A novel effector CfEC92 of Colletotrichum fructicola contributes to glomerella leaf spot virulence by suppressing plant defences at the early infection phase

Abstract The ascomycete fungus Colletotrichum fructicola causes diseases on a broad range of plant species. On susceptible cultivars of apple, it induces severe early defoliation and fruit spots, named glomerella leaf spot (GLS), but the mechanisms of pathogenicity have remained elusive. Phytopathogens exhibit small secreted effectors to advance host infection by manipulating host immune reactions. We report the identification and characterization of CfEC92, an effector required for C. fructicola virulence. CfEC92 is a Colletotrichum‐specific small secreted protein that suppresses BAX‐triggered cell death in Nicotiana benthamiana. Accumulation of the gene transcript was barely detectable in conidia or vegetative hyphae, but was highly up‐regulated in appressoria formed during early apple leaf infection. Gene deletion mutants were not affected in vegetative growth, appressorium formation, or appressorium‐mediated cellophane penetration. However, the mutants were significantly reduced in virulence toward apple leaves and fruits. Microscopic examination indicated that infection by the deletion mutants elicited elevated deposition of papillae at the penetration sites, and formation of infection vesicles and primary hyphae was retarded. Signal peptide activity, subcellular localization, and cell death‐suppressive activity (without signal peptide) assays suggest that CfEC92 could be secreted and perform virulence functions inside plant cells. RNA sequencing and quantitative reverse transcription PCR results confirmed that the deletion mutants triggered elevated host defence reactions. Our results strongly support the interpretation that CfEC92 contributes to C. fructicola virulence as a plant immunity suppressor at the early infection phase.


| INTRODUC TI ON
In natural environments, plants are attacked by numerous pathogens and evolve sophisticated immunity surveillance systems to protect themselves (Jones and Takemoto, 2004;Chisholm et al., 2006). A "zigzag" model summarizes the innate immunity in plant-pathogen interactions (Jones and Dangl, 2006). The pattern recognition receptors (PRRs) of plant cell surface receptors can detect pathogen-associated molecular patterns (PAMPs) in the first line of defence, called PAMP-triggered immunity (PTI). In turn, successful pathogens interfere with PTI by means of secreting effectors into host cells, resulting in effector-triggered susceptibility (ETS). After that, plants activate the second layer of immune system through resistance (R) proteins that recognize pathogen effectors directly or indirectly, known as effector-triggered immunity (ETI). However, successful pathogens can evolve new effectors that translocate into host cells to interfere with the plant immune system, facilitating pathogen colonization (Dou and Zhou, 2012;Dangl et al., 2013;Wang and Wang, 2018).
Colletotrichum is a large genus of ascomycete plant pathogens infecting diverse field crops, vegetables, and fruits worldwide (Bailey and Jeger, 1992;Prusky et al., 2000). Most Colletotrichum species have a hemibiotrophic lifestyle. They develop penetration pegs from specialized melanized appressoria to invade living host cells, and then produce bulbous biotrophic infection vesicles and primary hyphae, before switching to a necrotrophic stage featuring secondary hyphae formation (Perfect et al., 1999). Generally, phytopathogens deploy effectors to manipulate the host physiological metabolism and to suppress immune responses in order to establish biotrophy at an early phase of infection (Bozkurt et al., 2012;Rafiqi et al., 2012).
Later, toxins and lytic enzymes are produced to kill and degrade host tissues in the necrotrophy mode (Gan et al., 2012;O'Connell et al., 2012).
Using genome sequencing and bioinformatics, a large number of effector candidate genes have been predicted genome-wide in hemibiotrophic Colletotrichum species, including C. gloeosporiodes, C. graminicola, C. higginsianum, and C. orbiculare (Crouch et al., 2014). It has been suggested that each Colletotrichum species has evolved a unique set of effector genes to adapt to its own hosts (Crouch et al., 2014). Furthermore, transcriptomic analysis of the infection process in C. higginsianum and C. orbiculare has indicated dynamic effector gene expression, supporting infection phase-specific virulence roles of effector genes (Gan et al., 2012;Kleemann et al., 2012;O'Connell et al., 2012). However, functional characterization of identified effector candidate genes has been lagging in Colletotrichum. So far, only a few effectors have been demonstrated to be important for virulence (Bhadauria et al., 2012;Sanz-Martin et al., 2016), and in even fewer cases have the functional mechanisms been determined (Irieda et al., 2014;Azmi et al., 2018).
Glomerella leaf spot (GLS) is a disease of apple trees (Malus domestica) in Brazil, China, Japan, and the USA (Leite et al., 1988;González and Sutton, 1999;Wang et al., 2012;Yokosawa et al., 2017). Under favourable conditions, GLS can result in more than 90% defoliation in susceptible cultivars such as Gala and Golden Delicious, seriously weakening the trees and reducing yield; fruits with spots are reduced in quality (Wang et al., 2015b). In China, the disease occurs commonly from mid-July to early August, a period of high temperature and humidity. In addition, the latent period is as short as 2 days and the disease spreads rapidly, especially after prolonged periods of rain. Therefore, control of GLS is challenging; to date, there are no consistently effective management approaches (Leite et al., 1988;Wang et al., 2014).
Based on the results of a transient expression assay in Nicotiana, we identified a novel effector, CfEC92, that inhibited BAX-induced programmed cell death (PCD) and was highly expressed at the early infection phase. We also demonstrated that CfEC92 gene deletion did not affect fungal vegetative growth or development, but strongly reduced virulence. This virulence reduction was related to elevated host defence reactions at the early infection phase and reduced efficiency in differentiation of infection vesicles and primary hyphae. Our results demonstrate that the CfEC92 is important for GLS virulence and provide insights regarding the molecular interactions between C. fructicola and its apple host.

| CfEC92 is a Colletotrichum genus-specific effector candidate with cell death-suppressive activity
Among the effector candidates previously identified in C. fructicola (Liang et al., 2018), CfEC92 (1,104|04,129) showed in planta specific expression compared with conidia, in vitro appressoria, and infectious hyphae on cellophane. The gene encodes a putative protein with 85 amino acid residues and has a high cysteine content (9.41%). The protein was predicted to contain an N-terminal signal peptide (SP) and showed 73.81% amino acid identity with ChEC34, a candidate effector gene of C. higginsianum whose expression is up-regulated in appressoria . A BlastP search against the National Center for Biotechnology Information (NCBI) non-redundant (nr) database identified CfEC92 homologs in a range of additional Colletotrichum species (Figure 1), with amino acid identity ranging from 60.5% to 100%. However, no CfEC92 homologs were found in genera other than Colletotrichum (E-value cut-off = 1 × 10 −5 ). We determined that CfEC92 is a lineage-specific effector candidate that is conserved across the Colletotrichum genus.

Transient overexpression of an effector candidate (EC) in
Nicotiana benthamiana is a test method commonly used to investigate EC functions . We have tested whether CfEC92 could suppress BAX-induced cell death by co-injecting N. benthamiana leaves with Agrobacterium tumefaciens overexpressing CfEC92 and A. tumefaciens overexpressing BAX. As shown in Figure 2a, BAX gene expression strongly induced leaf necrosis at 7 days post-inoculation. However, this cell death induction was totally suppressed by CfEC92 co-infiltration. The same results were obtained in multiple independent infiltration assays.
Western blot assays showed that co-infiltration did not interfere with BAX gene expression (Figure 2b,c). These results strongly suggest that CfEC92 is a Colletotrichum genus-specific effector with cell death-suppressive activity.
F I G U R E 1 Sequence similarities between CfEC92 proteins among Colletotrichum species. (a) Phylogenetic analysis of CfEC92 and its homolog proteins in Colletotrichum. The full-length protein sequences were analysed by MEGA 5 using unrooted neighbour-joining bootstrap (1,000 replicates). The black box indicates CfEC92. The scale bar corresponds to a genetic distance of 0.05. (b) Multiple sequence alignment of CfEC92 and its homolog proteins. Full-lengths of protein sequences were aligned using ClustalW and the alignment was edited using JalView. Blue shading intensity reflects the level of amino acid identity at each position

| CfEC92 transcript accumulation is highly upregulated at the early infection phase
Quantitative reverse transcription PCR (RT-qPCR) was performed to compare the relative expression pattern of CfEC92 among conidia, vegetative hyphae, and appressoria on apple leaves at different time points after inoculation. The transcript accumulation of CfEC92 was barely detected in conidia, vegetative hyphae, and appressoria induced on cellophane ( Figure 3). However, on inoculated apple leaves, CfEC92 gene transcripts were induced significantly and reached a peak at 24 hr post-inoculation (hpi) by around 600-fold up-regulation relative to conidia (Figure 3). During apple leaf infection, 24 hpi corresponded to appressorium formation and appressorium-mediated penetration. We therefore concluded that CfEC92 may play a role in virulence during appressorium formation at the early infection stage.

| CfEC92 is specifically required for full expression of C. fructicola virulence
Two CfEC92-deletion mutants (ΔCfEC92-27 and ΔCfEC92-52) were obtained by PCR screening of >20 hygromycin-resistant transformants ( Figure S1b). Reverse transcription (RT)-PCR confirmed the lack of CfEC92 transcript accumulation in these mutants during apple leaf infection ( Figure S1c). The replacement of the CfEC92 coding region by a hygromycin-resistance gene was confirmed by RT-PCR ( Figure S1c) and Southern blotting ( Figure S1d). Genetic complementation of the ΔCfEC92-27 mutant was achieved by polyethylene glycol (PEG)-mediated transformation and the obtained construct (ΔCfEC92-27-C) was also validated by RT-PCR ( Figure S1c).
The knockout mutants showed no significant differences from the wild-type strain 1104-7 (WT) in vegetative growth on potato dextrose agar (PDA) (Figure 4a). In addition, the knockout mutants and WT were highly similar in appressorial development on cellophane ( Figure 4b). By 36 hr after inoculation on cellophane, the frequency of infection hyphae from the knockout mutants was indistinguishable from WT and complementation strains ( Figure 4b).
These results demonstrated that CfEC92 is not required for fungal vegetative growth, appressorium development, or appressorium-mediated cellophane penetration.
We further tested the virulence of the knockout mutants and

F I G U R E 2 Transient expression of CfEC92 in
Nicotiana benthamiana. (a) CfEC92 with or without signal peptide (sp) suppressed BAX-induced cell death. Leaves were infiltrated with Agrobacterium tumefaciens carrying the pGR107-GFP vector inserted the BAX gene or the CfEC92 gene with or without the signal peptide. Leaves were inoculated with empty pGR107-GFP vector as a control. Red circles indicate the infiltrated region. Photographs were taken at 7 days after infiltration. The experiments were repeated nine times. (b) Western blot analysis of agroinfiltrated leaves expressing CfEC92 with a green fluorescent protein (GFP) tag. The total proteins of infiltrating pGR107-GFP vector were used as a control. (c) Western blot analysis of agroinfiltrated leaves expressing BAX F I G U R E 3 Quantitative reverse transcription PCR analysis of CfEC92 expression patterns in conidia (con), mycelia (myc), appressoria (app) on cellophane, and inoculated leaves (4,8,12,24,36,48, and 72 hr post-inoculation) using β-tubulin gene for normalization. Results are presented as the average fold-change of three independent experiments compared to the conidia sample. Error bars represent SE. Asterisks indicate statistically significant differences (one-way analysis of variance, p < .05)

| CfEC92 suppresses plant defence reactions at the early infection phase
To determine the infection stage at which CfEC92 functions, we com-

| Plant defence genes suppressed by CfEC92
The increased deposition of papillae elicited by ∆CfEC92 infection led us to perform an RNA sequencing (RNA-Seq) experiment to compare the global expression patterns of host defence genes between WT and the ∆CfEC92 mutant. Infected leaves were sampled at 24 hpi and subjected to RNA sequencing. A total of six libraries were constructed (three for WT and three for ∆CfEC92) and sequenced. In total, 52.41 Gb reads were obtained and mapped to the apple cv. Golden Delicious genome. A hierarchical clustering F I G U R E 5 The deletion of CfEC92 reduced the pathogenicity of Colletotrichum fructicola on apple leaves and fruits. (a) ΔCfEC92 mutant strains reduced virulence on apple leaves compared to the wild type (WT) and complemented mutant strains. The virulence was estimated by fungal biomass relative to foliar biomass using quantitative PCR analysis of the C. fructicola β-tubulin fragment relative to the apple UBQ fragment. The total DNA was extracted from apple leaves inoculated with WT, ΔCfEC92-27, ΔCfEC92-52, and ΔCfEC92-27-C (complemented) strains at 72 hr post-inoculation, respectively. (b) ΔCfEC92 mutant strains showed reduced virulence on apple fruits compared to wild-type (WT) and complemented mutant strains. The virulence was evaluated by disease index. All experiments were performed for three replicates, each with nine leaves or fruits from three different apple trees. Values are presented as the means ± SE. The double asterisk indicates statistically significant differences (one-way analysis of variance, p < .01) heatmap showed two distinct groups: one group containing the mutant strain and the other containing the WT strain, indicating that gene expression levels had low biological variability within the same treatment samples ( Figure S2). We identified 310 differentially expressed genes that were expressed at least two-fold higher in mutant infection relative to WT infection (Table S2) The putative SP sequence of CfEC92 was fused in-frame to the invertase gene in the pSUC2 vector and transformed into yeast strain YTK12. The YTK12 strains containing the empty pSUC2 and the first 25 amino acids of Mg87 protein from Magnaporthe oryzae were used as negative controls. The SP sequence of Avr1b from Phytophthora sojae was used as positive control. The transformed YTK12 strains that are able to secrete invertase were able to grow in both CMD-W and YPRAA media. (b) Agrobacterium tumefaciens carrying green fluorescent protein (GFP), CfEC92-GFP or CfEC92Δsp-GFP fusion proteins was infiltrated into leaves of 5-week-old, Nicotiana benthamiana. Bar = 10 μm than with the WT strain ( Figure S4). In order to determine whether these flavonoids have a toxic role to C. fructicola, we measured the hyphal growth rate of C. fructicola after treatment by seven flavonoids on PDA. We found the tested flavonoids had different levels of antifungal properties against C. fructicola hyphae (Table 1 and Figure S5). The values of EC 50 showed that 2′,3,4,4′,6′-pentahydroxychalcone caused the greatest growth inhibition and 7,4′-dihydroxyflavone was similar in inhibitory effect. These results suggests that CfEC92 is required to suppress the expression of a specific subset of defence-related genes during the infection of C. fructicola on apple leaves.

| D ISCUSS I ON
In most Colletotrichum-host interaction systems, Colletotrichum species exhibit a hemibiotrophic lifestyle. The early biotrophic infection F I G U R E 8 CfEC92 could suppress host immune responses. The total RNA was extracted from leaves inoculated with ΔCfEC92-27 and wildtype (WT) strains at 24 hr post-inoculation and subjected to quantitative reverse transcription PCR analysis of 10 defence-related genes. The vertical axis represents the relative fold changes of transcripts compared to the WT-inoculated leaf sample and apple leaf MdUBQ was used as the reference gene. Error bars represent SE. Asterisks indicate statistically significant differences (one-way analysis of variance, p < .05) stage is indispensable for completion of their life cycle. In this stage, these fungi often secret effectors to interfere with host defences and generate biotrophic infection structures for absorbing host cell nutrients (Gan et al., 2012;O'Connell et al., 2012). In addition, the genome of C. fructicola possess a large number of effector candidates and some of them show specifically up-regulated expression during host infection (Liang et al., 2018). Identification of effector candidates functioning at early infection stages will provide insights into the molecular mechanism of C. fructicola-apple leaf interaction.
Recognition of phytopathogenic PAMPs or effectors by host receptors would elicit strong host cell defence mechanisms such as the hypersensitive response (HR). The BAX protein, a member of the death-promoting Bcl-2 family, can trigger cell death when expressed in tobacco leaves using a virus vector (Lacomme and Cruz, 1999). The physiological process of induction of PCD in plants by the BAX protein is similar to the defence-related HR (Lacomme and Santa Cruz, 1999). Therefore, it has been regarded as a valuable reference for pathogen effectors (Dou et al., 2008). To date, many bacterial, oomycete, fungal, and nematode effectors have been identified using transient expression assays in the N. benthamiana model system employing agroinfiltration (Guo et al., 2009;Wang et al., 2011;Ali et al., 2015). In the study, we demonstrated that CfEC92 suppressed BAXtriggered cell death using a tobacco transient expression system, implying a defence-suppressive function of CfEC92.
In the current study, the Colletotrichum-genus specific effector CfEC92 was shown to be important for C. fructicola virulence, and its function in suppressing plant defence reactions was demonstrated.
CfEC92 was strongly induced in the early infection stage of the host plant, during progression from formation of appressoria to appearance of biotrophic infection vesicle. The knockout of CfEC92 significantly reduced C. fructicola virulence in apple, indicating that CfEC92 contributed specifically to overcoming apple defences. In addition, secretion and subcellular assays suggested that CfEC92 was secreted from fungal cells and delivered into the host cells to influence the host. Further histological observations revealed that lack of CfEC92 significantly reduced the success rate of appressorium penetration due to the accumulation of papillary callose in host epidermal cells. Papillary callose is produced by callose synthase, and its deposition enhances the host cell's mechanical resistance to fungal appressorial pegs (Suzuki et al., 2003;Chen and Kim, 2009;Zhang et al., 2011). In accordance with the histological results, the callose synthase gene was highly up-regulated in the host inoculated with a CfEC92 mutant compared to inoculation with the WT. Papillary callose deposition acts as a key factor for successful plant defence against microbial pathogens. Many plant pathogens, including bacteria, oomycete, and fungi, can secrete effectors to inhibit papillary callose formation (Tao et al., 2003;Truman et al., 2006;van den Burg et al., 2006;Fabro et al., 2011). For example, effector PSTha5a23 from Puccinia striiformis f. sp. tritici is involved in suppression of plant papillary callose formation (Cheng et al., 2017). The core effector Cce1 contributes to early stages of infection of maize by Ustilago maydis via interfering with papillary callose deposition (Seitner et al., 2018). Moreover, papillary callose deposition is an output of the PTI response. The transcriptome and RT-qPCR experiments verified that the CfEC92 could suppress the expression of PTI marker genes, including PR1a, WRKY29, and MAPKKK3 (Meng et al., 2018;Sun et al., 2018;Yang et al., 2019).
These findings suggest that CfEC92 is able to promote appressorium penetration through suppression of host PTI-associated response.
Previous studies have reported that effector candidates are stage-specifically expressed in the hemibiotrophic Colletotrichum stage switches, implying that the effectors may play an important role in the production of each infection structure of Colletotrichum species (Gan et al., 2012;O'Connell et al., 2012). In this study, we found an effector, CfEC92, that was associated with differentiation of primary hyphae from infection vesicles, and CfEC92 could inhibit host PTI response. These results suggest that host PTI-related defence could restrain the formation of C. fructicola primary hyphae in apple leaf cells. It was also suggested that the formation of infection structures is the result of interaction between pathogen and host, which is not only related to pathogenesis genes (Dufresne et al., 2000), but also to host resistance.
CfEC92 inhibited a subset of defence-related gene expression.
Comparative transcriptome analysis results showed that several defence genes involved in pathogenisis-related (PR) proteins, the MAPK signalling pathway, and flavonoid synthesis were enriched in up-regulated differential gene expression. The production and accumulation of PR proteins in plants are very important to resist phytopathogens (Ebrahim et al., 2011). For example, PR3 can directly degrade chitin, a primary component of the fungal cell wall, to reduce the pathogenicity of an invading pathogen (Sharma et al., 2011). In the study, the up-regulated expressions of MAPKKK3 and WRKY29 were confirmed by RT-qPCR. The two genes, as components of the MAPK cascades, are central to the transmission of biotic stress signals from membrane receptors into the nucleus, activating the downstream resistance responses to pathogens (Asai et al., 2002;Pitzschke et al., 2009). Flavonoids are associated with plant development and plant defence against pathogen and herbivores (Stafford, 1997;Shirley, 1998;Aoki et al., 2000;Kliebenstein, 2004). For example, sakuranetin is a major phytoalexin in rice, effective against spore germination of Pyricularia oryzae (Kodama et al., 1992). In this study, we found remarkably up-regulated genes involving the biosynthesis of seven flavonoids that appeared to inhibit C. fructicola growth on PDA. Of these, naringenin has been reported to exert antimicrobial effects against Staphylococcus epidermidis, Lactococcus lactis, and Candida krusei, and phlorizin could directly inhibit growth of Botrytis cinerea in vitro (Nishino et al., 1987;Uzel et al., 2005;Mandalari et al., 2007;He et al., 2018); other flavonoids can exhibit pharmacological behaviour, such as anti-inflammatory, antioxidant, and anticancer activity (Katavic et al., 2007;Mersereau et al., 2008;Rasul et al., 2013).
In summary, we identified a Colletotrichum-specific effector

| Fungal strains and plants culture conditions
The WT isolate of C. fructicola, 1104-7, used in this study was maintained on PDA at 25 °C in the dark. For long-term preservation, the iso-

| RNA extraction and RT-qPCR
To clone and characterize the CfEC92 expression during infection, the total RNA of conidia, mycelia, appressoria on cellophane, and inoculated Malus domestica 'Gala' leaves (4, 8, 12, 24, 36, 48, and 72 hpi) were extracted using an RNAprep Pure Plant Kit (TianGen Biotech) following the manufacturer's instructions. First-strand complementary DNA (cDNA) was synthesized using the One-Step gDNA Removal and cDNA Synthesis SuperMix (TransGen Biotech) following the recommended protocol.

RT-qPCR was performed using an Applied Biosystems
StepOnePlus Three independent technical replicates were performed per sample and the entire experiment (from leaf sample preparation to RT-qPCR) was performed three times.

| Plasmid construction
The CfEC92 (accession number: MN604401) open reading frame (ORF) was amplified from cDNA (inoculated leaves) using FastPfu DNA polymerase (TransGen Biotech) and cloned into pEASY-Blunt Zero vector (TransGen Biotech). In order to determine the SP secretion activity, the predicted SP sequence of CfEC92 was fused in-frame into the vector pSUC2 (Jacobs et al., 1997). The SP of Avr1b and the N-terminal 25 amino acids of Mg87 were joined with the vector pSUC2 as positive and negative controls, respectively (Shan et al., 2004;Gu et al., 2011). To determine the subcellular localization in N. benthamiana, the ORF sequence of CfEC92 was cloned into the pBin-eGFP vector. To perform the cell-death inhibition assay in N. benthamiana, the coding sequence of CfEC92 with or without the putative SP was fused with GFP sequence and then cloned into the pGR107 vector (Jones and Baulcombe, 1999). To complement the ΔCfEC92 mutant, the entire CfEC92 sequence containing the coding region and 1.5kb promoter was amplified and cloned into the vector pHZ-100 containing G418 resistance gene. The primers used in this study are listed in Table S1.

| Domain prediction, sequence alignment, and phylogenetic tree construction
The SP of CfEC92 was predicted using SignalP v. 4.1 (Nielsen, 2017  The same leaves were infiltrated with A. tumefaciens carrying an empty vector pGR107 containing GFP as control. Cell death symptoms were observed and photographed at 7 days post-infiltration.

| Cell-death inhibition assay
All tests were performed in triplicate.

| Protein extraction and western blot
The total proteins of the leaf infiltrated area were extracted with a protein extraction kit (BB-3124-1, BestBio) in accordance with manufacturer's protocols. Western blotting was carried out to verify protein expression following standard procedures. Thirty microlitres of total protein per sample was loaded on 15% SDS-polyacrylamide gel for electrophoresis and then transferred to a polyvinylidene difluoride (PVDF) membrane (Millipore). For detecting effector and BAX proteins, mouse anti-GFP monoclonal antibody (TransGen Biotech) and mouse anti-BAX monoclonal antibody (Beyotime Biotechnology) were used as the primary antibodies, respectively.
Goat anti-mouse IgG-peroxidase conjugate secondary antibody (TransGen Biotech) was used to detect the primary antibodies. The membrane was treated with an Easysee western blot kit (TransGen Biotech) for 1 min. Subsequently, the chemoluminescence signal of the PVDF membrane was exposed to BioMax (Kodak) light film and imaged using a ChemiDoc XRS system (Bio-Rad).

| CfEC92 gene knockout and complementation
The upstream and downstream flanking sequences of CfEC92 were amplified. The resulting PCR products were fused with the hygromycin phosphotransferase (hph) fragments by double-jointed PCR ( Figure S1a).

| Plant inoculation
To prepare a conidial suspension, potato dextrose broth (PDB) was inoculated with two 9-mm diameter mycelial PDA plugs and shake cultured for 5 days at 28 °C and 180 rpm on a shaker (DHZ-DA, Taichang Experimental Instrument). The culture medium was then filtered through a layer of Miracloth (Millipore) and centrifuged at 8,000 rpm for 3 min. Harvested conidia were washed three times, resuspended, and adjusted to a concentration of 5 × 10 6 conidia/ml in sterile distilled water (SDW).
Leaves and fruit were sterilized by rubbing with 70% ethanol using cotton swabs, followed by wiping with sterile distilled water.
After the water evaporated, the leaves and fruit were sprayed with the conidial suspension, then incubated for 48 hr in humid chambers (IKEA) at 25 °C in the dark.
Lesion diameters were measured after 6 days at 25 °C to compare the pathogenicity differences between WT, ΔCfEC92, and complemented mutant strains. To further assess differences in pathogenicity, we quantified the fungal biomass relative to apple leaf biomass among inoculated leaves at 72 hpi. Next, 50 ng genomic DNA was used to amplify the C. fructicola β-tubulin fragment (CfβTUB) and DNA amounts were normalized to the M. domestica ubiquitin extension fragment (MdUBQ, MDU74358). At 5 dpi, the diameters of the fruit lesions were graded on the following scale: 1: ≤1 mm; 2: >1 mm, ≤2 mm; 3: >2 mm.

| Microscopy
To evaluate the rate of appressoria formation, and infection vesi-  (Khan and Hsiang, 2003) and observed with 1,000× magnification under an X-51 microscope (Olympus Corporation). Fifty spores were counted for each leaf disk. Simultaneously, their conidial suspensions were spread on cellophane appressed to the surface of water-agar medium, and amounts of appressoria formation and infection hyphae were assessed at 12 and 36 hpi, respectively. For observation of the papillary callose of the host epidermal cell, inoculated leaves were sampled and treated as described by Arroyo et al. (2005). Semithin sections were cut to 0.5 μm thickness by an ultramicrotome (Ultra cut R; Leica), stained for 1 min with aniline blue, and examined using an X-51 microscope.

| Subcellular localization assay
To determine the subcellular localization of CfEC92, A. tumefaciens GV3101 containing the vector pBin-eGFP with CfEC92 was transiently expressed in N. benthamiana in an infiltration medium with an OD 600 of 0.6. GV3101 carrying the empty pBin-eGFP vector was used as control. At 2 days post-infiltration, epidermal tissues of tobacco leaves were sampled and observed with an FV3000 laser scanning confocal microscope (Olympus Corporation) with a 488 nm emission filter.

| Transcriptome analysis
To perform RNA-Seq analysis, three independent biological samples of apple leaves inoculated with WT or ΔCfEC92 mutant strain were collected for RNA extraction and sequencing. Paired-end sequencing with a read length of 150 bp was performed by an Illumina HiSeq X10. Filtered reads were mapped to the genome of cv. Golden Delicious using TopHat v. 2.0.9 with the parameter "-g 1" (Trapnell et al., 2009). The profile of gene expression levels was analysed using Cufflinks v. 2.2.1 (Trapnell et al., 2010). Fragments per kilobase of transcript per million fragments mapped (FPKM) of transcripts that were greater than a two-fold change (p < .05) were considered to be differentially expressed. Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis was performed in an Omcishare cloud platform (http://www.omics hare.com). Genes of interest were selected for further confirmation by RT-qPCR according to the procedure described above.

| Fungal inhibition assays
Flavonoids were dissolved in 250 μl ethanol and diluted to 12.5 mL with PDA to different final concentrations (

ACK N OWLED G EM ENTS
This work was supported by National Natural Science Foundation of China (31772113, 31972220) and China Agriculture Research System (CARS-27).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author on reasonable request.