Functional Analysis of Mating Type Genes and Transcriptome Analysis during Fruiting Body Development of Botrytis cinerea

ABSTRACT Botrytis cinerea is a plant-pathogenic fungus producing apothecia as sexual fruiting bodies. To study the function of mating type (MAT) genes, single-gene deletion mutants were generated in both genes of the MAT1-1 locus and both genes of the MAT1-2 locus. Deletion mutants in two MAT genes were entirely sterile, while mutants in the other two MAT genes were able to develop stipes but never formed an apothecial disk. Little was known about the reprogramming of gene expression during apothecium development. We analyzed transcriptomes of sclerotia, three stages of apothecium development (primordia, stipes, and apothecial disks), and ascospores by RNA sequencing. Ten secondary metabolite gene clusters were upregulated at the onset of sexual development and downregulated in ascospores released from apothecia. Notably, more than 3,900 genes were differentially expressed in ascospores compared to mature apothecial disks. Among the genes that were upregulated in ascospores were numerous genes encoding virulence factors, which reveals that ascospores are transcriptionally primed for infection prior to their arrival on a host plant. Strikingly, the massive transcriptional changes at the initiation and completion of the sexual cycle often affected clusters of genes, rather than randomly dispersed genes. Thirty-five clusters of genes were jointly upregulated during the onset of sexual reproduction, while 99 clusters of genes (comprising >900 genes) were jointly downregulated in ascospores. These transcriptional changes coincided with changes in expression of genes encoding enzymes participating in chromatin organization, hinting at the occurrence of massive epigenetic regulation of gene expression during sexual reproduction.

biological function is in dispersal. Sexual spores of plant-pathogenic fungi can be dispersed by rain or wind, sometimes over long distances, and can subsequently serve as inoculum for new infections (3)(4)(5)(6). In ascomycete fungi, sexual compatibility is determined by two opposite mating type (MAT) loci, designated MAT1-1 and MAT1-2 (7)(8)(9). Ascomycetes can have either heterothallic or homothallic sexual reproduction systems. In a heterothallic species, mating occurs between two isolates with opposite mating types. The opposite MAT loci of ascomycetes are generally present at the same chromosomal location and therefore genetically allelic; however, they lack sequence similarity and are often referred to as "idiomorphs" rather than alleles (10). Each idiomorph contains at least one gene encoding a transcription factor. By convention, the MAT1-1 idiomorph encodes a protein with an ␣ domain, while MAT1-2 encodes a protein with a high-mobility-group box (HMG box) (9,11). Homothallic species are capable of self-fertilization, as they carry both core MAT genes (MAT1-1-1 and MAT1-2-1) in one genome (12)(13)(14)(15), usually next to each other, although the model fungus Aspergillus nidulans serves as an exception since its core MAT genes are located on separate chromosomes (16).
Botrytis cinerea is a heterothallic ascomycete fungus in the class Leotiomycetes, order Helotiales, family Sclerotiniaceae. A gapless community-annotated genome sequence is available (17). As a member of the Helotiales, B. cinerea develops a fruiting body called an apothecium (18,19), consisting of an exposed hymenium on top of a stipe. Apothecia are carpogenic from a sclerotium, a melanized survival structure that contains a reservoir of nutrients to support apothecium development (20) and usually resides in plant residues in the topsoil layer. By their emergence from sclerotia and their phototropic growth (21), apothecia of Sclerotiniaceae function as an elevated launch platform that facilitates the discharge of ascospores (22), which act as primary inoculum for plant infection, especially in early spring when the asexual conidia are scarce.
Several studies have focused on unraveling the functions of mating type genes in ascomycetes by targeted mutagenesis, primarily in model fungi (26)(27)(28)(29)(30)(31). Doughan and Rollins (32) recently described a functional analysis of genes in the MAT locus of S. sclerotiorum and reported that mutants in the MAT1-1-1, MAT1-1-5, and MAT1-2-1 genes were entirely sterile, while mutants in the MAT1-2-10 gene were delayed in carpogenic germination and formed apothecia with aberrant morphology. So far, the functions of the four genes in the B. cinerea MAT locus in apothecium development have not been studied. The genome sequence and gene annotations of B. cinerea (15,17) enabled the molecular dissection of apothecium development in B. cinerea, at both the transcriptional and the functional level.
This study aimed to identify the functions of B. cinerea MAT genes in apothecium development by targeted deletion and to perform a genome-wide transcriptome analysis during sexual reproduction using RNA sequencing (RNA-seq), which currently is the preferred method because of its sensitivity and quantitative accuracy (33)(34)(35) as well as its affordability. Expression profiling by RNA-seq was reported in numerous filamentous fungi (e.g., references 36 to 40); however, RNA-seq analyses in the context of sexual development are less abundant (41)(42)(43)(44)(45). Expression profiles observed during B. cinerea fruiting body development revealed concerted upregulation in the sexual ascospores of numerous genes, many of which are involved in interactions with plants, suggesting developmental priming for host invasion during sexual reproduction. Hints that epigenetic changes play a role in this regulation were obtained.

RESULTS AND DISCUSSION
Development of Botrytis cinerea apothecia. Fertilization of vernalized sclerotia of B. cinerea isolate SAS405 (MAT1-2) with microconidia of B. cinerea isolate SAS56 (MAT1-1), followed by incubation under appropriate conditions (18,19), resulted in the formation of apothecia with asci and ascospores. Crosses were effective reciprocally, i.e., regardless of which isolate was used as the maternal parent (sclerotia) or the paternal parent (microconidia). Sexual structures began to emerge from sclerotia at 20 to 30 days postfertilization (dpf) and reached maturity at 30 to 60 days postemergence. Apothecial development was divided into six stages ( Fig. 1): primordia emerging from sclerotium (stage 1, 20 to 30 dpf); primordia extending into stipes (stage 2, 24 to 35 dpf); fully extended stipes before onset of tip swelling (stage 3, 30 to 45 dpf); stipes with swollen tips before apothecial disk expansion (stage 4, 35 to 60 dpf); immature apothecial disk with a diameter of Ͻ3 mm and a pale color (stage 5, 40 to 70 dpf); and mature disk with a diameter of Ͼ5 mm and a light brown color and filled with asci containing eight ascospores (stage 6, 50 to 90 dpf). Functional analysis of the MAT genes by targeted deletion. In order to study the function of mating type genes MAT1-1-1, MAT1-1-5, MAT1-2-1, and MAT1-2-10 in apothecium development, deletion mutants were generated by replacing the coding region of each gene with a hygromycin resistance cassette in wild-type strains SAS56 and SAS405 (see Fig. S1 in the supplemental material). The MAT1-1-5 gene was also deleted in the genetic background of strain B05.10. Between three and six independent deletion mutants were obtained for each of the four MAT genes. All phenotypic analyses were carried out on three independent deletion mutants for each gene, and results were identical for independent mutants. To analyze whether the mutants had additional ectopic integrations, the copy number of the hygromycin resistance cassette was determined by quantitative PCR (qPCR) on genomic DNA and normalized to the single-copy housekeeping gene Bcrpl5. The deletion mutants had a single copy of the hph gene with the exception of one MAT1-2-10 mutant (out of three tested) which FIG 1 Different stages of sexual reproduction in Botrytis cinerea. Following fertilization of asexual resting structures (sclerotia), apothecium development is divided into six stages: primordia emerging (stage 1), primordia extending (stage 2), extended stipes before tip swelling (stage 3), stipes with swollen tips (stage 4), immature apothecium (stage 5), and mature apothecium with asci and ascospores (stage 6). Pure ascospores sampled from mature apothecial disks are shown in the right-hand image. The five samples used for transcriptome analyses are indicated at the top. contained~10 additional copies in an unknown location(s). Growth rates and morphology of asexual structures (mycelium, sclerotia, and conidia) of all ⌬MAT1-1-1, ⌬MAT1-1-5, ⌬MAT1-2-1, and ⌬MAT1-2-10 deletion mutants (including the ⌬MAT1-2-10 mutant with additional copies of the hph gene) were indistinguishable from those of the corresponding wild type. Reciprocal crosses were set up using two wild-type strains, SAS56 and SAS405, and four single-gene mutant strains (⌬MAT1-1-1, ⌬MAT1-1-5, ⌬MAT1-2-1, or ⌬MAT1-2-10) in all relevant combinations (Table 1). In the control cross of wild-type sclerotia (acting as maternal parent) fertilized with wild-type microconidia (as paternal parent), apothecia developed as described above. In contrast, apothecia never developed and there was no sign of outgrowth of primordia when crosses were performed between wild-type strain SAS405 and the ⌬MAT1-1-1 mutant or between wild-type strain SAS56 and the ⌬MAT1-2-1 mutant (Fig. 2). Failure to develop sexual structures was also observed in reciprocal crosses, indicating that the MAT1-1-1 and MAT1-2-1 genes are essential for the initiation of sexual development, in both maternal and paternal tissues. Loss of fertility was also reported for deletion mutants in the MAT1-1-1 or the MAT1-2-1 gene of S. sclerotiorum (a close relative of B. cinerea), though ascogonia were normally formed in these mutants (32). We did not investigate the presence of ascogonia in these B. cinerea mutants, since it is unknown when and where the fertilization occurs in the sclerotia during the 4-week incubation period following fertilization.
Sexual behavior of the ⌬MAT1-1-5 and ⌬MAT1-2-10 deletion mutants was distinct from that of ⌬MAT1-1-1 and ⌬MAT1-2-1 mutants (Fig. 2). Aberrant development of apothecia was observed in crosses between wild-type strain SAS405 and the ⌬MAT1-1-5 mutant, as well as in crosses between wild-type strain SAS56 and the ⌬MAT1-2-10 mutant. The emergence of primordia and development of stipes occurred similarly as in crosses between two wild-type strains; however, the stipes failed to develop into disks. Stipes did swell at the tip similarly to the wild type but failed to expand laterally. After several weeks of extended incubation, the stipes developed lobed, indented  structures at the top (Fig. 3A), clearly distinct from wild-type apothecial disks (Fig. 3B). In these defective structures, obtained in crosses with ⌬MAT1-1-5 and ⌬MAT1-2-10 mutants, the development of croziers, asci, and ascospores was never observed. The failure of mutant stipes to make the developmental switch to apothecial disks was observed for three independent knockout mutants for either gene, and it occurred likewise in reciprocal crosses. Crosses between a ⌬MAT1-1-5 mutant and a ⌬MAT1-2-10 mutant yielded an identical defective phenotype as crosses between either single mutant and their corresponding wild-type mating partner (Fig. 2), suggesting that these two genes jointly control the transition from stipe to disk development. The phenotype of these B. cinerea mutants differed from that of the corresponding mutants in S. sclerotiorum, in which the ⌬MAT1-1-5 mutant was entirely sterile and the ⌬MAT1-2-10 mutant showed severely delayed carpogenic germination and aberrant apothecium morphology (32). The transition from stipe to apothecial disk is typified by the formation of crozier cells in which karyogamy occurs (46,47), and the resulting diploid nuclei subsequently enter meiosis to eventually form eight ascospores in an ascus. We propose that in B. cinerea, MAT1-1-5 and MAT1-2-10 act as (possibly dimeric) regulators that directly or indirectly control the formation of crozier cells and that the absence of either of these proteins results in failure to proceed to karyogamy, which blocks further downstream processes such as apothecial disk expansion. It remains to be demonstrated whether the MAT1-1-5 and MAT1-2-10 proteins indeed physically interact and how they impact on transcription and sexual development.
Attempts were made to complement the phenotype of MAT gene mutants. With the exception of the MAT1-1-5 gene, complementation constructs in which wild-type genes, including flanking regulatory sequences, were cloned into vector pNR4 containing a nourseothricin resistance cassette (48) appeared unstable in Escherichia coli. Complementation could thus not be performed for the ⌬MAT1-1-1, ⌬MAT1-2-1, and ⌬MAT1-2-10 mutants. For the MAT1-1-5 gene, the complementation construct was transformed into a ⌬MAT1-1-5 mutant. Three independent transformants with an ectopic insertion of the intact MAT1-1-5 gene (including flanking sequences) in the ⌬MAT1-1-5 mutant background were tested in crosses, and all failed to show recovery of normal mating behavior. Ectopic integration of the intact MAT1-1-5 gene may have resulted in an inadequate level or temporal pattern of transcription, either because of the wrong chromatin context or because of the occurrence of meiotic silencing by unpaired DNA (MSUD) in an ectopic location, as was reported for Neurospora crassa crosses using a mutant strain that contains the matA gene in an ectopic location (49). The MAT1-1-5-complemented transformants were not analyzed further.
Sampling for analysis of gene expression during sexual development. B. cinerea apothecia are an attractive resource for transcriptome analysis compared to other ascocarps such as cleistothecia, perithecia, and pseudothecia. Apothecia develop from sclerotia in an aqueous environment (18,19), and the distinct developmental stages are in the millimeter-to-centimeter size range (Fig. 1). Hence, fairly pure tissue samples could easily be obtained. The fruiting body tissue samples were devoid of vegetative mycelium and sclerotial tissue. The use of laser microdissection to obtain ascocarp tissues, as performed in Sordaria macrospora (50), was thus not required for B. cinerea. Samples representing five different stages of sexual development were used for RNA extraction ( Fig. 1): (i) 4-week-old sclerotia prior to vernalization, (ii) primordia of stages 1 and 2 that were dissected from sclerotia, (iii) stipes of stages 3 and 4, (iv) apothecial disks of stages 5 and 6, and (v) pure ascospores obtained from mature apothecia. The samples were designated Scl, Apo12, Apo34, Apo56, and Asc, respectively. In addition, we sampled for RNA extraction the elongated stipes with a swelling (stage Apo34) from a cross between wild-type strain SAS405 and the ⌬MAT1-1-5 deletion mutant and from a cross between wild-type strain B05.10 and the ⌬MAT1-2-10 deletion mutant, which are defective in apothecial disk development (Fig. 2). Two biological replicates of each RNA sample were used for cDNA synthesis, quantitative real-time PCR (qRT-PCR), and RNA sequencing.
Expression of mating type genes during development of apothecia. First, qRT-PCR analysis was performed to determine the transcript levels of the MAT1-1-1, MAT1-1-5, MAT1-2-1, and MAT1-2-10 genes over three stages of apothecium development (Apo12, Apo34, and Apo56) in a wild-type cross between SAS56 and SAS405. Transcript levels were normalized to the constitutively expressed ␤-tubulin gene BctubA. Figure 4A shows that the transcript levels of all four MAT genes were lowest in primordia (Apo12), slightly increased in stipes (Apo34), and peaked in apothecial disks (Apo56). Transcript levels of all MAT genes were also quantified in stipes defective in apothecial disk development, obtained from crosses between a wild-type strain and a ⌬MAT1-1-5 or ⌬MAT1-2-10 deletion mutant (Fig. 4B). In ⌬MAT1-1-5 stipes, the transcript of the MAT1-1-5 gene was undetectable, as expected, but also transcript levels of the MAT1-1-1 and MAT1-2-1 genes were lower than in stipes from the cross between two wild-type isolates. In ⌬MAT1-2-10 stipes, the transcript of the MAT1-2-10 gene was undetectable, but also the transcript levels of the other three MAT genes were lower than in stipes from the cross between two wild-type isolates. As discussed above, it is assumed that the mutant stipes failed to undergo karyogamy and therefore remained blocked in the dikaryotic stage. The expression profiles of MAT genes in the mutant stipes indicate that the absence of one MAT gene in a nucleus derived from one of the mating partners affects transcript levels of MAT genes in the nucleus derived from the opposite partner. How transcriptional cross talk between opposite MAT alleles in separate nuclei is accomplished remains to be studied.
Whole-genome transcriptome analysis during development of apothecia. RNA sequencing was performed on the five samples, each in two biological replicates. The library size and mapping efficiency of reads on the B. cinerea genome are listed in Table S1. Over 94% of the 11,700 gene models in the B. cinerea genome (17) were represented by at least one read in each individual sample; in every sample, at least 79% of all genes had counts per million (CPM) of Ն1. The transcriptome of sample Apo56 was the most diverse, with Ͼ90% of the gene models displaying CPM of Ն1. This sample also contained the greatest diversity of cell types, as material taken from apothecial disks was composed of cell layers of excipulum and hymenium, of dikaryotic cells, and of diploid cells undergoing meiosis as well as asci and ascospores in different stages of development and maturity. Sclerotia and ascospore samples, on the other hand, represent more homogeneous tissue types; it thus is logical that their expression profile was less diverse.
Expression of B. cinerea orthologs to genes involved in sexual development of model fungi. Transcript levels of B. cinerea homologs to genes in Aspergillus nidulans and Sordaria macrospora that were previously listed as activators or repressors of sexual development in these fungi were examined (51,52). Blast searches identified 88 B. cinerea homologs for 70 of these A. nidulans and S. macrospora genes (Table S2). Expression of these B. cinerea genes was detectable in all samples tested, with CPM values ranging from 0.1 to 6,282. Table S2 presents the expression profiles of the B. cinerea genes, grouped on the basis of their proposed role (in A. nidulans) in perception of environmental signals, mating processes and signal transduction, transcription factors, endogenous physiological processes, and ascospore production and maturation, according to the work of Dyer and O'Gorman (51). Expression profiles of a subset of these genes are discussed below in more detail. The Bcwcl2 gene and the Bcvel1 gene are putative positive regulators of sexual reproduction, and their transcript levels increased 2-to 3-fold in B. cinerea primordia and mature stipes, compared to sclerotia and ascospores, while transcripts of Bcwcl1 and Bcvel2 genes were less abundant and barely changed during sexual development. Genes encoding components of the B. cinerea heterotrimeric G protein complex (G␣1, G␤, and G␥ but not G␣2 and G␣3) showed slight increases in transcript levels in apothecial tissues compared to sclerotia and ascospores. Furthermore, the G protein signaling regulator and a phosducin-like chaperone showed~8-fold increases in transcript levels at the transition from sclerotia to apothecial primordia. The transcript levels of transcription factor genes BcnsdD, BcfhpA, and BcnrdA/msnA (positive regulators of sexual reproduction in A. nidulans) increased strongly in all stages of apothecial development and decreased in ascospores, while expression patterns of BcflbC and BcstuA (negative regulators of sexual reproduction in A. nidulans) showed an almost inverse pattern. The expression profiles of B. cinerea genes orthologous to S. macrospora genes required for sexual development (52) were also examined ( Table S2). The transcript level of the gene encoding ATP citrate lyase subunit ACL1 increased 3-fold in the transition from sclerotia to primordia, remained constant in apothecia, and further increased in ascospores. The ortholog of S. macrospora pro4/leu1 (␤-isopropylmalate dehydrogenase) was expressed abundantly in primordia and stipes but at low levels in sclerotia, apothecial disks, and ascospores; BcnoxD, the ortholog of S. macrospora pro41, was highly expressed during all stages of apothecial development but at low levels in sclerotia and ascospores.
In summary, a number of B. cinerea genes which, based on functions described in A. nidulans, likely act as positive regulators of sexual development had high transcript levels in early stages of apothecium development, which dropped in mature apothecial disks and ascospores. Conversely, several B. cinerea genes that, based on functions described in A. nidulans, likely act as negative regulators of sexual development had high transcript levels in sclerotia and primordia, which dropped in later stages of apothecium development and ascospores. Although such patterns would be in agreement with proposed positive or negative regulatory roles of these genes in sexual development, validation of their role in sexual reproduction would require functional analysis. Such expression patterns were, however, not observed for all B. cinerea genes that (based on functions in A. nidulans) might act as positive or negative regulators in sexual development. In several cases, the expression level was constant or showed a distinct pattern. It should, in this context, be considered that the role of a particular gene as a regulator of sexual development does not necessarily imply that such a gene is differentially transcribed during sexual development.
Genes differentially expressed between consecutive stages of sexual development. Changes in gene expression during developmental transitions were analyzed by comparing expression levels between subsequent stages of apothecial development (Scl and Apo12, Apo12 and Apo34, Apo34 and Apo56, and Apo56 and Asc). Differentially expressed genes [adjusted P value of Ͻ0.05 and log 2 (fold change [FC]) of Ͼ2 or ϽϪ2) that were identified are recorded in Table 2 and listed in Tables S3 to S6. Pronounced changes in transcript levels were detected in the transition from sclerotia to primordia (Ͼ2,500 differentially expressed genes) and even more so between apothecial disks and ascospores, in which Ͼ3,900 genes were differentially expressed (i.e., one-third of the 11,700 genes in the B. cinerea genome). In contrast, only 75 genes were differentially expressed between primordia and stipes, whereas between stipes and apothecial disks, the number of differentially expressed genes was 594.
In order to examine whether the expression patterns during sexual development were distinct from asexual development, we determined pairwise correlations between CPM values in the five stages of sexual development reported here and three published data sets of asexual tissues, representing RNA from young germlings (conidia inoculated in liquid medium and grown for 12 h) (53) or from mycelium grown in polygalacturonate or glucose (54) (Fig. 5A). The correlations between samples were generally low (Ͻ0.5) with some exceptions. The transcriptomes in stages Apo12 and Apo34 were very similar to one another, and these tissues are morphologically alike. All other pairwise correlations between sexual and asexual stages were Ͻ0.75, with sclerotia displaying the most dissimilar expression profile.
The expression levels of all 3,084 genes that are differentially expressed during at least one of the five stages of sexual development were plotted in a heat map and compared to levels in the asexual tissues. The heat map (Fig. 5B) shows that the

Sex in Botrytis
® expression patterns of this gene set differed markedly between each of the sexual stages and the three nonsexual stages. Features of the sets of differentially expressed genes are discussed below. Gene ontology (GO) enrichment analysis was carried out on the four sets of differentially expressed genes (adjusted P value of Ͻ0.05, any fold change). Differentially expressed genes in the transition from sclerotia to primordia. The transition from sclerotia to primordia was associated with significantly increased transcript levels of 1,570 genes and reduced transcript levels of 1,011 genes (Table S3). A major change in secondary metabolism was observed. Among the upregulated genes (Table S3a) were 10 distinct secondary metabolite (SM) gene clusters, involved in production of polyketides (6 clusters), sesquiterpenes (2 clusters), and nonribosomal peptides (2 clusters), for all of which the SMs that they produce are unknown. Concomitantly, there was downregulation (Table S3b) of 5 SM clusters involved in the production of polyketides (2 clusters) and nonribosomal peptides (3 clusters). Among the upregulated genes (Table S3a) were also several regulatory genes, such as the MAT1-1-1 and MAT1-1-5 genes, two homeobox genes (Bchox4 and Bchox8 [55]), the Velvet complex subunit Bcvel4, the Vivid-like putative light sensor, the Medusa-like transcriptional regulator, and the light-responsive GATA-type transcription factor Bcltf1 (56); several genes involved in histone modification and DNA methylation (see below); hydrophobin genes Bhp1, Bhp2, and Bhl1 (57); five laccase genes; and 19 transporter genes of various classes. Among the downregulated genes (Table S3b) were, besides genes in five SM clusters mentioned above, regulatory genes such as HMG-box transcription factor-encoding gene Bcste11 (58) and the mitogen-activated protein (MAP) kinase gene Bmp3 (59), several genes involved in histone modification (see below), four glutathione S-transferases, the hydrophobin gene Bhp3, the melanin biosynthetic tetrahydroxynaphthalene reductase gene Bcbrn1 (60,61), the sclerotiumspecific lectin gene Bcssp1, and 28 transporters of various classes.
GO term enrichment analysis on the upregulated genes showed overrepresentation of (among others) the terms ribosome biogenesis, DNA replication, microtubule-based movement, fatty acid biosynthetic process, lipid biosynthetic process, oxidationreduction process, tricarboxylic acid cycle, and aerobic respiration (Table S3c). GO term enrichment analysis on the downregulated genes showed an overrepresentation of (among others) the terms oxidation-reduction process, hexose metabolic process, and transmembrane transport (Table S3d).
Differentially expressed genes in the transition from primordia to stipes. The transition from primordia (Apo12) to stipes (Apo34) was associated with significantly different transcript levels of 75 genes (Tables 2 and S4a and b). Among the 38 upregulated genes were 13 genes (34%) of unknown function that are specific to B. cinerea or the family Sclerotiniaceae. Among the 37 downregulated genes were a gene (Bcin04g02570) encoding a small secreted cysteine-rich protein (86-amino-acid mature protein with 13 Cys residues of which four occur in two Cys-Cys pairs) that does resemble, but not precisely match, the pattern for hydrophobin-like proteins; two genes encoding glycosyltransferases (CAZY GT2 and GT4 family proteins); and two genes in an SM gene cluster comprising polyketide synthase gene Bcpks15, of which the product is unknown. GO term enrichment analysis detected overrepresentation of the term response to stress in the set of upregulated genes (Table S4c). GO term enrichment analysis on the set of downregulated genes (Table S4d) showed an overrepresentation of (among others) the terms ribosome biogenesis, rRNA processing, and oxidation-reduction process.
GO term enrichment analysis on the set of upregulated genes showed an overrepresentation of (among others) the terms cellular chemical homeostasis, amino acid transmembrane transport, heme metabolic process, oxidation-reduction process, cellular biogenic amine metabolic process, dephosphorylation, carbohydrate metabolic process, autophagy, organic anion transport, phosphatidylinositol metabolic process, and aspartate family amino acid biosynthetic process (Table S5c). GO term enrichment analysis on the set of downregulated genes showed an overrepresentation of (among others) the terms oxidation-reduction process, carbohydrate metabolic process, potassium ion transport, sterol metabolic process, and response to oxidative stress (Table S5d).
Gene expression in mutant stipes blocked in transition to the apothecial disk. The transcriptomes of mutant stipes, obtained in crosses with ⌬MAT1-1-5 and ⌬MAT1-2-10 deletion mutants, were determined by RNA-seq and compared to one another and to the transcriptome of wild-type stipes at the equivalent stage (Apo34). A total of 1,310 genes were differentially expressed between mutant stipes from the SAS405 ϫ ⌬MAT1-1-5 mutant cross and stipes from a wild-type cross (Tables 3 and S6). Among this total set, 854 genes were downregulated and 456 genes were upregulated in the mutant stipes (blocked in transition to the apothecial disk), compared to stipes from the cross between two wild-type strains (Table S6a and b). In the cross between wild-type strain B05.10 and the ⌬MAT1-2-10 mutant, 985 genes were downregulated and 152 genes were upregulated in mutant stipes, compared to stipes in a cross between wild-type strains (Table S6c and d).
There was some overlap between the differentially expressed genes in both mutant stipes: compared to wild-type stipes, 87 genes were downregulated in both the SAS405 ϫ ⌬MAT1-1-5 cross and the B05.10 ϫ ⌬MAT1-2-10 cross, while 26 genes were upregulated in both mutant stipes (Table S6e and f). This list of 87 genes was further compared with the genes that are, in crosses between two wild-type isolates, upregulated in the transition from stipes (stage Apo34) to mature apothecial disks (stage Apo56). Among the genes that were not upregulated in both mutant stipes while they were upregulated in wild-type apothecia were one Zn 2 Cys 6 transcription factor-encoding gene, four major facilitator superfamily (MFS) transporter genes, the MAT1-1-1 and MAT1-2-1 genes, a HIT/MYND domain containing gene, and a G-protein-coupled receptor (BcG-PCR2), as well as 24 genes of unknown function that are specific either to Botrytis or to the family Sclerotiniaceae.
Differentially expressed genes in the transition from apothecial disks to ascospores. Between the apothecial disks (Apo56) and the ascospore sample, as many as 3,909 genes were differentially expressed, of which 37% were upregulated and 63% were downregulated (Tables 2 and S7a and b). Among the upregulated genes were 57 genes encoding ribosomal proteins and 24 other genes involved in ribosome biogenesis, as well as 10 translation initiation and elongation factor-encoding genes and 9 mitochondrial protein-encoding genes. The gene set also contains the Bcnop1 gene, encoding a histone glutamine methyltransferase; the heterotrimeric G protein ␤ subunit gene Bcgbl1; and five calcineurin-dependent genes (67) as well as the calcineurin regulator calcipressin gene (68). GO term enrichment analysis on the set of upregulated genes showed an overrepresentation of Ͼ200 terms (Table S7c), while GO term enrichment analysis on the downregulated genes showed an overrepresentation of 42 terms (Table S7d). A substantial number of genes upregulated in ascospores were of special interest, despite not being reflected in the GO enrichment analysis by lack of annotation in the "biological process" domain. These genes are discussed in more detail below.
Spatially clustered coregulation of transcription during sexual development. Using sliding windows of Ն12 genes, more than 150 cases were observed in which at least 50% of genes were significantly coregulated (either jointly up-or jointly downregulated) during the same developmental transition (Table 4). At the onset of sexual development, during the transition from sclerotia to primordia, 35 coregulated clusters were identified with an average length of 44.7 kb and containing on average 8 upregulated genes. Even more notably, there were 99 clusters of downregulated genes at the completion of sexual reproduction (the transition from apothecial disks to ascospores), with an average length of 56.5 kb and containing on average 9.3 downregulated genes (Table 4; Fig. 6). The largest of these clusters was 135 kb in length and contained as many as 24 genes. Altogether, these 99 clusters contained 923 genes (Table S8), representing 37% of the 2,486 genes that were downregulated during the transition from apothecial disks to ascospores. Of the 35 clusters that were upregulated in the transition from sclerotia to primordia, 23 (66%) overlapped at least partially with the 99 clusters that were downregulated in ascospores, which was especially notable in chromosomes 2, 3, and 13 (Fig. 6). The spatial pattern in transcriptional changes, especially during the onset and the end of sexual development, suggested the occurrence of changes in chromatin organization during the developmental transitions in apothecium development. The importance of histone acetylation and chromatin remodeling in meiotic recombination was reported in the fission yeast, Schizosaccharomyces pombe (82). Furthermore, the histone chaperone ASF1 in Sordaria macrospora was shown to be essential for the formation of mature perithecia (83). We examined the transcriptional profiles of 39 genes encoding histone-modifying enzyzmes or DNA methylases in the five stages of sexual development (Fig. 7). Half of these genes showed the highest expression level in the mature apothecia, in which meiosis was taking place. Notably, all four genes encoding C-5 cytosine-specific DNA methyltransferase and six of the 10 histone lysine deacetylase genes showed a clear peak of expression in this sample. In contrast, genes encoding two histone lysine methyltransferases, a histone lysine acetyltransferase, and two other histone lysine deacetylases, as well as a histone chaperone, displayed a strong peak of expression in the ascospores. Finally, one histone arginine methyltransferase gene was predominantly expressed in sclerotia. Chromatin immunoprecipitation sequencing (ChIP-seq) and bisulfite sequencing experiments in the different stages of B. cinerea apothecium development will be required to explore the changes in chromatin architecture during these developmental transitions. The roles of histonemodifying enzymes and DNA methylases in sexual reproduction can be validated by targeted mutagenesis (if such mutants are viable).
Conclusion. This study presents the first transcriptional analysis of sexual development in the Helotiales and a unique view on the transcriptome of pure ascospores of a filamentous ascomycete. The comparison of expression profiles of consecutive stages of sexual development revealed massive changes in transcript levels in the transition from sclerotia to primordia (the onset of sexual development) and even more so in the transition from mature disks to ascospores (completion of sexual reproduction). The number and nature of genes that were upregulated in the ascospores indicate that these spores are transcriptionally primed for the production of virulence factors prior to their first encounter with a host plant. B. cinerea ascospores thus do not land on plant tissues totally naive but possess a reservoir of virulence factors at their arrival, which facilitates a fast and effective invasion. Such an observation does not preclude that virulence genes might be further upregulated during germination of the ascospores on host tissue. It will be interesting to study whether the induction of virulence genes also occurs during development of asexual conidia, which are as infectious as ascospores and form important propagules for dispersal of B. cinerea to neighboring host plants. The present study also provided evidence for a clear spatial clustering in transcriptional expression data under a negative binomial distribution, weighting count values for replicate effects and sample dispersion. Resulting P values were corrected for multiple testing using the Benjamini-Hochberg method (90). The functional annotations of genes with a log 2 (FC) of Ͼ2 (or ϽϪ2) between two samples, at an adjusted P value of Ͻ0.05, were manually inspected. For visual inspection and comparison of expression profiles, the CPM expression values were scaled using a Z-score transformation: Z ϭ (X Ϫ )/.
Here, X is the CPM value of the respective gene at a particular stage, and and represent the mean and standard deviation for this gene over the five stages, respectively. Enrichment tests. To test for overrepresented gene ontology (GO) terms in each set of differentially expressed genes, GO enrichment analysis was performed. The total set of proteins from B. cinerea was annotated with GO terms using InterProScan v5.18.57 (91). Using the R package TopGO v2.18.0 (92), Fisher's exact tests were performed on the GO terms associated with the significantly up-and downregulated genes (adjusted P value of Ͻ0.05) between every consecutive pair of conditions. The resulting lists of significantly overrepresented GO terms were trimmed using REVIGO (93).
Differentially expressed gene clusters. To identify coexpressed gene clusters on the genome, a sliding window analysis was performed using CROC software (94). For the sliding window, a step size of one gene and a window size of 12 genes were applied, of which at least 6 genes must be either up-or downregulated for these to be considered for further analysis. Consecutively, on these windows a hypergeometric test was performed to test for a nonrandom distribution of the up-or downregulated genes, which were considered coexpressed gene clusters. The resulting P values were corrected using a Benjamini-Hochberg multiple testing correction method (88).