QTL detection and candidate gene analysis of grape white rot resistance by interspecific grape (Vitis vinifera L. × Vitis davidii Foex.) crossing

Abstract Grape white rot, a devastating disease of grapevines caused by Coniella diplodiella (Speg.) Sacc., leads to significant yield losses in grape. Breeding grape cultivars resistant to white rot is essential to reduce the regular use of chemical treatments. In recent years, Chinese grape species have gained more attention for grape breeding due to their high tolerance to various biotic and abiotic factors along with changing climatic conditions. In this study, we employed whole-genome resequencing (WGR) to genotype the parents of ‘Manicure Finger’ (Vitis vinifera, female) and ‘0940’ (Vitis davidii, male), along with 101 F1 mapping population individuals, thereby constructing a linkage genetic map. The linkage map contained 9337 single-nucleotide polymorphism (SNP) markers with an average marker distance of 0.3 cM. After 3 years of phenotypic evaluation of the progeny for white rot resistance, we confirmed one stable quantitative trait locus (QTL) for white rot resistance on chromosome 3, explaining up to 17.9% of the phenotypic variation. For this locus, we used RNA-seq to detect candidate gene expression and identified PR1 as a candidate gene involved in white rot resistance. Finally, we demonstrated that recombinant PR1 protein could inhibit the growth of C. diplodiella and that overexpression of PR1 in susceptible V. vinifera increased grape resistance to the pathogen.


Introduction
Grapevine is a prominent global horticulture crop that has been grown for millennia to provide fresh fruit, wine, juice, and raisins. However, the grape is subject to a variety of diseases, including grape powdery mildew, white rot, black rot, and downy mildew [1][2][3][4]. Grape white rot, a devastating disease of grapevines caused by Coniella diplodiella (Speg.) Sacc., attacks leaves, branches, and berries, leading to a reduction in grape output and quality [5,6] ( Fig. 1). Surveys of grape white rot in China revealed that the disease is widespread in significant grape-growing regions [7]. Despite efforts of growers to limit the occurrence of white rot through thinning, leaf removal, and pruning operations to increase air and light circulation, the disease can still manifest under unfavorable environmental conditions such as heavy rain and hail [7]. As a result, the control of grape white rot in grape production requires regular application of chemicals, which not only increases production cost but is also harmful to the environment and is against the policies of food safety measures [8]. Breeding of grape white rot disease-resistant varieties is beneficial for the healthy development of the grape industry. Traditional crossbreeding procedures are time-consuming and inefficient. Adopting the marker-assisted selection strategy, which enables the targeted selection of progeny with resistance loci, is one way to significantly expedite the breeding process [9,10].
Plants are constantly exposed to a variety of pathogenic agents, including bacteria, fungi, oomycetes, and viruses, which pose significant challenges to their growth and development. To protect themselves, plants have evolved two types of immune response: pathogen-associated molecular pattern-triggered immunity (PTI) and effector-triggered immunity (ETI). PTI is initiated when a pattern recognition receptor (PRR) recognizes pathogen-associated molecular patterns (PAMPs) and activates the plant immune system. On the other hand, ETI is activated when resistance (R) proteins recognize pathogen effector proteins (PEPs). ETI triggers the hypersensitive response (HR), which induces cell death at the site of infection, limiting local pathogen dissemination [26][27][28]. Furthermore, ETI can elicit systemic transfer of immune signals, triggering the synthesis of pathogenesis-related (PR) proteins with antimicrobial activity, thereby enhancing plant defense against subsequent pathogenic assaults [28][29][30]. Among the PR family, pathogenesis-related 1 (PR1) proteins are essential and are produced abundantly during defense responses [31,32]. Su et al. [33] demonstrated that the expression of PR1 is regulated by crosstalk between salicylic acid and jasmonic acid, which is crucial for grapevine resistance against white rot.
In this work, we generated a linkage map using a population of 'Manicure Finger' (Vitis vinifera, susceptible) × '0940' (V. davidii, resistant). The map, in combination with phenotypic data, was able to identify one stable white rot resistance QTL on Chr 3. According to the mapped QTL and the RNA-seq results, one potential PR1 gene was analyzed that may be related to white rot.

Phenotypic evaluation
In July of 2019, 2020, and 2021, 101 hybrid progeny were evaluated for resistance to white rot. Their susceptibility to white rot ranges from 1 (most resistant) to 5 (most susceptible). Vd0940 (male) displayed a distribution of resistance at level 1, while VvMF (female) showed a susceptibility distribution at level 4. Most of the offspring showed a susceptibility level of 4 (45.5-56.4%), followed by level 3 (26.7-34.7%) (Fig. 2, Supplementary Data S1). Moreover, there was a significant correlation between the grape white rot resistance scores in three years according to Pearson's correlation coefficient (P < .001) (Supplementary Data Fig. S1).

Construction of genetic linkage map
We conducted whole-genome resequencing of the F 1 population and both parents, generating 232.04 Gb of clean data. The average coverage depths of the genomes were 34-fold for VvMF, 27-fold for Vd0940, and 6-fold on average for their offspring. After filtering the single-nucleotide polymorphism (SNP) markers, we constructed separate genetic maps of the two parents and an integrated genetic map using high-quality SNP markers. The female parent VvMF had a genetic map distance of 3229 cM, consisting of 7230 SNP markers distributed over 19 linkage groups (LGs). The total number of SNP markers ranged between 199 and 572 across all linkage groups. We observed the greatest gap, of 55.9 cM, in LG 2, while the smallest was 6.3 cM in LG 8. The percentage of gap < 5 cM (adjacent marker distance < 5 cM) ranged between 97.0% and 99.7% among 19

Quantitative trait locus analysis
We performed QTL mapping by utilizing the integrated genetic map in conjunction with phenotypic data collected over a 3-year period (2019-21) and mapped them to QTL on Chr 3, 8, 12, and 18 ( Fig. 4). Intriguingly, only one stable QTL, Rcd1 (resistance to C. diplodiella 1), was consistently detected on LG 3 over the 3-year period. The maximum LOD value in Rcd1 was 3.98, explaining up to 17.9% of the phenotypic variation (Table 2). Notably, a stable expression QTL was also detected on LG 3 using two parental maps. By mapping SNP marker positions in the parental QTL intervals to the grape physical map, we observed that the QTL mapping results using the Vd0940 map were consistent with the physical positions of the integrated map. The physical location of QTL mapping using the VvMF map is also within the physical location interval of the integrated map (Supplementary Data Table S5). Furthermore, we detected several disease resistance QTLs in 2019-21 using the integrated map, with LOD values ranging from 3.07 to 4.16 and explaining 14.3 and 18.6% of phenotypicvariation, respectively (Supplementary Data Table S6).
In 3 years, 22 markers that co-segregated with C. diplodiella resistance were found using a rank sum test based on the Kruskal-Wallis method. Among them, Marker 663881 was found to be  most significantly associated with C. diplodiella resistance, and was also located in close proximity to the LOD peak in the Rcd1 region ( Table 2, Supplementary Data Table S7). The position of this marker on Chr 3 was 7 653 321 bp. Further analysis of the raw sequencing data linked with this marker revealed that the nucleotides in Vd0940 were T/T, whereas the nucleotides in VvMF were T/A. In comparison with T/A individuals with a susceptible phenotype, progeny carrying T/T usually showed a resistant phenotype, and 3-year findings are generally consistent (Fig. 5).

Quantitative trait locus candidate gene analysis
Based on the grapevine genome sequence, Rcd1 was located at physical positions of 6 282 673-9 416 779 bp on Chr 3 and contained 185 predicted genes (Supplementary Data S4). Of these predicted genes, 40 were annotated to be associated with disease. To further investigate the potential involvement of these genes in disease resistance, we performed expression gene analysis on grape leaves of VvMF and Vd0940 at 24 and 48 h after C. diplodiella infection during RNA-seq analysis. Only genes with FPKM (fragments per kilobase of transcript per million mapped reads) values ≥1 were considered expressed genes (Table 3). For these 19 genes, one gene (100246419) had higher expression in Vd0940 compared with VvMF at 24 and 48 h after C. diplodiella infection, indicating a significant role in grape white rot resistance.

Functional analysis of PR1
The pathogenesis-related protein 1 gene (PR1, gene ID 100246419) was found on Chr 3 with a full-length coding sequence of 483 bp, encoding 160 amino acids and including a CAP conserved  domain, according to the Grape Genome Browser at NCBI (Fig. 6A). To understand the molecular basis for PR1 expression, we investigated the genomic sequence and promoter activation of the PR1 gene. The PR1 genomic sequence analysis indicated that the VdPR1 (Vd0940-PR1) coding region was identical to the VvPR1 (VvMF-PR1) coding region at the amino acid level  pathogens and stress responses. Nonetheless, some differences were observed in the PlantCARE database analysis (Fig. 6C, Supplementary Data S5). We examined the promoter activities of the VvPR1 and VdPR1 genes by histochemical staining to detect β-glucuronidase (GUS) activity in agro-infiltrated tobacco leaves. The GUS-staining blue colors in VdPR1 were more intense than those in VvPR1, indicating that the promoter activity for VdPR1 is stronger than that for the VvPR1 genes examined (Fig. 6D).
Whether PR1 protein plays a role in white rot resistance is not known. We selected the VdPR1 gene for functional studies because the coding sequences of VvMF and Vd0940 were completely consistent. We first purified the PR1 protein and subjected it to an inhibition assay against C. diplodiella. Results showed that recombinant VdPR1 protein incubated at 40 μg/ml for 4 days had a strong inhibitory effect on the growth of C. diplodiella, as compared with the no-protein buffer (control) and boiled recombinant VdPR1 protein, which had no effect on C. diplodiella growth ( Fig. 7A and B). To further validate the role of VdPR1 in enhancing resistance against C. diplodiella, we transiently transformed the PR1 gene into diseasesusceptible grape (Jingxiu) leaves and inoculated them with agar disks containing mycelium of white rot. Transcript levels of PR1 were analyzed at 24 h after infiltration, and results showed that the overexpressed PR1 gene had a 3-fold higher expression level compared with untransformed leaves and empty vector, confirming successful overexpression of PR1 (Fig. 7C). Lesion diameters were then measured 3 days after pathogen inoculation, and we found that overexpressed VdPR1 significantly reduced lesion diameter compared with the overexpressed empty vector, indicating that transient VdPR1 overexpression enhances C. diplodiella resistance in disease-susceptible grape varieties (Fig. 7D).

Discussion
In grapevine, genetic maps are typically generated by crossing two heterozygous parents and evaluating their segregating markers in the offspring [54]. The genetic maps based on first-generation and second-generation markers, such as RFLP, AFLP, RAPD, and SSR, have resulted in low map marker density, large marker spacing across linkage groups, and uneven marker distribution throughout the genome [11]. With the advancement of sequencing technology, high-density genetic maps using SNP markers have greatly increased the level of fine localization [54][55][56]. Constructed genetic maps using the PCR method had a density range of 3.9-12.7 cM, whereas those using the SNP method had a density range of 0.41-1.81 cM [55,[57][58][59]. Although the genetic distance between markers is decreasing, the populations used to construct the genetic map had an average of 166 offspring individuals [54]. Here, we constructed a high-density linkage genetic map with an average marker interval of 0.3 cM using two parents (Vd0940 and VvMF) and a mapping population of 101 individuals. Additionally, the average Spearman correlation coefficient of positions in genetic and physical maps was 0.94 ( Supplementary Data Fig. S5). The high collinearity indicates that the markers accurately cover the 19 chromosomes and sufficiently represent the Vitis genome. Therefore, our genetic map provides valuable information for QTL analysis, gene mapping, and marker-assisted selection to enhance grape resistance to white rot.
In resistance studies, the accuracy of resistance assessment and phenotypic data is crucial. The evaluation of grapevine resistance under in vitro conditions enables an increased degree of bioassay replication and allows for the execution of one or more experiments per year, rendering the in vitro phenotyping strategy an efficient and practical tool for identifying resistance QTLs [60]. Phenotypic analyses in resistance mapping studies usually use visual ratings of disease symptoms. Particularly in the study of resistance loci for downy mildew (DM) and powdery mildew (PM), DM and PM sporulation is evaluated for incidence and severity using different visual scales, such as the OIV452-1 and 455-1 descriptors [60,61]. In the pathogenicity evaluation of white rot fungal strains, Chethana et al. [7] recorded lesion length 3 days after inoculation to determine that the most pathogenic strain was JZB3700012. Although visual evaluation is considered subjective, the impact of human inf luence can be mitigated by employing biological replication and multi-year testing [60]. In this study, we ensured the accuracy of the phenotypes by collecting four leaves from each plant and inoculating four spots on each leaf for 3 years. Importantly, there was a significant correlation between the phenotypic data in each year. Identifying QTLs and candidate genes associated with white rot resistance in grapes is of utmost importance for the industry. In this study, we confirmed one stable QTL for white rot resistance on Chr 3, explaining up to 17.9% of the phenotypic variation. Only one report on QTL for grapevine white rot resistance has been published so far. Su et al. [25] identified one stable QTL for white rot resistance locus on chromosome 14 through the 'Zhuosexiang' × 'Victoria' population, explaining 13.43% of the phenotypic variation. Various localization results are frequently found utilizing different populations for the same disease. Notably, we found that individuals carrying the homozygous T/T genotype were usually resistant to white rot. Similarly, Su et al. [25] found that individuals carrying the homozygous G/G (Chr14_3929380) genotype had smaller lesion areas and were resistant to C. diplodiella, while those carrying the heterozygous G/A genotype showed susceptibility. In QTL mapping of DM resistance, Bhattarai et al. [12] found that individuals carrying the homozygous genotype showed higher resistance to DM. Additionally, a marker closely associated with sugarcane yellow leaf virus (SCYLV) resistance was found in sugarcane, with homozygous (C/C) progeny effectively suppressing the incidence of SCYLV [62]. Several disease-resistance-related genes, including disease-resistance protein (NBS-LRR class), protein trichome birefringence, pathogenesis-related protein, zinc finger protein, E3 ubiquitin-protein and serine/threonine-protein kinase, were found in the QTL region. Some of the 386 R genes predicted for the grapevine are related to resistance to powdery mildew, white rot, downy mildew, and anthracnose [63][64][65][66]. Numerous NBS-LRR-like R genes were also identified in previously identified disease resistance QTL regions. For example, 11 NBS-LRR genes were found in the comparable region of Cgr1 [57]. Two of the seven TIR-NB-LRR genes in the relevant region of the Rpv1/Run1 locus have been demonstrated to increase disease resistance using transgenic verification [67]. Additionally, an NBS-LRR-like R gene that may be crucial in grape white rot resistance was discovered [25]. Upregulation of PR1 expression is often used as a marker of resistance responses, such as PTI and SAR [68]. The PR1 gene was found to be important for grape resistance to white rot [33].
In this study, we also identified QTLs located in Chr 8, Chr 12, and Chr 18 with lower reproducibility. This may be attributed to the fact that QTLs with substantial effects can be identified in small populations, whereas QTLs with weaker effects require larger population sizes to be detected [69]. We found these loci present in multiple disease-resistance genes. The receptor-like kinases on the cell membrane serve as PRRs that are crucial for the detection of PAMPs, a process commonly known as PTI [26]. Wang has reported that the lectin receptor-like kinases (LecRKs) can enhance resistance to powdery mildew in wheat [70]. Pathogens can counteract PTI by secreting effectors, but R proteins in plants can recognize these effectors and trigger an immune response (ETI), leading to programmed cell death, inhibiting pathogen spread [28,29]. Both PTI and ETI can cause downstream responses such as reactive oxygen species (ROS) burst, MAPK cascade activation, and PR protein induction [28,71]. These intricate defense responses involve the expression of numerous defense genes. The expression of defense-related genes is regulated by transcription factors (TFs), which work in concert with other proteins to bind to the DNA-binding sites of target genes [72]. Our results indicate that the expression level of the receptor-like kinase gene (gene ID 100854743, Chr 8; 109121483, Chr 12), peroxidase (gene ID 100854817, Chr 12), WRKY49 (gene ID100254510, Chr 8), E3 ubiquitin-protein ligase (gene ID 100257429, Chr 12) was higher in Vd0940 than VvMF induced by C. diplodiella after 24 and 48 h.
Based on the results of QTL mapping and RNA-seq analysis, we hypothesized that the PR1 gene plays a significant role in white rot resistance. Even though the coding sequences of PR1 genes in Vd0940 and VvMF are identical, their expression patterns in response to C. diplodiella infection were significantly different. During the infection phase, the level of induced expression of VdPR1 was significantly higher than that of VvPR1, which may be a contributing factor to the much higher resistance of Vd0940 to C. diplodiella (Fig. 6B). Previous studies have demonstrated that the cis-acting sequence in the plant promoter is a critical element in controlling plant gene expression, and GUS reporter geneaided histochemistry can be utilized to study gene expression at the transcriptional level [73,74]. Thus, the stronger promoter activity for VdPR1 compared with VvPR1 was the primary reason for the difference in gene expression (Fig. 6D). Our findings indicated that the PR1 protein can inhibit the growth of C. diplodiella (Fig. 7B). Recent research has shown that the in vitro activity of PR1 orthologs from various plant species can inhibit the growth of many pathogens [75][76][77]. Furthermore, we observed that overexpression of PR1 in susceptible V. vinifera enhanced plant resistance to the pathogen (Fig. 7D). Similarly, overexpression of PR1 in tobacco improved resistance to Botrytis cinerea, while overexpression of PR1 in transgenic Arabidopsis exhibited higher resistance to Sclerotinia sclerotiorum [78,79]. Additionally, the effector protein of the pathogen can be targeted by PR1. For example, Wheat PR1 proteins were shown to interact with the effector proteins of Stagonospora nodorum (SnToxA and SnTox3), barley powdery mildew effector (CSEPP0055) targeted PR1a and PR1b, and S. sclerotiorum effector (SsCP1) interacted with PR1. These findings suggested that PR1 is a common host protein implicated in host defense against pathogens [79][80][81][82][83]. Therefore, further studies are required to determine whether C. diplodiella effector proteins can target PR1.

Plant materials
The F 1 population was generated by crossing 'Manicure Finger' (V. vinifera, female, VvMF) with '0940' (V. davidii, male, Vd0940) in 2015. The crosses were manually performed in the field by removing the f loral caps on the VvMF parent and applying dried pollen collected from the Vd0940 parent. To prevent unintended pollination, inf lorescences were covered with paper bags for 4 weeks. Subsequently, seeds were collected from berries in October of the same year. The germination of seeds took place in the greenhouse during the spring of 2016, and the resulting seedlings were transferred to the National Grape Germplasm Resource Garden (Zhengzhou) for field management. The seedlings were cultivated in accordance with the appropriate field management procedures during their early growth stages. Eventually, 101 individual plants survived and were used for the mapping population. Nevertheless, the F 1 populations were not treated with fungicides in the current-year trials to evaluate their resistance to white rot.

Disease evaluation
The C. diplodiella (Speg.) Sacc. strain WR01 was obtained from the Institute of Plant Protection, Chinese Academy of Agricultural Sciences. The C. diplodiella was grown in the dark at 28 • C on potato dextrose agar (PDA) medium until the mycelium covered the whole medium. Mature leaves were collected from each individual from the vineyard. Leaf surface sterilization was performed as described by Chethana et al. [7], [10]. Wound inoculation experiments were conducted using five leaves per plant (including one control) and each leaf was wounded on the upper epidermal layer and divided into four parts on average. PDA blocks (9 mm in diameter) containing fungal mycelium were placed on the wounds

Rating
Average lesion diameter (mm) 1 ≤10  2  10-20  3  20-30  4 30-40 5 >40 while sterile agar blocks (9 mm in diameter) were placed on the punctured leaves in the control treatment. The leaves were placed on moist filter paper within Petri dishes and incubated at 28 • C. Three days after inoculation, we measured the average lesion diameter of necrotic patches using a slide caliper rule and rated disease severity on a scale of 1-5 (Table 1). In each experiment, both parental leaves were inoculated as resistant and susceptible controls. We conducted the experiment in July of 2019, 2020, and 2021 and averaged phenotypic data for lesion diameter from four biological replicates per individual (Supplementary Data S1).

DNA extraction and library building
Genomic DNA was extracted from both the parent and progeny leaves using a DP360 genomic DNA extraction kit (Tiangen Biotech, China). The purity of the DNA was determined through agarose gel analysis. Next, 200-to 500-bp fragments were randomly cut using sonication, and 3 A and sequencing adapters were added. The resulting samples were purified and amplified via PCR to develop the library. Subsequently, the HiSeq2500 system (Illumina, San Diego, CA, USA) was utilized to re-sequence the entire genome. The average coverage depths of the genomes were 34-fold for VvMF, 27-fold for Vd0940, and 6-fold on average for their offspring (Supplementary Data S2).

Genotyping and genetic map construction
A 12× assembly of the grape PN40024 genome was used as a reference [34]. Prior to alignment, the reads underwent Fastp [35] processing to eliminate adapters, N >10% of read pairs, and low-quality bases (Q <10) >50% of read pairs. Using the Burrows-Wheeler Aligner (BWA) tool, the reference genome was aligned with clean reads. SNP marker detection and filtering were performed using GATK [36]. The SNP markers were discovered based on parental sequence depths exceeding 6-fold and offspring sequence integrity >75%. Additionally, map development initially excluded markers with significant segregation distortion (P < .01). High-quality SNP markers were partitioned into 19 groups. After determining the modified logarithms of odds (MLOD) values of markers for each linkage group, markers with MLOD values >3 were ranked. A LOD value of >3.0 implies that the chances are greater than 1000:1 that the markers are linked for a given recombination estimate [37]. Finally, we utilized HighMap [38] software to evaluate the arrangement of markers in each linkage group and estimate the genetic distance between nearby markers, resulting in the final genetic map (Supplementary Data S3).

Quantitative trait locus analysis and candidate gene prediction
To detect QTLs, we employed the MapQTL6.0 [39] software on the integrated map, combining it with the phenotypic data used for each year alone. We repeated this analysis with separate  parental maps for comparison. The initial step involved using interval mapping (IM) to detect potential QTLs, and a significant QTL was defined based on LOD >3.0 [40,41]. After identifying a potential QTL using interval mapping analysis, we utilized markers that linked to the QTL as cofactors. For more precise positioning of QTLs, we employed multiple QTL model (MQM) mapping with selected cofactors [41,42]. To determine the LOD threshold required for a QTL to be present in a specific genomic region, we conducted a permutation test with 1000 cycles. We considered a locus to be present if its confidence interval was higher than the LOD threshold. Finally, all SNP markers located in the QTL intervals of the integrated map were utilized to identify candidate genes by locating their physical position on the grape PN40024 genome [43,44]. MapChart 2.2 was used to show the QTL regions of the linkage map [45].

Transcriptome analysis and qRT-PCR validation
The collected leaves were inoculated as described previously. We chose infected grape leaves 24 and 48 h post-infection of VvMF and Vd0940, and non-infected leaves (0 h) as a control, using three replicates per cultivar (Supplementary Data Fig. S2

Cloning and analysis of gene sequences
The PR1 reference sequences were downloaded from the GenBank database of the NCBI (https://www.ncbi.nlm.nih.gov/ gene/100246419). To clone the PR1 coding sequence, primers (PR1CDS) were designed using the PR1 coding sequence from the grapevine database. Using SignalP6.0 (https://services.healthtech. dtu.dk/service.php?SignalP-6.0), signal peptides were predicted. The promoter sequence of PR1primer (PR1 promoter) was designed based on the upstream sequences 2000 bp of PR1 in the grapevine database. The promoter sequence was analyzed with the PlantCARE database [48]. Aligning multiple sequences was performed by DNAMAN (Lynnon Biosoft; San Diego, CA, USA). The primers used are provided in Supplementary Data Table S1.

Construction of β-glucuronidase vectors and histochemical β-glucuronidase staining
The GUS reporter gene was fused to the PR1 promoter using primer (pBI121-VvPR1promoter and pBI121-VdPR1promoter) in the expression vector pBI-121. pBI-121, an expression vector containing the CaMV35 strong promoter, and pBI-101, a nonpromoter expression vector, served as positive and negative controls, respectively. Transient expression assay using Agrobacterium was as described by Rahman et al. [49]. The Agrobacterium suspension was infiltrated into the tobacco leaf, which was placed in a dark room at 26 • C for 24 hours after infiltration, then moved to a growth chamber. Histochemical staining was used for the measurement of β-glucuronidase (GUS) expression for qualitative analysis of promoter activity [50]. Following the manufacturer's protocol, agro-infiltrated tobacco leaves were stained with the Gusblue kit (GT0931, Huayueyang Biotech, China) and chlorophyll was removed from tobacco leaves by using ethanol. The primers used are provided in Supplementary Data Table S1.

Purification of recombinant PR1 protein
In order to express the PR1 protein with the His tag protein, the coding region without the signal peptide sequence was amplified using primer (PR1-no signal peptide) and cloned into the pEASY-Blunt1 (CE111-01, TransGen Biotech, China) vector, which was then introduced into BL21.After incubating transformed BL21 bacteria in Luria-Bertani (LB) broth at 37 • C to an optical density of OD 600 = 0.6, a final dosage of 0.6 mM of isopropyl d-1thiogalactopyranoside (IPTG) was added to promote PR1 production. At 12 hours after induction at 19 • C, bacterial cells were extracted by centrifugation and suspended in a balanced buffer (300 mM NaCl, 30 mM NaH 2 PO 4 , and 10 mM imidazole, pH 8) and lysed with an ultrasonic cell crusher. The manufacturer's instructions were followed to purify His6-tagged proteins from bacterial extracts using Ni-NTA resin (DP101-01, TransGen Biotech, China). Using BSA as a standard, the Lowry method was used to measure protein concentration. The used primers are provided in Supplementary Data Table S1.

Antifungal activity determination in vitro
The in vitro antifungal activities of PR1 protein were investigated by disk diffusion assays with modifications described by Zandvakili et al. [51]. Recombinant PR1 protein (40 μg/ml) was applied to the solid PDA medium's surface. Boiled recombinant PR1 protein and protein-free buffer were incubated as control plates. After 4 days of incubation at 28 • C without light, the mycelial growth of each sample was assessed.

Analysis of transient expression of PR1 in grapevine leaf tissue
The PR1 coding sequence was cloned with the GFP reporter gene into the plant expression vector pCAMBIA1302 using primer PBI1302-VdPR1 and then injected into GV3101. Cultures were kept at 28 • C and 180 rpm for 12 hours in a liquid medium made of lysogenic broth. After 20 minutes of centrifuging at 7000 g, the bacteria that had settled out were resuspended in infiltration buffer (10 mM MES, pH 5.6, 10 mM MgCl 2 , 100 μM acetosyringone) until the OD 600 reached 0.6. For agro-infiltration, we used leaves from in vitro-grown 'Jingxiu' plants with similar ages and sizes. The agro-infiltration was carried out as specified by Guan et al. [52]. One day after agro-infiltration, the expression of PR1 was examined in the infiltrated leaves. After 1 day of agro-infiltration, leaves were infected with C. diplodiella, as reported by Santos-Rosa et al. [53]. The diameter of the lesions was determined 3 days after pathogen inoculation. Three independent biological repeats were performed and each repeat contained four grape leaves. The primers used are provided in Supplementary Data Table S1.