The V. mali wild-type strain LXS080601 was grown on the potato dextrose agar (PDA, 200 g of potato, 20 g of dextrose, and 15 g of agar per liter) at 25°C in the dark. The gene deletion and complemented strains were cultured on PDA supplemented with 100 mg/mL of hygromycin B or geneticin G418 (Sigma, St. Louis, MO, United States). The Escherichia coli strains were grown in the lysogeny broth (LB) with appropriate antibiotics at 37°C. The measured colony diameter was used to calculate the growth rate of different strains on PDA medium.

The genomic DNA and total RNA were extracted from the V. mali mycelium. The cDNA was synthesized using the Prime Script RT reagent Kit with gDNA Eraser (TaKaRa, Dalian, China) using an oligo (dT)_12–18 primer. The gene VmGlu2, with putative β-glucosidase activity and high transcript levels during V. mali infection was cloned. Primer pairs for the cloning of VmGlu2 were synthesized by TsingKe (Beijing, China) (Supplementary Table 1). The PCRs were performed with PrimerSTAR Max DNA polymerase and cloned to T-Vector pMD 19 Simple (TaKaRa, Dalian, China), according to the manufacturer’s instructions.

The amino acid sequences of β-glucosidase from other strains in this study were obtained from the NCBI GenBank. All the homology searches were carried out on the NCBI BLAST server. The obtained sequences were compared with the sequences from V. mali (KX013493). Maximum likelihood (ML) method, as implemented in MEGA7, was used to infer the phylogenetic tree with 1000 bootstrapping replicates. Multiple sequence alignments of VmGlu2, and other well characterized β-glucosidase genes were performed using the DNAMAN (version 6.0) with all the parameters set at the default values.

The cDNA fragment of VmGlu2 was subcloned into the pET-32a vector by homologous recombination using the ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). The vector construct was transformed into E. coli strain Rosetta while the soluble recombinant protein of VmGlu2 was obtained after induction with 0.5 mM isopropyl β-D-thiogalactopyranoside (IPTG) for 16 h at 15° (Shi et al., 2015). The purification of recombinant VmGlu2 protein from the culture was performed using Ni-NTA Spin Column (Qiagen, Beijing, China). The expression and purification of recombinant protein were detected by the sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot.

The β-glucosidase activities were determined using pNPG as a substrate according to a previously established method (Alconada and Martínez, 1996). The enzyme activity was measured at 50° in 50 mmol/L sodium-citrate buffer with a pH of 5.5. One unit of enzyme activity was defined as the amount of product (μmol) formed per min under the assay conditions.

The activities of β-glucosidase were also measured using the phloridzin as a substrate by high performance liquid chromatography (HPLC) according to Gaucher et al. (2013) with some modifications. Activity was assayed in 1 mL of reaction mixture containing 0.5 mmol/L phloridzin and about 20 μg purified recombinant protein in 50 mmol/L phosphate buffer (pH 7.0) at 25° under continuous stirring conditions. The samples were analyzed by HPLC on an Agilent 1200 (Agilent Technologies, Santa Clara, United States), and separated on a C18 column (250 mm × 4.6 mm, 5 μm) at 30°, with a detection wavelength at 285 nm.

The mycelia plugs of V. mali were used to inoculate apple twigs described by Yu et al. (2018). The lesion border of apple bark tissues was sampled at different time points (0, 6, 12, 24, 48, 72, and 120 hpi). For samples at 0 hpi, bark tissues around inoculation sites containing mycelium plugs were collected. The RNA was extract from bark tissues using the RNAiso Plus Kit (TaKaRa, Dalian, China), and then the cDNA was synthesized with a Reverse Transcription Kit. All of the qRT-PCR experiments were conducted in a LightCycler 480II PCR Detection System (Roche, Mannheim, Germany) with SYBR Master Mix (TaKaRa, Dalian, China) following the manufacturer’s protocol. The EF1-a of V. mali was used as an internal control, while primers are given (Supplementary Table 1). The relative expression level of VmGlu2 was calculated using the 2^–ΔΔCT method (Livak and Schmittgen, 2001). Data from three biological replicates were used to calculate the means and standard deviation. The whole experiment were repeated twice.

To obtain the VmGlu2 deletion mutants, the polyethylene glycol (PEG)-mediated homologous recombination was performed as described previously (Yu et al., 2018). VmGlu2 and about 1200 bp flanking sequences of the gene were amplified from the genomic DNA of the wild-type strain LXS080601. The hygromycin B phosphotransferase gene (HPH) gene was amplified with primers HPH-F and HPH-R from the vector pBS (Supplementary Table 1). Two flanking fragments and HPH resistance cassette were constructed into a fusion fragment (3,383 bp) using a nested PCR reaction. The fusion fragment was later transformed into the protoplasts of V. mali strain LXS080601, and the transformants were screened by culturing on medium with 100 μg/mL of hygromycin (Yu et al., 2018). More than 600 transformants were detected by PCR using the detection primer pairs VmGlu2-I-F/R. The putative gene deletion mutants were then validated by PCR using four primer pairs (Supplementary Table 1). The gene deletion mutants were finally verified by Southern hybridization with the DIG DNA Labeling and Detection Kit II (Roche, Mannheim, Germany) following the instruction manual.

To generate a complemented strain, the entire VmGlu2 gene with upstream fragment was amplified from genomic DNA using the primer pair 1491-CM-F/R (Supplementary Table 1). The PCR products were cloned into the plasmid pYF11 using yeast gap repair and then verified by the sequencing analysis. The recombinant vector pYF11-Glu2 was then transformed into the gene deletion mutant via PEG-mediated method. The transformants were selected using G418, and confirmed by PCR with the corresponding primers (Supplementary Table 1).

The full-length fragment of VmGlu2 was amplified from the genomic DNA of V. mali and then cloned into plasmid pYF11 using the ClonExpress II One Step Cloning Kit (Vazyme, Nanjing, China). The construct was verified by sequencing and then transformed into the wild-type strain LXS080601 as described above. Transformants were screened by PCR with corresponding primers outside the cloning sites of the pYF11. Relative transcript levels of VmGlu2 in mycelia grown on PDA and inoculated apple twigs were determined as described above. Transformants were further verified by Western blot with Anti-GFP antibody (Abcam, United Kingdom).

The vegetative growth of VmGlu2 deletion mutants and overexpression transformants was determined as described by Yu et al. (2018). Colony shape and color were observed, and then the colony diameters were measured. For determination of the dry weight of mycelia, plugs were inoculated into PDB media (potato dextrose broth) at 150 rpm and 25° for 7 days. For pycnidia formation, all the strains were grown on PDA under UV-light (365 nm) at 25° for 4 weeks. The assays were carried out three times and each experiment with four replicates.

For pathogenicity assays, we selected healthy apple leaves and 1-year old twigs (Malus domestica Borkh. cv. “Fuji”), which were sterilized with 75% ethanol, and spaced wounds were made as described by Yu et al. (2018). The mycelia plugs of all strains were used to inoculate the wounds. The inoculated leaves and twigs were placed in trays to maintain humidity at 25° in the dark. Then, the lesion length was measured and the photography was performed at different time intervals. The assays were repeated three times, with at least 10 leaves and twigs per treatment.

For toxins production, the mycelium plugs of different strains were inoculated into apple branch extract media. After culturing for 7 days at 150 rpm/min and 25°C with a 12-h photoperiod, the supernatant was collected. The liquid media inoculated with PDA was used as a control. Toxins were extracted and detected by HPLC according to the method described by Wang et al. (2014). Each experiment was done in triplicate.

All statistical analysis was conducted using SPSS software (Version 19.0, SPSS Inc., Shanghai, China). All the data collected were subjected to analysis of variance (ANOVA) followed by Duncan’s multiple range tests. The asterisks indicate a statistically significant difference with the wild-type strain (P < 0.05).

The VmGlu2 gene was amplified by PCR using cDNA of V. mali as template, and then confirmed by the sequencing (Supplementary Figure 1A). The full-length cDNA of VmGlu2 is 1,677 bp containing 5°- and 3°-non-coding regions, as well an open reading frame of 1,491 bp that encodes a protein with a calculated molecular mass of 54.6 kDa. The DNA fragment was also 1,491 bp, with no intron. Using the Signal P5.0 server, VmGlu2 was predicted that no signal peptide is present. NCBI’s conserved domain database showed that VmGlu2 had a β-glucosidase (Bg1B) conserved domain, which belonged to a GH1 family (Supplementary Figure 1B). The sequence comparison demonstrated that VmGlu2 exhibited high homology with the well-characterized β-glucosidase. A phylogenetic tree was constructed with β-glucosidase from other strains (Figure 1A). The sequence alignments of VmGlu2 with four characterized GH1 β-glucosidases showed that VmGlu2 had the catalytic residues among these enzymes that are related to their activity, in which Glu395 acts as a nucleophile and the Glu181 acts as an acid/base (de Giuseppe et al., 2014). The sequence alignment is shown (Figure 1B).

In order to determine the enzymatic activity of Bg1B, the VmGlu2 was cloned into pET-32a for heterologous expression in E. coli with a six histidine tag fused to the N-terminus part. The VmGlu2 recombinant protein with the expected molecular weight (MW) was obtained and verified by Western blot with Anti-His antibody (Figure 2A). We got the purified recombinant protein through Ni-nitrilotriacetic acid (Ni-NTA) Spin Column, and checked the protein by SDS-PAGE (Figure 2B).

The β-glucosidase activity of VmGlu2 protein was measured using pNPG as substrate at different temperatures (from 30 to 70°C) and pH values (pH 2.5–7.5). The enzyme activity remained constant from 30 to 40°C, dramatically increased to the highest value of 1.74 U/mL at 50°C, then declined abruptly to 36% of the maximal activity at 70°C. The optimal pH of VmGlu2 is around pH 5.5 as the enzyme maintained high activity between pH 3.5 and pH 6.5, while its activity declined rapidly to 30 and 37% of its optimum at pH 2.5 and pH 8.5, respectively (Figure 2C). In addition, we determined its activity when phloridzin as substrate by high performance liquid chromatography (HPLC). The Figure 2D showed that VmGlu2 had high activity to convert phloridzin into phloretin, with 78 and 100% conversion rate at 24 and 72 h, respectively.

To examine the expression level of VmGlu2 in different apple tissues, we sampled infected apple phloem and xylem tissues at 72 h post inoculation (hpi) and compared the expression level with that of mycelia grown on PDA. qRT-PCR analysis showed that the expression level of VmGlu2 were significantly up-regulated in the apple phloem (9.7-fold change) as compared to its expression in the mycelia grown on PDA. During the infection stage, the up-regulation trend of VmGlu2 in apple xylem (1.7-fold change) had no significant difference relative to its expression levels in PDA (Figure 3A).

We also determined the expression level of VmGlu2 during the interaction of V. mali with apple twigs (Figure 3B). To do this, we sampled infected apple phloem tissues at seven different time points (0, 6, 12, 24, 48, 72, and 120 hpi), and then compared gene expression levels to those of mycelia grown on PDA. The transcript level of VmGlu2 was significantly up-regulated from 12 hpi, gradually increased and reached the highest with a 13.5-fold change at 48 hpi, and then declined sharply from 72 to 120 hpi (4.8-fold change). Overall, the high induction of VmGlu2 during infection suggests its potential role in the pathogenicity of V. mali.

To verify the functional role of VmGlu2 in V. mali, we successfully knocked out the gene by PEG-mediated protoplasts transformation method. The mutants were examined via PCR assays and confirmed by Southern blot (Supplementary Figure 2). When probed with HPH fragment, the wild-type (WT) sample showed no hybridization signal. However, a clear hybridization signal of the predicted size was appeared on analysis of the gene deletion mutants. For the complementation of VmGlu2 deletion mutant, a complementary construct was generated and transformed into the gene deletion mutant. All the complementation transformants were confirmed by PCR assays (Supplementary Figure 3A). In addition, RT-PCR analysis were also performed to confirm that the target VmGlu2 was knocked out in the gene deletion mutants and that the complementation strains contained the target gene (Supplementary Figure 3B).

To determine whether VmHbhs plays a role in V. mali growth and development, we investigated the colony, growth rate, and pycnidia formation of the VmGlu2 deletion mutants and the wild-type strain. However, no obvious differences either in colony morphology or growth rate were observed between the VmGlu2 deletion mutants and the wild-type strain (Figures 4A,B). Assessment of the pycnidia formation by ΔVmGlu2 was performed on PDA under UV-light (365 nm) at 25°. The VmGlu2 deletion mutants produced less than 13 pycnidia per plate, while the wild-type strain produced more than 126 pycnidia per plate. Furthermore, the complementation strain ΔVmGlu2-C by reintroducing VmGlu2 restored the wild-type strain pycnidia formation (Figure 4C).

To investigate whether four VmGlu2 play role in the virulence, we performed pathogenicity assays of the wild-type strain and gene deletion mutants on detached apple leaves and twigs. Our results showed that VmGlu2 deletion strains were significantly less virulent toward the apple leaves and twigs at 3 or 5 dpi, as compared to the wild-type stain, which typically displayed diseased symptoms of necrosis and canker (Figure 5). However, smaller lesions were found on the ΔVmGlu2 inoculated apple leaves and twigs than those caused by the wild-type strain (Figures 5A,B). The ΔVmGlu2 strains demonstrated a more than 60 and 76% reduction in the average lesion size on apple leaves at 3pi and twigs at 5 dpi. The complementation strain ΔVmGlu2-C recovered the high virulent phenotype that showed same symptoms on the apple leaves and twigs after reintroduction of VmGlu2 into the ΔVmGlu2 mutant (Figure 5C).

To further determine the function of VmGlu2 on the pathogenicity of V. mali, the gene overexpression transformants were generated and detected by PCR (Supplementary Figure 4). The overexpression transformants were confirmed by Western blot analysis. When probed with Anti-GFP antibody, the wild-type strain showed no GFP protein signal was found, however, two transformants of VmGlu2 overexpression (OE) displayed specific GFP detected bands with the expected size. Moreover, when probed with Anti-GAPDH antibody, all the strains exhibited specific immune signals (Figure 6A). Two VmGlu2 overexpression transformants (VmGlu2-OE-78 and VmGlu2-OE-92) were selected for further analysis. Compared with the wild-type stain, the transcript levels of VmGlu2 in VmGlu2-OE-78 and VmGlu2-OE-92 increased 8.6- and 7.6-fold, respectively, during the infection apple twigs (Figure 6B). Further analysis showed that overexpression of VmGlu2 did not affect vegetative growth and pycnidia formation (Figures 7A–C). However, VmGlu2 overexpression transformants significantly enhanced the virulence of V. mali. Larger lesions were observed on the VmGlu2 overexpression transformants inoculated apple leaves (17% increase), especially apple twigs (25% increase) than those caused by the wild-type strain (Figures 7D–F). The phenotype and pathogenicity of the empty vector transformants and the wild-type strain were comparable (data not shown). These results verify that VmGlu2 is a major virulence factor involved in the V. mali pathogenicity.

To detect VmGlu2 activity, the wild-type strain, gene deletion mutants, and overexpression transformants were cultured in the apple branch extract media. We determined β-glucosidase activity in the culture filtrates of all strains and uses pNPG as a substrate. The gene deletion strain ΔVmGlu2 showed 50 and 39% reduction in the β-glucosidase activity at 1 and 3 dpi, respectively. The complementation strain ΔVmGlu2-C with the native gene restored β-glucosidase activity to the wild-type levels (Figure 8A). In contrast, the VmGlu2 overexpression transformants exhibited an enhanced β-glucosidase activity, with 3.0- and 3.3-fold increases compared with the wild-type strain (Figure 8B). Thus, the results in enzyme activity assays suggest the deletion of VmGlu2 significantly reduced the β-glucosidase activity, however, the gene overexpression enhanced the β-glucosidase activity.

To further reveal the mechanism underlying the virulence of VmGlu2 deletion and overexpression, we examined the toxin levels in the apple branch extract media for different strains. All tested strains produced five kinds of toxins, however, the ΔVmGlu2 mutants exhibited different degrees of reduction in the levels of five toxic compounds (Figure 8C). The VmGlu2 deletion strains exhibited 20 (phloroglucinol) to 71% (protocatechuic acid) reduction in each toxic compounds than the wild-type strain. Moreover, the total toxins produced by the ΔVmGlu2 mutants showed 48% decrease than the wild-type strain, whereas the complementation strain with the native gene restored the phenotype. In contrast, the toxin levels produced by the VmGlu2 overexpression transformants were not significantly different than those produced by the wild-type strain (Figure 8D).