Sclerotinia sclerotiorum Response to Long Exposure to Glucosinolate Hydrolysis Products by Transcriptomic Approach

ABSTRACT White mold disease, caused by the necrotrophic fungus Sclerotinia sclerotiorum, affects Brassica crops. Brassica crops produce a broad array of compounds, such as glucosinolates, which contribute to the defense against pathogens. From their hydrolysis, several products arise that have antimicrobial activity (GHPs) whose toxicity is structure dependent. S. sclerotiorum may overcome the toxic effect of moderate GHP concentrations after prolonged exposure to their action. Our objective was to identify the molecular mechanism underlying S. sclerotiorum response to long exposure to two chemically diverse GHPs: aliphatic GHP allyl-isothiocyanate (AITC) and indole GHP indol-3-carbinol (I3C). We found that the transcriptomic response is dependent on the type of GHP and on their initial target, involving cell membranes in the case of AITC or DNA in the case of I3C. Response mechanisms include the reorganization of chromatin, mediated by histone chaperones hip4 and cia1, ribosome synthesis controlled by the kinase-phosphatase pair aps1-ppn1, catabolism of proteins, ergosterol synthesis, and induction of detoxification systems. These mechanisms probably help S. sclerotiorum to grow and survive in an environment where GHPs are constantly produced by Brassica plants upon glucosinolate breakdown. IMPORTANCE Brassica species, including important vegetable crops, such as cabbage, cauliflower, or broccoli, or oil crops, such as rapeseed, produce specific chemical compounds useful to protect them against pests and pathogens. One of the most destructive Brassica diseases in temperate areas around the world is Sclerotinia stem rot, caused by the fungus Sclerotinia sclerotiorum. This is a generalist pathogen that causes disease over more than 400 plant species, being a serious threat to economically important crops worldwide, including potato, bean, soybean, and sunflower, among many others. Understanding the mechanisms utilized by pathogens to overcome specific plant defensive compounds can be useful to increase plant resistance. Our study demonstrated that Sclerotinia shows different adaptation mechanisms, including detoxification systems, to grow and survive when plant protective compounds are present.


RESULTS
Phenotypic effect of GHP treatments on S. sclerotiorum. S. sclerotiorum mycelia took more time to reach the end of the petri dish when they were treated with GHPs than control samples. This effect was clearer when samples were treated with I3C ( Fig. 1A). Once AITC and control sample mycelia reached the end of petri dishes, they started producing sclerotia on the edge. However, sclerotia from I3C-treated samples also were produced in the midpoint of petri dishes (Fig. 1B). 100 mM AITC provoked membrane permeabilization, and it was half that by 70% ethanol (Fig. 1C). No permeabilization effect was observed for I3C (Fig. 1D).
Transcriptomic analysis. Averages of 14,960,380, 16,415,912, and 10,776,800 tag sequences were obtained from the control and samples treated with AITC and I3C. To offer a general result overview, GO analysis with all of the annotated transcripts was conducted for each comparison (GHP versus control and AITC versus I3C) with GO-Tool2. We describe the most important features found under the biological process (BP) category (see Table S1 in the supplemental material).
AITC versus control. Transcript analysis resulted in 212 significantly enriched GO terms (P # 0.05 and 21 # log 2 fold change $ 1), from which 181 were induced and 31 were repressed. The most significant induced GO terms were related to mitosis and cell division (i.e., GO:1901989, GO:1901992, GO:1903047, and GO:0051301). Other categories were related to the organonitrogen compound metabolic process ( GO:1901564) and the small-molecule metabolic process (GO:0044281). Otherwise, there is a large set of significantly enriched terms associated with the positive regulation of the proteasomal protein catabolic process (GO:1901800 and GO:0032436). The first six most significantly repressed GO terms are related to toxin biosynthetic pathways (GO:1901376, GO:1901378, GO:0045461, GO:0045460, GO:0009403, and GO:0009404). Several terms were related to apoptosis and cell death (GO:0006915, GO:0012501, GO:0008219, and GO:0016265). Other significant terms included processes related to mitochondrial transmembrane transport (GO:1990542 and GO:0045041), mitochondrial DNA repair (GO:0043504), and maturation of LSU-rRNA (GO:0000463 and GO:0000470).
I3C versus control. GO analysis of I3C versus control treatment rendered 197 significant GO terms, from which 190 were repressed and 7 were induced. Significantly induced GO terms were mostly related to amino acid synthesis, including serine (GO:0009070 and GO:0009092), threonine (GO:0009088), and sulfur amino acids (GO:0000097). The ubiquitin-dependent protein catabolic process via the multivesicular body sorting pathway (GO:0043162) and glyoxylate metabolic process (GO:0046487) were also significant. A large part of significant GO repressed terms were related to translation and ribosome biogenesis and assembly (GO:0002181, GO:0042273, and GO:0042255). Among the most significant GO downregulated terms, we could find several terms related to stress and stimulus response (GO:1901700, GO:0042221, GO:0006970, GO:0009408, and GO:0010035). Other significantly repressed terms were involved in the small-molecule metabolic process and nucleobase-containing small-molecule metabolic process (GO:0044281 and GO:0055086) and toxin biosynthetic process (GO:0009403).
AITC versus I3C. Expression of 296 GOs was higher in mycelia treated with AITC than in mycelia treated with I3C. Among them, we can find terms related to cell division, such as positive regulation of metaphase/anaphase transition of the cell cycle (GO:1902101), positive regulation of cell cycle phase transition (GO:1901989), or actin polymerization or depolymerization (GO:0008154). Other terms are related to catabolic processes, such as positive regulation of the proteasomal protein catabolic process (GO:1901800) and positive regulation of proteolysis (GO:0045862). Translation (GO:0006412), ribonucleoprotein complex assembly (GO:0022618), endoplasmic reticulum (ER)-nucleus signaling pathway (GO:0006984), cellular response to sterol depletion (GO:0071501), and fungus-type cell wall biogenesis (GO:0009272) were also Differentially expressed genes. There was a higher number of DEGs in response to AITC than in response to I3C compared to control samples. Thirty genes were commonly induced by both treatments, and 98 were repressed ( Fig. 2A and B). In the AITC versus I3C comparison, 291 DEGs were induced in response to AITC, and 427 were induced in response to I3C ( fig. 2C). Relationships among DEGs from each comparison were deduced by using the web tool STRING (v10.5). One hundred twenty-three DEGs from the AITC versus control comparison interact in a single network (Fig. 3). DEGs interact in several clusters related to sterol and lipid synthesis, proteasome, translation, cell division, and DNA replication (Fig. 3, Table S2). Twenty-one DEGs interact in two different networks in the I3C versus control comparison (Fig. 4, Table S2). Part of network 1's structure (including DEGs mad2, klp9, mug190, ish1, hsp9, and SPAC32A11) is shared with the network of the AITC versus control comparison. In both comparisons, mad2 and klp9 are induced and mug190, ish1, hsp9, and SPAC32A11 are repressed.
Seventy-four DEGs are interacting in the AITC versus I3C comparison (Fig. 5, Table S2) in two different networks. In network 1, there is a cluster of structural proteins related to ribosome synthesis, which are more highly induced by AITC than by I3C. All these structural proteins are related to protein Aps1, which is repressed by I3C and induced by AITC. In this network, a cluster of DEGs related to ergosterol synthesis (induced by AITC) and leucine synthesis (induced by I3C) also appears, interacting with protein SPAC22A12.16.1. In network 2, several DEGs, hip4, hhf1, mst1, and ngg1, are induced by AITC, whereas DEGs rhp6, ubc7, rev1, and rhp14 are induced by I3C. Both branches are connected by DEG cia1, which is induced by I3C and repressed by AITC.
DEGs in the three comparisons that were related to detoxification of GHPs and toxin biosynthesis are shown in Table 1.

DISCUSSION
GHPs alter fungal growth and cell cycle. Treatment with both GHPs produced slower mycelium growth than control conditions. This effect was more evident in fungi treated with I3C than with AITC. Moreover, in I3C-treated samples, the number of transcripts found was lower than that in AITC samples or the control group, indicating that, at the same concentrations, I3C is more toxic than AITC on S. sclerotiorum.
Several common transcripts in both treatments have an antiapoptotic function, of which only casA and crz1 were significantly repressed by both GHPs and ppk31 was significantly repressed by AITC. casA is a metacaspase-1, which is a positive regulator of ER stress-induced cell death (12) in other fungi (Aspergillus nidulans and A. fumigatus). crz1 is a target of the calcineurin protein that is related to resistance to antifungal molecules and homeostasis of cations in Saccharomyces cerevisiae (13). ppk31 is involved in the induction of autophagy and life span regulation. ppk31 increases the chronological life span of the S. pombe DY8531 strain (14). A life span can be modified by changing the number of mitotic divisions that a cell can undergo during its lifetime or by modifying the time a cell remains viable. In the first case, the term employed is replicative life span; in the second case, it is called chronological life span  Table S2. (14). Therefore, by inhibiting ppk31, fungi treated by AITC may stop the chronological life span to increase the replicative life span.
Unlike AITC, no GOs associated with mitosis were found in mycelia treated with I3C. Chronological life span in S. cerevisiae is related to amino acid availability. The addition of isoleucine, valine, and leucine extended the average chronological life span in S. cerevisiae with the limitation of nitrogen and carbon sources (15). Accordingly, in the AITC versus I3C comparison, DEGs leu1 and leu2, key genes in leucine synthesis, were induced by I3C. Furthermore, these DEGs coexpressed with oac1. oac1 exchanges isopropylmalate by oxalacetate from mitochondria to cytosol. Leucine synthesis is related to this transporter, since its precursor pyruvate can be obtained by oxaloacetate, once it is in the mitochondria (16). Tsang et al. (17) proved that oac1 is related to life span extension in S. cerevisiae through NAD 1 regulation. In the GO analysis, the glyoxylate cycle was upregulated by I3C compared to control samples. When simple sugars are scarce, lower eukaryotes employ the glyoxylate cycle to obtain energy from two-carbon compounds (18). There is a link between the glyoxylate cycle and chronological life span in S. cerevisiae. icl1 is a key gene in the glyoxylate cycle and was induced by I3C. Deletion of this gene in S. cerevisiae under calorie restriction provokes a decrease in chronological life span (19). Following reference 19, the chronological life span is favored by glyoxylate/gluconeogenesis flux, contrary to the ethanol and acetate flux via the Krebs cycle and mitochondrial respiration. One of the responses of fungus treated with I3C may be to increase the chronological life span of mycelium cells. Therefore, it seems that fungi are following two different growth strategies to cope with the toxic effects of both GHPs.
Both GHPs provoke a delay in mycelium growth. However, once fungi treated with AITC start growing, the shape of the graph of growth over time is like that of control samples. However, I3C-treated samples show a different growth pattern. The differential expression on genes related to the cell cycle may underlie differences in growth patterns.
There are several transcripts related to the cell cycle that were commonly regulated in AITC versus control and I3C versus control comparisons. DEGs mad2 and klp9 were  Table S2. induced by both ITCs. They are related to spindle dynamics during mitosis (20,21). ish1 and hsp9 are part of the mitogen-activated protein kinase (MAPK) pathway, and both are downregulated by both GHPs. The expression of ish1 increases after S. pombe exposure to osmotic stress or glucose deprivation (22). The function of hsp9 is related to the regulation of the cell cycle by increasing the length of S. pombe cells and decreasing their growth rate (23).
Other common DEGs were regulated differentially by both treatments in the AITC versus I3C comparison. DEGs from network 2 are mostly related to histone modification and DNA repair. Synthesis of histone H4 (DEG hhf1) is induced by AITC. Histone chaperones modulate nucleosome assembly and, therefore, the structure of chromatin (24). hip4 and cia1 in S. pombe (homologous to hpc2 and asf1 of S. cerevisiae) are chaperones acting on histones H3 and H4. hip4 is a component of the HIR complex. Deletion of hip4 provokes cycle delay (25). This gene is necessary for correct chromosome segregation. cia1 is a repressor of the HIR-dependent histone genes. Mutants in cia1 cause double-strand breaks in DNA, increase DNA sensitivity to nuclease, and  Table S2. subsequently activate the DNA damage checkpoint pathway. Consequently, life span is severely reduced (25,26). AITC induces hip4, and I3C induces cia1.
The function of histone chaperones can be regulated by posttranscriptional modifications, leading to the regulation of various processes involving DNA, such as transcription, reparation, or condensation (27). DNA replication is regulated by the acetylation of histones H3 and H4 (28). This function is carried out by DEGs ngg1 and mst1 (29) when fungi are treated with AITC. On the other hand, ubiquitination of DNA is induced by I3C, via rhp6 and rhp14. Both DEGs, together with the DNA polymerase rev1, act in DNA repair (30)(31)(32), contributing to genome stability. Therefore, regulation of chromatin structure is dependent on GHPs. In the case of AITC-treated samples, this reorganization seems to be involved in mitosis and DNA replication, where, in the case of I3C, it may be related to DNA repair. Differences in growth performance between both samples may be due to DNA reparation, being a possible direct or indirect target of I3C.
GHPs alter protein and lipid metabolism. Translation and ribosome synthesis were induced by AITC and repressed by I3C. Since mitosis of AITC treated cells is induced, they need to produce new proteins. However, in the case of I3C cells, with slower growth and the need to repair their DNA, ribosome synthesis is a highly energy-consuming process. Eukaryotic cells cease the production of ribosomes very rapidly upon unfavorable conditions to ensure survival (33). In the network constructed with DEGs of the AITC versus I3C comparison, a big cluster of ribosome-associated proteins is induced by AITC. All these proteins are linked to asp1, which is induced by I3C.
The inositol pyrophosphates (PP-InsPs) play a central role in signaling and reprogramming cell metabolism through their phosphorylation/dephosphorylation. Asp1 is an inositol kinase that participates in the synthesis of the diphosphate group of PP-InsP5 and (PP)2-InsP4 (34), and previous authors (34) found that a considerable amount of proteins isolated from S. cerevisiae that interact with 5PP-InsP5 are related to ribosome biogenesis and maturation. asp1 coexpressed with ppn1, which is an endopolyphosphatase, and catalyzes the hydrolysis of polyphosphate chains. This last DEG was induced by AITC. Therefore, it seems that the system aps1-ppn1 is controlling ribosome synthesis in fungi treated with GHPs.
Both treatments induced protein catabolism. AITC induced the ubiquitin-proteasomal system, whereas I3C induced the multivesicular (MVB) body sorting pathway. The ubiquitin-proteasome system has an essential role in cell growth and differentiation, survival, stress tolerance, and adaptation to environmental changes (9,35). This mechanism is critical for cell survival (36). Formation of MVB takes place by invagination of the endosome membrane. MVB is degraded by fusing to lysosomes (35). Both mechanisms serve to regulate protein turnover and degrade the damaged ones. Proteins to be degraded by both systems are joined to ubiquitin as a signal (35). In the AITC versus I3C comparison, two ubiquitination-related DEGs were significantly expressed. ubc11 was induced by AITC, whereas ubc7 was induced by I3C. Catabolism of proteins may be associated with mitosis and damage caused by GHP treatments. Secretory proteins are primarily assembled and folded in the ER. An increment in misfolded proteins inside ER lumen may be provoked by an increment in the protein load, but it can also respond to oxidative stress and modifications in lipids and calcium (37). Lipid perturbation seems to happen in cells treated with AITC, because this treatment diminished the permeabilization of S. sclerotiorum cells, whereas I3C did not affect it. The sterol biosynthetic process was induced in fungi treated with AITC compared to I3C. Sterols play an important structural role in regulating membrane fluidity and permeability (38). Besides, DEG wA was induced by AITC while repressed by I3C. This gene is related to melanin synthesis in fungi, which gives strength and rigidity to the cell wall. Consequently, it is related to cell permeability and turgor (39). In the AITC versus I3C comparison, DEG plh1 was induced by I3C. This DEG codes for a phospholipid, diacylglycerol acyltransferase, related to triacylglycerol synthesis (40). Fungi treated with AITC probably modify the composition of cellular membranes by increasing the level of sterols compared to control samples and I3C treatments and decreasing triacylglycerols to increase their stability. Similarly, the ergosterol synthesis pathway and, specifically, gene erg11 is a target of azole and non-azole fungicides in different fungal species (10,41,42).
GHPs induce detoxification systems and repress toxin biosynthesis. As a detoxification mechanism, glutathione S-transferases were induced by S. sclerotiorum (8) and A. brassicicola (4, 43) after exposure to several ITCs. In line with this, several glutathione transferases were induced by both treatments; coincidentally, SS1G_10295 was also induced after treatment of S. sclerotiorum with AITC (8). Exposure to AITCs causes oxidative damage in fungi (4,8,10). Conjugation of glutathione with GHPs may cause a drop in free glutathione and the inhibition of glutathione reductase. Hog1 is a MAP kinase related to the regulation of the oxidative stress response and was induced after exposure of A. brassicicola to AITC (4,44). However, in this work, sty1 (homologous to hog1) was repressed after AITC treatment, probably because, when samples were taken, oxidative stress caused by AITC was reduced after several days of adaptation.
Induction of MFS and ABC transporters is a common defense mechanism to detoxify fungus cells (10,42,45). In this case, several MFS and one ABC transporter were induced by GHP treatments. Induced transporters were not coincident between treatments, indicating the substrate specificity. MFS transporter SS1G_12101 was also induced in S. sclerotiorum after the invasion of B. napus plants (46). SS1G_12040 was induced by AITC. Following reference 47, it is a saxA gene, responsible for the hydrolysis of ITCs to amines and allowing S. sclerotiorum to tolerate ITCs in the medium.
In GO analysis, the toxin biosynthetic process was downregulated by both GHPs. This GO term includes genes that are similar to sterigmatocystin coding genes. Sterigmatocystin is a polyketide compound produced by several Aspergillus species and is a precursor of aflatoxin. Repression of toxin-related genes by fungi in the presence of fungicides has been previously described (42) and may respond to a need to derive energy to prime metabolism.
Conclusions. Moderate concentrations of GHPs inhibit the growth of S. sclerotiorum in vitro. However, after long exposure, fungi may overcome their toxic effects and grow at the same level as nontreated fungi. We give an overview of the adaptation mechanisms of S. sclerotiorum to two chemically diverse GHPs, the aliphatic AITC and the indole I3C. We show that adaptation mechanisms are dependent on the type of GHP and on their initial target, involving cell membranes in the case of AITC or DNA in the case of I3C. Adaptation mechanisms include the reorganization of chromatin, mediated by histone chaperones hip4 and cia1, ribosome synthesis controlled by the kinase-phosphatase pair aps1-ppn1, catabolism of proteins, ergosterol synthesis, and the induction of detoxification systems. These mechanisms probably help S. sclerotiorum to grow and survive in an environment where GHPs are being produced by Brassica plants upon glucosinolate breakdown.

MATERIALS AND METHODS
Fungal growth and treatment with GHPs. We collected S. sclerotiorum isolate MBG-Ss2, employed in the experiments on B. napus in Spain, and it is currently maintained at MBG-CSIC. To provide fresh colonies for the experiments, surface-sterilized sclerotia were cultivated on potato dextrose agar (PDA; Difco) plates and incubated at 24°C in the dark for 4 days. Fresh colonies were obtained through routine growth of mycelium-agar plugs from the margin of the fungal colony on PDA and incubated at 24°C for 72 h. GHPs were purchased from Sigma-Aldrich, including allyl-ITC (AITC) (36682) and indole-3-carbinol (I3C) (I7256-5G). Fungal growth in 2Â YPG medium was monitored spectrophotometrically for 4 days at 25°C with the addition of each GHP in concentrations ranging from 0 to 250 mM. A sublethal concentration of 100 mM for both compounds then was selected to be used in the transcriptomic analysis.
Fungi were grown in PDA medium supplemented with each GHP, which was added to the medium before solidification to reach a concentration of 100 mM. Petri dishes were coated with one layer of cellophane membrane (catalog no. 165-0963; Bio-Rad). Agar plugs of 2.5-mm diameter containing fresh fungal colonies then were transferred to petri dishes and incubated at 24°C in darkness. After 5 days, mycelium mass was harvested from cellophane membranes and homogenized in liquid nitrogen. The resulting material was immediately used to isolate total RNAs. Two replicates per treatment and control sample were carried out.
In another experiment, the diameter of fungal mycelium growing radially from the center was measured for 8 days, until it reached the end of petri dishes.
RNA isolation, library preparation, and sequencing. RNA was extracted with a Spectrum plant total RNA kit (Sigma, MO, USA) according to the manufacturer's protocol. In order to remove any traces of genomic DNA, the RNA was treated with DNase by following the manufacturer's instructions. Massive analysis of 39-cDNA end (MACE) libraries was generated by using GenXPro's MACE kit (GenXPro GmbH, Frankfurt, Germany) as described previously (48). Briefly, cDNA from 2 mg of total RNA was randomly fragmented, and biotinylated 39 ends were captured after binding to a streptavidin matrix. A library ready for high-throughput sequencing was prepared by using TrueQuant adapters included in the kit. The library consisted of 50-to 700-bp-long fragments derived from the 39 end from cDNAs. The 59 ends from the libraries were sequenced by using an Illumina HiSeq 2000 version 4 chemistry (Illumina, Inc., San Diego, CA, USA).
NextSeq-derived sequence reads were processed with GenXPro's online in-house MACE analysis pipeline. Briefly, libraries were sorted according to their respective index, and all duplicate reads detected by the TrueQuant technology were removed from raw data sets. Additionally, quality filtering (by software Cutadapt [https://github.com/marcelm/cutadapt/]) and poly(A) tails were clipped by an inhouse python script. Reads were aligned to reference sequences using bowtie2. On this genome, since the functional annotation was poor, a cluster_bed step was performed. Reads that successfully map the genomic regions with no annotation (feature) were clustered. Clusters (consensus sequences) were annotated via BLASTX to the UniProt database (Swiss-Prot and TrEMBL). Reads that did not match the genome were taken into a de novo assembly step and annotated via BLASTX to the UniProt database (Swiss-Prot and TrEMBL). The assembly was completed with Trinity RNA-Seq (https://github.com/ trinityrnaseq/trinityrnaseq/wiki). As for the functional annotation, UniProt (http://www.uniprot.org/) and NCBI fungal sequences were employed.
Bioinformatic analysis. In order to have a general result overview, Gene Ontology (GO) analysis with all the annotated transcripts was done for each comparison (GHP versus control and AITC versus I3C) with GO-Tool2. The likelihood of a specific GO term was calculated by using Fisher's exact test.
Those transcripts with a false discovery rate (FDR) of #0.05 and 21 # log 2 fold change (FC) $ 1 were considered to be differentially expressed genes (DEGs) between treatments. The web tool STRING (v. 10.5) was used to study the interconnections between DEGs. In addition to well-supported protein-protein interactions experimentally observed, this resource includes indirect and predicted interactions on top (49). However, there is little information about the interactions among proteins in the S. sclerotinia genome. Relationships among proteins were deduced by employing the genome of Schizosaccharomyces pombe as a reference. Ten percent of annotations were found, and there is much more experimental evidence of interactions among proteins than in S. sclerotiorum.
Determination of membrane permeabilizing potential in S. sclerotiorum treated with GHPs. The SYTOX green uptake assay was carried out by following reference 50. When plasma membrane integrity is compromised, the influx of the dye and binding to DNA causes a significant increase in fluorescence. In brief, 10 ml of fungal mycelium suspension (optical density [OD], 0.37) was added to 100 ml of 2Â YPG liquid medium (pH 4.5) and incubated for 48 h at 27°C. AITC or I3C then was added to get the final concentrations of 50, 75, and 100 mM; 0.2 mM SYTOX green was added to each sample. Samples showing 100% permeabilization were prepared by degrading cells with 70% ethanol. Fluorescence reads were carried out over a time course of t = 0 to t = 210 min using a Promega fluorescence measurement system (GloMax) at an excitation wavelength of 480 nm and an emission wavelength of 530 nm. Fluorescence values were corrected by the absorbance of samples measured at 600 nm. Measurements were carried out in triplicate.
Data availability. Data supporting the findings of this study are openly available in the Gene Expression Omnibus repository under reference number GSE144344.