Dynamic Changes in Plant Secondary Metabolites Induced by Botrytis cinerea Infection

In response to pathogen infection, some plants increase production of secondary metabolites, which not only enhance plant defense but also induce fungicide resistance, especially multidrug resistance (MDR) in the pathogen through preadaptation. To investigate the cause of MDR in Botrytis cinerea, grapes ‘Victoria’ (susceptible to B. cinerea) and ‘Shine Muscat’ (resistant to B. cinerea) were inoculated into seedling leaves with B. cinerea, followed by extraction of metabolites from the leaves on days 3, 6, and 9 after inoculation. The extract was analyzed using gas chromatography/quadrupole time-of-flight mass (GC/QTOF) combined with solid-phase microextraction (SPME) for volatile and nonvolatile metabolomic components. Nonvolatile metabolites γ-aminobutyric acid (GABA), resveratrol, piceid, and some carbohydrates or amino acids, coupled with volatile metabolites β-ocimene, α-farnesene, caryophyllene, germacrene D, β-copaene, and alkanes, accumulated at a higher level in grape leaves infected with B. cinerea compared to in noninoculated leaves. Among the established metabolic pathways, seven had greater impacts, including aminoacyl-tRNA biosynthesis, galactose metabolism, valine, leucine, and isoleucine biosynthesis. Furthermore, isoquinoline alkaloid biosynthesis; phenylpropanoid biosynthesis; monobactam biosynthesis; tropane, piperidine, and pyridine alkaloid biosynthesis; phenylalanine metabolism; and glucosinolate biosynthesis were related to antifungal activities. Based on liquid chromatography/quadrupole time-of-flight mass (LC/QTOF) detection and bioassay, B. cinerea infection induced production of plant secondary metabolites (PSMs) including eugenol, flavanone, reserpine, resveratrol, and salicylic acid, which all have inhibitory activity against B. cinerea. These compounds also promoted overexpression of ATP-binding cassette (ABC) transporter genes, which are involved in induction of MDR in B. cinerea.


Introduction
Plant pathogens interact with their hosts in several ways in the infection course. One consequence of this interaction is that a pathogen will elicit production of plant secondary metabolites (PSMs), such as terpenoids, phenylpropanoids, fatty acids, and alkaloids. These compounds protect plants from biotic and abiotic stresses [1].

Plants and Pathogens
Grape leaves were selected as the experimental material since they are susceptible tissues to B. cinerea infection. Grapes 'Victoria' (sensitive to B. cinerea) and 'Shine Muscat' (resistant to B. cinerea) were grown in a greenhouse from February to April 2022. Botrytis cinerea multidrug-resistant strain 242 was collected from Yinlong Soil Farm in Shanghai in April 2016. Botrytis cinerea standard strain B05.10 was obtained from Germany (Institut für Botanik, Westfälische Wilhelms-Universität, Münster). The two strains were cultivated on potato dextrose agar (PDA) in the dark at 18 • C. To prepare spore suspension as an inoculum, the fungi were cultured in the dark at 18 • C on carrot agar (CA) dishes [21] for 3 to 5 d and placed under a blacklight lamp for 5 d. Conidia were washed off the culture with sterile water and filtered through two-layer lens-wiping papers to remove mycelia. The conidial concentration was adjusted to 5 × 10 6 /mL with a hemacytometer. The suspension with 1.5% glucose and 0.5% tween 20 (Beijing Soleibao Technology Co., Ltd., Beijing, China) was also added.

Nonvolatile Metabolites of Grape Leaves
Grape seedlings were either inoculated, by spraying B. cinerea B05.10 and 242 at 10 6 conidia/mL, or noninoculated. The treatments included 'Victoria' without inoculation, inoculated with B05.10, and inoculated with 242 and 'Shine Muscat' without inoculation, inoculated with B05.10, and inoculated with 242. Disease severity and incidence were evaluated at the end of the trial, followed by collection of all plant samples into a Ziplock bag that was stored at −80 • C for later use.
The sampled seedlings were processed for metabolomic analysis following the previously reported procedure [22]. Briefly, the sample was prechilled with liquid nitrogen and ground in a ball mill (MM400; Verder Shanghai Instruments and Equipment Co., Ltd., Shanghai, China) at 30 Hz for 1 min. Each treatment had four replicates (plants). Each sample (200 ± 2 mg) was dissolved in 1.8 mL of extraction solvent (methanol/water, v/v = 8/2) with 10 µg/mL of ribitol as an internal standard and was distributed into 4 tubes as 4 technical replicates. Each sample was treated with 100 Hz ultrasound for 20 min, followed by centrifugation at 13,800× g for 15 min; 0.4 mL of supernatant was collected, desiccated at 45 • C in a vacuum concentrator, and stored at −20 • C until use. Methoxyaminen hydrochloride solution (20 mg/mL) in 100 µL was added to the sample, which was incubated at 30 • C for 2 h. The sample was added to 100 µL containing BSTFA (1% TMS) and incubated on a dry-bath block at 37 • C for 6 h for derivatization. After centrifugation, 120 µL of liquid supernatant was transferred into a sample vial sealed with a rubber cap, and metabolome analysis was performed within 48 h.
Metabolites of grape leaves were separated and detected using an HP-5MS capillary column (30 m × 0.25 mm × 0.25 µm) coupled with the 7890A-5975C GC-MS system (Agilent, CA, USA). Helium was used as a carrier gas, with a 1.1 mL/min flow rate. Each injection volume was 1 µL. A GC oven was heated to 60 • C for 1 min, then raised to 325 • C at 10 • C/min for 2 min. The auxiliary heater was set at 290 • C. The ion source (EI) temperature was set to 250 • C. Electron impact ionization (70 eV) was set in a full scan mode (m/z 50 to 600) to 0.2 s/scan.
To analyze metabolites using liquid chromatography-mass spectrometry (LC-MS), the sample was prechilled with liquid nitrogen and ground in a ball mill (MM400; Verder Shanghai Instruments and Equipment Co., Ltd., Shanghai, China) at 30 Hz for 1 min. For each treatment, six samples were collected as replicates; 500 mg of sample was dissolved in 3 mL of extraction solvent (methanol/acetonitrile, v/v = 3/1) and divided into 6 technical replicates (6 tubes). Each sample was vortexed at 2500 rpm for 2 min, followed by centrifugation at 13,800× g for 15 min at 4 • C, and then filtered through a 0.22 µm filter membrane.
LC-MS chromatographic analyses were performed on a Thermo Vanquish UHPLC system coupled with an ACQUITY UPLC HSS T3 (2.1 mm× 150 mm, 1.7 µm) column and Thermo QE HF-X ESIs. For positive ion analysis, the eluent was a mixture of methanol and H 2 O at a flow rate of 0.3 mL/min. The mobile phase consisted of 0.1% formic acid/H 2 O (solvent A), 0.1% formic acid/methanol (solvent B), 0.1 acetic acid/H 2 O (solvent C), and 0.1% acetic acid/methanol (solvent D). Gradient steps for positive ion analysis were applied as follows (using solvents A and B): 0 to 0.5 min, B: 2%; 0.5 to 6 min, B: 2% to 50%; 6 to 10 min, B: 50% to 98%; 10 to 14 min, B: 98%; 14 to 16 min, B: 98% to 2%; 16 to 21 min, B: 2%. For negative ion analysis, solvent C was used to replace A and D to replace B. The MS system was operated using both positive and negative ESIs in multiple reaction monitoring (MRM) mode to identify the analytes of interest. The operational parameters for MS analysis were as follows: ESI source, sheath gas flow at 50 L/min and auxiliary gas at 13 L/min; spray voltage at 2.5 KV (+)/2.5 KV (−); a capillary temperature maintained at 325 • C; an auxiliary gas temperature of 300 • C; secondary collision energy (NCE) at 30 V; an isolation window at 1.5 m/z; top N = 10; and a scan range from 70 to 1050 m/z. The detected metabolites were qualified using CD software and analyzed with the Thermo Fisher database, the NIST database, and a self-built database.

Volatile Metabolites of Grape Leaves
Solid phase microextraction-gas chromatography-mass spectrometry (SPME-GC-MS/MS) was applied to analyze volatile metabolites. To extract volatiles, healthy grape leaves with similar sizes were selected. A 500 mL beaker was lined with cotton and tin foil, with an appropriate amount of distilled water added at the bottom to wet the cotton. The petiole of a grape leaf was inserted into the wet cotton, along with a 5 mm mycelial disk of B. cinerea. Volatile gases were collected at 3, 6, and 9 dpi using solid-phase microextraction (SPME) coupled with 50/30 µm DVB/CAR/PDMS fiber after aging at 250 • C for 30 min. All samples were analyzed in triplicate.
The extracted samples were examined with a GC-MS system, as previously described [23]. The program was set as splitless with 50 mL/min, an injection port of 250 • C and an ion source temperature of 230 • C. The oven program was 40 • C for 5 min and 5 • C/min to 180 • C (33 min), then 10 • C/min to 280 • C. Electron impact ionization (70 eV) was set in a full scan mode (m/z 50 to 300) to 0.3 s/scan.

Effects of Secondary Metabolites on Botrytis cinerea Growth
Secondary metabolites eugenol, flavanone, reserpine, resveratrol, and salicylic acid were used to examine inhibitor activity against B. cinerea at final concentrations of 10, 25, 50, and 100 mg/L, with three replicates, following the Song et al. method [24]. The expressions of the three major ABC genes BcatrB, BcatrD, and BcatrK were assessed through a quantitative polymerase chain reaction (qPCR) conducted on an ABI7500 sequence detection system (Applied Biosystems, California, USA) using the FastSYBR Mixture Kit (Beijing ComWin Biotech Co., Ltd., Beijing, China), with specific primers (Table 1) and cDNA as templates. The program of the qPCR was set as denaturation at 95 • C for 2 min, followed by 40 cycles of 95 • C for 10 s and 60 • C for 34 s. The reaction system contained 10 µL of FastSYBR × 2, 0.4/0.4 µL of a mixture of forward/reverse primer, 1 µL of template cDNA, and 8.2 µL of ddH 2 O. The template cDNA was obtained with reverse transcription from RNA of B05.10 using a PrimeScript RT Reagent Kit with gDNA Eraser (Takara, Beijing, China). The relative expression of the genes was calculated using the 2 −∆∆Ct method [25], and the actin gene was used as a reference to normalize the quantification of the ABC geneexpression levels. This experiment was conducted twice, and each experiment included three replicates for each treatment.

Statistical Analysis
Mass Hunter Qualitative Analysis B.07.00 (Agilent Technologies, Santa Clara, CA, USA) was used for data processing and deconvolution, with the parameter of 20,000 absolute peak height, and the NIST14 and Fiehn mass spectrometry databases were used as references for qualitative analysis. Agilent MassHunter Mass Profiler Professional 13.1.1 was used for principal component analysis (PCA), cluster analysis, and variance analysis. The results of the LC-MS were analyzed with Xcalibur 4.1 (Thermo Fisher Scientific, Inc., Waltham, MA, United States) and Compound Discovery. Metabo Analyst online analysis software (https://www.metaboanalyst.ca/home.xhtml, accessed on 1 August 2022) was used to conduct metabolic pathway analysis [26]. Analysis of variance (ANOVA) was performed, and metabolites were compared between B. cinerea-treated and non-treated groups. A significant difference was determined with fold changes >2 and p < 0.05.

Nonvolatile Metabolites
A chromatogram was generated on the extract of the nonvolatile metabolites of grape leaves inoculated with B. cinerea or noninoculated using gas chromatography-mass spectrometry (GC-MS) (Figure 1a). In total, 287 to 312 compounds were deconvolved from the grape leaves, selected for absolute peak heights higher than 200,000, of which 206 metabolites were identified and normalized. Principal component analysis (PCA) was performed to explore the differential metabolites between six groups. A PCA score plot and PCA loading were performed to investigate the overall differences in the nonvolatile metabolites ( Figure 2). All groups were distinctly separated. In both grape varieties, the noninoculated groups were separated from the inoculated groups by the largest distance. The distance between the 242-and B05.10-infected groups was small in the susceptible variety 'Victoria' but much larger in the resistant variety 'Shine Muscat', which showed a big difference in metabolites between the two varieties. Otherwise, 238 to 330 compounds were deconvolved from the samples in all groups through liquid chromatography-mass spectrometry (LC-MS), selected for absolute peak heights higher than 2000. software (https://www.metaboanalyst.ca/home.xhtml, accessed on 1 August 2022) was used to conduct metabolic pathway analysis [26]. Analysis of variance (ANOVA) was performed, and metabolites were compared between B. cinerea-treated and non-treated groups. A significant difference was determined with fold changes >2 and p < 0.05.

Nonvolatile Metabolites
A chromatogram was generated on the extract of the nonvolatile metabolites of grape leaves inoculated with B. cinerea or noninoculated using gas chromatography-mass spectrometry (GC-MS) (Figure 1a). In total, 287 to 312 compounds were deconvolved from the grape leaves, selected for absolute peak heights higher than 200,000, of which 206 metabolites were identified and normalized. Principal component analysis (PCA) was performed to explore the differential metabolites between six groups. A PCA score plot and PCA loading were performed to investigate the overall differences in the nonvolatile metabolites ( Figure 2). All groups were distinctly separated. In both grape varieties, the noninoculated groups were separated from the inoculated groups by the largest distance. The distance between the 242-and B05.10-infected groups was small in the susceptible variety 'Victoria' but much larger in the resistant variety 'Shine Muscat', which showed a big difference in metabolites between the two varieties. Otherwise, 238 to 330 compounds were deconvolved from the samples in all groups through liquid chromatography-mass spectrometry (LC-MS), selected for absolute peak heights higher than 2000.

Characterization of Significantly Changed Nonvolatile Differential Metabolites
To further analyze the effects of different treatments on nonvolatile profiles, differential metabolites of treated groups relative to non-treated groups were determined by the criteria of fold changes >2 and p < 0.05. There were 52 differential metabolites in 'Victoria' under three treatments (Table S1). Twenty-nine metabolites were upregulated and twenty-three were downregulated in B05.10-inoculated leaves compared to noninoculated leaves. Twenty-seven metabolites were upregulated and twenty-six were downregulated in 242-inoculated leaves compared to noninoculated leaves. Among these, 24 compounds were upregulated and 20 were downregulated after inoculation with both B. cinerea strains in the 'Victoria' variety.
There were 17 differential metabolites in 'Shine Muscat' under these treatments (Table S2). In the B05.10-inoculated leaves, five metabolites were upregulated and twelve were downregulated. In 242-inoculated leaves, fourteen metabolites were upregulated and three were downregulated. Among these, only two compounds were upregulated in the 'Shine Muscat' variety inoculated with either of the two strains.
Meanwhile, 400 to 700 compounds were detected with LC-MS and 36 were identified by standards through liquid chromatography/quadrupole time-of-flight mass (LC/QTOF), which showed differences between the B. cinerea-inoculated and noninoculated grape leaves (

Characterization of Significantly Changed Nonvolatile Differential Metabolites
To further analyze the effects of different treatments on nonvolatile profiles, differential metabolites of treated groups relative to non-treated groups were determined by the criteria of fold changes >2 and p < 0.05. There were 52 differential metabolites in 'Victoria' under three treatments (Table S1). Twenty-nine metabolites were upregulated and twenty-three were downregulated in B05.10-inoculated leaves compared to noninoculated leaves. Twenty-seven metabolites were upregulated and twenty-six were downregulated in 242-inoculated leaves compared to noninoculated leaves. Among these, 24 compounds were upregulated and 20 were downregulated after inoculation with both B. cinerea strains in the 'Victoria' variety.
There were 17 differential metabolites in 'Shine Muscat' under these treatments (Table S2). In the B05.10-inoculated leaves, five metabolites were upregulated and twelve were downregulated. In 242-inoculated leaves, fourteen metabolites were upregulated and three were downregulated. Among these, only two compounds were upregulated in the 'Shine Muscat' variety inoculated with either of the two strains.
Meanwhile, 400 to 700 compounds were detected with LC-MS and 36 were identified by standards through liquid chromatography/quadrupole time-of-flight mass (LC/QTOF), which showed differences between the B. cinerea-inoculated and noninoculated grape leaves (

Identification of Volatile Metabolites
Grape leaves inoculated with B. cinerea showed obvious disease symptoms compared to noninoculated leaves (Figure 3). About 163 to 219 volatile compounds were detected in 'Victoria' and 'Shine Muscat' filtrated for absolute peak heights higher than 2000 (Figure 1b). Among them, 133 metabolites were selected and normalized by excluding outliers and qualitative analysis via the NIST14 and Fiehn mass spectrometry databases in each group of samples. Principal component analysis (PCA) analysis revealed two major differences within the metabolic profiles of 12 treated groups (Figure 4). In 'Victoria,' the metabolites were separated by sampling time, regardless of inoculation. In 'Shine Muscat,' the metabolites were not separated by inoculation, but 3 and 6 d postinoculation (dpi) samples were clustered in one group and 9 dpi was in another group.

Identification of Volatile Metabolites
Grape leaves inoculated with B. cinerea showed obvious disease symptoms compared to noninoculated leaves (Figure 3). About 163 to 219 volatile compounds were detected in 'Victoria' and 'Shine Muscat' filtrated for absolute peak heights higher than 2000 ( Figure  1b). Among them, 133 metabolites were selected and normalized by excluding outliers and qualitative analysis via the NIST14 and Fiehn mass spectrometry databases in each group of samples. Principal component analysis (PCA) analysis revealed two major differences within the metabolic profiles of 12 treated groups (Figure 4). In 'Victoria,' the metabolites were separated by sampling time, regardless of inoculation. In 'Shine Muscat,' the metabolites were not separated by inoculation, but 3 and 6 d postinoculation (dpi) samples were clustered in one group and 9 dpi was in another group.

Crucially Changed Volatile Metabolites
In the 'Victoria' group, eleven different metabolites at 3 dpi, six different metabolites at 6 dpi, and twenty-nine different metabolites at 9 dpi were identified (Table 3). For the 'Shine Muscat' group, five different metabolites at 3 dpi, eleven different metabolites at 6 dpi, and eleven different metabolites at 9 dpi were found (Table S4).

Crucially Changed Volatile Metabolites
In the 'Victoria' group, eleven different metabolites at 3 dpi, six different metabolites at 6 dpi, and twenty-nine different metabolites at 9 dpi were identified (Table 3). For the 'Shine Muscat' group, five different metabolites at 3 dpi, eleven different metabolites at 6 dpi, and eleven different metabolites at 9 dpi were found (Table S4).   ucially Changed Volatile Metabolites the 'Victoria' group, eleven different metabolites at 3 dpi, six different metabolites i, and twenty-nine different metabolites at 9 dpi were identified (Table 3). For the Muscat' group, five different metabolites at 3 dpi, eleven different metabolites at 6 d eleven different metabolites at 9 dpi were found (Table S4).
.  rucially Changed Volatile Metabolites n the 'Victoria' group, eleven different metabolites at 3 dpi, six different metabolites pi, and twenty-nine different metabolites at 9 dpi were identified (Table 3). For the Muscat' group, five different metabolites at 3 dpi, eleven different metabolites at 6 nd eleven different metabolites at 9 dpi were found (Table S4).  ), noninoculation at 9 dpi (C

Crucially Changed Vo
In the 'Victoria' gro at 6 dpi, and twenty-nin 'Shine Muscat' group, fi dpi, and eleven differen

Crucially Changed V
In the 'Victoria' gro at 6 dpi, and twenty-nin 'Shine Muscat' group, f dpi, and eleven differen

Crucially Changed Volatile Metabolites
In the 'Victoria' group, eleven different metabolites at 3 dpi, six different metabolites at 6 dpi, and twenty-nine different metabolites at 9 dpi were identified (Table 3). For the 'Shine Muscat' group, five different metabolites at 3 dpi, eleven different metabolites at 6 dpi, and eleven different metabolites at 9 dpi were found (Table S4).

Effect of Botrytis cinerea on Metabolic Pathways of Grapes
Metabolic pathways in grape seedlings were analyzed using the platform at Metabo Analyst. A total of 77 out of 92 nonvolatile differential compounds were categorized into 40 networks of plant metabolic pathways ( Table 4). The schematic diagram indicated a global disturbance of metabolomes in grape leaves under infection of B. cinerea. The regulated metabolites were involved in forty-one pathways, among which seven had impacts greater than 0.1, such as aminoacyl-tRNA biosynthesis; galactose metabolism; and valine, leucine, and isoleucine biosynthesis. Among these pathways, isoquinoline alkaloid biosynthesis; phenylpropanoid biosynthesis; monobactam biosynthesis; tropane, piperidine, and pyridine alkaloid biosynthesis; phenyl-alanine metabolism; and glucosinolate biosynthesis were important in producing antifungal compounds. Moreover, there were two different networks of volatile differential metabolic pathways in the two cultivars, which were sulfur metabolism and sesquiterpenoid and triterpenoid biosynthesis in 'Victoria' and monoterpenoid biosynthesis and sesquiterpenoid and triterpenoid biosynthesis in 'Shine Muscat'.

Discussion
We found a trend that B. cinerea infection induced production of several PSMs to be upregulated from those in noninoculated grape leaves, although the increase was moderate. In inoculated leaves, many volatile PSMs were upregulated, such as β-ocimene, α-farnesene, caryophyllene, germacrene D, β-copaene, and alkanes. Nonvolatile PSMs showed a complex network of relationships. These included GABA, resveratrol, piceid, and some amino acids. Among them, eugenol, flavanone, reserpine, resveratrol, and salicylic acid were highly upregulated in two varieties of grapes inoculated with both isolates of B. cinerea. These compounds, with antifungal activities against B. cinerea, promoted overexpression of ABC transporter genes, which is associated with induction of MDR of B. cinerea. These findings are supported by previous results [31,32].
Botrytis cinerea infection induced or increased upregulation of nonvolatile compounds, including 3-pyridinol, butanoic acid, threitol, GABA, resveratrol, piceid, flavanone, catechin, eugenol, and some amino acids, regardless of grape variety. γ-aminobutyric acid (GABA) plays as a signal molecule, eliciting plant defense against abiotic stresses [44], as it is involved in the expressions of genes for plant signal transduction, transcrip-tional control, hormone biosynthesis, reactive oxygen species generation, and polyamine metabolism [31,43,44]. Therefore, GABA is an important metabolite, contributing to disease resistance. The Lys catabolite pipecolic acid (Pip) is a critical metabolic mediator of several forms of inducible resistance in Arabidopsis, which accumulates following pathogen recognition [45]. Then, N-hydroxypipecolic acid (NHP) is the actual regulator of SAR rather than Pip [46,47]. This result was consistent with the downregulated Pip in the leaves of the treated group compared to the non-treated group, probably attributable to hydroxylation of Pip to NHP. Unfortunately, the compound directly involved in SAR was not detected.
Plants have developed several systems to regulate levels of ROSs. Abiotic stresses cause accumulation of ROSs and reduce photosynthetic activity, which triggers production of hydrogen peroxide (H 2 O 2 ) and compounds such as proline, glutathione, ascorbic acid, carotenoids, flavonoids, and tocopherols to alleviate oxidative damage by neutralizing ROSs [11]. Among the PSMs detected in this study, ascorbic acid was upregulated after B. cinerea infection regardless of grape variety. This indicates that ascorbic acid may act as a deoxidizer.
Plant secondary metabolites (PSMs) possessing antifungal activities can be a potential resource for fungicide development. For example, since resveratrol and its derivative are linked to plant immunity [32,48,49], they have been registered as an effective fungicide in China (http://www.chinapesticide.org.cn/, accessed on 31 December 2021) for controlling cucumber gray mold. In this study, resveratrol, reaveratroloside, and trans-piceid showed significant upregulation in B. cinerea-inoculated grape leaves. Resveratrol resulted in up to 47% inhibition of B. cinerea. This moderate efficacy can be increased by changing the chemical structure, which was demonstrated by in vivo methylation of hydroxyphenyl groups in phenolics [50].
Differential metabolites were more numerous in the resistant variety 'Shine Muscat' than in the susceptible variety 'Victoria'. 'Shine Muscat' had more compounds with higher levels of antifungal activity regardless of B. cinerea inoculation. These compounds might suppress infection and proliferation of pathogens [51]. Interestingly, some metabolites were regulated differently when the grape was infected with wild-type B. cinerea or with an MDRcontaining strain. While B. cinerea 242 can develop MDR, it may have a cost of fitness and reduced virulence toward the host plant, which was confirmed not only in this study but also in several other studies on B. cinerea. Therefore, acquisition of high-level resistance is associated with a decrease in fitness [52]. Meanwhile, grape plants with different resistance have different metabolomic activities, as shown in the susceptible variety 'Victoria' versus the resistant variety 'Shine Muscat.' This indicates that plant defense against pathogen infection through metabolomic regulation is dependent on its resistance level.
On one hand, PSMs, such as phytoalexins, can be induced by pathogens and increase plant defense against the pathogens [53,54]. On the other hand, the pathogens may be adaptive to and detoxify PSMs by enhancing the ABC efflux system [55][56][57][58]. In fungi, ABC transporters are involved in transportation of PSMs such as phenylpropanoids, lignans, flavonoids, and condensed tannins [59,60] and are also upregulated when exposed to some PSMs, such as eugenol and reserpine [61]. These studies support our results. We found that eugenol, flavanone, reserpine, resveratrol, and salicylic acid inhibited B. cinerea but promoted, in the fungus, overexpression of three major ABC transporter genes: BcatrB, BcatrD, and BcatrK. As such, we speculate that when exposed to antimicrobial PSMs, the pathogen would carry out the efflux of PSMs, impeding their antimicrobial activity. All tested PSMs exerted growth inhibition but could potentially elicit preadaptation in B. cinerea.

Conclusions
In this study, we found that B. cinerea infection triggered grapes to increase production of some secondary metabolites that have antifungal bioactivities and therefore enhanced plant resistance to diseases such as B. cinerea. We identified indicator compounds in grapes during infection and discovered key secondary metabolites that are associated with the defensive activities of grapes in inhibiting B. cinerea. The downside of these PSMs is that the pathogen may be adaptive to and detoxify PSMs by enhancing the ABC efflux system, which impedes the antimicrobial activity of PSMs. These findings enhanced our understanding of plant resistance in host plants and provided a new perspective for development of MDR of B. cinerea. PSM adaptation with the pathogen should be considered when a fungicide is to be developed using PSMs and their derivatives.

Conflicts of Interest:
The authors declare no conflict of interest.