Botrytis cinerea identifies host plants via the recognition of antifungal capsidiol to induce expression of a specific detoxification gene

Abstract The gray mold pathogen Botrytis cinerea has a broad host range, causing disease in >400 plant species, but it is not known how this pathogen evolved this polyxenous nature. Botrytis cinerea can metabolize a wide range of phytoalexins, including the stilbenoid resveratrol in grape, and the sesquiterpenoids capsidiol in tobacco and rishitin in potato and tomato. In this study, we analyzed the metabolism of sesquiterpenoid phytoalexins by B. cinerea. Capsidiol was dehydrogenated to capsenone, which was then further oxidized, while rishitin was directly oxidized to epoxy- or hydroxyrishitins, indicating that B. cinerea has separate mechanisms to detoxify structurally similar sesquiterpenoid phytoalexins. RNA-seq analysis revealed that a distinct set of genes were induced in B. cinerea when treated with capsidiol or rishitin, suggesting that B. cinerea can distinguish structurally similar phytoalexins to activate appropriate detoxification mechanisms. The gene most highly upregulated by capsidiol treatment encoded a dehydrogenase, designated Bccpdh. Heterologous expression of Bccpdh in a capsidiol-sensitive plant symbiotic fungus, Epichloë festucae, resulted in an acquired tolerance of capsidiol and the ability to metabolize capsidiol to capsenone, while B. cinerea Δbccpdh mutants became relatively sensitive to capsidiol. The Δbccpdh mutant showed reduced virulence on the capsidiol producing Nicotiana and Capsicum species but remained fully pathogenic on potato and tomato. Homologs of Bccpdh are found in taxonomically distant Ascomycota fungi but not in related Leotiomycetes species, suggesting that B. cinerea acquired the ancestral Bccpdh by horizontal gene transfer, thereby extending the pathogenic host range of this polyxenous pathogen to capsidiol-producing plant species.


Introduction
The antimicrobial secondary metabolites produced in plants during the induction of resistance against pathogens are collectively termed phytoalexins (1,2). Several hundred phytoalexins of diverse structures have been identified from a wide range of plant species, which include terpenoids, flavonoids, and indoles (3,4). Many of these phytoalexins are considered to exhibit their antimicrobial activity by targeting the cell wall or cell membrane of pathogens (5,6), but their mechanisms of action remain largely unknown.
In plants belonging to the Solanaceae family, the major phytoalexins are sesquiterpenoids, such as capsidiol in Nicotiana and Capsicum species and rishitin in Solanum species (2,7,8). In Nicotiana sp., production of capsidiol is strictly controlled by regulating the gene expression for capsidiol biosynthesis, such as EAS (5-epi-aristolochene synthase) and EAH (5-epi-aristolochene dihydroxylase), encoding the enzymes dedicated to the production of capsidiol (9,10). In N. benthamiana, expression of EAS and EAH genes is hardly detected in healthy tissues, but their expression is rapidly induced during infection by pathogens (11,12). Produced phytoalexins are secreted locally via ATP-binding cassette (ABC) transporters at the site of pathogen attack (13,14). In N. benthamiana, silencing of genes involved in capsidiol production and secretion significantly reduces the resistance to the oomycete pathogen Phytophthora infestans, suggesting that the inhibition of pathogen growth by capsidiol plays an important role in disease resistance of capsidiol-producing plants (10,12,13). For plants, temporal and spatial control of phytoalexins is probably critical as the toxicity of phytoalexins is often not specific to microorganisms but is also harmful to plant cells (13,15,16). For example, rishitin affects the permeability of plant liposomal membranes and disrupts chloroplasts (17). Therefore, timely production and efficient transport of phytoalexins to the site of pathogen attack are important for plants to apply these double-edged weapons effectively.
To counter the production of antimicrobial substances in plants, various phytopathogenic fungi can metabolize and detoxify phytoalexins (18). Although many studies have shown an approximate correlation between the ability of strains to detoxify phytoalexins and their virulence (19)(20)(21), the importance of phytoalexin detoxification for pathogen virulence is largely unproven for most plant-pathogen interactions. The best-studied phytoalexin metabolizing enzyme is pisatin demethylase (PDA) of Nectria haematococca. Deletion of the PDA gene resulted in reduced virulence of N. haematococca on pea, directly proving the importance of PDA for the virulence of this pathogen (22). It has also been reported that sesquiterpenoid phytoalexins are metabolized by pathogenic fungi. Capsidiol is metabolized to less toxic capsenone via dehydration by pathogens such as the gray mold pathogen Botrytis cinerea and Fusarium oxysporum (23). Gibberella pulicaris, the dry rot pathogen of potato tubers, oxidizes rishitin to 13-hydroxyrishitin and 11,12-epoxyrishitin (24). However, the enzymes involved in the detoxification of sesquiterpenoid phytoalexins have not been identified to date, and their importance for pathogenicity has not been demonstrated.
In this study, we first investigated some pathogens isolated from Solanaceae and non-Solanaceae plants on their tolerance to capsidiol and rishitin, as well as their ability to detoxify/metabolize these phytoalexins. Among the pathogens that can metabolize both capsidiol and rishitin, we chose B. cinerea for further analysis to investigate the importance of its ability to detoxify capsidiol for the virulence on plant species that produce capsidiol.

Metabolization of sesquiterpenoid phytoalexins capsidiol and rishitin by fungal plant pathogens
Four oomycetes and twelve fungal species were evaluated for their ability to detoxify/metabolize sesquiterpenoid phytoalexins capsidiol and rishitin, produced by Solanaceae plant species. Four Phytophthora species isolated from Solanaceae host plants, including potato late blight pathogen P. infestans, P. nicotiana isolated from tobacco, P. capsici isolated from green pepper, and P. cryptogea isolated from nipplefruit (Solanum mammosum), are all sensitive to capsidiol and rishitin. The amount of capsidiol and rishitin after the incubation with these oomycete pathogens did not decrease, indicating that they are neither capable of metabolizing nor tolerant to capsidiol and rishitin ( Fig. 1 and Fig. S1). Among 12 fungal plant pathogens (eight isolated from Solanaceae plants), 7 and 8 fungal strains can metabolize capsidiol and rishitin, respectively, and showed increased resistance to phytoalexins. In some cases, pathogens can metabolize phytoalexins that are not produced by their host, indicating that the ability to detoxify phytoalexins does not always correlate with their host range (Figs. S1-S3 and Supplementary Notes 1 and 2). Botrytis cinerea and Sclerotinia sclerotiorum are well known as pathogens with a wide host range and are capable of metabolizing both capsidiol and rishitin. In this study, B. cinerea was selected as the pathogen to further investigate the role of detoxification of phytoalexins in the virulence of this polyxenous pathogen.

Metabolization of capsidiol and rishitin by B. cinerea
Metabolization of capsidiol and rishitin by B. cinerea was evaluated by LC/MS. Under the experimental conditions of this study, capsidiol was metabolized to capsenone by dehydrogenation within 12 h, and oxidized capsenone was detected at 24 h, whereas oxidized capsidiol was not detected (Fig. 1C and Fig. S4). In contrast, rishitin was directly oxidized within 6 h, and at least four different forms of oxidized rishitin were detected within 24 h ( Fig. 1C and Fig. S5), indicating that despite the structural similarity of these phytoalexins, B. cinerea detoxifies/metabolizes capsidiol and rishitin by different mechanisms (Fig. 1C).

Unique sets of genes are upregulated in B. cinerea treated with capsidiol, rishitin, or resveratrol
To identify B. cinerea genes upregulated during the detoxification of sesquiterpenoid phytoalexins, RNA-seq analysis was performed for mycelia of B. cinerea cultured in minimal media supplemented with capsidiol or rishitin. A stilbenoid phytoalexin, resveratrol, produced in grape was also used for comparison. Mycelia of B. cinerea were grown in liquid CM media supplemented with either 500 μM rishitin, 500 μM resveratrol, or 100 μM capsidiol, as 500 μM capsidiol caused significant growth defects in B. cinerea ( Fig. 1A and Fig. S6). The mycelial tissue was then used to perform an RNA-seq analysis. Among 11,707 predicted B. cinerea genes (25), 25, 27, or 23 genes were significantly upregulated (log 2 fold change > 2, P < 0.05) by the treatment with capsidiol, rishitin, or resveratrol, respectively. Unexpectedly, distinctive sets of genes were upregulated even between B. cinerea treated with capsidiol and rishitin, which resemble each other structurally ( Fig. 2A and Tables S1-S3), indicating that B. cinerea can either distinguish the structural difference of capsidiol and rishitin or the damage caused by these sesquiterpenoid phytoalexins. For instance, two genes Bcin08g00930 and Bcin12g01750 encoding hypothetical proteins containing a motif for dehydrogenases were specifically induced by capsidiol, while the treatment of rishitin significantly induced Bcin07g05430, encoding a cytochrome P450 gene ( Fig. 2B and C). Expression of BcatrB encoding an ABC transporter involved in the tolerance of B. cinerea against structurally unrelated phytoalexins resveratrol and camalexin and fungicides fenpiclonil and fludioxonil (26,27) is upregulated by treatment with rishitin and resveratrol, but not by capsidiol (Table S2). In contrast, Bcin15g00040, encoding a major facilitator superfamily (MFS)-type transporter, was upregulated specifically by capsidiol (Fig. 2B). Interestingly, treatment of B. cinerea with phytoalexins also activated genes predicted to be involved in pathogenicity to plants. For example, capsidiol treatment induced expression of a hydrophobin gene Bcin06g00510.1 (Bhp3), while rishitin treatment activated the expression of Bcin01g00080.1 (Bcboa8), encoding an enzyme for biosynthesis of a phytotoxin, botcinic acid ( Fig. 2B and C). Given that capsidiol is metabolized to capsenone by a dehydrogenation reaction in B. cinerea ( Fig. 1C and Fig. S4), the function of Bcin08g00930 and Bcin12g01750, both encoding a predicted shortchain dehydrogenase/reductase (SDR), was further analyzed in this study. Data marked with asterisks are significantly different from control as assessed by the two-tailed Student's t-test: * * P < 0.01, * P < 0.05.

Bcin08g00930 encodes a short-chain dehydrogenase for the detoxification of capsidiol
To investigate the function of SDR genes induced by capsidiol treatment, an endophytic fungus Epichloë festucae was employed for the heterologous expression of these genes. Bcin08g00930 and Bcin12g01750 were expressed in E. festucae under the control of the constitutive TEF promoter (28). The growth of wildtype and control E. festucae expressing DsRed was severely in-hibited in the 100 μM capsidiol ( Fig. 3A and B). Epichloë festucae transformants expressing Bcin12g01750 also hardly grew in 100 μM capsidiol; however, E. festucae became tolerant to capsidiol by the expression of Bcin08g00930 ( Fig. 3A and B). The amount of capsidiol was not altered by wild type, DsRed or Bcin12g01750 expressing E. festucae, while capsidiol was metabolized to capsenone by the E. festucae transformant expressing Bcin08g00930 (Fig. 3C). These results indicated that . Asterisks indicate a significant difference from the control (CM) as assessed by two-tailed Student's t-test, * * P < 0.01, * P < 0.05.
Bcin08g00930 encodes a dehydrogenase that can detoxify capsidiol; thus, the SDR encoded by Bcin08g00930 was designated as BcCPDH standing for B. cinerea capsidiol dehydrogenase.

Expression of bccpdh is specifically induced by capsidiol
To further investigate the expression pattern of Bccpdh, B. cinerea was transformed with a reporter construct for GFP expression under the control of the 1 kb proximal Bccpdh promoter (P_Bccpdh:GFP). P_Bccpdh:GFP transformants showed no obvious expression of GFP in water, but a significant increase of GFP fluorescence was detected in 500 μM capsidiol (Fig. 4A). The P_Bccpdh:GFP transformant was incubated with other antimicrobial terpenoids produced by Solanaceae species, including rishitin, debneyol (29), sclareol (30), and capsidiol 3-acetate (31). Among the terpenoids tested, treatment of capsidiol and capsidiol 3-acetate induced expression of GFP in the P_Bccpdh:GFP transformant. This result further indicated that B. cinerea specifically reacts to capsidiol and its derivative capsidiol 3-acetate. As the Bccpdh promoter was activated by capsid-iol 3-acetate, we investigated whether BcCPDH could metabolize capsidiol 3-acetate to capsenone 3-acetate. However, capsidiol 3-acetate was not metabolized by E. festucae expressing Bccpdh (Bcin08g00930) (Fig. S7A). Consistently, capsidiol 3-acetate was directly oxidized by B. cinerea, and at least two different forms of oxidized capsidiol 3-acetate, but not capsenone 3acetate, were detected (Fig. S7B), indicating that capsidiol and capsidiol 3-acetate were metabolized in B. cinerea by distinctive pathways.
Analysis of different lengths of Bccpdh promoters indicated that 250 bp of promoter sequence is sufficient for specific activation of the promoter by capsidiol treatment, and the ciselement required for capsidiol-specific expression was shown to be within −250 and −200 bp upstream of the start codon (Figs. S8 and S9). By further analysis using B. cinerea transformant P_Bccpdh:Luc using luciferase as a quantitative marker, it was shown that Bccpdh promoter was activated within 2 h of capsidiol treatment, and the expression activity decreased as capsidiol was metabolized (Fig. S10, Supplementary Notes 3), suggesting that the Bccpdh promoter is under strict control of capsidiol recognition.

Expression of bccpdh is specifically induced during the infection of capsidiol producing plant species
Leaves of N. benthamiana, which produce capsidiol as the major phytoalexin (13), were inoculated with conidia of the B. cinerea transformant P_Bccpdh:GFP. No GFP signal was detected in germinating B. cinerea conidia grown on the surface of N. benthamiana leaves 2 h after the inoculation, but obvious GFP expression was detected in the appressoria at 8 h after the inoculation (Fig. 4B). Growing hyphae in the leaf tissue of N. benthamiana showed GFP fluorescence, and the intensity of GFP fluorescence in hyphae was stronger near the edge of the lesion compared with that of hyphae growing in the areas of dead tissue within the lesions (Fig. 4B), probably because capsidiol was detoxified in these areas, which are heavily infected with B. cinerea. Induction of Bccpdh during the early stage of infection (8 h) and reduced expression of Bccpdh in hyphae in the necrotic plant tissue were confirmed by qRT-PCR ( Fig. S11A and Supplementary Notes 4). Accumulation of capsenone was detected both in necrotic tissue and near the edge of the lesion and only trace amounts of capsidiol were detected in infected tissue (Fig. S11B), suggesting that B. cinerea immediately metabolizes capsidiol produced by the host plant during the infection.
To examine the possibility whether polyxenous B. cinerea activates Bccpdh for detoxification of other phytoalexin or antimicrobial compounds produced in other plant species, the B. cinerea P_Bccpdh:GFP transformant was inoculated on a wide variety of plants. Over 50 plant species were used for the inoculation test including nine Solanaceae, six Brassicaceae, six Rosaceae, five Fabaceae, and six Asteraceae plants, and development of disease symptoms was observed in all tested plant species. Among the tested plants, expression of GFP under the control of Bccpdh promoter was only detected during the infection in three Nicotiana and two Capsicum species (Fig. 4C, Fig. S12, and Table S4), all of which are reported to produce capsidiol (8,(32)(33)(34). qRT-PCR confirmed that the expression of Bccpdh is hardly detected in potato leaves ( Fig. S11A and Supplementary Notes 4). These results further indicated that B. cinerea specifically recognizes capsidiol for the induction of Bccpdh.

BcCPDH metabolizes and detoxifies capsidiol to capsenone in B. cinerea
Botrytis cinerea Bccpdh KO mutant ( bccpdh) was produced to investigate the function of BcCPDH (Fig. S13). Mycelia of B. cinerea wild type and bccpdh were incubated with capsidiol and the metabolites were detected by LC/MS. While capsidiol was metabolized to capsenone in the wild type strain, most of the capsidiol remained unmetabolized 2 d after incubation with the bccpdh strain (Fig. 5A). Instead, oxidized capsidiol, which was not detectable in the wild type strain, was detected in bccpdh incubations ( Fig. S14 and Supplementary Notes 5). While the growth of B. cinerea hyphae was not affected by the disruption of the Bccpdh gene in 100 μM capsidiol, growth of bccpdh was diminished compared with that of wild type B. cinerea at higher concentrations of capsidiol (Fig. 5B). Colony growth, conidial gemination, and appressoria-mediated penetration were not significantly af-  To investigate the effect of bccpdh disruption on the ability of B. cinerea mycelia to expand lesions, leaves, tubers, or fruits of indicated plant were inoculated with mycelial plugs (5 × 5 mm) of B. cinerea WT or bccpdh-52 and lesion size was measured between 4 and 7 d after inoculation (dpi). Asterisks indicate a significant difference from WT as assessed by two-tailed Student's t-test. * * P < 0.01. Lines and crosses (×) in the columns indicate the median and mean values, respectively. fected by the disruption of bccpdh (Fig. S15). These results confirmed that BcCPDH is the enzyme responsible for the detoxification of capsidiol in B. cinerea.

BcCPDH is required for the pathogenicity of B. cinerea on plant species producing capsidiol
To investigate the effect of bccpdh disruption on the ability of B. cinerea mycelia to expand lesions in host plants producing different phytoalexins, leaves, tubers, or fruits of several plant species were inoculated with mycelial plugs of B. cinerea wild type and bccpdh. On plant species that produce capsidiol, such as N. benthamiana, N. tabacum, and C. annuum, the development of disease symptoms by bccpdh was significantly reduced compared with those caused by wild type B. cinerea (Fig. 5C). Consistently, capsenone was detected in N. benthamiana leaves inoculated with wild type, but not that with bccpdh (Fig. S16). In contrast, both wild type and bccpdh strains caused comparable symptoms in potato, tomato, grape, and eggplant, which is consistent with the finding that expression of Bccpdh is not induced during the infection in these plant species (Fig. 5C and Fig. S17). In the complementation strain, the ability to metabolize capsidiol to capsenone was restored, as was virulence against the capsidiol producing plants (Fig. S18). These results indicated that BcCPDH is an enzyme dedicated to virulence of B. cinerea on capsidiol-producing plant species.

Epoxidation of capsidiol and capsenone is mediated via Bcin16g01490
The incubation of the bccpdh knockout strain in capsidiol revealed the presence of a potential secondary capsidiol degradation pathway, which resulted in the accumulation of oxidized capsidiol. Since we were unable to find any indication for such a pathway in our RNA-seq results for 100 μM capsidiol treated incubations, we extended our search to a preliminary RNA-seq analysis, which used 500 μM capsidiol treatments. We identified the gene Bcin16g01490 encoding a cytochrome P450, which was significantly upregulated under these elevated capsidiol concentrations (Table S1 and Fig. S19A). Heterologous expression of Bcin16g01490 in E. festucae resulted in the conversion of capsidiol to oxidized capsidiol as detected during bccpdh strain incubations (Figs. S14 and S19B and Supplementary Notes 5). Moreover, the sequential incubation of capsidiol with the Bccpdh expressing E. festucae transformant followed by a Bcin16g01490 expressing transformant reproduced the reactions detected in B. cinerea (Figs. S4 and S19B). Structural analysis of oxidized capsenone indicated that the end product of these reactions is capsenone 11,12-epoxide (Fig. 1, Fig.  S19C and S20-S22, and Supplementary Notes 6).

Distribution of bccpdh homologues in the fungal kingdom
The distribution of Bccpdh homologs in taxonomically related fungal species (Ascomycota, Leotiomycetes) was investigated. A search for Bccpdh homologs in the genome sequences of six hostspecialized phytopathogenic Botrytis species, such as B. tulipae (pathogen of tulip), B. hyacinthi (hyacinth), and B. paeoniae (peony), indicated that among Botrytis species, Bccpdh is a unique gene only found in B. cinerea (Fig. S23). Consistently, four Botrytis species, not including B. cinerea, were incubated with capsidiol, but no CPDH activity was detected for the tested strains (Fig. S24). Search in 11 draft genomes of Leotiomycete fungi also indicated the absence of Bccpdh homologs in these species. Comparison of the corresponding genome regions between Botrytis species and S. sclerotiorum (another polyxenous pathogen belonging to Leotiomycetes) revealed that the ∼4.9 kb sequence surrounding Bccpdh is unique to B. cinerea (Fig. 6). This unique sequence shows a lower GC content compared to the surrounding region (Fig. S25), suggesting that Bccpdh might have been obtained via horizontal gene transfer (35). A blast search using BcCPDH as query sequence revealed that probable orthologs can be found only in some Pezizomycotina fungi belonging to Ascomycota (Fig. S23). Orthologs were found from a taxonomically diverse range of fungal species, including animal and insect pathogens, and there was no correlation between their homology and taxonomic relationship, which might indicate multiple horizontal gene transfer events of the cpdh gene in the diversification of Ascomycota fungi (Figs. S23 and S26-S28 and Supplementary Notes 7). Among plant pathogenic fungi, Bccpdh homologs were found in Fusarium species, all of which are pathogens that infect plants that do not produce capsidiol. This result is consistent with a previous report showing that some Fusarium species can metabolize capsidiol to capsenone (36).

CPDH activity is conserved among B. cinerea strains
Although B. cinerea is a polyxenous fungus, it has been reported that different strains of B. cinerea exhibit various degrees of virulence on different host plants (37,38). Hence, we investigated whether BcCPDH is conserved among strains isolated from diverse plant species. A total of 24 B. cinerea strains isolated from 14 different plant species were incubated with capsidiol and resultant capsenone was detected by GC/MS. All tested B. cinerea strains can metabolize capsidiol to capsenone, indicating that Bc-CPDH is highly conserved among B. cinerea strains even though their (most recent) host was not a producer of capsidiol ( Fig. 7 and Table S5). Similarly, 17 strains of F. oxysporum isolated from 7 plant species were subjected to the analysis for CPDH activity. Altogether 8 out of 17 F. oxysporum strains showed CPDH activity, and the activity was not related to the natural host of the strains ( Fig. 7 and Table S6). This result is consistent with the finding that Bccpdh orthologues can be found in 5 out of 14 available genomes of F. oxysporum (Fig. S28). These results indicate that the cpdh gene is randomly distributed in F. oxysporum strains, whereas it is highly conserved among B. cinerea strains.

Discussion
To survive the exposure to microorganisms in the environment, different plant species have developed diverse resistance mechanisms over the course of evolution. The structural variety of phytoalexins is a prime example of such diversity. Different plant families produce phytoalexins with relatively similar basic structures, but their side-chain structures often differ from species to species (4). These differences in structure between species may have contributed to the determination of host specificity between plants and pathogens, as in some cases a pathogen may have acquired resistance to a particular phytoalexin, but an analogous substance is not overcome by the same resistance mechanism (39). However, plant pathogens with a broad host range such as B. cinerea must employ strategies to counter a multitude of diverse phytoalexins. The prompt killing of plant cells, or the presence of efflux pumps of extremely low specificity may present effective strategies for such pathogens. Indeed, B. cinerea produces hostnonspecific phytotoxins (such as botrydial and botcinic acid, 40, 41) upon infection, and activates transporters capable of effluxing diverse substances such as PDR-type ABC transporter BcatrB  and MFS transporter Bcmfs1 (26,27,42). More recently, it has been shown that B. cinerea suppresses the immune response of different plant species via gene silencing by delivering various small RNAs into host cells (43,44).
In addition to effective infection mechanisms that generally facilitate infection of a wide range of plants, this study revealed that B. cinerea responds to chemical cues from the host plant to adapt its infection strategy to overcome specific plant resistance mechanisms. Botrytis cinerea strictly distinguishes structurally similar phytoalexins, such as capsidiol and rishitin, and activates appropriate detoxification responses. Interestingly, treatment of B. cinerea with phytoalexins activated not only genes involved in detoxification and efflux of toxic compounds, but also genes predicted to be involved in pathogenicity to plants. These results support the notion that B. cinerea utilizes phytoalexins as a means to identify a given host and adjust the method of infection.
Expression of the Bccpdh gene is activated specifically during infection of capsidiol-producing plants, and the pathogenicity of the bccpdh mutant strains is compromised on plants producing capsidiol, but does not suffer any disadvantages on plants that do not produce capsidiol. These results suggest that BcCPDH is a critical component that specifically enables B. cinerea to infect capsidiolproducing plants. Despite capsidiol-producing plants representing only a small fraction of >400 host plants of B. cinerea, CPDH activity was maintained in all investigated B. cinerea strains isolated from plants that do not produce capsidiol. This may hint at the presence of a selection pressure against the loss of CPDH, despite it only affecting a limited number of host plants. This poses the question of whether B. cinerea is able to maintain acquired resistances for a prolonged time, which may explain how it evolved and established itself as the polyxenous pathogen that it is.

How does B. cinerea recognize phytoalexins with similar structures?
Although B. cinerea is known to have the ability to metabolize a wide variety of phytoalexins (18), only a small portion of these detoxification mechanisms are required for the infection of any particular plant. Thus, B. cinerea needs to strictly control those various detoxification mechanisms, since maintaining sufficient levels of all these detoxification mechanisms represents a considerable waste of resources. Elucidating the mechanism by which B. cinerea recognizes phytoalexins and activates a specific set of genes is one of the important subjects for further research. Two major scenarios can be envisioned: first, B. cinerea may possess receptors that can identify the chemical structures of phytoalexins. In plants, various terpenes with diverse structures are employed as hormones recognized by specific receptors, such as gibberellins, cytokinins, and abscisic acid (39). However, the possibility that B. cinerea maintains receptors to perceive all of the myriad of phytoalexins may not be realistic. The second scenario could be that B. cinerea recognizes the damage caused by phytoalexins. However, as for capsidiol and rishitin, both are presumed to have cell membranes as their targets, and these phytoalexins are relatively unspecific toxicants that cause damage even to plant cells (5,13,17,45), so it is not certain whether there exists a specific target by which these two compounds can be distinguished. Immediate downregulation of the bccpdh promoter after the metabolization of capsidiol (Fig. S10) also suggests that it is unlikely that bccpdh expression is primarily controlled by cell damage, since the damage signals would remain present even after capsidiol was detoxified. Recently, we have reported that KO transformants of BcatrB (which is activated by rishitin but not by capsidiol) became sen-sitive to rishitin, but not to capsidiol. Moreover, bcatrB KO showed reduced virulence on tomato (producing rishitin) but can infect N. benthamiana (producing capsidiol) same as the wild type does (46). BcatrB not being induced by capsidiol further strengthens the notion that B. cinerea relies on specific triggers, rather than on general, indirect cues to mount its defense against the various phytoalexins. Contrary to this notion is, however, that BcatrB is also induced by structurally unrelated phytoalexins as well as fungicide treatments (26,27,46). It would therefore be reasonable to assume that the perception of phytoalexins may rely on the integration of multiple signals, and that specific signals may be attenuated by additional unspecific triggers, such as cell damage. Using the reporter system developed in this study, mutant strains defective in capsidiol response could be isolated to elucidate the mechanism by which B. cinerea distinctly identifies phytoalexins.

Does B. cinerea possess other capsidiol resistance mechanism besides detoxification by BcCPDH?
Although the bccpdh knockout mutants showed reduced virulence on plant species producing capsidiol, the mutant can develop the disease symptoms on these plants and showed tolerance to 100 μM capsidiol, same as the wild type. In contrast, Phytophthora spp., Alternaria spp., and E. festucae are sensitive to the same concentration of capsidiol. Although we also isolated Bcin16g01490, which we found to oxidize capsidiol, its activity is fairly limited, indicating that it is unlikely to confer sufficient protection against capsidiol. We therefore presumed that B. cinerea has another mechanism for capsidiol tolerance other than detoxification. RNA-seq analysis of B. cinerea genes upregulated by capsidiol treatment identified two genes (Bcin15g00040.1 and Bcin14g02870.1/Bcmfs1) encoding MFS transporters and a gene (Bcin01g05890.1/Bcbmr1) encoding an ABC transporter (Table S1). Bcmfs1 has been reported to be involved in the resistance of B. cinerea to structurally different natural toxicants (camptothecin produced by the plant Camptotheca acuminata and cercosporin produced by the plant pathogenic fungus Cercospora kikuchii) and fungicides (sterol demethylation inhibitors) (42). The bcbmr1 mutants showed an increased sensitivity to fungicides, polyoxin (a chitin synthetase inhibitor) and iprobenfos (a choline biosynthesis inhibitor) (47). The expression of one or more of these transporters may be involved in capsidiol efflux from B. cinerea cells, and regulated by signaling that is common to Bccpdh.

Why and how can B. cinerea strains stably maintain BcCPDH?
Phylogenetic analysis of CPDH orthologs indicates that sequence similarity does not necessarily correlate with the taxonomic relationship. Rather, CPDH orthologs of the same cluster type tend to form a clade in the phylogenetic tree, which suggest CPDH orthologs and surrounding genes were transferred via multiple horizontal gene transfer events (Supplementary Note 7). While no CPDH ortholog was found in other Botrytis species, the Bccpdh gene is conserved in the genomes of published B. cinerea strains, and CPDH activity was detected in all tested B. cinerea strains isolated from plants that do not produce capsidiol. In F. oxysporum, in contrast, there were strains with and without CPDH activity regardless of the host plant, and consistently, some publicly available F. oxysporum genomes contain cpdh homologs and others do not. As shown in this study, metabolic capacity (and tolerance) to capsidiol and rishitin can be detected in some fungal strains regardless of the natural host of these pathogens, and perhaps unused detoxification activities are maintained in the population of phytopathogenic filamentous fungi for future use.
It is theoretically implausible that all B. cinerea strains maintain an enzyme required only upon infection of capsidiol-producing plants, even if the said gene is completely repressed in its expression under normal circumstances. Given that B. cinerea has an extremely wide host range, some strains will go for long periods where capsidiol detoxification ability is not a selection pressure. One possibility is that BcCPDH has functions other than capsidiol degradation. For example, BcCPDH might be involved in the detoxification of antimicrobial substances produced by insects, because it has been reported that spores of B. cinerea are transmitted between plants via insects such as thrips (48). Since homologous genes for Bccpdh are found in several insect-infecting fungi, it is expected that CPDH-like enzymes in these species can metabolize insect-derived substances. We, therefore, performed preliminary experiments to determine whether Bccpdh expression can be induced by inoculation of B. cinerea P_Bccpdh:GFP transformant with several insects, but to date, no induction of GFP expression has been detected. Alternatively, B. cinerea may have acquired the trait of having many hosts because of its ability to maintain rarely used virulence factors. Future clarification of the phytoalexin recognition mechanisms and comparative analysis of the genome with that of closely related Botrytis species are anticipated to elucidate unknown features of B. cinerea that have led to its evolution as a polyxenous fungi.

Materials and methods
Comprehensive descriptions of the materials and methods used in this study, including biological materials (Tables S7-S9 and S13), vector construction (Table S10), primer sequences (Tables S11 and S12), transformation of E. festucae and B. cinerea, confocal microscopy, pathogenicity tests, and detection and structural analysis of phytoalexins and their derivatives, are available in supplementary material.