Hydroxybenzoate hydroxylase genes underlying protocatechuic acid production in Valsa mali are required for full pathogenicity in apple trees

Abstract Valsa mali is the causative agent of apple tree valsa canker, which causes significant losses in apple production. It produces various toxic compounds that kill plant cells, facilitating infection. Among these, protocatechuic acid exhibits the highest phytotoxic activity. However, those genes involved in toxin production have not been studied. In this study we identified four hydroxybenzoate hydroxylase genes (VmHbh1, VmHbh2, VmHbh3, and VmHbh4) from the transcriptome of V. mali. The VmHbh protein had high enzymatic activities of hydroxybenzoate hydroxylase, which could convert 4‐hydroxybenzoate to protocatechuic acid. These four VmHbh genes all had highly elevated transcript levels during the V. mali infection process, especially VmHbh1 and VmHbh4, with 26.0‐ and 53.4‐fold increases, respectively. Mutants of the four genes were generated to study whether VmHbhs are required for V. mali pathogenicity. Of the four genes, the VmHbh1 and VmHbh4 deletion mutants considerably attenuated V. mali virulence in apple leaves and in twigs, coupled with much reduced toxin levels. The VmHbh2 and VmHbh3 deletion mutants promoted the transcript levels of the other VmHbhs, suggesting functional redundancies of VmHbhs in V. mali virulence. The results provide insights into the functions of VmHbhs in the production of protocatechuic acid by V. mali during its infection of apple trees.

genes involved in plant cell wall degradation and biosynthesis of secondary metabolites (Feng et al., 2020;Yin et al., 2015). Previous studies have reported that genes encoding the CWDEs play important roles during the pathogenicity of V. mali. For instance, pectinase secreted by V. mali causes apple bark tissue maceration and is thus essential to pathogen virulence (Xu et al., 2016). The role in the virulence played by xylanase, involved in the hydrolysis of xylan, was confirmed in plant pathogens such as Botrytis cinerea and Sclerotinia sclerotiorum (Brito et al., 2006;Yu et al., 2016). One xylanase gene, VmXyl1, contributes to the production of pycnidia and is required for the full pathogenicity of V. mali (Yu et al., 2018). Inactivation of feruloyl esterases in V. mali leads to reduced virulence . V. mali can produce several kinds of nonspecific toxic compounds, both in vivo and in vitro, that are associated with the pathogenicity (Natsume et al., 1982;Wang et al., 2014). However, the genes involved in the production of those toxins remain unknown.
Among the five toxic compounds produced by V. mali, protocatechuic acid exhibits the greatest phytotoxic activity .
According to the hypothesized toxin production pathway in V. mali, 4-hydroxybenzoate is hydroxylated and then converted to protocatechuic acid (Natsume et al., 1982). Previous studies in bacteria suggest that 4-hydroxybenzoate is predominantly hydroxylated at the third position by 4-hydroxybenzoate 3-hydroxylase to yield 3,4-dihydroxybenzoic acid (protocatechuic acid) (Huang et al., 2008). In Xanthomonas campestris, 4-hydroxybenzoate is hydroxylated by 4-hydroxybenzoate 3-hydroxylase for conversion to protocatechuic acid, which can be further metabolized into other products. The 4-hydroxybenzoate degradation pathway in X. camperstris is required for its full pathogenicity . In Pseudomonas sp. and Alcaligenes sp., 4-hydroxybenzoate is also degraded by 4-hydroxybenzoate 3-hydroxylase via the protocatechuic acid route (Deveryshetty et al., 2007).
The aim of the present study was to identify those hydroxybenzoate hydroxylase (Hbh) genes that convert 4-hydroxybenzoate to protocatechuic acid, and to evaluate their potential functions during infection of apple trees by V. mali. The published genomic sequence predicted 10 candidate genes that encode hydroxybenzoate hydroxylase (Yin et al., 2015). Transcriptome profiling showed that four of the 10 genes were upregulated during V. mali infection (VmHbh1, VmHbh2, VmHbh3, and VmHbh4), indicating that these four genes may play important roles in fungal virulence. In this study, we obtained the purified fusion protein of VmHbh1 and constructed gene deletion mutants to evaluate their roles in the production of protocatechuic acid and pathogenicity of V. mali. Of the four genes, two (VmHbh1 and VmHbh4) were required for the full virulence of V. mali in apple trees.

| Sequence identification and analysis of V. mali Hbh genes
Four candidate Hbh genes were amplified by PCR using cDNA of V. mali as a template and were confirmed by sequencing.
The candidate Hbhs all have a monooxygenase flavin-adenine dinucleotide (FAD)-binding domain that belongs to the family of group A flavoprotein monooxygenases, which includes 4-hydroxybenzoate 3-hydroxylases and 3-hydroxybenzoate 6-hydroxylase (Montersino et al., 2011(Montersino et al., , 2017. A phylogenetic tree was constructed with the characterized proteins of 3-hydroxybenzoate 6-hydroxylases and 4-hydroxybenzoate 3-hydroxylases from other strains, and confirmed that the four VmHbhs were all closely related to the 3-hydroxybenzoate 6-hydroxylases ( Figure 1a). The sequence alignment of VmHbhs and four characterized 3-hydroxybenzoate 6-hydroxylases are given in Figure 1b. VmHbh1, VmHbh3, VmHbh4, and the four reference sequences all have the three flavin-binding motifs, such as GxGxGG, DG, and GD, whilst VmHbh2 has the two conserved motifs DG and GD (Eppink et al., 1997).

| Conversion of 4-hydroxybenzoate to 3,4-dihydroxybenzoic acid by VmHbh1
The purified VmHbh1 protein was obtained using a prokaryotic expression technique (Figure 2a). High-performance liquid chromatography (HPLC) analysis showed that 4-hydroxybenzoate decreased with the corresponding increase in protocatechuic acid (Figure 2b). Further analysis based on the liquid chromatography-mass spectrometry (LC-MS) confirmed that 4-hydroxybenzoate and protocatechuic acid are the substrate and product, respectively (Figure 2c,d). During the conversion of 4-hydroxybenzoate by VmHbh1, 2,4-dihydroxybenzoic acid, which has the same molecular weight as protocatechuic acid, was not detected. Similarly, 2,4-dihydroxybenzoic acid was not detected in V. mali culture on apple branch medium nor in V. mali -infected twigs (Figure 2d). These results suggest that VmHbh encode V. mali Hbh, which converts 4-hydroxybenzoate to protocatechuic acid.

| Transcription levels of VmHbh genes
Infected apple phloem and xylem tissues were sampled at 72 hr postinoculation (hpi) to compare the transcript levels of the four VmHbhs in the host tissues with those of mycelia grown on potato dextrose agar (PDA). The quantitative reverse transcriptionpolymerase chain reaction (RT-qPCR) analysis showed that the transcript levels of the four VmHbhs were significantly upregulated in the apple phloem compared to their respective expression levels in the apple xylem ( Figure 3a). This was particularly true for VmHbh1: the transcription level in apple phloem was 17.1-fold higher than that in apple xylem.
The transcript levels of VmHbhs during infection of apple twigs by V. mali were determined for infected apple phloem tissues sampled at seven time points (0,6,12,24,48,72, and 120 hpi) as well as for mycelia grown on PDA (Figure 3b). The transcript level of VmHbh1 was upregulated from 24 to 120 hpi with 16.2-to 26.0-fold increases. Similarly, VmHbh4 was upregulated from 12 to 120 hpi with 7.5-to 53.4-fold increases. For VmHbh2, upregulation was observed only at 48 and 72 hpi with 2.7-and 5.3-fold increases, respectively. The transcript level of VmHbh3 was upregulated from 6 to 120 hpi with 2.0-to 3.1-fold increases. Overall, the high induction of VmHbhs during infection suggested their potential role in the pathogenicity of V. mali.

| VmHbhs are not required for growth and pycnidia formation of V. mali
For functional analysis of the four VmHbhs in V. mali, we knocked out each of the four genes by the polyethylene glycol (PEG)mediated protoplast transformation method. The mutants were initially examined via PCR assays, and then were confirmed by (a)

(b)
Southern blotting ( Figure S2). For the complementation of each VmHbh deletion mutant, a complementary construct was generated and transformed into the corresponding mutants. All the complementation transformants were selected using geneticin G-418 and confirmed by PCR assays with primer pairs ID-F/ID-R (Table S1). In addition, RT-PCR analysis was performed to confirm that the target VmHbh was knocked out in the gene deletion mutants, and the complementation strains contained their corresponding target genes ( Figure S4).
Colony morphology, growth rate, dry mycelia weight, and pycnidia formation of the VmHbh deletion mutants and the wildtype strain were measured to determine whether any VmHbh played a role in V. mali development. The VmHbh deletion mutants did not differ statistically from the wildtype strains in colony morphology

| Deletion of VmHbh1 or VmHbh4 reduce V. mali virulence
Pathogenicity assays on the detached apple leaves and twigs were conducted to investigate whether the four VmHbh genes play roles in disease development. Pathogenicity was measured in terms of the lesion size following inoculation ( Figure 5). Deletion of VmHbh1 led to at least 65% reduction in lesion sizes on the apple leaves and twigs compared to the wildtype strain. Similarly, the ΔVmHhb4 mutant also reduced the lesion size on apple leaves and twigs, but to a lesser extent (about 40%), compared to ΔVmHhb1. In contrast, the ΔVmHhb2 and ΔVmHhb3 mutants did not differ from the wildtype strain in lesion size. To confirm that the reduced pathogenicity of the mutants was caused by the deletion of VmHbhs, each complementary construct was generated and transformed into their respective deletion mutants. The previously observed phenotype was rescued in the complementation strains. These results suggested that VmHhb1 and VmHhb4 are essential for the pathogenicity of V. mali in apple leaves and twigs.

| Deletion of VmHbh1 and VmHbh4 reduce toxin production by V. mali
To identify possible mechanisms underlying the reduced pathogenicity of the ΔVmHhb1 and ΔVmHhb4 mutants, toxins in the apple branch extract medium under the exposure of the wildtype and mutant strains were quantified. All strains produced five types of toxins, but the ΔVmHhb1 mutant produced 79% and 31% less protocatechuic acid and 4-hydroxybenzoate, respectively, in the apple branch extract medium (Figure 6a). Similarly, the ΔVmHhb4 mutant produced less protocatechuic acid (43% less) and 4-hydroxybenzoate (38% less) compared to the wildtype strain. Toxins during fungal infection of apple twigs with the wildtype and the deletion strains were also quantified ( Figure 6b). The ΔVmHhb1 and ΔVmHhb4 mutants had 72% and 48% reductions in protocatechuic acid, respectively; the corresponding values were 42% and 34% for 4-hydroxybenzoate. There were no significant differences in the quantities of the other three toxic compounds between the wildtype and mutant strains. When VmHhb1 or VmHhb4 was reintroduced into the gene deletion mutants, the phenotype was rescued. Neither the ΔVmHhb2 nor ΔVmHhb3 mutants differed from the wildtype strain in the toxin levels (data not shown).

| Redundancies of VmHbhs in the virulence of V. mali
The RT-qPCR analysis showed that, when grown on PDA, the transcript levels of VmHbhs did not differ between the deletion mutants and the wildtype strain ( Figure S5). During infection, the transcript levels of VmHhb2, VmHhb3, and VmHhb4 remained at a similar level in the ΔVmHhb1 mutant relative to that of the wildtype strain ( Figure 7a). However, VmHhb1, VmHhb3, and VmHhb4 were significantly upregulated (with 4.7-, 3.2-, and 6.0-fold increases, respectively) in the ΔVmHhb2 mutant relative to the wildtype strain F I G U R E 1 Phylogenetic tree and alignment of the amino acid sequences of hydroxybenzoate hydroxylase (Hbh). (a) Phylogenetic analysis of 3-hydroxybenzoate 6-hydroxylase and 4-hydroxybenzoate 3-hydroxylase. The phylogram was generated by the maximumlikelihood algorithm as implemented in MEGA 7. Numbers beside each node indicate bootstrap values as a percentage of 1,000 bootstraps. Multiple sequence alignment of the four Hbhs from Valsa mali and known 3-hydroxybenzoate 6-hydroxylase. Excerpts of sequences were used for multiple sequence alignment, and the aligned sequences are from the phylogenetic tree. Black rectangles indicate conserved motifs of 3-hydroxybenzoate 6-hydroxylase

| D ISCUSS I ON
Protocatechuic acid, 4-hydroxybenzoate, 4-hydroxyacetophenone, 4-hydrobenzene propanoic acid, and phloroglucinol produced by V. mali have been described as phytotoxins that can cause necrosis in plants (Natsume et al., 1982;Wang et al., 2014). In V. mali, strains producing more toxins are more virulent than strains producing fewer toxins Wang et al., 2014). The five studied toxic compounds are believed to be the degradation products of the phloridzin, but little is known about functional genes that are involved in the phloridzin degradation process. Among these toxic compounds, protocatechuic acid is the most phytotoxic and the last toxic compound in the production pathway of these toxins. Previous studies in bacteria have shown that 4-hydroxybenzoate is hydroxylated by 4-hydroxybenzoate 3-hydroxylase to yield protocatechuic acid (Deveryshetty et al., 2007;Donoso et al., 2011;Romero-Silva et al., 2013). In the present study, we identified four VmHbh genes and demonstrated the role of hydroxybenzoate hydroxylase in the Interestingly, amino acid sequence analysis suggested that these four hydroxybenzoate hydroxylases from V. mali are more closely related to the 3-hydroxybenzoate 6-hydroxylases than to the 4-hydroxybenzoate 3-hydroxylases. Moreover, these four VmHbhs have similar conserved motifs to those of 3-hydroxybenzoate 6-hydroxylases, namely GxGxGG, DG, and GD . Previous studies demonstrated that 3-hydroxybenzoate 6-hydroxylases can convert 3-hydroxybenzoate to 2,5-dihydroxybenzoic acid and show activity when salicylate or 4-hydroxybenzoate is used as the substrate (Gao et al., 2005;Montersino & van Berkel, 2012). By sequence analysis, we found few functional annotations of 4-hydroxybenzoate 3-hydroxylases in fungi in the NCBI database. Until now, the functional analysis and crystal structure of 4-hydroxybenzoate 3-hydroxylases and 3-hydroxybenzoate 6-hydroxylase are all from bacteria or yeast (van Berkel et al., 1994;Hiromoto et al., 2006;Holesova et al., 2011;Montersino et al., 2013;Schreuder et al., 1988). This is the first report that hydroxybenzoate hydroxylase (VmHbh) from V. mali can convert 4-hydroxybenzoate to protocatechuic acid but does not have enzymatic activities when the substrate is 3-hydroxybenzoate (data not shown). Therefore, the present results indicate that VmHbh1 performs the function of 4-hydroxybenzoate 3-hydroxylase in V. mali.
During infection by V. mali, all four VmHbhs genes showed high transcript levels, especially VmHbh1 and VmHbh4, indicating that these genes may play a role in pathogenicity toward apples. Furthermore, RT-qPCR analysis also demonstrated that the greatest upregulation of their expression occurred at different time points: VmHbh1, VmHbh3, and VmHbh4 were upregulated at 6-24 hpi, whereas VmHbh2 was F I G U R E 3 Transcript levels of VmHbhs at different conditions and time points postinoculation of apple tissues or twigs with Valsa mali as determined by quantitative reverse transcription PCR. The transcript level of V. mali EF1-a was used as an endogenous control, and the transcript level of each VmHbh gene in the mycelia grown on potato dextrose agar was normalized to 1. The means and standard deviation of the relative expression levels were calculated from three independent biological replicates. (a) Relative expression levels of VmHbhs in apple tissues at 72 hr postinoculation (hpi). Asterisks represent significant differences (p < 0.05) in transcript levels in the apple phloem and xylem tissues. (b) Relative expression levels of VmHbhs at 0, 6, 12, 24, 48, 72, and 120 hpi of apple twigs. Asterisks represent significant differences (p < 0.05) in transcript levels as compared to that at 0 hpi (a) (b) upregulated at 48 hpi. The differential temporal dynamics in the transcripts may reflect that these genes play important roles at different time points during infection. In addition, the four VmHbh genes showed significantly higher expression levels in phloem than in xylem tissues. Previous studies indicated that V. mali can grow rapidly and survive for a long time in xylem, but it could not develop disease symptoms until the pathogen reached the phloem (Chen et al., 2016;Wang et al., 2018). Thus, the transcript levels of the VmHbh genes may reflect the pathogenicity level of V. mali in apple tissues.

In this study, we generated gene deletion mutants of all four
VmHbh genes and demonstrated that VmHbhs are not necessary for mycelial growth and pycnidia formation in V. mali. In contrast, the pathogenicity-related gene BcKMO in B. cinerea, encoding kynurenine 3-monooyxgenase (KMO) with a monooxygenase FADbinding domain, is known to be important for fungal development in B. cinerea, and its mutant does not produce conidia or sclerotia, and has a lower rate of growth (Zhang et al., 2018). These data indicate that the contribution of monooxygenases to fungal growth and development may depend on the pathogen species.  (Xu et al., 2016).
An important discovery of our work is the finding that the deletion of VmHbh1 and VmHbh4 significantly reduced levels of toxin, which may explain the corresponding decrease in the virulence of V. mali. This is consistent with the previous results of the V. mali xylanase gene VmXyl1, which contributes significantly to both the total xylanase activity and virulence (Yu et al., 2018). Similarly, the deletion of feruloyl esterase genes decreases the capacity of V. mali to utilize ethyl ferulate, which is consistent with reduced virulence of the gene deletion mutants . In contrast, polyketide synthase BcPKS6 and BcPKS9 are key enzymes for the biosynthesis of botcinic acid, an important phytotoxin during infection by B. cinerea: when one of theses two genes is deleted, the loss of botcinic acid production does not affect the virulence of B. cinerea in plant tissues (Dalmais et al., 2011). In this study, the fact that deletion of VmHbh2 or VmHbh3 did not affect the toxin levels demonstrated redundant functions in toxin production.
In the present study, we demonstrated that VmHbh1 can convert 4-hydroxybenzoate to protocatechuic acid. VmHbhs are required for full pathogenicity of V. mali against apple trees. The present findings provide a new perspective for the contribution of Hbhs to the growth, development, and virulence of phytopathogenic fungi.

| Strains and culture conditions
The V. mali wildtype strain LXS080601 was grown on potato dextrose agar (PDA, 200 g potato, 20 g dextrose, 15 g agar per L) at 25 ℃ in the dark. The gene deletion and complemented mutants were cultured on PDA supplemented with 100 mg/ml hygromycin B or geneticin G-418 (Sigma). Escherichia coli strains were grown in lysogeny broth (LB) with appropriate antibiotics at 37 ℃. Colony diameter was measured to calculate the growth rate of different strains on PDA.

| Identification of Hbh genes in V. mali
To identify the four candidate Hbh genes in V. mali genome, BLAST searches with well-characterized Hbh genes were conducted. The conserved domains of the candidate Hbh proteins were confirmed using the BLAST-based NCBI conserved domain search engine (Marchler-Bauer et al., 2017). Primer pairs for the cloning of the four candidate Hbh genes were synthesized by TsingKe (Beijing, China) (Table S1). Total RNA was extracted from fungal mycelia with the RNAiso Plus Kit, and cDNA was synthesized with the Prime Script RT reagent kit with the gDNA Eraser (TaKaRa) with an oligo (dT) [12][13][14][15][16][17][18] primer. PCRs were performed with PrimerSTAR Max DNA polymerase and the PCR products were cloned to T-Vector pMD 19 Simple (TaKaRa), according to the manufacturer's instructions.

| Sequence alignment and phylogenetic analysis
The amino acid sequences of Hbhs from other microorganisms in this study were obtained from the NCBI GenBank. All the homology searches were carried out on the NCBI BLAST server. These

| Hydroxybenzoate hydroxylase activity assay
To assay the hydroxybenzoate hydroxylase activity of VmHbhs, the cDNA fragments were subcloned into pET-32a or pMal-c2× by homologous recombination using the ClonExpress Ⅱ One Step Cloning Kit (Vazyme). The vector constructs were transformed into E. coli

| Detection of gene expression by RT-qPCR
Mycelia grown on PDA for 3 days were used to inoculate apple twigs. Apple bark around the lesion margin was sampled at 0, 6, 12, 24, 48, 72, and 120 hpi. For samples at 0 hpi, the bark tissues around inoculation sites containing mycelium plugs were collected (Yu et al., 2018). The RNA was extract from bark tissues using the RNAiso Plus Kit (TaKaRa), and then the cDNA was synthesized. All RT-qPCR experiments were conducted in a LightCycler 480ⅠⅠ PCR detection system (Roche) with SYBR Master Mix (TaKaRa), following the manufacturer's protocol. The V. mali EF1-a gene was used as an endogenous control, and the primers are given in Table S1. The relative expression levels of each gene were calculated using the 2 −ΔΔCt method. Data from three biological replicates were used to calculate the means and standard deviation. The whole experiment was repeated twice.

| Generation of gene deletion and complementation strains
To obtain the VmHbh gene deletion mutants, the PEG-mediated homologous recombination was performed as described previously (Yu et al., 2018). The gene deletion cassette with three components used the hygromycin B phosphotransferase gene (HPH) as a selective marker for the gene deletion ( Figure S1). Upstream and downstream fragments of VmHbh genes were amplified from genomic DNA of the wildtype strain LXS080601 using the gene-specific primer pairs Up-F/Up-R and Down-F/Down-R (Table S1), respectively. The HPH gene was amplified with primers HPH-F and HPH-R from the vector pBS (Table S1). The gene deletion cassette was generated by double-joint PCR, which was confirmed by the sequencing. The cassettes were later transformed into the protoplasts of V. mali LXS080601, and the transformants were screened by culturing on medium with 100 μg/ml hygromycin (Yu et al., 2018). The putative gene deletion mutants were validated by the PCR using four primer pairs ( Figure S1 and Table S1). Primers I-F and I-R were used to detect the fusion segment; HPH-P-F and HPH-P-R to detect the resistance gene (HPH). The Up-F and Full-R were used to confirm if the upstream sequence of the replaced resistance gene was fused at the right position, and Full-F and Down-R to confirm whether the downstream sequence of the introduced resistance gene was fused at the right position. The gene deletion was finally verified by Southern hybridization with a DIG DNA Labeling and Detection Kit Ⅱ (Roche), following the instruction manual. Genomic DNA was digested with appropriate restriction enzymes and separated via agarose gel electrophoresis. The HPH gene used for the replacement cassette served as probe.
To construct the gene complement vector, the entire VmHbh genes and predicted promoter sequences were amplified from genomic DNA using the primer pair CM-F/CM-R (Table S1). The PCR products were cloned into pYF11 using yeast gap repair, as described previously (Yu et al., 2014). Fusion constructs were verified by the sequencing analysis, and plasmids were transformed into the corresponding gene deletion mutant using the PEG-mediated method.
The transformants were selected using G-418 and confirmed by PCR with the corresponding primers (Table S1).

| RT-PCR detection of VmHbhs in the wildtype, gene deletion mutants, and complementation mutants
The RNA extraction from cultured mycelia of the wildtype strain, VmHbh deletion mutants, and complementation mutants were performed as described previously. The cDNA was synthesized and the primers Full-F/Full-R were employed to detect the VmHbh genes (Table S1). The PCR conditions were as follows: 30 cycles of 94 ℃ for 30 s, 55 ℃ for 45 s, and 72 ℃ for 60 s; with a final extension at 72 ℃ for 5 min. The housekeeping gene EF1-a was used as a control.

| Vegetative growth, pycnidia formation, and pathogenicity assays
Mycelium plugs (5 mm diameter) cut from actively growing colony edges of the wildtype strain, gene deletion, and complemented mutants were transferred to new PDA plates. The plates were then incubated at 25 ℃ before colony shape, colour, and colony diameters were assessed. For the dry weight of mycelia, plugs were inoculated into potato dextrose broth at 150 rpm and 25 ℃ for 7 days.
Pathogenicity assays were performed on the apple leaves and 1-year-old twigs (Malus domestica 'Fuji') were taken from the greenhouse at Qingdao Agricultural University, Qingdao, China. The detached leaves and twigs were sterilized with 75% ethanol and the wounds were made as described by Yu et al. (2018). Mycelium plugs were used to inoculate the wounds. The inoculated leaves and twigs were placed in trays to maintain high humidity at 25 ℃ in the dark.
Then, the lesion length was measured and lesion development was photographed at several time points. The assays were repeated three times, with at least 15 leaves and twigs per treatment.

| Determination of toxins production
For toxins production in the liquid media, mycelium plugs of the wildtype strain, gene deletion mutants, and complemented mutants were inoculated into apple branch extract medium. After culturing for 7 days under 150 rpm/min and 25 ℃ with a 12-hr photoperiod, the supernatant was collected. The medium inoculated with PDA was used as a control. For toxin production in apple tissues, detached twigs were inoculated according to the method in section 4.8. After 7 days, the diseased areas and margins of the twigs were collected and crushed into small pieces. Toxins were extracted according to a method described by Wang et al. (2014). The final samples were filtered through a 0.22μm pore membrane for the HPLC analysis.
Each experiment was repeated two times.

ACK N OWLED G EM ENTS
We thank Professor Xiangming Xu at NIAB East Malling Research for helpful suggestions and revising the manuscript. This study was

CO N FLI C T O F I NTE R E S T
The authors declare that no competing interests exist.

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 upon reasonable request.