An increase in the number of peroxisomes is coupled to the initial infection stage and stress response of Botrytis cinerea

Peroxisomes are very important organelles in eukaryotic cells and participate in various biological processes, including pathogen–host interactions. A variety of proteins involved in peroxisome proliferation and metabolic activity within peroxisomes have been shown to be essential for full virulence of several phytopathogenic fungi. However, the effects of changes in the number of peroxisomes and proteins involved in the peroxisome pathway on the pathogenicity of Botrytis cinerea have rarely been reported. In this study, by analysing transcriptome data and RT-qPCR validation, we found that more than half of the genes annotated to the peroxisome pathway in B. cinerea were upregulated more than twofold between mycelial samples cultured in medium with tomato leaves and without tomato leaves. A strain of B. cinerea with fluorescently labelled peroxisomes was obtained by overexpression of GFP fused to peroxisomal targeting signal 1 (the tripeptide ‘SKL’). The addition of tomato leaves to the liquid medium induced a significant increase in the number of peroxisomes, β-oxidation level, H2O2 content, and acetyl-CoA level in B. cinerea mycelia. When B. cinerea was cultured with oleic acid as the sole carbon source, the formation of infection-related structures and their penetration into plant cells were found to be associated with peroxisome pathway activity. Furthermore, peroxisome proliferation and lipid metabolism increased in response to different extracellular stresses in B. cinerea. Taken together, our results confirmed that activation of the peroxisome pathway in B. cinerea contributes to the initial infection and the ability to cope with environmental stress.


Background
Botrytis cinerea, the causal agent of grey mould disease, is one of the most invasive pathogens and affects more than 1000 species of 586 plant genera, including numerous agriculturally important crop cultivars (Dean et al. 2012;Elad et al. 2016a). As a typical necrotrophic plant pathogen, B. cinerea can attack different plant tissues, causing enormous pre-and post-harvest crop loss and leading to incalculable economic loss annually (Elad et al. 2016b;Soltis et al. 2019). During pathogenesis, B. cinerea releases a large number of virulence factors, most of which are proteins that regulate the production or catabolism of reactive oxygen species (ROS), various proteases (cutinases, lipases, glycosyl hydrolases, pectin methylesterases, etc.) that degrade components of the host cuticle and cell wall or generate assimilable nutrients, or various pathogen-associated molecular patterns that suppress the host immune system (González et al. 2016;Siegmund and Viefhues 2016). Currently, chemical control based on the application of chemical fungicides remains the easiest and most effective way to manage grey mould epidemics in many crop species (Fillinger and Walker 2016). However, under long-term selective pressure, B. cinerea has developed varying degrees of Han et al. Phytopathology Research (2022) 4:25 tolerance to almost all fungicides, which severely weakens the control efficacy of these fungicides on grey mould (Hu et al. 2016). Point mutations in drug targets of B. cinerea are mainly responsible for high levels of resistance to fungicides with site-specific activity (Chatzidimopoulos et al. 2016). Therefore, new targets need to be identified to explore more effective fungicides to control the occurrence of B. cinerea.
Peroxisomes are a class of ubiquitous and crucial organelles that perform a variety of important cellular functions in nearly all eukaryotic cells (Wanders 2013;Islinger et al. 2018). In accomplishing cellular tasks or in response to various environmental stimuli, the abundance of peroxisomes can be changed and coordinated through several cellular processes, including peroxisome biogenesis, proliferation, and degradation (Smith and Aitchison 2013;Chen et al. 2016). The assembly of peroxisome biogenesis proteins (collectively called peroxins) is required for the biogenesis and proliferation of peroxisomes and the import of proteins into the peroxisomal matrix (Distel et al. 1996;Smith and Aitchison 2013). To date, more than 30 peroxins have been shown to function in peroxisome formation; among them, 3 peroxins have been characterized as import receptors that recognize peroxisomal matrix proteins, including peroxisomal targeting signal 1 or 2 (PTS1 or PTS2) (Stanley et al. 2007;Kong et al. 2019). The conserved functions of peroxisomes include β-oxidation of fatty acids and metabolism of hydrogen peroxide (H 2 O 2 ). In filamentous fungi, peroxisomes are also involved in many other metabolic pathways, such as the biogenesis of Woronin bodies (WBs) and secondary metabolites, all of which are important for fungal development and pathogenesis (van der Klei and Veenhuis 2013; Chen et al. 2016). β-Oxidation of fatty acids in peroxisomes can lead to an abundance of acetylcoenzyme A (acetyl-CoA), which can be used for the biosynthesis of polyketides and biotin, the production of glucose via the glyoxylate cycle and gluconeogenesis, and the synthesis of melanin and cell wall components (van der Klei and Veenhuis 2013; Chen et al. 2016).
The functions of peroxisomes have been studied in various organisms, and increasing amounts of evidence have shown that peroxisomes and several peroxins are essential for the pathogenicity of several plant pathogenic fungi (Kubo 2013;Chen et al. 2016). In the anthracnosecausing fungus Colletotrichum orbiculare, appressorial melanization and penetration depend on β-oxidation and the peroxins PEX6, PEX13, and MEF (Kimura et al. 2001;Fujihara et al. 2010;Asakura et al. 2012). In Fusarium graminearum, the peroxins PEX5, PEX6, PEX13, PEX14, PEX33, PEX4, and PEX2 have been reported to be associated with pathogenicity (Min et al. 2012;Chen et al. 2018;Zhang et al. 2019;Wang et al. 2020). An overview of the relationship between peroxisomes and pathogenicity in the rice blast fungus Magnaporthe oryzae has also been provided (Chen et al. 2016). Multiple proteins that play important roles in the maintenance of peroxisomal functions in M. oryzae have been reported to be associated with pathogenicity. Specifically, single-deletion mutants of several of these proteins, such as PEX19, PEX5, PEX6, and PEX7, exhibited complete loss of pathogenicity (Chen et al. 2016). The role of peroxisomal proteins in pathogenesis of various plant pathogens has gradually become clear. However, most relevant studies have focused on the phenotypic changes resulting from the deletion of a certain peroxisomal protein, and little attention has been given to the relationship between the number of peroxisomes and pathogenicity. Moreover, the roles of peroxisomes in B. cinerea have not been studied extensively.
In this study, we focused on the overall changes in peroxisomes in B. cinerea. Using reverse transcriptionquantitative PCR (RT-qPCR), we measured the expression levels of genes annotated to the peroxisome pathway and labelled peroxisomes with green fluorescent protein (GFP) to observe quantitative changes. Our results demonstrate that the number of intracellular peroxisomes and β-oxidation in B. cinerea is significantly increased during the initial infection stage and in response to different culture conditions and environmental stimuli.

Genes annotated to the peroxisome pathway in B. cinerea were differentially expressed at the initial stage of infection
To investigate the involvement of the peroxisome pathway in the early stages of B. cinerea infection, we analysed transcriptomic data to identify the corresponding genes. A total of 66 genes were annotated to the peroxisome pathway (Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway bfu04146), of which 38 were differentially expressed by more than twofold (P value < 0.05) between B. cinerea cultured in medium with tomato leaves (BcT) and without tomato leaves (BcU) according to the transcriptomic data of three biological replicates ( Fig. 1 and Additional file 1: Table S1). Among these 38 genes, 34 and 4 genes were upregulated and downregulated, respectively, at 6 h after induction with tomato leaves versus without tomato leaves (BcU vs. BcT). These differentially expressed genes (DEGs) were mainly involved in the process of peroxisome formation (biogenesis, elongation, and fission), fatty acid β-oxidation, and ROS homeostasis in the peroxisome pathway (Fig. 1).
The relative expression levels of 12 representative DEGs (10 upregulated and 2 downregulated) were validated by RT-qPCR. Hierarchical cluster analysis was performed based on the relative expression levels of these 12 genes in 9 samples (conidia at 0 h and germinated conidia collected at 3, 6, 9, and 12 h post-inoculation (hpi) from medium with or without tomato leaves). The results showed that the samples induced by the addition of tomato leaves clustered together, while the samples without leaf induction clustered together (Fig. 2a). The expression levels of the 12 genes were significantly upregulated at the germination stage (3, 6, 9, and 12 hpi) compared with those at the conidial stage (0 h) (Fig. 2b). Importantly, the expression levels of the 10 upregulated DEGs in the samples with leaf induction were significantly higher than those in the samples without leaf induction at the germination stage (Fig. 2b). In addition, the expression levels of 2 downregulated DEGs in the samples with leaf induction were lower than those in the samples without leaf induction at 3, 6, and 9 hpi, and the levels with and without leaf induction were the same at 12 hpi (Fig. 2b).

Bioinformatics analysis of B. cinerea genes annotated to the peroxisome pathway
To further understand the functions of these DEGs, Gene ontology (GO) enrichment analysis, BLAST searches of the pathogen-host interaction (PHI) database, and protein-protein interaction (PPI) analysis were performed. Three DEGs were enriched in the biological process category, 5 DEGs were enriched in the cellular component category, and 9 DEGs were enriched in the molecular function category, with a threshold of P<0.05 (Additional file 2: Figure S1a and Additional file 1: Table S2). Among the proteins encoded by 37 DEGs, 27 were assigned to the PHI database, with an e-value < 1E-5 cut-off. Of these proteins, 7 belonged to the loss of pathogenicity category, and 16 and 1 were associated with reduced virulence and increased virulence, respectively (Additional file 2: Figure  S1b and Additional file 1: Table S3). Twenty-two proteins were mapped to an interaction network with a medium confidence score of 0.4 and were shown to cluster into three groups via the Markov cluster (MCL) algorithm (Additional file 2: Figure S1c).

Visualization of peroxisomes in B. cinerea via SKL-tagged GFP without affecting fungal growth and development
To visualize the peroxisomes, we tagged the C-terminus of GFP with the PTS1 tripeptide SKL, a widely used peroxisomal targeting signal, and then transformed the construct into the protoplasts of B. cinerea (Fig. 3a). There was no significant difference in mycelial growth and conidial production between the wild-type strain B05.10 and the transformants GFP-SKL-1 and GFP-SKL-2 ( Fig. 3b-d). Under fluorescence microscopy, the fluorescence produced by these two transformants (represented by GFP-SKL-1) was present in a punctate pattern throughout the cytoplasm, while the fluorescence produced by GFP only (GFP with no SKL tag) was distributed evenly throughout the cell (Fig. 3e). These punctate structures did not overlap with the nucleus (stained with 4' ,6-diamidino-2-phenylindole (DAPI)) and were smaller than the nucleus (Fig. 3e). Western blot analysis showed that both GFP-SKL and GFP were expressed in the corresponding strains (Fig. 3f ).
Lipids have been shown to induce an increase in the number of peroxisomes (Chen et al. 2017). When Tween-80 was used as the sole carbon source, the number of fluorescent punctate structures in conidia and hyphae was increased at 6 and 9 hpi compared with that cultured in yeast extract peptone dextrose (YEPD) liquid medium (Additional file 2: Figure S2a, b). In particular, the number of fluorescent punctate structures per unit of hyphal length (0.1 mm) in strain GFP-SKL grown in Tween-80 was significantly higher than that in YEPD medium (Additional file 2: Figure S2c, d). As expected, the distribution and the increase in the number of fluorescent punctate structures were consistent with the known marker for peroxisomes, indicating that the peroxisomes of B. cinerea were indeed marked by GFP and could be observed under fluorescence microscope.

The peroxisome pathway in B. cinerea was activated at the initial stage of infection
Changes in the number of peroxisomes were observed and quantified in the B. cinerea strain GFP-SKL cultured in YEPD medium with or without tomato leaves ( Fig. 4a, b). The results showed that when tomato leaves were added to the medium, the number of peroxisomes in the hyphae of GFP-SKL was significantly (P<0.01) greater than that in the medium without leaves at 6 hpi, but in the conidia, no significant difference (P>0.05) in peroxisome numbers was found between these two treatments; at 9 hpi, the number of peroxisomes in both the hyphae and the conidia under the induction with leaves was slightly higher than that without leaves. However, there was no statistically significant difference (P>0.1) in hyphal length between the two groups ( Fig. 4c). Malondialdehyde (MDA) is a common lipid peroxidation product and is widely used as a biomarker for lipid peroxidation and ROS accumulation (Miret et al. 2018). When tomato leaves were added to the medium, the MDA content in the mycelia was significantly (P<0.01) increased at 9 hpi compared with that in the medium with no leaves (Fig. 4d), indicating that the presence of tomato leaves increased lipid oxidation in B. cinerea mycelia. The lipid β-oxidation process is accompanied by the generation of large amounts of ROS, which are scavenged in peroxisomes to maintain ROS homeostasis in cells (Knoblach et al. 2013;Tripathi et al. 2016). However, when tomato leaves were added to the medium, the H 2 O 2 content in B. cinerea mycelia was significantly (P<0.01) higher than that in the medium with no tomato leaves at both 6 and 9 hpi (Fig. 4e). In addition, under induction with tomato leaves, the accumulation level of acetyl-CoA in mycelia was significantly higher than that without tomato leaves ( Fig. 4f ), which may allow B. cinerea to synthesize more secondary metabolites to infect the host (van der Klei and Veenhuis 2013; Chen et al. 2016).

Fatty acids induced an increase in the number of peroxisomes and the formation of infection structures of B. cinerea
It has been reported that fructose can promote the formation of appressorium (AP)-like structures in B. cinerea (Liu et al. 2018). To understand the potential relationship between the activation of the peroxisome pathway and the infection ability of B. cinerea at the early infection stage, the numbers of peroxisomes and AP-like structures were quantified under different culture conditions (Fig. 5). Under culture with glucose or fructose as the sole carbon source compared with the group cultured in YEPD medium, the total number of peroxisomes showed no significant difference in either the hyphae or conidia at 6 and 9 hpi, but the number per unit length (0.1 mm) of hyphae was greater because the hyphae were shorter in the glucose or fructose medium than in the YEPD medium ( Fig. 5a-c). Conversely, under culture with glucose or fructose as the sole carbon source, the number of AP-like structures formed by hyphae was significantly higher than that in YEPD medium (Fig. 5d). In particular, under culture in fructose, the number of AP-like structures was at least 5 times and twice as high as that under culture in YEPD medium and glucose, respectively. When oleic acid was used as the sole carbon source, the number of peroxisomes in the conidia, hyphae, and per unit hyphae increased significantly (Fig. 5a-c). Consistent with this, culturing in oleic acid compared with that in YEPD medium or in glucose also increased APlike structure formation (Fig. 5d). Taken together, these results suggest that the increase in AP-like structure formation may be independent of peroxisome pathway activation in B. cinerea when cultured with sugars but may be positively correlated with peroxisome pathway activation in response to fatty acid induction. The transcript levels of 6 DEGs annotated to the peroxisome pathway in mycelia of B. cinerea cultured in different media were evaluated by RT-qPCR. Most genes exhibited higher transcript levels in mycelia cultured in oleic acid at both time points (6 and 9 hpi) than those cultured in YEPD, glucose, or fructose (Fig. 5e). Onion epidermal cell infection experiments demonstrated that the hyphae were more likely to invade plant cells under culture conditions that facilitated the production of more AP-like structures (fructose > oleic acid > glucose) (Fig. 6a, b), consistent with the aforementioned result that more peroxisomes were produced in hyphae at the initial infection stage in the presence of oleic acid ( Fig. 6c-e). Meanwhile, the transcript levels of these 6 DEGs in mycelia of B. cinerea collected from onion epidermal cells were higher than those in mycelia collected from cellophane (Fig. 6f ).

B. cinerea peroxisomes may function in response to extracellular stress
Peroxisome fission affects the cell wall integrity of M. oryzae (Chen et al. 2017). To test the responses of B. cinerea peroxisomes to different extracellular stresses, we cultured B. cinerea in the presence of different stress agents in YEPD medium, including the cell wall stressor Congo red (CR), the oxidative stressor H 2 O 2 , and biological stressors from the biocontrol bacterium Bacillus velezensis QSE-21 (Xu et al. 2021). These reagents inhibited the hyphal growth of B. cinerea to varying degrees at 6 and 9 hpi (Fig. 7a, b). Compared with the control group cultured in YEPD medium without adding stress reagents, the number of peroxisomes produced in B. cinerea in response to these three stress reagents showed no significant difference in the conidia, a small increase in the hyphae at 9 hpi, and a significant increase in hyphae per unit length (Fig. 7a, c). Meanwhile, the transcript levels of the 6 DEGs annotated to the peroxisome pathway in B. cinerea cultured in YEPD medium or YEPD medium containing QSE-21 were measured by RT-qPCR. The results showed that most of these genes were transcriptionally upregulated in response to QSE-21 (Fig. 7d), indicating that the peroxisome pathway in B. cinerea may be activated in response to various extracellular stressors.

Discussion
Research on the functions of peroxisomes has been ongoing for decades, and peroxisomes are known to be widely involved in multiple biological processes.
In recent years, studies have shown that peroxisomes are related to several human genetic diseases (Wanders 2013) and perform functions in pathogen-host plant interactions (Kubo 2013). In the present study, we found that approximately 52% (34/66) of the genes in the peroxisome pathway (bfu04146) were upregulated more than twofold at the initial stage of B. cinerea infection (Figs. 1, 2). We also observed that peroxisomal functions in B. cinerea can be enhanced by an increase in peroxisome number during the initial infection stage, during growth in medium with oleic acid as a sole carbon source or in onion epidermal cells, and in response to extracellular stress (Figs. 4-7). Interestingly, we found that the formation of AP-like structure and the ability of B. cinerea to penetrate plant cells when using oleic acid as the sole carbon source may be positively correlated with an increase in peroxisome number (Figs. 5, 6). These results indicated that lipid metabolism in peroxisomes plays important roles in B. cinerea initial infection. Intracellular storage lipids, mainly triacylglycerols (TAGs), are universally used for energy storage in fungal conidia (Both et al. 2005;Ma et al. 2021), and the expression of lipid metabolism-related genes is strongly induced during infection of plant hosts by fungal pathogens (Chen et al. 2018;Lanver et al. 2018). TAGs can be transported from conidial cells to the incipient APs and then decomposed by intracellular lipases to release fatty acids and glycerol, after which the fatty acids can be metabolized through the peroxisome pathway, which is essential for turgor generation and appressorial penetration during fungal pathogen infection (Thines et al. 2000;Chen et al. 2017). In peroxisomes, fatty acid β-oxidation begins with a reaction catalysed by acyl-CoA oxidase (ACOX) and then by other multifunctional β-oxidation proteins, including 2-enoyl-CoA hydratase, 3-hydroxacyl-CoA dehydrogenase, and 3-ketoacyl-CoA thiolase; finally, a two-carbon unit acetyl-CoA is produced (Wang et al. 2007). In M. oryzae, MFP1, a protein that functions as a peroxisomal β-oxidation enzyme, and PTH2 and CRAT2, which function in acetyl-CoA transport, have been shown to play important roles in fatty acid metabolism and pathogenesis (Bhambra et al. 2006;Chen et al. 2016). The produced acetyl-CoA is then used for gluconeogenesis and synthesis of other metabolites, including polyketides, biotin, melanin, and cell wall components, in peroxisomes and other organelles (van der Klei and Veenhuis 2013; Chen et al. 2016).
The metabolism of intracellular fatty acids requires the peroxisome pathway, and exogenous addition of fatty acids can also induce the activation of the peroxisome pathway (Chen et al. 2017;Kong et al. 2019). Plant cell surfaces and intercellular spaces are rich in various lipids (Li et al. 2007;Keymer and Gutjahr 2018). Regardless of whether they synthesize fatty acids independently, both symbiotic and parasitic fungi in plants can exploit host lipids or lipid sources to sustain colonization or facilitate invasion (Jiang et al. 2017;Keymer and Gutjahr 2018). The assimilation of plant structural lipids by fungi benefits from multiple secreted lipases and phospholipases, which function as hydrolytic or lytic enzymes to degrade host waxes, cuticles, cell walls, and plasma membranes (Bravo-Ruiz et al. 2013). These extracellular lipolytic enzymes are associated with adhesion to and penetration of the plant surface by fungal pathogens and the generation of nutrients during certain stages of the fungal life cycle (Feng et al. 2009;Bravo-Ruiz et al. 2013;González et al. 2016). In B. cinerea, a secreted lipase, Lip1, has been identified and shown to be induced in the early stages of infection (Comménil et al. 1995;CommÉNil et al. 1999), but deletion of Lip1 has no effect on the pathogenicity of the fungus on tomato and bean leaves (Reis et al. 2005). The function of host lipids and secreted lipases in the interaction of B. cinerea with its hosts therefore needs further investigation.
Oxidative reactions in peroxisomes are very important sources of intracellular ROS, mainly including superoxide radicals (O 2 ▪− ) and H 2 O 2 (Sandalio et al. 2013;Chen et al. 2016). Peroxisomes are enriched with antioxidant enzymes, including superoxide dismutase (SOD) and catalase; thus, excess ROS that are toxic to living cells can be rapidly scavenged in this organelle (Sandalio et al. 2013;Kong et al. 2019). Our results showed that the transcript level of a few catalase-coding genes was downregulated and that the H 2 O 2 content increased at the initial stage of B. cinerea infection (Figs. 1, 4). This may be related to the dual function of ROS in mediating pathogen-host interactions. In M. oryzae-rice interactions, the accumulation of ROS is important for the AP development of the pathogen and for the host plant to initiate local programmed cell death to restrict the spread of invasive hyphae (Liu and Zhang 2021). When attacked by pathogens, plant hosts generally react with rapid and transient accumulation of ROS, referred to as an oxidative burst, to activate their defence system against pathogen invasion. However, the oxidative burst seems to be essential for successful infection by B. cinerea (Siegmund and Viefhues 2016). The absence of BcSOD1, a peroxisomal matrix protein, was shown to reduce the virulence of B. cinerea on different hosts (Rolke et al. 2004

Conclusions
In summary, the working model of peroxisomal metabolism in B. cinerea during the initial stage of infection or in response to stress is shown in Fig. 8. The metabolism of intracellular or extracellular host lipids increases via stimulation of peroxisome proliferation. Accompanying the oxidation of fatty acids in peroxisomes, a large amount of acetyl-CoA is produced, which can be used to synthesize other compounds, form infection structures or cope with stress; and a high level of H 2 O 2 is accumulated, which may be used to damage host tissues. Of course, there remain many issues that need to be addressed in future research, such as how B. cinerea uses host lipids to complete infection, how excessive H 2 O 2 is transported to host cells, and how the expression of genes involved in peroxisome formation and matrix metabolism is coordinated. As the peroxisome pathway may play important roles in the initial stage of infection by fungal pathogens, the key proteins in this pathway may be good targets for the development of new fungicides.

Bioinformatics analysis of DEGs annotated to the peroxisome pathway
In the genome of B. cinerea, a total of 66 genes were annotated to the peroxisome pathway (bfu04146: https:// www. kegg. jp/ pathw ay/ bfu04 146) in the KEGG database. These 66 genes were used to search a transcriptome sequencing data of B. cinerea mycelia collected from medium with or without tomato leaves induction (BcT or BcU) at 6 hpi, which has been deposited in NCBI Sequence Read Archive (SRA) with an accession number PRJNA837395. The transcriptome data contained 2,975 DEGs listed in Additional file 1: Table S4 with a threshold of P value < 0.05 and foldchange ≥ 2. Hierarchical clustering of DEGs annotated to the peroxisome pathway was performed using Multiexperimental Viewer (MeV) software (Saeed et al. 2006) according to their expression patterns from the transcriptome data with three biological replicates or from RT-qPCR results. GO annotation was implemented on the online server QuickGo (https:// www. ebi. ac. uk/ Quick GO/), and the genes were classified by GO annotation based on the categories: biological process, cellular component, and molecular function. Protein sequences, encoded by DEGs annotated into peroxisome pathway, were downloaded from the B. cinerea genome database (ASM83294v1) in EnsemblFungi (http:// fungi. ensem bl. org/ Botry tis_ ciner ea/ Info/ Index). And then sequences were used to blast the PHI database (http:// www. phi-base. org/) (Urban et al. 2020) with an e-value < 1E-5 cut-off as previously mentioned (Bashyal et al. 2017). PPI analysis was performed using the Search Tool for the Retrieval of Interacting Genes/ Proteins (STRING version 11.5, https:// cn. string-db. org/) with a confidence score larger than 0.4 and clustered by the MCL algorithm.

RNA extraction and RT-qPCR
The mycelia cultured under the above-mentioned conditions was collected by centrifugation and blotted dry with filter paper. Total RNA was extracted from mycelia or conidia using the OMEGA fungal RNA kit (E.Z.N.A. R6840), and 1 μg of total RNA of each sample was used for the first strand cDNA synthesis using the M5 Super plus qPCR RT kit with gDNA remover (Mei5bio, MF166plus  Table S5.

Fungal strains and culture conditions
The wild-type strain B05.10 of B. cinerea and its derivative transformants were grown and maintained on PDA medium (200 g potato, 20 g glucose, 15 g agar, and 1 L water) under conditions as described previously (Giesbert et al. 2011). To mimic the initial stage of infection, conidia suspension of B. cinerea were inoculated into 10% YEPD liquid medium (10 g yeast extract, 20 g peptone, 20 g glucose, and 1 L water) supplemented with 3-week-old tomato leaves (about 20 g leaves per 100 mL) and cultured on a rotary shaker at 25 ℃ and 120 rpm. The single carbon source tests were performed in minimal liquid medium (MM: 0.5% (NH 4 ) 2 SO 4 , 0.1% KH 2 PO 4 , 0.05% MgSO 4 ) supplemented with glucose, fructose, or oleic acid at a final concentration of 10 mM under the above-mentioned conditions. To simulate the extracellular stress, conidia of B. cinerea were inoculated into 10% YEPD liquid medium and cultured on a rotary shaker at 25 ℃ and 180 rpm for 3 h, followed by adding CR (final concentration 0.3 mg/mL), H 2 O 2 (final concentration 5 mM), or cell free fermentation broth of Bacillus velezensis QSE-21 (CFB-Q) (final concentration 10%), and then were incubated for another 3 h or 6 h. The CFB-Q was prepared as described previously (Xu et al. 2021). B. cinerea conidia harvested from 7-day-old PDA plate was suspended in the above-mentioned liquid medium at a final concentration of 1×10 5 conidia/mL.

Generation of GFP-SKL labelled B. cinerea strains and fungal development assays
The coding sequence of the tripeptide 'SKL' , TCA AAG TTG, was added to the 3` end of GFP sequence by PCR using primers listed in Additional file 1: Table S5. PCR product was then ligated into the NcoI/NotI-digested pNDT-OGG vector (Schumacher 2012). Protoplast preparation and transformation were performed as described previously (Gronover et al. 2001). The resulting hygromycin-resistant transformants were preliminarily validated by PCR followed by several rounds of single conidium isolation. Fungal growth and conidiation assays on PDA were performed as described previously (Feng et al. 2017).

Western blot analysis
B. cinerea conidia harvested from 7-day-old PDA plates were inoculated in YEPD medium and cultured on a rotary shaker at 25 ℃ and 180 rpm for 16 h. The mycelia were harvested by filtration through filter paper and washed several times with sterile distilled water. Total proteins were extracted from mycelia as described previously (Gu et al. 2015). Equal volume of protein samples was loaded and separated on 12% sodium dodecylsulphate-polyacrylamide gel electrophoresis (SDS-PAGE), and then transferred to polyvinylidene fluoride (PVDF) membrane with a Bio-Rad electroblotting apparatus. The anti-GFP antibody (Abcam, Cambridge, Cat#Ab32146) was used to detect GFP and GFP-SKL protein. The experiment was performed three times independently.

Microscopic examinations
For the observation of GFP signal, fresh hyphae cultured in different liquid media were transferred to a glass slide and then examined with a fluorescence microscope under the GFP channel. For the observation of nuclei in hyphae, fresh hyphae were stained with 10 μg/mL of DAPI and then examined with a fluorescence microscope under the UV channel. The morphology of hyphae was examined in the bright field. The number of fluorescent spots was counted and the length of fungal hyphae was measured using ImageJ software from images taken under the same conditions. For AP-like structure formation assay, 20 μL-droplets of conidial suspension (1×10 5 conidia/mL) were placed onto a glass slide and incubated in a moistened box at 25 °C. The proportion of AP-like formation inoculated into different media was determined by microscopic examination as described previously (Feng et al. 2017). The infection assay in onion epidermal cells was performed as described previously using conidial droplets (1×10 4 conidia/mL) suspended in different media. The proportion of invasive hyphae to total conidia in each image was determined using ImageJ software.

Detection of lipid β-oxidation level, H 2 O 2 content and acetyl-CoA level
Mycelia cultured in liquid medium was collected by centrifugation and blotted dry with filter paper. An amount of 0.1 g mycelia was homogenized in a glass tissue homogenizer containing 1 mL of ice-cold PBS buffer (8.0 g NaCl, 0.2 g KCl, 1.44 g Na 2 HPO 4 , 0.24 g KH 2 PO 4 , 1 L H 2 O). The homogenate was transferred to a tube followed by centrifuging for 10 min at 12,000 × g at 4 °C and then the supernatant was transferred to a new tube and kept at 4 °C before use. Lipid β-oxidation level was then determined by detecting the content of MDA using a Lipid Peroxidation MDA Assay Kit (Beyotime, S0131S) based on thiobarbituric acid (TBA) test (Heath and Packer 1968). Measurement of H 2 O 2 content was carried out using a Hydrogen Peroxide Assay Kit (Beyotime, S0038) based on the Ferric Xylenol Orange method (Gay et al. 1999). The measurement of acetyl-CoA level was carried out using an Acetyl-CoA Content Assay Kit (Solarbio, BC0980) based on the formation rate of NADH, which is directly proportional to the acetyl-CoA level through the coupled reaction between malate dehydrogenase and citrate synthase. And the quantification of NADH was further carried out by spectrophotometry. For comparison between samples, the concentration of MDA and H 2 O 2 was normalized with the protein concentration in the sample and expressed as μmol/g protein. The protein concentration was determined using BCA method performed by a BCA Protein Assay Kit (Beyotime, P0012S). Three biological replicates for each experiment and three technical replicates for each biological replicate were performed.

Statistical analysis
All data were expressed as the means ± standard deviations obtained from more than three replicates per experimental condition. The least significant differences (LSD) method for multiple comparisons was carried out with Statistical Analysis Systems (SAS) software to determine differences between means.