. These structural differences point to an evolution Pheromone receptor from tetrapolar to bipolar mating behavior, which might have occurred several independent times during Homeodomain fungal speciation. Sporisorium reilianum Ustilago hordei Ó 2008 Elsevier Inc. All rights reserved. Malassezia globosa

1. Introduction vanced our understanding of the biology of smut fungi. The U. maydis genome sequence is publicly available through the Smut fungi are important cereal crop pathogens that depend on Broad Institute (http://www.broad.mit.edu/annotation/fungi/usti- sex to cause disease. There are approximately 1200 smut species lago_maydis/) and has been manually annotated by the Munich known that together can infect more than 4000 different plant spe- Institute for Protein Sequences (MIPS; http://MIPS.gsf.de/genre/ cies. The vast majority of plant species serving as hosts are from proj/ustilago/). Genome sequencing efforts are under way for the grass family (Graminaceae) and includes the world’s most S. reilianum (R. Kahmann and J. Schirawski, unpublished) and for important crops: corn, barley, wheat, oats, sorghum, sugarcane U. hordei (R. Kahmann, J. Schirawski and G. Bakkeren, unpublished) and forage grasses. Smut symptoms are characterized by the for- that will allow whole genome comparison of these related smut mation of fruiting structures containing black masses of teliospores species and is expected to lead to insights into the determinants that give the infected tissue a ‘‘sooty” or ‘‘smutted” appearance. of host selection and symptom formation. While most smut fungi develop sexual spores exclusively in the Common to all smut species investigated so far is their need to inflorescence and symptoms are visible only late in the infection undergo a successful mating reaction to form dikaryotic hyphae upon heading, Ustilago maydis, the corn smut pathogen, is a notable before being able to infect a host plant. On the other hand, the exception that can induce tumors, in which the fungal spores de- infectious dikaryon requires a host for proliferation and for the for- velop, on all above-ground parts of the plant. Since fungal biomass mation of sexual spores. Thus, sex and pathogenesis are intimately usually develops in inflorescences of infected plants, considerable intertwined in these species. yield reductions are suffered. Therefore, it is not surprising that For mating to occur, two haploid cells of different mating-type smuts have been the subject of intense study for the last century. need to recognize each other and fuse to form the infectious dikar- However, molecular insight on mating systems and their function yon. Mating is regulated by two loci, a and b, which harbor con- has so far only been obtained for three smut species, the corn served genes. At the a locus, these genes encode pheromones and smuts U. maydis and Sporisorium reilianum and the barley smut pheromone receptors while at the b locus two subunits of a hete- Ustilago hordei. All three species show a close phylogenetic rela- rodimeric transcription factor are encoded. In S. reilianum and tionship (Bakkeren et al., 2000) that is reflected by the ease with U. maydis that have a tetrapolar mating system, these genetic loci which molecular tools developed for one species can be transferred segregate independently, while in bipolar species, such as to the other species. The elucidation of the complete genome se- U. hordei, the a and b loci are linked and MAT segregates as one lo- quence of U. maydis (Kämper et al., 2006) has enormously ad- cus. Despite the similarity in gene function and sequence, in the three smuts that have been analyzed at a molecular level (U. may- * Corresponding author. Fax: +49 6421 178609. dis, U. hordei and S. reilianum), the mating-type loci differ substan- E-mail address: [email protected] (J. Schirawski). tially in locus structure. These differences will be reviewed to cover

1087-1845/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.fgb.2008.04.005 S16 G. Bakkeren et al. / Fungal Genetics and Biology 45 (2008) S15–S21 function, structure and evolution of mating complexes in compar- different mating-type alleles are known (Bakkeren and Kronstad, ison to a few more distantly related fungal basidiomycete species 1994), five different b alleles have been described in S. reilianum for which molecular data exist. (Schirawski et al., 2005) and at least 19 exist in U. maydis (J. Kämper and R. Kahmann, unpublished). Interestingly, both bE 2. Structure of mating-type loci and bW mating-type genes are present in the genome of the Usti- laginomycete Malassezia globosa (Table 1), a human pathogenic The b mating-type genes encode two subunits of a homeodo- fungus involved in dandruff disease, for which no sexual cycle is main (HD) transcription factor, consisting of an HD1 class and an known (Xu et al., 2007). Even the distantly related, opportunistic HD2-class protein. In general, the HD1 and HD2 proteins are not human pathogen, the Tremellomycete Cryptococcus neoformans related to each other in primary sequence; however, they share a (Table 1), carries HD1- and HD2-encoding genes within the mat- common domain organization. The N-terminal regions of the HD ing-type region (Loftus et al., 2005). For C. neoformans, two mating proteins contain the highest degree of variation when different al- types are known but in this species each mating type has retained leles are compared and are thus designated as variable regions, only one of the b mating-type genes, with the HD1 protein being while the C-terminal regions of the proteins, including the homeo- specific for the MATa mating type and the HD2 protein being pres- domains, are highly conserved (Gillissen et al., 1992; Kronstad and ent only in the MATa mating-type region (Fig. 1; Fraser et al., 2004; Leong, 1990; Schulz et al., 1990). The tetrapolar species U. maydis Lengeler et al., 2002). and S. reilianum as well as the bipolar species U. hordei possess In addition to the b mating-type complexes, smut fungi contain one divergently transcribed gene pair encoding the homeodomain genes necessary for cell–cell recognition located in the a mating- proteins bE (HD1) and bW (HD2; Fig. 1). For the b mating-type type loci, encoding pheromones and pheromone receptors. The genes of U. maydis, S. reilianum and the U. hordei MAT-1 locus, gene detailed structure of these loci has been determined for both a order, orientation, as well as the genomic context are conserved alleles of U. maydis, for all three a alleles of S. reilianum and (Fig. 1). The orientation of the divergently transcribed b gene pair for the MAT-1 allele of U. hordei (Fig. 1). For the MAT-2 allele of seems inverted in the U. hordei MAT-2 locus. In the absence of U. hordei only partial information is available (Fig. 1). additional sequence information for the MAT-2 locus of U. hordei, Both U. maydis and U. hordei have two alleles of an a mating sys- it cannot be assessed, whether this gene pair is part of an inversion tem with one pheromone receptor (pra) and one functional phero- covering a large genomic context. While from U. hordei two mone gene (mfa) per locus (Anderson et al., 1999; Bakkeren and

Fig. 1. Genetic organization of the mating-type loci of selected basidiomycetes. Genes are indicated by arrows with the arrow denoting direction of transcription. Related genes are denoted by the same color and respective gene functions are explained in the lower part of the Figure. Ã indicates that the relative order and orientation of these genes has not been determined. In the tetrapolar species U. maydis and S. reilianum the a and b specific sequences reside on different chromosomes while they are linked by spacer regions (which are not drawn to scale and whose length is indicated) in the bipolar species U. hordei and C. neoformans, as well as in M. globosa. The black bars on top of the figure indicate the regions of the b locus, which covers the two homeodomain protein genes bE and bW, and the a locus (that expands to different length in the different loci, indicated by a broken line) from the lba gene to the rba gene. Sequence information was obtained from the following Accession Nos. AF043940, AM118080, AAC- P01000083, AACP01000013, AJ884588, AJ884583, AJ884590, AJ884585, AJ884589, AJ884584, U37796, M84182, AF184070, AF184069, Z18531, AAYY01000003, AF542530, and AF542531. G. Bakkeren et al. / Fungal Genetics and Biology 45 (2008) S15–S21 S17

Table 1 Taxonomic information on the basidiomycete fungi compared

Class Species Mating system Reference Agaricomycotina Agaricomycetes Coprinopsis cinerea (syn. Coprinus cinereus) Tetrapolar Casselton and Kües (2007) Pholiota nameko Bipolar Aimi et al. (2005), Ratanatragooldacha et al. (2002) Pleurotus djamor Tetrapolar James et al. (2004) Schizophyllum commune Tetrapolar Fowler and Vailllancourt (2007), Kothe (1996) Laccaria bicolor Tetrapolar Martin et al. (2008) Tremellomycetes Filobasidiella (syn. Cryptococcus) neoformans; Cryptococcus gattii Bipolar Fraser et al. (2004), Lengeler et al. (2002)

Ustilaginomycotina Ustilaginomycetes Malassezia globosa Bipolar Xu et al. (2007) Sporisorium reilianum Tetrapolar Schirawski et al. (2005) Ustilago hordei Bipolar Anderson et al. (1999), Bakkeren et al. (2006) Ustilago maydis Tetrapolar Kämper et al. (2006)

Kronstad, 1996; Bölker et al., 1992). The sequences of the phero- Overall, sequence analysis and comparison of the mating-type mone and receptor genes in the a2 locus differ from those in the regions of tetrapolar and bipolar smut fungi revealed that they a1 locus. Additionally, sequences immediately surrounding the are not fundamentally different. Both bipolar and tetrapolar smuts pheromone and receptor genes are unrelated between different a and related species contain the genes for the a and b mating-type mating types. Thus, the a1 locus of U. maydis encompasses a 4-kb complexes. In the tetrapolar species U. maydis and S. reilianum, the region, while the a2 locus extends over 8 kb and contains two addi- a and b mating-type genes reside on different chromosomes and tional a2-specific genes, rga2 and lga2, with a possible function in therefore segregate independently during meiosis giving rise to uniparental mitochondrial inheritance (Bortfeld et al., 2004). progeny with four different mating types. In contrast, the The a1 and a2 loci of S. reilianum are syntenic to the a1 and a2 mating-type complexes of the bipolar species U. hordei and C. neo- loci of U. maydis, respectively, with conservation of gene content, formans are encoded on the same chromosome and in a recombi- order, position and genomic context (Fig. 1). However, in S. reilia- nation-suppressed region ensuring genetic linkage (Bakkeren num, both loci contain an additional pheromone gene that allows et al., 2006; Bakkeren and Kronstad, 1994; Fraser et al., 2004). recognition of S. reilianum strains of a3 mating type (Schirawski The recently published genome of the dandruff-causing fungus et al., 2005). Correspondingly, the a3 mating-type locus of S. reilia- M. globosa reveals an average of 50% amino acid similarity with num also contains one pheromone receptor and two pheromone predicted proteins in U. maydis, although gene synteny among genes, one each for recognition of a1 and a2 mating partners (Schi- these species is only 15% for adjacent genes and <3% for groups lar- rawski et al., 2005). ger than two genes (Xu et al., 2007). In M. globosa, the identified a Interestingly, the genome of the anamorphic species M. globosa and b-like mating-type gene sequences were shown to reside on a also contains one pheromone and one pheromone receptor gene 173-kb long contig, which possibly defines a MAT locus. This would (Xu et al., 2007), raising the interesting possibility that this species suggest the existence of a bipolar mating system in M. globosa sim- is mating competent. Pheromone and pheromone receptor genes ilar to the one in U. hordei if this species is indeed mating compe- are also present in both the MATa and the MATa mating-type tent. The M. globosa MAT locus reveals 82 predicted genes but with regions of C. neoformans. Each region contains one pheromone the exception of the a and b mating-type proteins mentioned receptor gene and three pheromone genes (Fraser et al., 2004; above none seem to have orthologs in the U. hordei MAT-1 region Lengeler et al., 2002). However, while the two pheromone genes and only one gene is shared with the C. neoformans MATa locus. in the S. reilianum a locus encode different pheromones, the three In addition, the region between the a and b mating-type complexes pheromone genes of the C. neoformans MAT locus code for identical on the M. globosa genome does not contain the high number of pheromones (McClelland et al., 2002). repetitive elements found on U. hordei MAT-1. It is therefore unli- Comparison of the 527-kb large MAT-1 locus of U. hordei with kely that these distantly related bipolar species share a recent the genome of U. maydis revealed that more than 90% of the genes common ancestor, possibly illustrating evolutionary independent encoded in the U. hordei MAT-1 region have orthologs in the vicin- genetic linkage of a and b mating-type complexes (see below). ity of either the U. maydis a or the b mating-type genes (Bakkeren et al., 2006). Gene order is conserved over several regions of the 3. Functions of mating-type complexes U. hordei MAT-1 locus, but inversions are apparent and in some areas the U. hordei genes have homologs in unrelated regions of The initial recognition of two compatible sporidia is mediated the U. maydis genome. One of the hallmarks of the U. hordei via the pheromone/receptor system encoded by the a mating-type MAT-1 locus is an accumulation of repetitive elements covering locus. In its simplest form, the a locus consists of two different al- more than 50% of this large region. These repetitive elements are leles, each comprising genes for a peptide pheromone and a not found in the syntenic regions (or any other part of the genome) pheromone receptor. The pheromone genes within the a loci of of U. maydis. Many of the elements show long terminal repeat (LTR) U. maydis, U. hordei and S. reilianum encode small pheromone structures reminiscent of (retro-) transposable elements, some of precursors of approximately 40 amino acids that are posttransla- which are of the copia and gypsy type (Bakkeren et al., 2006). Many tionally modified by isoprenylation and carboxymethylation at of the classes of elements are scattered over the U. hordei genome the C-terminal cysteine residue, and by cleavage of a large part but certain classes seem to have accumulated disproportionately in of the N-terminus, resulting in mature peptides of 9–14 amino and near the MAT-1 region. Specifically, the repetitive elements are acids (Kosted et al., 2000; Schirawski et al., 2005; Spellig et al., often found in-between syntenous gene-rich stretches and border- 1994). The mature pheromones are secreted as small lipopeptide ing inversions in the MAT-1 locus, suggesting their involvement in molecules, and removal of the lipid moiety drastically reduces genome rearrangement during evolution (Bakkeren et al., 2006). their activity (Spellig et al., 1994). In S. reilianum, one of the two S18 G. Bakkeren et al. / Fungal Genetics and Biology 45 (2008) S15–S21 pheromones produced by a1 strains is identical to one of the pher- The formation of the heterodimeric bE/bW complex is the sole omones of a2 strains, while the other is identical to one of the determinant for the initiation of pathogenic development in U. pheromones of a3 strains (Schirawski et al., 2005). Thus, the six maydis. This has been shown by the construction of haploid strains pheromone genes of S. reilianum give rise to only three different carrying hybrid bE1/bW2 alleles. These strains are solopathogenic pheromones with one each for the recognition of each of the three and can infect the host without the need of a mating partner a mating types. (Bölker et al., 1995). Pheromone recognition by the cognate pheromone receptor The bE/bW heterodimer binds to a conserved homeodomain leads to a series of events that have been most extensively ana- binding motif (b binding site, bbs) in the promoter regions of lyzed in U. maydis. Exposure of U. maydis cells to pheromone from b-responsive genes. Such directly b-regulated genes were desig- a compatible mating partner results in a G2 cell cycle arrest and nated as class I genes, while class II genes are indirectly regulated the formation of conjugation tubes (Garcia-Muse et al., 2003; via a b-dependent regulatory cascade. Using the Affymetrix micro- Spellig et al., 1994). Upon binding of the pheromone to a compat- arrays, profiling of gene expression upon bE/bW induction within a ible receptor, transmission of the pheromone signal is mediated by 12 h time course has led to the identification of approximately a mitogen-activated protein kinase (MAPK) module that has been 350 b-responsive genes in U. maydis (M. Scherer and J. Kämper, shown to be essential for both mating and virulence (Müller unpublished). The bioinformatic analysis of these genes allowed et al., 2003). The module consists of the MAPK kinase kinase visualizing the processes controlled by the b locus. Among the Kpp4/Ubc4 (Andrews et al., 2000; Müller et al., 2003), the MAPK b-regulated genes in U. maydis are several predicted to be involved kinase Fuz7/Ubc5 (Andrews et al., 2000; Banuett and Herskowitz, in the synthesis and modification of cell wall components, suggest- 1994), and the MAPK Kpp2/Ubc2 (Mayorga and Gold, 2001; Müller ing that the cell walls of the yeast-like sporidia and of the filamen- et al., 1999). In addition, the mating response requires active pro- tous dikaryon are dissimilar in their overall composition. A large tein kinase A (PKA) signalling, which most likely integrates both number of b-regulated genes affect cell cycle, mitosis and DNA rep- nutritional and environmental cues (Feldbrügge et al., 2004). How- lication, consistent with the observed cell cycle block. Several ever, the connection between the pheromone receptors and the b-regulated genes with potential regulatory functions have been MAPK and PKA pathways is currently unknown. Both the MAPK shown to be required for pathogenic development: biz1 (Flor-Parra and the PKA pathways regulate the activity of Prf1, the key tran- et al., 2006) and rbf1 (Scherer et al., 2006), encoding two zinc finger scription factor for the induction of pheromone-responsive genes. transcription factors, as well as hdp2 and clp1 (Scherer et al., 2006), Prf1 is an HMG-box transcription factor that binds to specific pher- encoding a homeodomain transcription factor and a protein with- omone response elements (PREs) in the promoter regions of a-reg- out recognizable structural motifs, respectively. Rbf1 is required ulated genes (Hartmann et al., 1996; Urban et al., 1996). Among for filamentous growth, cell cycle arrest and pathogenic develop- the a-regulated genes are both the pheromone and receptor genes, ment. This factor plays a crucial role in the expression of a large as well as the genes of the b mating-type locus (see below). fraction of the b-regulated genes and thus has a central role within The first draft sequence of the U. maydis genome served as a the b regulatory network: Expression of the transcription factors template for the construction of an Affymetrix oligonucleotide ar- Biz1, Hdp1 and Hdp2 is regulated by Rbf1 (Scherer et al., 2006). ray, covering about 90% of the currently predicted 6902 genes Similar to other basidiomycetes, the dikaryotic filament of (Kämper et al., 2006). The use of these microarrays has allowed U. maydis develops clamp cells that participate in the distribution the identification of 375 pheromone-regulated genes that could of nuclei during cell division (Scherer et al., 2006). In clp1 mutant be classified into subgroups implicated in different processes such strains, dikaryotic filaments penetrate the plant cuticle, but devel- as pheromone maturation, signal transduction, transcriptional reg- opment is stalled before the first mitotic division, and the clamp- ulation, and cell cycle progression (R. Kahmann and M. Feldbrügge, like structures are not formed. The induction of either bE/bW, unpublished). rbf1, biz1 or hdp1 leads to a switch to filamentous growth that is While the a locus mediates fusion of two cells, the b locus con- in each case accompanied by a cell cycle arrest. The induction of trols the events after fusion, including sexual and pathogenic clp1, on the other hand, strongly interferes with b-dependent gene development. Fusion of two compatible cells leads to the formation regulation, blocks b-dependent filament formation and b-depen- of a dikaryon that grows filamentously; however, at the initial dent cell cycle arrest (Scherer et al., 2006). The proposed role for stage only the apical cell of the hyphae is filled with cytoplasm, Clp1 is therefore the modulation of transcription factor functions while the distal part consists of visually empty sections. Similar during biotrophic growth of U. maydis. This is similar to the role to a-dependent pheromone signalling, the b locus mediates a G2 suggested for Clp1 protein function of C. cinerea that is thought cell cycle block that is released after the dikaryon has invaded to initiate clamp formation by repression of the clamp cell repres- the plant via a specialized infection structure, the appressorium. sor Pcc1, a potential HMG-box transcription factor (Kamada, 2002). Only after plant penetration is the "true" filament with multiple The genome sequence and the availability of microarrays will septated compartments formed. The b locus of U. maydis encodes greatly facilitate to unravel the complex regulatory network con- the two unrelated HD transcription factors bE and bW. These trolling pathogenic development in U. maydis. two proteins can form a heterodimeric complex, but only when the proteins are derived from different alleles (Gillissen et al., 4. Evolution of mating-type complexes 1992; Kämper et al., 1995), a situation which naturally occurs after mating of two compatible partners. The current model proposes 4.1. Pheromones that dimerization is achieved via a limited number of hydrophobic and polar interactions within the variable N-terminal regions of The three different pheromones of S. reilianum show only very the bE and bW proteins (Kämper et al., 1995; Yee and Kronstad, little sequence identity to each other. However, the pheromones 1998). This heterodimerization to form active transcription factors involved in recognition by the a2 receptors of U. hordei, U. maydis is the general principle for interaction between the b proteins of and S. reilianum are related in primary sequence, as are the three smut fungi. Functional dimer formation has even been observed pheromones recognized by the a1 receptors in the three species between b proteins from different species (Bakkeren and Kronstad, (Fig. 2; Schirawski et al., 2005). This indicates that these phero- 1993). However, natural interspecies hybridization will also de- mones share a common ancestor. The lack of sequence relatedness pend on recognition of the other species pheromone(s), and has of the S. reilianum pheromone involved in recognition of the third a been observed (see below; Bakkeren and Kronstad, 1996). mating-type might indicate that this pheromone was acquired by G. Bakkeren et al. / Fungal Genetics and Biology 45 (2008) S15–S21 S19

Fig. 2. Comparison of predicted unprocessed pheromones using ClustalX (version 1.81; Thompson et al., 1997). Molecular phylograms of protein homologies with bootstrap values in percent at the nodes (1000 replicates, values P50%). Phero- mone sequences were from [species (protein name, accession number)]: U. maydis (Um_mfa1, AAA99765; Um_mfa2, AAA99771), U. hordei (Uh_mfa1, AAC02682; Uh_mfa2, AAD56043), S. reilianum (Sr_mfa1.2, CAI59747; Sr_mfa1.3, CAI59748; Sr_mfa2.1, CAI59758; Sr_mfa2.3, CAI59754; Sr_mfa3.1, CAI59764; Sr_mfa3.2, CAI59762), M. globosa (Mg_mfa1, EDP44481), C. neoformans (Cn_mfa, AAN75621; Cn_mfalpha, AAG25675). The ba pheromone of S. commune (Szc_bap1_2, AAC49155) served as outgroup.

recombination after mating with an as yet unidentified species. In line with this idea is the fact that while the a1 and a2 mating-type loci of U. maydis are unrelated, the three a loci of S. reilianum show several stretches of high sequence conservation between different Fig. 3. Comparison of mating-type proteins in ClustalX (version 1.81; Thompson a alleles (Schirawski et al., 2005). This sequence conservation is et al., 1997). Molecular phylograms of protein homologies with bootstrap values in localized around the pheromone genes, which indicates that they percent at the nodes (1000 replicates, values P50%). These non-comprehensive were likely exchanged between loci by recombination. Interest- analyses serve to indicate groupings among some basidiomycetes, with a focus on ingly, the a2 locus of U. maydis contains a pheromone pseudogene the smuts. Only representative alleles were included. (A) Pheromone receptor protein sequences of basidiomycetes. Pheromone receptor sequences were C-ter- in a location similar to one of the two pheromone genes in the S. minally truncated to exclude the cytoplasmic tail and to optimise the alignment, reilianum a2 locus (Fig. 1; Urban et al., 1996). This could indicate and were from [species (protein name, accession number, number of amino acids)]: that the ancestral a mating-type locus had more than one active U. maydis (Um_pra1, P31302, 296; Um_pra2, P31303, 298), U. hordei (Uh_pra1, pheromone gene. During speciation these genes might have been CAJ41875, 296; Uh_pra2, AAD56044, 300), S. reilianum (Sr_pra1, CAI59749, 296; recombined or lost, giving rise to simple loci as found in U. maydis. Sr_pra2, CAI59755, 299), M. globosa (Mg_pra1, EDP44482, 298), C. neoformans (Cn_ste3a, AAN75624, 295; Cn_ste3alpha, AAN75724, 295), C. gattii (Cg_ste3a, The presence of such pseudogenes/remnants might signify active AAV28758, 295; Cg_ste3alpha, AAV28793, 295), C. cinerea (Ccin_rcb1, AAF01418, evolutionary processes warranting the analysis and comparison 293; Ccin_rcb2, AAQ96344, 298), P. nameko (Pnam_rcb1, BAE47138, 301), P. djamor of a mating-type loci from other smut fungi. (Pdja_ste3, AAS46748, 292), S. commune (Szc_bbr1, AAB41858, 293; Szc_bbr2, The sequence identity between the various pheromones pro- AAD35087, 300). STE2 of S. cerevisiae (Sc_ste2, AAD56044, 302) served as outgroup. (B) Homeodomain-containing protein sequences of basidiomycetes. Homeodo- duced by smut fungi and those of other basidiomycetes is low main-containing protein sequences were from [species (protein name, accession (Fig. 2). In general, specificity seems to have evolved along species number)]: U. maydis (Um_bE3, P22017; Um_bW1, XP_756725; Um_bW18, lines to safeguard proper mate selection. Indeed, intercompatibility CAF34004), U. hordei (Uh_bE1, CAA79218; Uh_bW1, CAA79219), S. reilianum (Sr_bE3, has been used as one of the defining characters of the species con- CAI59736; Sr_bW1, CAI59727; Sr_bW4, CAI59739), M. globosa (Mg_bE1, EDP44402; cept (Boidin, 1986). Nevertheless, promiscuity or cross-reactivity Mg_bW1, EDP44401), C. neoformanns (Cn_sxi1, AAN75718; Cn_sxi2, AAV98474), C. gattii (Cg_sxi1, AAV28797; Cg_sxi2, AAZ07733), S. commune (Szc_Aalpha-Z3, between pheromones and receptors from different species, leading AAB01369; Szc_Aalpha-Z4, AAB01372; Szc_Aalpha-Y1, P37936; Szc_Aalpha-Y3, to intercompatibility (though not necessarily fecundity), has been AAB01370; Szc_Aalpha-Y4, AAB01373), C. cinerea (Ccin_a2-1, CAA56131; Ccin_b2-1, reported. For example, mating has been observed between U. hor- CAA56132; Ccin_b1-4, AAD33326; Ccin_b1-9, AAD33323), P. nameko (Pnam_hox1, dei and Sporisorium scitamineum (formerly known as Ustilago sci- BAE47137), P. djamor (Pdja_HD1, AAS46737; Pdja_HD2, AAS46736), L. bicolor (Lbic_HD1, XP_001873385). STE12 of S. cerevisiae (Sc_ste12, P13574) served as taminea, the causative agent of sugarcane smut), and between U. outgroup. maydis and S. scitamineum (Bakkeren and Kronstad, 1996).

4.2. Pheromone receptors the pheromone receptors have evolved in concordance with the established phylogenetic placement of the species and genus they We compared a selection of pheromone receptors from the belong to (Table 1). For example, receptors from the smuts belong- three smuts U. maydis, U. hordei and S. reilianum, and from more ing to the Ustilaginomycotina seem to cluster, as do the ones from distantly related basidiomycete fungi, including M. globosa, C. neo- the higher mushrooms (Fig. 3A). When sequence information for formans and C. gattii, as well as the higher mushrooms Schizophyl- both receptor types are available, the two types seem to fall into lum commune, Pleurotus djamor, Coprinopsis cinerea, Laccaria bicolor two respective classes; i.e. Pra1-like or Pra2-like among the Usti- and Pholiota nameko (Table 1; Fig. 3A). This analysis suggests that laginomycotina and STE3a- and the STE3a-like for the higher S20 G. Bakkeren et al. / Fungal Genetics and Biology 45 (2008) S15–S21 mushrooms (Fig. 3A). In contrast to a combination of STE2- and for which the complete DNA sequence has not yet been deter- STE3-like receptors present in the ascomycetes, the basidiomyce- mined (Fig. 1; Lee et al., 1999), could also have contributed to de- tes seem to have retained only the STE3-like receptors (Fraser creased recombination. Lack of ‘‘purifying recombination” thought et al., 2007). These evolved into several allelic versions, allowing to reduce the deleterious effect of transposable element activity for discriminatory interaction among ancestral partners in their and possibly one advantage of sex, could also have allowed the early struggles for sexual procreation. While co-evolution between accumulation of repetitive elements. Sequencing would reveal receptor and pheromone likely has occurred on many occasions, whether the MAT-2 region also contains an accumulation of MAT- segmental duplications and mutation, as well as conversion/ 2-specific repeats. It is also possible that the lack of recombination recombination processes likely have contributed to the present at the MAT-region is caused by the presence of a centromere. One day allelic diversity within the mating-type genes of the higher of the repetitive elements identified in the MAT-1 region (Bakkeren mushrooms. The mechanism(s) by which this expansion was et al., 2006) displays similarity to the U. maydis retrotransposon achieved might have varied among species and genera during evo- HobS. For U. maydis it has been shown that mobile elements of lution. A simple biallelic system as found in U. maydis and U. hordei the HobS-type cluster around the presumed centromeres in one might represent the epitome of streamlining genomes in organ- location on each chromosome, and it is these regions that allow isms occupying a successful niche in nature. autonomous replication of plasmids in functional assays (Kämper et al., 2006). This could indicate that the MAT-1 region of U. hordei 4.3. Homeodomain-containing mating-type proteins has evolved to become a centromeric region after the presumed initial translocation event. It would be interesting to functionally A similar molecular comparison was performed on a selection determine whether this region (or parts of it) provides activity of homeodomain-containing mating-type proteins (Fig. 3B). In for autonomous replication. spite of only limited sequence identity, each protein clearly groups A scenario for the evolution toward bipolar mating systems, the to one of two classes, HD1 or HD2 (Fig. 3B; Stankis and Specht, genesis of ‘‘sex chromosomes”, was also presented for Cryptococcus 2007). Among HD1-harboring proteins are bE-like proteins from species (Fraser et al., 2004, 2007). Similarly to U. hordei, in these the smuts and M. globosa, SXI1 from C. gattii and C. neoformans, basidiomycetes, remnants of transposable elements as possible the HD1 protein of L. bicolor and P. djamor, the Z proteins from drivers of gene shuffling were found with over 5-fold ‘‘enrichment” S. commune, the b1 proteins from C. cinerea and the Hox1 protein of transposons in the MAT alleles (15% of the DNA sequence) of P. nameko (Fig. 3B). Proteins of the HD2-class include the smut compared to the genome at large. In another somewhat related bW series, SXI2 from C. gattii and C. neoformans, the Y proteins basidiomycete, the anther smut Microbotryum violaceum (a repre- from S. commune, the a2 and b2 proteins from C. cinerea and the sentative of the Pucciniomycotina, but previously considered a HD2 protein of P. djamor (Fig. 3B). Thus, in both the HD1 and the Ustilago species), many transposable elements have also been HD2 classes of homeodomain proteins, members of the smuts found, with an increased presence on the ‘‘sex chromosomes” har- and the higher mushrooms group together, and these groupings boring the mating-type loci (Hood, 2005). In this bipolar fungus, are supported by high bootstrap values. This points to a very an- the dimorphic ‘‘sex chromosomes” are also suppressed for recom- cient function of these homeodomain proteins indeed. bination (Hood et al., 2004). Two of the three major lineages of basidiomycetes, the smuts and the mushrooms (the rusts being the third group), display tet- 5. Summary and prospects rapolar mating systems. This is generally thought to indicate its likely ancient origin (Raper, 1966). Because many examples of The genome sequence of U. maydis has become a reference for bipolar systems are found in these lineages, such change from tet- the elucidation of many aspects of basidiomycete biology. Compar- rapolar ancestors must have occurred several times independently ison of its mating-type complex organization with that of other during evolution. This points to a selective advantage in certain related fungi has led to insight into the basic mechanism underpin- fungi that is possibly dependent on the ecological niche they occu- ning bipolar versus tetrapolar mating behavior, and into a possible py. Advantages of inbreeding versus outbreeding in conjunction evolution toward the genesis of a sex chromosome. To understand with niche formation, such as seen by specialized plant pathogens, the evolutionary processes shaping these essential sex determi- have been named to contribute to such evolutionary pressures nants and leading to speciation in smut fungi, the availability of (Aanen and Hoekstra, 2007). However, it is necessary to keep in complete sequences of mating-type loci of other smut (and related) mind that the terms tetrapolar and bipolar refer to a genetic species at critical branch points along the fungal tree of life would description of the respective mating systems, and do not imply be extremely beneficial. The analysis of the complete genome any molecular explanation for a tetrapolar or bipolar mating sequences of two more smut genomes (U. hordei and S. reilianum; behavior. The evolution from a tetrapolar to a bipolar mating R. Kahmann, J. Schirawski, G. Bakkeren, unpublished) is a step to- system could have involved different mechanisms, and physical ward this goal and will greatly advance our understanding of smut linkage of the two mating complexes is only one of them. Other biology and evolution. possible scenarios include the accumulation of mutations in either tetrapolar gene complex that lead to self-compatible interactions, or a gradual change in one factor assuming the function of the References other (Fraser et al., 2007; Raper, 1966). Within the smut fungi, the evolution from tetrapolar to bipolar Aanen, D.K., Hoekstra, R.F., 2007. Why sex is good: on fungi and beyond. In: mating systems could have involved a translocation event between Heitman, J. et al. (Eds.), Sex in Fungi, Molecular Determination and Evolutionary Implications. ASM Press, Washington, DC, pp. 527–534. the two chromosomes carrying the a and b loci in the progenitor of Aimi, T., Yoshida, R., Ishikawa, M., Bao, D., Kitamoto, Y., 2005. Identification and U. hordei, possibly aided or caused by transposable element activ- linkage mapping of the genes for the putative homeodomain protein (hox1) and ity. An accumulation of repetitive elements within the region sep- the putative pheromone receptor protein homologue (rcb1) in a bipolar basidiomycete, Pholiota nameko. Curr. Genet. 48, 184–194. arating the a and b mating-type genes could then have led to a Anderson, C.M., Willits, D.A., Kosted, P.J., Ford, E.J., Martinez-Espinoza, A.D., sufficiently diverged chromosome (chromosomal segment) to Sherwood, J.E., 1999. Molecular analysis of the pheromone and pheromone explain the observed suppression of recombination between receptor genes of Ustilago hordei. Gene 240, 89–97. Andrews, D.L., Egan, J.D., Mayorga, M.E., Gold, S.E., 2000. The Ustilago maydis ubc4 MAT-1 and MAT-2 (Lee et al., 1999). In addition, inversions found and ubc5 genes encode members of a MAP kinase cascade required for at the MAT-2 locus of U. hordei, which spans a 430-kb region but filamentous growth. Mol. Plant Microbe Interact. 13, 781–786. G. Bakkeren et al. / Fungal Genetics and Biology 45 (2008) S15–S21 S21

Bakkeren, G., Jiang, G., Warren, R.L., Butterfield, Y., Shin, H., Chiu, R., Linning, R., Kosted, P.J., Gerhardt, S.A., Anderson, C.M., Stierle, A., Sherwood, J.E., 2000. Schein, J., Lee, N., Hu, G., Kupfer, D.M., Tang, Y., Roe, B.A., Jones, S., Marra, M., Structural requirements for activity of the pheromones of Ustilago hordei. Kronstad, J.W., 2006. Mating factor linkage and genome evolution in Fungal Genet. Biol. 29, 107–117. basidiomycetous pathogens of cereals. Fungal Genet. Biol. 43, 655–666. Kothe, E., 1996. Tetrapolar fungal mating types: sexes by the thousands. FEMS Bakkeren, G., Kronstad, J.W., 1993. Conservation of the b mating-type gene complex Microbiol. Rev. 18, 65–87. among bipolar and tetrapolar smut fungi. Plant Cell 5, 123–136. Kronstad, J.W., Leong, S.A., 1990. The b mating-type locus of Ustilago maydis Bakkeren, G., Kronstad, J.W., 1994. Linkage of mating-type loci distinguishes bipolar contains variable and constant regions. Genes Dev. 4, 1384–1395. from tetrapolar mating in basidiomycetous smut fungi. Proc. Natl. Acad. Sci. Lee, N., Bakkeren, G., Wong, K., Sherwood, J.E., Kronstad, J.W., 1999. The mating-type USA 91, 7085–7089. and pathogenicity locus of the fungus Ustilago hordei spans a 500-kb region. Bakkeren, G., Kronstad, J.W., 1996. The pheromone cell signaling components of the Proc. Natl. Acad. Sci. USA 96, 15026–15031. Ustilago a mating-type loci determine intercompatibility between species. Lengeler, K.B., Fox, D.S., Fraser, J.A., Allen, A., Forrester, K., Dietrich, F.S., Heitman, J., Genetics 143, 1601–1613. 2002. Mating-type locus of Cryptococcus neoformans: a step in the evolution of Bakkeren, G., Kronstad, J.W., Lévesque, C.A., 2000. Comparison of AFLP fingerprints sex chromosomes. Eukaryot. Cell 1, 704–718. and ITS sequences as phylogenetic markers in Ustilaginomycetes. Mycologia 92, Loftus, B.J., Fung, E., Roncaglia, P., Rowley, D., Amedeo, P., Bruno, D., Vamathevan, J., 510–521. Miranda, M., Anderson, I.J., Fraser, J.A., Allen, J.E., Bosdet, I.E., Brent, M.R., Chiu, Banuett, F., Herskowitz, I., 1994. Identification of fuz7,aUstilago maydis MEK/ R., Doering, T.L., Donlin, M.J., D’Souza, C.A., Fox, D.S., Grinberg, V., Fu, J., MAPKK homolog required for a-locus-dependent and a-locus-independent Fukushima, M., Haas, B.J., Huang, J.C., Janbon, G., Jones, S.J., Koo, H.L., steps in the fungal life-cycle. Genes Dev. 8, 1367–1378. Krzywinski, M.I., Kwon-Chung, J.K., Lengeler, K.B., Maiti, R., Marra, M.A., Boidin, J., 1986. Intercompatibility and the species concept in the saprobic Marra, R.E., Mathewson, C.A., Mitchell, T.G., Pertea, M., Riggs, F.R., Salzberg, basidiomycotina. Mycotaxon 26, 319–336. S.L., Schein, J.E., Shvartsbeyn, A., Shin, H., Shumway, M., Specht, C.A., Suh, B.B., Bölker, M., Böhnert, H.U., Braun, K.H., Görl, J., Kahmann, R., 1995. Tagging Tenney, A., Utterback, T.R., Wickes, B.L., Wortman, J.R., Wye, N.H., Kronstad, J.W., pathogenicity genes in Ustilago maydis by restriction enzyme-mediated Lodge, J.K., Heitman, J., Davis, R.W., Fraser, C.M., Hyman, R.W., 2005. The integration (REMI). Mol. Gen. Genet. 248, 547–552. genome of the basidiomycetous yeast and human pathogen Cryptococcus Bölker, M., Urban, M., Kahmann, R., 1992. The a mating type locus of U. maydis neoformans. Science 307, 1321–1324. specifies cell signaling components. Cell 68, 441–450. Martin, F., Aerts, A., Ahren, D., Brun, A., Danchin, E.G., Duchaussoy, F., Gibon, J., Bortfeld, M., Auffarth, K., Kahmann, R., Basse, C.W., 2004. The Ustilago maydis a2 Kohler, A., Lindquist, E., Pereda, V., Salamov, A., Shapiro, H.J., Wuyts, J., Blaudez, mating-type locus genes lga2 and rga2 compromise pathogenicity in the absence D., Buee, M., Brokstein, P., Canback, B., Cohen, D., Courty, P.E., Coutinho, P.M., of the mitochondrial p32 family protein Mrb1. Plant Cell 16, 2233–2248. Delaruelle, C., Detter, J.C., Deveau, A., DiFazio, S., Duplessis, S., Fraissinet-Tachet, Casselton, L.A., Kües, U., 2007. The origin of multiple mating types in the model L., Lucic, E., Frey-Klett, P., Fourrey, C., Feussner, I., Gay, G., Grimwood, J., mushrooms Coprinus cinerea and Schizophyllum commune. In: Heitman, J. et al. Hoegger, P.J., Jain, P., Kilaru, S., Labbe, J., Lin, Y.C., Legue, V., Le Tacon, F., (Eds.), Sex in Fungi, Molecular Determination and Evolutionary Implications. Marmeisse, R., Melayah, D., Montanini, B., Muratet, M., Nehls, U., Niculita- ASM Press, Washington, DC, pp. 283–300. Hirzel, H., Oudot-Le Secq, M.P., Peter, M., Quesneville, H., Rajashekar, B., Reich, Feldbrügge, M., Kämper, J., Steinberg, G., Kahmann, R., 2004. Regulation of mating and M., Rouhier, N., Schmutz, J., Yin, T., Chalot, M., Henrissat, B., Kües, U., Lucas, S., pathogenic development in Ustilago maydis. Curr. Opin. Microbiol. 7, 666–672. Van de Peer, Y., Podila, G.K., Polle, A., Pukkila, P.J., Richardson, P.M., Rouze, P., Flor-Parra, I., Vranes, M., Kämper, J., Perez-Martin, J., 2006. Biz1, a zinc finger protein Sanders, I.R., Stajich, J.E., Tunlid, A., Tuskan, G., Grigoriev, I.V., 2008. The genome required for plant invasion by Ustilago maydis, regulates the levels of a mitotic of Laccaria bicolor provides insights into mycorrhizal symbiosis. Nature 452, cyclin. Plant Cell 18, 2369–2387. 88–92. Fowler, T.J., Vailllancourt, L.J., 2007. Pheromones and pheromone receptors in Mayorga, M.E., Gold, S.E., 2001. The ubc2 gene of Ustilago maydis encodes a putative Schizophyllum commune mate recognition: retrospective of a half-century of novel adaptor protein required for filamentous growth, pheromone response progress and a look ahead. In: Heitman, J. et al. (Eds.), Sex in Fungi, Molecular and virulence. Mol. Microbiol. 41, 1365–1379. Determination and Evolutionary Implications. ASM Press, Washington, DC, pp. McClelland, C.M., Fu, J., Woodlee, G.L., Seymour, T.S., Wickes, B.L., 2002. Isolation 301–316. and characterization of the Cryptococcus neoformans MATa pheromone gene. Fraser, J.A., Diezmann, S., Subaran, R.L., Allen, A., Lengeler, K.B., Dietrich, F.S., Genetics 160, 935–947. Heitman, J., 2004. Convergent evolution of chromosomal sex-determining Müller, P., Aichinger, C., Feldbrügge, M., Kahmann, R., 1999. The MAP kinase Kpp2 regions in the animal and fungal kingdoms. PLoS Biol. 2, e384. regulates mating and pathogenic development in Ustilago maydis. Mol. Fraser, J.A., Hsueh, Y.-P., Findley, K.M., Heitman, J., 2007. Evolution of the mating- Microbiol. 34, 1007–1017. type locus: the Basidiomycetes. In: Heitman, J. et al. (Eds.), Sex in Fungi, Müller, P., Weinzierl, G., Brachmann, A., Feldbrügge, M., Kahmann, R., 2003. Molecular Determination and Evolutionary Implications. ASM Press, Mating and pathogenic development of the smut fungus Ustilago maydis are Washington, DC, pp. 19–34. regulated by one mitogen-activated protein kinase cascade. Eukaryot. Cell 2, Garcia-Muse, T., Steinberg, G., Perez-Martin, J., 2003. Pheromone-induced G2 arrest 1187–1199. in the phytopathogenic fungus Ustilago maydis. Eukaryot. Cell 2, 494–500. Raper, J., 1966. Genetics of Sexuality in Higher Fungi. The Ronald Press, New York, Gillissen, B., Bergemann, J., Sandmann, C., Schroeer, B., Bölker, M., Kahmann, R., NY. 1992. A two-component regulatory system for self/non-self recognition in Ratanatragooldacha, S., Hui, C., Kitamoto, Y., Kumata, A., 2002. The constitution of Ustilago maydis. Cell 68, 647–657. incompatibility factor and mating characteristics of spore isolates in a bipolar Hartmann, H.A., Kahmann, R., Bölker, M., 1996. The pheromone response factor mushroom, Pholiota nameko. Mycoscience 43, 113–117. coordinates filamentous growth and pathogenicity in Ustilago maydis. EMBO J. Scherer, M., Heimel, K., Starke, V., Kämper, J., 2006. The Clp1 protein is required for 15, 1632–1641. clamp formation and pathogenic development of Ustilago maydis. Plant Cell 18, Hood, M.E., 2005. Repetitive DNA in the automictic fungus Microbotryum violaceum. 2388–2401. Genetica 124, 1–10. Schirawski, J., Heinze, B., Wagenknecht, M., Kahmann, R., 2005. Mating type loci of Hood, M.E., Antonovics, J., Koskella, B., 2004. Shared forces of sex chromosome Sporisorium reilianum: novel pattern with three a and multiple b specificities. evolution in haploid-mating and diploid-mating organisms: Microbotryum Eukaryot. Cell 4, 1317–1327. violaceum and other model organisms. Genetics 168, 141–146. Schulz, B., Banuett, F., Dahl, M., Schlesinger, R., Schäfer, W., Martin, T., Herskowitz, I., James, T.Y., Liou, S.R., Vilgalys, R., 2004. The genetic structure and diversity of the A Kahmann, R., 1990. The b alleles of U. maydis, whose combinations program and B mating-type genes from the tropical oyster mushroom, Pleurotus djamor. pathogenic development, code for polypeptides containing a homeodomain- Fungal Genet. Biol. 41, 813–825. related motif. Cell 60, 295–306. Kamada, T., 2002. Molecular genetics of sexual development in the mushroom Spellig, T., Bölker, M., Lottspeich, F., Frank, R.W., Kahmann, R., 1994. Pheromones Coprinus cinereus. Bioessays 24, 449–459. trigger filamentous growth in Ustilago maydis. EMBO J. 13, 1620–1627. Kämper, J., Kahmann, R., Bölker, M., Ma, L.J., Brefort, T., Saville, B.J., Banuett, F., Stankis, M.M., Specht, C.A., 2007. Cloning the mating-type genes of Schizophyllum Kronstad, J.W., Gold, S.E., Müller, O., Perlin, M.H., Wösten, H.A., de Vries, R., Ruiz- commune: a historical perspective. In: Heitman, J. et al. (Eds.), Sex in Fungi, Herrera, J., Reynaga-Pena, C.G., Snetselaar, K., McCann, M., Perez-Martin, J., Molecular Determination and Evolutionary Implications. ASM Press, Feldbrügge, M., Basse, C.W., Steinberg, G., Ibeas, J.I., Holloman, W., Guzman, P., Washington, DC, pp. 267–282. Farman, M., Stajich, J.E., Sentandreu, R., Gonzalez-Prieto, J.M., Kennell, J.C., Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The Molina, L., Schirawski, J., Mendoza-Mendoza, A., Greilinger, D., Münch, K., CLUSTAL_X windows interface: flexible strategies for multiple sequence Rössel, N., Scherer, M., Vranes, M., Ladendorf, O., Vincon, V., Fuchs, U., Sandrock, alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882. B., Meng, S., Ho, E.C., Cahill, M.J., Boyce, K.J., Klose, J., Klosterman, S.J., Deelstra, Urban, M., Kahmann, R., Bölker, M., 1996. Identification of the pheromone response H.J., Ortiz-Castellanos, L., Li, W., Sanchez-Alonso, P., Schreier, P.H., Häuser-Hahn, element in Ustilago maydis. Mol. Gen. Genet. 251, 31–37. I., Vaupel, M., Koopmann, E., Friedrich, G., Voss, H., Schlüter, T., Margolis, J., Platt, Xu, J., Saunders, C.W., Hu, P., Grant, R.A., Boekhout, T., Kuramae, E.E., Kronstad, J.W., D., Swimmer, C., Gnirke, A., Chen, F., Vysotskaia, V., Mannhaupt, G., Güldener, U., Deangelis, Y.M., Reeder, N.L., Johnstone, K.R., Leland, M., Fieno, A.M., Begley, Münsterkötter, M., Haase, D., Oesterheld, M., Mewes, H.W., Mauceli, E.W., W.M., Sun, Y., Lacey, M.P., Chaudhary, T., Keough, T., Chu, L., Sears, R., Yuan, B., DeCaprio, D., Wade, C.M., Butler, J., Young, S., Jaffe, D.B., Calvo, S., Nusbaum, C., Dawson Jr, T.L., 2007. Dandruff-associated Malassezia genomes reveal Galagan, J., Birren, B.W., 2006. Insights from the genome of the biotrophic convergent and divergent virulence traits shared with plant and human fungal plant pathogen Ustilago maydis. Nature 444, 97–101. fungal pathogens. Proc. Natl. Acad. Sci. USA 104, 18730–18735. Kämper, J., Reichmann, M., Romeis, T., Bölker, M., Kahmann, R., 1995. Multiallelic Yee, A.R., Kronstad, J.W., 1998. Dual sets of chimeric alleles identify specificity recognition: nonself-dependent dimerization of the bE and bW homeodomain sequences for the bE and bW mating and pathogenicity genes of Ustilago maydis. proteins in Ustilago maydis. Cell 81, 73–83. Mol. Cell. Biol. 18, 221–232.