The MAT Locus Genes Play Different Roles in Sexual Reproduction and Pathogenesis in Fusarium graminearum

Sexual reproduction plays a critical role in the infection cycle of Fusarium graminearum because ascospores are the primary inoculum. As a homothallic ascomycete, F. graminearum contains both the MAT1-1 and MAT1-2-1 loci in the genome. To better understand their functions and regulations in sexual reproduction and pathogenesis, in this study we assayed the expression, interactions, and mutant phenotypes of individual MAT locus genes. Whereas the expression of MAT1-1-1 and MAT12-1 rapidly increased after perithecial induction and began to decline after 1 day post-perithecial induction (dpi), the expression of MAT1-1-2 and MAT1-1-3 peaked at 4 dpi. MAT1-1-2 and MAT1-1-3 had a similar expression profile and likely are controlled by a bidirectional promoter. Although none of the MAT locus genes were essential for perithecium formation, all of them were required for ascosporogenesis in self-crosses. In outcrosses, the mat11-1-2 and mat11-1-3 mutants were fertile but the mat1-1-1 and mat1-2-1 mutants displayed male- and female-specific defects, respectively. The mat1-2-1 mutant was reduced in FgSO expression and hyphal fusion. Mat1-1-2 interacted with all other MAT locus transcription factors, suggesting that they may form a protein complex during sexual reproduction. Mat1-1-1 also interacted with FgMcm1, which may play a role in controlling cell identity and sexual development. Interestingly, the mat1-1-1 and mat1-2-1 mutants were reduced in virulence in corn stalk rot assays although none of the MAT locus genes was important for wheat infection. The MAT1-1-1 and MAT1-2-1 genes may play a host-specific role in colonization of corn stalks.


Introduction
Fusarium graminearum (teleomorph Gibberella zeae) is a causal agent of Fusarium head blight (FHB) or scab of wheat, barley, and other small grains worldwide [1,2]. This fungal pathogen overwinters on plant debris and produces ascospores in the spring to initiate infection of flowering tissues of wheat and other host plants. Unlike most pathogenic fungi, sexual reproduction plays a critical role in the infection cycle of F. graminearum because ascospores are the primary inoculum [3]. Flowering wheat heads are susceptible to Fusarium infection from anthesis through the milk stages. Under favorable environmental conditions, FHB can cause severe yield losses. In addition, F. graminearum is a producer of deoxynivalenol (DON) and zearalenone. Infested grains are often contaminated with these harmful mycotoxins [4].
The functions of MAT locus genes have been characterized in several Sordariomycetes, including Neurospora crassa, Podospora anserina, and Sordaria macrospora [10,11]. In general, the MAT TFs are dispensable for vegetative growth and asexual reproduction. In most filamentous ascomycetes, the MAT1-1-1 and MAT1-2-1 orthologs are essential for mating and mating type specificity. In N. crassa, the matA-1 and mata-1 mutants were sterile but the matA-2 and matA-3 mutants were only slightly reduced in fertility [12].
Even the matA-2 matA-3 double mutant was still fertile although it was significantly reduced in perithecium and ascospore formation, suggesting that these genes may have overlapping functions in sexual development. In S. macrospora, a homothallic ascomycete closely related to N. crassa, the SmtA-1 (MAT1-1-1) and SmtA-3 (MAT1-1-3) deletion mutants produced the same number of perithecia and mature, viable ascospores as the wild type. In contrast, mature perithecia were not observed in the SmtA-2 (MAT1-1-2) deletion mutant, which was blocked in sexual development at the stage of early protoperithecium formation [10].
In Saccharomyces cerevisiae, only the MAT locus genes in the MAT locus are transcribed. The MAT genes in the HML and HMR loci are silenced [13,14]. The silenced HMLa or HMLa loci function as the donor sequences for gene conversion during mating type switching. In MATa cells, Mata1 interacts with Mcm1 to express a-specific genes and MATa2 inhibits the a-specific genes by binding with Mcm1/Tup1/Ssn6. In diploid cells, Mata1 interacts with Mata2 to inhibit haploid-specific genes, such as aor a-specific pheromone and pheromone receptor genes [15,16,17]. Mating type switching has not been observed in filamentous fungi although some homothallic fungi contain both MAT idiomorphs in the genome. In N. crassa, the interaction between matA-1 and mata-1 proteins was detected [18]. The matA-2 and matA-3 proteins weakly interact with each other in yeast two-hybrid assays [19]. In S. macrospora, SmtA-1 is known to interact with Mcm1 and Smta-1 but not with SmtA-2 or SmtA-3 [20,21]. In the heterothallic ascomycete P. anserina, no direct interactions among MAT locus TF proteins encoded by the opposite mating type loci were detected in yeast two-hybrid assays. Nevertheless, the mat-A specific transcription factors FMR1 and SMR2 that are orthologous to matA-1 and matA-3 of N. crassa interacted with each other [22].
The MAT locus genes are well conserved in F. graminearum. Because proper regulation of the expression, activation, and interaction of MAT locus genes is important for growth and differentiation in a homothallic fungus, in this study we functionally characterized individual MAT locus genes. All four MAT locus genes in F. graminearum had induced expression during sexual reproduction and were required for ascosporogenesis. However, none of them was essential for growth and infection of flowering wheat heads, which is similar to a recent report on functional characterization of MAT locus genes [9] that was published during the preparation of this manuscript. Although we failed to detect the direct interaction between Mat1-1-1 and Mat1-2, Mat1-1-2 interacts with all other MAT locus genes in yeast two hybrid-assays, suggesting that they may form protein complexes during sexual reproduction. Both MAT1-1-1 and MAT1-2-1 were rapidly induced during early stages of sexual development then gradually declined. The expression of MAT1-1-2 and MAT1-1-3 peaked at 4 days post-perithecial induction, which was later than the up-regulation of MAT1-1-1 and MAT1-2-1. MAT1-1-2 and MAT1-1-3 had similar expression profiles and they may have overlapping functions during late stages of perithecium development. In addition to the male-and femalespecific defects observed in the mat1-1-1 and mat1-2-1 deletion mutants, respectively, we found that MAT1-1-1 and MAT1-2-1 are important for colonization of corn stalks, suggesting that they play a host-specific role in pathogenesis.
To determine the expression profiles of MAT locus genes during sexual reproduction, we isolated RNA from aerial hyphae scrapped off 7-day-old carrot agar plates and fungal biomasses harvested from carrot agar cultures 1, 2, 4, 6, and 10 days postperithecial induction (dpi). In aerial hyphae harvested before perithecial induction, the expression of MAT locus genes was not detectable or relatively low (Fig. 1B), suggesting that their expression was suppressed during vegetative growth. The expression of MAT1-1-1 and MAT1-2-1 increased approximately 11and 10-fold, respectively at 1 dpi, and decreased after that (Fig. 1B). Therefore, these two MAT locus TF genes are likely important for earlier stages of sexual reproduction. For the MAT1-1-2 and MAT1-1-3 genes, their expression also increased after perithecial induction but peaked at 4 dpi (Fig. 1B). A similar expression profile of MAT1-1-2 and MAT1-1-3 is consistent with the fact that they share a common promoter (Fig. 1). These two genes may play more important roles in later stages than in earlier stages of perithecium development. When compared with the expression levels of other MAT TF genes, MAT1-2-1 had the highest level at 0, 1, 4, and 7 dpi (Fig. 1C). MAT1-1-1 had a higher expression level than MAT1-1-2 and MAT1-1-3 at 0, 1, 4, and 10 dpi (Fig. 1C). Similar trends were observed in an earlier study [9].

Expression and Subcellular Localization of MAT Locus Genes
To determine their expression and subcellular localization during sexual development, we generated the knock-in GFP fusion transformants for the four MAT locus TF genes of F. graminearum (Fig. S1). All the resulting GFP knock-in transformants (Table 1) were normal in vegetative growth and sexual reproduction, indicating that GFP insertion had no obvious effects on their functions. However, GFP signals were very weak ( Fig. 2A) or not detectable (Fig. S1) in vegetative hyphae or ascogenous tissues in all these knock-in transformants, including the MAT1-1-3-GFP strain ( Fig. 2A). Weak GFP signals were too faint to be distinguished from the auto-fluorescence of perithecium tissues and we failed to detect the localization of GFP signals in the nucleus in any of these GFP knock-in transformants.
In addition to the knock-in transformants, we also generated transformants expressing the P TrpC -MAT1-1-3-GFP and P TrpC -MAT1-2-1-GFP fusion constructs ectopically. In the P TrpC -MAT1-1-3-GFP transformant, GFP signals in the cytoplasm were still relatively weak but appeared to be stronger than the knock-in transformant (Fig. 2B). Localization of GFP-signals to the nucleus was observed in vegetative hyphae of the MAT1-2-1-GFP transformant (Fig. 2B). Nevertheless, we still failed to observe GFP signals that were stronger than the fluorescent background in perithecium tissues of the P TrpC -MAT1-1-3-GFP and P TrpC -MAT1-2-1-GFP transformants. It is possible that the GFP fusion proteins of these MAT locus gene are not stable or degraded rapidly during sexual reproduction. To determine the functions of MAT TFs, we generated gene replacement mutants deleted of individual MAT locus genes in the wild-type strain PH-1 (Fig. S2). For each gene, at least three independent knockout mutants were isolated ( Fig. S2; Table 1) and found to have the same phenotypes described below (data not shown). None of the resulting mutants (Table 1) had obvious defects in the growth rate or colony morphology (Fig. S3). Conidiation and conidium morphology also were normal in these mutants (Fig. S4). Nevertheless, all the MAT locus gene deletion mutants were defective in sexual reproduction in self-crosses. At 7 days post-perithecial induction, all the mutants produced darkpigmented perithecia that were smaller than the wild-type perithecia (Fig. 3A). These mutant perithecia were sterile and failed to produce ascospore cirrhi. When cracked open, asci or ascospores were not observed in perithecia formed by the mutants (Fig. 3B).
To further characterize the defects of MAT locus TF deletion mutants in sexual reproduction, thick-sections of the wild-type and mutant perithecia were examined under a light microscope (Fig. 3C). In contrast to ascogenous hyphae and asci at the base The Mating Type Genes in the Wheat Scab Fungus PLOS ONE | www.plosone.org of the wild-type perithecia, perithecia formed by the MAT locus gene deletion mutants appeared to be blocked in the development of ascogenous hyphae and tended to have thicker layers of perithecium wall (Fig. 3C). No asci or ascospores were observed. These results indicate that none of the MAT locus genes is essential for the initiation of perithecium development. However, the development of asci or ascogenous hyphae and enlargement of young perithecia require proper functions of all four MAT locus genes in self-crosses.
The mat1-1-1 and mat1-2-1 Deletion Mutants are Defective in Infection of Corn Stalks We also assayed DON production and virulence on flowering wheat heads with the MAT locus TF gene deletion mutants. In comparison with the wild type, none of these mutants had obvious changes in virulence in infection assays with flowering wheat heads, which was consistent with a previous report [9]. However, in infection assays with corn stalks, we noticed that the mat1-1-1 and mat1-2-1 deletion mutants, but not the mat1-1-2 and mat1-1-3 mutants, were reduced in virulence (Fig. 4). When the stalk rot areas were measured, the virulence of the mat1-1-1 and mat1-2-1 deletion mutants were reduced 66% and 77% in comparison with that of the wild type. In nature, corn also is a host plant for F. graminearum, which often produces abundant perithecia on corn stalks. The defects of the mat1-1-1 and mat1-2-1 deletion mutants in corn stalk rot assays suggest that the mating type locus may play a host-specific role in plant infection.

Mat1-1-2 Interacts with All other MAT Locus TF Genes in Yeast Two-hybrid Assays
Because of similar defects of the MAT TF mutants in sexual reproduction, it is likely that these MAT locus genes interact with each other and function in a protein complex (similar to the Mata-Mata interaction in S. cerevisiae) during sexual reproduction or ascosporogenesis inside perithecia. To test this hypothesis, we amplified the cDNA clones of all the MAT locus genes and assayed for their pair-wise interactions. In yeast two-hybrid assays, Mat1-1-1 interacted with Mat1-1-2 but not with Mat1-2 or Mat1-1-3. Mat1-1-2 and Mat1-1-3 interacted with each other and both of them interacted with Mat1-2 ( Fig. 5). Although it is surprising that the direct interaction between Mat1-1-1 and Mat1-2 was not observed, Mat1-1-2 interacted with all the other MAT locus TF proteins, which may bring Mat1-1-1 and Mat1-2 together in a protein complex in F. graminearum.
In attempt to further characterize the interactions among the MAT locus genes, we generated the BiFC constructs and transformed them in pairs into PH-1. In the resulting transformants (Table 1), we assayed for YFP signals in conidia, germlings, hyphae, and perithecium tissues. Unfortunately, we failed to observe obvious YFP signals in repeated experiments in all the resulting transformants, including the MAT1-1-2-YFPC MAT1-1-1-YFPN transformant (Table 1). In addition to the possibility that these MAT locus genes are expressed at a relatively low level, their YFP fusion proteins may be unstable or their interactions are too transient.

Mat1-1-1 Interacts with FgMcm1
In the budding yeast, Mcm1 interacts with Mata or Mata to regulate different processes in haploid or diploid cells [25,26]. We also assayed the interaction of different MAT locus genes with the orthologs of Mcm1 in F. graminearum. In yeast two-hybrid assays, only Mat1-1-1 interacted with FgMcm1 (Fig. 6). Although Mat1-1-2 interacted with all other MAT locus genes, it did not interact with FgMcm1.
In crosses with the mat1-1-2 and mat1-1-3 mutants, fertile perithecia were formed only when the mat1-1-1 or mat1-1 deletion mutant [6] was used as the female. When the mat1-1-1 mutant was used as the male, only sterile, small perithecia were formed,  suggesting that MAT1-1-1 is required for male fertility but dispensable for female fertility ( Table 2; Fig. 6). In contrast, MAT1-2-1 may play a more important role in female fertility than in male fertility because the mat1-2-1 deletion mutant displayed a female-specific defect in crosses with the mat1-1-2 and mat1-1-3 mutants ( Table 2; Fig. 6). In the crosses with the mat1-1-2 and mat1-1-3 mutants, normal, fertile perithecia were formed when the mat1-2-1 mutant was used as the male. When the mat1-2-1 mutant was used as the female, most of the perithecia, similar to those of the mat1-2-1 mutant formed in self-crosses, were small and sterile.

The FgSO Gene and Hyphal Fusion are Required for Sexual Reproduction
Because the SO gene is known to regulate hyphal fusion and female fertility in N. crassa [27,28], we generated the Fgso deletion mutant in F. graminearum by the split marker approach (Fig. S5). The Fgso mutant was sterile in self-crosses. When crosses with the mat1-1-1 and mat1-2-1 mutants, the Fgso mutant was female sterile but retained male fertility (Fig. 7A). Fertile perithecia and normal ascospores were observed only in crosses with the mat1-1-1 and mat1-2-1 mutants when the Fgso mutant functioned as the male. The Fgso1 (male) x mat1-2-1 (female) cross produced fewer fertile  perithecia than the Fgso1 (male) x mat1-1-1 (female) cross, which is consistent with the female-specific defects of the mat1-2-1 mutant.
Similar to the so mutant in N. crassa, the Fgso deletion mutant was defective in hyphal fusion (Fig. 7B). In comparison with the wild type, hyphal fusion was normal in the mat1-1-1 mutant (Fig. 7B). In the mat1-2-1 deletion mutant, hyphal fusion still occurred but at a reduced frequency (Fig. 7B). When assayed by qRT-PCR, the expression level of FgSO was significantly reduced in the mat1-2-1 mutant but not in the mat1-1-1 mutant (Fig. 7C). Therefore, the defects of the mat1-2-1 mutant in hyphal fusion may be directly related to the down-regulation of FgSO expression. MAT1-2-1 may be involved in regulating the expression of FgSO during sexual reproduction in F. graminearum.    Deletion of one MAT Gene Affects the Expression of Other MAT Locus Genes One possible explanation for similar phenotypes of the MAT locus gene deletion mutants is that deletion of any one of them may affect the expression of the others. To test his hypothesis, we isolated RNA from different mutants and assayed for the expression levels of individual MAT locus TF genes by qRT-PCR (Fig. 8). MAT1-1-1 expression was significantly reduced in all the MAT locus gene deletion mutants assayed in comparison with the wild type, which may be related to similar mating defects in self-crosses of these mutants because MAT1-1-1 is important for male fertility. The expression of MAT1-1-2 and MAT1-1-3 was up-regulated in the mat1-1-3 and mat1-1-2 mutants, respectively (Fig. 8). Thus, we concluded that deletion of one of these two genes increased the transcription of the other, which may be related to the fact that they share the same bidirectional promoter. Both MAT1-1-2 and MAT1-1-3 were down-regulated in the mat1-1-1 and mat1-2-1 mutants (Fig. 8), indicating that their expression may be positively regulated by MAT1-1-1 and MAT1-2-1, two MAT locus genes with up-regulated expression during earlier stages of sexual reproduction (Fig. 1B). Because their close proximity, it is possible that deletion of individual MAT locus genes affect the expression of nearby genes by altering chromatin structures or promoter activities. The other possibility is that lack of one mating type protein may affect the expression of other MAT genes because they may interact with each other to form a protein complex in F. graminearum.

Increased Expression of Pheromone or Pheromone Receptor Genes in the MAT Locus Gene Mutants
PRE1, PRE2, PPG1, and PPG2 are the pheromone and pheromone receptor genes in F. graminearum [29,30]. When assayed by qRT-PCR with RNA isolated from hyphae grown on 7-day-old carrot agar cultures, all the PRE and PPG genes were up-regulated in individual MAT locus gene deletion mutants (Fig. 9). In the mat1-1-1 mutant, the expression level of PRE1, PRE2, PPG1, and PPG2 was increased 7-, 9-, 10-, and 14-fold, respectively. The up-regulation of these pheromone and pheromone receptor genes varied between 1.7-and 3.8-fold in the other MAT locus gene deletion mutants (Fig. 9). These results indicate that MAT1-1-1, and possibly other MAT locus genes, negatively regulate the expression of the PRE1, PRE2, PPG1, and PPG2 genes during vegetative growth in F. graminearum.

Discussion
In F. graminearum, the MAT1-1-3 and MAT1-1-2 genes are likely under the control of the 259-bp bidirectional promoter and they had a similar expression profile during sexual reproduction (Fig. 1B). In N. crassa, the mata-1 gene encodes a protein with the core binding site of CTTTG [24]. The palindromic GAAAGCTTTC sequence in this bidirectional promoter region has two CTTTG-like sequences in different strands. MAT1-2-1, a homolog of mata-1, may have similar recognition site in F. graminearum and bind to this palindromic sequence to regulate the expression of MAT1-1-3 and MAT1-1-2. Other transcription factors, such as Mcm1, may also bind to this bidirectional promoter region to negatively regulate the expression of MAT1-1-3 and MAT1-1-2 during vegetative growth or positively regulate their expression during sexual reproduction. The 259-bp promoter region has two sequences similar to the consensus Mcm1-binding site CCNNNWWRGG (rulai.cshl.edu/cgi-bin/SCPD/getfactor?MCM1).
During the preparation of this manuscript, a paper on functional characterization of the expression and knockout mutants of individual MAT locus genes in F. graminearum was published [9]. In this study, fungal biomass used for RNA isolation were harvested by directly scraping off hyphae and fruiting bodies together from the surface of mating cultures at 1, 4, 7, and 10 days post-perithecial induction (dpi). We found that the expression of MAT1-1-1 and MAT1-2-1 peaked at 1 dpi but MAT1-1-2 and MAT1-1-3 had the highest expression level at 4 dpi. Kim and colleague conducted qRT-PCR analysis with RNA isolated from mating cultures at 2, 4, 6, 8, 10, and 12 dpi [9]. Although there were slight differences in the expression levels of individual MAT genes among different strains of F. graminearum, they found that the expression of MAT1-1-1 and MAT1-2-1 peaked at 2 dpi and MAT1-1-2 and MAT1-1-3 had the highest expression level at 4 dpi in PH1 [9], which is similar to the overall expression profiles of these four MAT TFs reported in this study. However, the exact fold changes for individual MAT TF genes were different between what was reported recently [9] and our observations, which may be related to different methods used to harvest fungal biomass for RNA isolation. We noticed that Kim and colleagues separated perithecia from hyphae [9]. Because perithecium development is not synchronous, our approach may have perithecia of different development stages mixed with undifferentiated hyphae. In addition, we used 0.1% Tween 20 instead of 2.5% Tween 60 to Figure 9. Effects of MAT locus gene deletion on the expression of pheromone and pheromone receptor genes. The expression levels of PRE1, PRE2, PPG1, and PPG2 in PH-1 or the mat1-1-3, mat1-1-2, mat1-1-1, and mat1-2-1 deletion mutants were assayed by qRT-PCR. The relative expression level of each gene in the wild type was arbitrarily set to 1. RNA samples were isolated from aerial hyphae harvested from 7day-old carrot agar cultures. Mean and standard error were calculated from three independent biological replicates. doi:10.1371/journal.pone.0066980.g009 induce sexual reproduction by pressing down aerial hyphae (Kim et al., 2012).
In this study, the expression levels of PPG1, PPG2, PRE1, and PRE2 in mutants deleted of individual MAT TF genes were assayed with RNA samples isolated from 7-day-old carrot agar cultures before perithecial induction. We found that the expression of all the PRE and PPG genes was significantly up-regulated in the mat1-1-1 mutant (Fig. 9). Their expression levels also were increased in the other MAT TF deletion mutants but not as significantly as in the mat1-1-1 mutant (Fig. 9). These results suggested that deletion of individual MAT TF genes may stimulate the expression of PPG and PRE genes in vegetative hyphae of F. graminearum. In F. graminearum, it has been reported that the PPG and PRE genes are not essential for sexual reproduction [30]. Whereas the expression of PPG2 increased at different stages of perithecial development in the mat1-1deletion mutant, up-regulation of PPG1, PRE1, and PRE2 was not observed in the mat1-1 and mat1-2 deletion mutants. Whereas Kim and colleagues [30] assayed the expression levels of the PPG and PRE genes in cultures after sexual induction, we assayed their expression in vegetative hyphae. Also, we used mutants deleted of individual MAT TF genes instead of the mat1-1 and mat1-2 locus deletion mutants.
One of our original goals was to determine the expression and subcellular localization of these four MAT locus genes during the switching from vegetative growth to sexual reproduction. Therefore, we generated knock-in GFP-fusion transformants in this study. Unfortunately, GFP signals were too faint or not detectable in the knock-in GFP fusion transformants of these MAT locus genes. The MAT1-1-3 knock-in transformant had the strongest GFP signals but we failed to detect fluorescence signals in the nucleus during sexual reproduction. One likely explanation is that their expression at the protein level was relatively low and perithecium tissues had a strong fluorescence background. It is also possible that MAT locus proteins only localize to the nucleus transiently. Interestingly, GFP signals were observed in the nucleus in vegetative hyphae and conidia but not in perithecium tissues of the P TrpC -MAT1-2-1-GFP transformant. In the P TrpC -MAT1-1-3-GFP, GFP signals in vegetative hyphae were stronger than the GFP knock-in transformant but were not localized to the nucleus (Fig. 2B).
Similar to what were reported [9], the mat1-1-1, mat1-1-2, and mat1-1-3 mutants produced smaller perithecia with thicker perithecium walls than the wild type but did not form asci or ascospores in self-crosses, indicating that all these four genes in the MAT loci are required for ascospore production. Unlike the mat1-2-1 mutant of GZ3639 [9], although it was reduced in perithecium formation, the mat1-2-1 mutant generated in this study still produced numerous small, sterile perithecia. Close examination revealed that ascogenous hyphal growth inside perithecia was blocked or defective in the mat1-2-1 and other MAT locus gene deletion mutants (Fig. 3).
In out-crosses, we found that the mat1-1-1 mutant displayed male-specific defects in mating although it was normal in female fertility. Kim and colleagues did not assay the defects of these MAT locus gene mutants in outcrosses [9]. To our knowledge, no other genes are known to be essential for male fertility in F. graminearum. In M. oryzae, the mcm1 mutant was reported to be male sterile and it was blocked in the production of microconidia [32]. F. graminearum, unlike many other Fusarium species, does not produce microconidia. It will be important to determine the role of MAT1-1-1 in male fertility.
Interestingly, the mat1-2-1 deletion mutant displayed a femalespecific defect in the production of normal perithecia and ascospores in out-crosses. Most of the perithecia produced in crosses with the mat1-2-1 mutant as the female were small and sterile. Although a few of them were fertile, they failed to produce cirrhi. Therefore, MAT1-2-1 is not essential but plays an important role in perithecium development, likely before the formation of dikaryotic hyphae or diploid cells. There are many mutants known to be defective in female fertility in F. graminearum. Most of them are blocked in the production of proto-perithecia, including the mgv1 and Gpmk1 mutants [33,34,35,36]. The MAT1-1-2 and MAT1-1-3 genes were dispensable for mating either as the male or female in out-crosses. It is possible that MAT1-1-2 and MAT1-1-3 are functionally redundant. It will be of interest to generate the mat1-1-2 mat1-1-3 double mutant and determine its defects in outcrosses with other strains.
In yeast, MATa interacts with MATa in diploid cells to suppress the expression of haploid-specific genes [14,25]. In F. graminearum, the dikaryotic and diploid stages are transient and occur only inside perithecia. In yeast two-hybrid assays, we observed the interaction of MAT1-1-2 with all other MAT locus genes. The interactions of MAT1-1-3 with MAT1-1-1 and MAT1-2-1 also were detected. To detect the interaction of different MAT locus genes by BiFC assays, we generated the transformants expressing different YFPC-and YFPN-fusion constructs (Table 1). Unfortunately, we failed to observe YFP signals in any of these transformants during vegetative growth or sexual reproduction. The expression levels of these MAT locus genes may be too low and fluorescence background may be too high in perithecium tissues. It is also possible that the interactions among the MAT locus TF genes may be too transient to be detected by BiFC assays.
In N. crassa, SO is important for hyphal fusion. The so mutant displayed pleiotropic defects in growth, conidiation, and sexual reproduction [27,28]. Orthologs of SO are well conserved in filamentous ascomycetes but it is absent in yeast and its function in plant pathogenic fungi is not clear. In F. graminearum, the Fgso mutant was defective in hyphal fusion and sterile in self-crosses. It was female sterile but male fertile in outcrosses. In infection assays, the Fgso1 mutant was significantly reduced in virulence. Because FgSO expression and hyphal fusion were reduced in the mat1-2-1 mutant, it is possible that MAT1-2-1 may directly or indirectly regulate the expression of FgSO in F. graminearum.
Interestingly, although the MAT locus gene deletion mutants were normal in wheat head infection, we noticed that the mat1-1-1 and mat1-2-1 mutants, but not the mat1-1-2 and mat1-1-3 mutants, were reduced in virulence in corn stalk rot assays. The same results were obtained in repeated experiments. The mat1-1-1 and mat1-2-1 mutants may be defective in adaptation to or colonization of corn stalks, which are substrates for F. graminearum to produce perithecia in the field. It will be important to further characterize the defects of the mat1-1-1 and mat1-2-1 mutants in corn stalk infection and determine the underlying mechanism. Corn stalks are known to accumulate diterpenoid phytoalexins [37]. The mat1-1-1 and mat1-2-1 mutants may have enhanced sensitivity to these phytoalexins or may be defective in overcoming other plant defensive responses in corn stalks.

Strains and Culture Conditions
The wild-type strain and mutants of F. graminearum used in this study are listed in Table 1. Cultures were routinely grown on PDA plates at 25uC [38,39]. Conidiation in 5-day-old CMC cultures and growth rate on PDA plates were measured as described [40,41]. For DNA extraction, vegetative hyphae were harvested from 2-day-old YEPD (1% yeast extract, 2% peptone, 2% glucose) cultures. For genetic crosses, aerial hyphae of 7-day-old carrot agar cultures were pressed down with 300 ml of sterile 0.1% Tween 20 or conidium suspensions (10 5 conidia/ml) of the male strains and incubated under black light [42]. Perithecium formation and cirrhi production were assayed after incubation at 25uC for 2 weeks. Protoplast preparation and fungal transformation were performed as described [33,43,44]. For transformation, hygromycin B (Calbiochem, La Jolla, CA) and geneticin (Sigma-Aldrich, St. Louis, MO) were added to the final concentration of 250 and 150 mg/ml, respectively [45].

Plant Infection Assays
For wheat head and corn stalk infection assays, freshly harvested conidia were re-suspended to 10 5 spores/ml in sterile distilled water. Flowering wheat heads of cultivar Xiaoyan22 were inoculated with 10 ml of conidium suspensions at the fifth spikelet from the base of the spike and scored for head blight symptoms as described [46,47]. Stalks of 8-week-old corn plants of cv. Pioneer 2375 were inoculated with toothpicks dipped in conidium suspensions as described [41,48]. Stalk rot symptoms were examined after splitting the stalks longitudinally along the inoculation site 14 dpi.

qRT-PCR Analysis
RNA samples were isolated from conidia, vegetative hyphae, and fungal biomass (hyphae and fruiting bodies) harvested from carrot agar cultures by scraping gently with a spatula with the TRIzol reagent (Invitrogen, Carlsbad, CA). For each experiment, at least three independent biological replicates were conducted. First-strand cDNA was synthesized with the Fermentas 1st cDNA synthesis kit (Hanover, MD) following the instructions provided by the manufacturer. For the internal control, the FgTUB2 betatubulin gene of F. graminearum was amplified with primers TubQF and TubQR [47].
Generation of the mat1-1-1, mat1-1-2, mat1-1-3, mat1-2-1, and Fgso Mutants The split-marker approach was used to generate the gene replacement constructs for the MAT locus and FgSO genes ( Fig. S3;  Fig. S5). Putative knockout mutants were identified by PCR and confirmed by Southern blot hybridizations to confirm the gene replacement event. All of the mutants generated in this study were preserved in 15% glycerol at 280uC.

Generation of GFP knock-in Transformants and GFP Fusion Constructs
A modified split-marker approach was used to generate the GFP knock-in transformants (Fig. S1). Primers 1F/2R and 3F/4R were used to amplify the flanking sequences of the stop codons of the target genes (Fig. S1). The GFP and geneticin resistance marker (Gen R ) fusion construct pGTP was generated in this study by overlapping PCR and used as the split marker for transformation. The resulting G418-resistant transformants were screened by PCR. In-frame fusions of GFP with MAT locus genes were then confirmed by sequencing analysis with PCR products.
To generate the MAT1-1-3-GFP and MAT1-2-1-GFP fusion constructs, PCR products containing the genomic fragments of the target genes were amplified and cloned into pFL2 [49] by the yeast gap repair approach [50]. All GFP fusion constructs were verified by sequencing analysis and transformed into protoplasts of the corresponding mutants. G418-resistant transformants harboring the transforming constructs were identified by PCR and confirmed by the presence of GFP signals.

Yeast Two-hybrid Assays
Protein-protein interactions were assayed with the Matchmaker yeast two-hybrid system (Clontech, Mountain View, CA). ORFs of the MAT1-1-2, MAT1-1-3, MAT1-2-1, and MCM1 were amplified from first-strand cDNA of PH-1 and cloned into pGBK7 (Clontech) as the bait constructs. For the MAT1-1-1, MAT1-2-1, and MCM1 genes, their ORFs were amplified and cloned into pGADT7 as the prey constructs. The resulting bait and prey vectors were co-transformed in pairs into yeast strain AH109 (Clontech). The Leu+ and Trp+ transformants were isolated and assayed for growth on SD-Trp-Leu-His medium and galactosidase activities with filter lift assays as described [51]. The positive and negative controls were provided in the Matchmaker library construction kit (Clontech).