CRISPR/Cas9-mediated VvPR4b editing decreases downy mildew resistance in grapevine (Vitis vinifera L.)

Downy mildew of grapevine (Vitis vinifera L.), caused by the oomycete pathogen Plasmopara viticola, is one of the most serious concerns for grape production worldwide. It has been widely reported that the pathogenesis-related 4 (PR4) protein plays important roles in plant resistance to diseases. However, little is known about the role of PR4 in the defense of grapevine against P. viticola. In this study, we engineered loss-of-function mutations in the VvPR4b gene from the cultivar “Thompson Seedless” using the CRISPR/Cas9 system and evaluated the consequences for downy mildew resistance. Sequencing results showed that deletions were the main type of mutation introduced and that no off-target events occurred. Infection assays using leaf discs showed that, compared to wild-type plants, the VvPR4b knockout lines had increased susceptibility to P. viticola. This was accompanied by reduced accumulation of reactive oxygen species around stomata. Measurement of the relative genomic abundance of P. viticola in VvPR4b knockout lines also demonstrated that the mutants had increased susceptibility to the pathogen. Our results confirm that VvPR4b plays an active role in the defense of grapevine against downy mildew.


Introduction
Grapevine is one of the most economically important fruit crops worldwide and is widely used for producing wine, juice, and dried and fresh fruit 1 . However, its yield and quality are adversely affected by abiotic and biotic factors. Grapevine downy mildew, caused by the oomycete P. viticola (Berk. et Curt.) Berlese and De Toni, is among the most serious fungal diseases of grapevine worldwide 2 . This pathogen infects green parts such as leaves, petioles, new shoots, and young fruit 3 .
Pathogenesis-related (PR) proteins comprise a large and heterogeneous class of plant disease defense proteins.
Transgenic plants expressing PR proteins at high levels show significantly higher resistance to various pathogens. For example, in tobacco, constitutive expression of PR-1a confers increased resistance to two oomycete pathogens, namely, Peronospora tabacina and Phytophthora parasitica var. nicotianae 4 . In rice, ectopic expression of PR-5 increases resistance to rice sheath blight 5 . One of the most extensively studied PR proteins is PR4, which is a chitinase and chitin-binding protein 6 . PR4 genes have been cloned from several plants. Overexpression of β-1,3-glucanase and chitinase genes in 'Crimson Seedless' grapes conferred markedly improved resistance to downy mildew 7 . Previous studies have shown that PR4 additionally has DNase activity and inhibits fungal mycelium growth. However, although the PR4 proteins can be classified based on the presence (Class I) or absence (Class II) of an amino-terminal chitin-binding domain (ChtBD), this does not affect its fungal inhibition activity 8 . In jelly fig (Ficus awkeotsang Makino), deletion of ChtBD from the Class I FaPR-4 protein does not affect the DNase activity or antifungal function of this protein 9 . In grapevine, transgenic lines overexpressing PR4 showed increased resistance to powdery mildew 10 . Due to the inherent difficulties of transformation in grapevine and other perennial fruit crops, gene functional analyses have generally been limited to transient expression or transgene integration, and targeted gene knockout approaches have rarely been utilized. In grapevine, the most efficient approaches for Agrobacterium-mediated transformation have utilized proembryonic masses and somatic embryos (SEs) 11,12 .
The clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) system (CRISPR/Cas9 system), comprising the Cas9 endonuclease and guide RNA (gRNA), can be used to introduce heritable mutations at target sites 13 . The sgRNA comprises an~20-bp target site sequence as well as a protospacer adjacent motif (PAM), required for the introduction of double-strand breaks (DSBs) by Cas9 14,15 . DSBs can be repaired by endogenous DNA-repair mechanisms, but this process frequently results in mutations.
The CRISPR/Cas9 system has been used for genome editing in many organisms 16 . Diverse CRISPR/ Cas9 systems have exhibited high efficiency and specificity for gene editing in plants [17][18][19] . Chen et al. conveniently identified transgenic tobacco lines without using antibiotics, and these plants were confirmed to be allelic mutants 20 . In rice, the CRISPR/Cas9 system has been applied to quickly regenerate deletion mutants targeting the Tos17 retrotransposon, providing rapid breeding techniques 21 . Recently, high-efficiency targeted deletion and insertion mutagenesis in both grapevine and apple has been enabled by the direct introduction of purified CRISPR/Cas9 ribonucleoproteins into protoplasts 22 .
There have been very few reports on the successful use of the CRISPR/Cas9 system in perennial fruit crops such as grapevine. Ren et al. employed CRISPR/Cas9 in grapevine (cv. "Chardonnay") suspension cells and plants to disrupt the L-idonate dehydrogenase gene related to tartaric acid 23 . Using two sgRNAs targeting distinct sites in the gene, they obtained a 100% mutation frequency in the transgenic cell mass (CM) 23 . In another study, Wang et al. employed the CRISPR/Cas9 system in the cultivar 'Thompson Seedless' to show that the transcription factor gene VvWRKY52 is required for susceptibility to Botrytis cinerea 24 . They used gRNAs targeting four sites, and obtained 15 lines with biallelic mutations and 7 lines with monoallelic mutations. The editing events included both large (≥5 bp) and small deletions 24 . Nakajima et al. successfully mutagenized the grapevine PHYTOENE DESA-TURASE (VvPDS) gene using embryogenic calli 25 . In that experiment, loss of gene function was apparent in albino tissue 25 . They found that in the regenerated shoots, the ratio of albino cells in lower and older leaves was higher than that in new, young, higher leaves, possibly due to repeated induction of DSBs or the imprecision of DSB repair in the older leaves 25 . In another study targeting the VvPDS gene, Ren et al. surveyed the relationship between gene editing efficiency and the GC content of the sgRNA, kinds of materials used for genetic modification and expression level of SpCas9 in the transgenic CM 26 . Their research showed that sgRNA with a GC content of 65% had the highest CRISPR/Cas9 editing efficiency 26 . In addition, the expression of the SpCas9 protein correlated with the editing efficiency, but this effect was not as strong as that seen for the GC content 26 . Ren et al. successfully used CRISPR/Cas9 in grapevine to show that the Carotenoid Cleavage Dioxygenase 8 (VvCCD8) gene is crucial for normal branching 27 .
Here, we utilized the CRISPR/Cas9 system to study the role of grapevine PR4 in the resistance to downy mildew.

Results
Cloning of VvPR4b from "Thompson Seedless" The VvPR4b gene was cloned from "Thompson Seedless" using oligonucleotide primers (Table S1) designed to be specific for VvPR4b as annotated in the "Pinot Noir" reference genome (12×; http://www.genoscope.cns.fr). The VvPR4b open reading frame is 432 bp in length and encodes a 143-amino-acid product containing an Nterminal 21-amino-acid signal peptide and a conserved 125-amino-acid barley wound-induced (Barwin) domain. VvPR4b lacks an amino-terminal ChtBD and thus is a Class II PR4.

Target selection and CRISPR/Cas9 vector construction
Three potential independent target gRNAs within the Barwin domain of VvPR4b were designed using the online tool CRISPR-P (http://cbi.hzau.edu.cn/cgi-bin/ CRISPR) ( Fig. 1a and Fig. S1). The target sequences were incorporated into the reverse primer (Table S1). Then, using the pCACRISPR/Cas9 expression vector as a template, forward primers were designed to amplify the AtU6 promoter, obtaining an expression cassette for AtU6 and VvPR4b-sgRNA (Table S1). Finally, the expression cassette was cloned into the linearized vector by homologous recombination to obtain the VvPR4b knockout vector, which contains a single guide RNA (Fig. 1b).

Generation and identification of transgenic plants
Prior to transformation, proembryogenic mass (PEM) cells were cultured on X6 medium (Fig. 2a). After transformation, the PEM was transferred to delayed screening medium (Fig. 2b). Subsequently, the PEM was cultured on regeneration medium containing hygromycin for  (e-g) selection on differentiation medium (X3 screening medium); (h) cotyledons turned green and roots appeared after culture on X1.5 medium; (i, j) regenerated plantlets cultured on WPM medium; (k, l) regenerated seedlings were transplanted into soil and cultured in a controlled-environment chamber selection for~3 months, during which time a faint yellow embryogenic callus differentiated (Fig. 2c, d). Hygromycin-resistant embryos were transferred to selection medium for further induction ( Fig. 2e-g). The above processes were carried out in the dark at 26°C. The induced cotyledons and hypocotyls were subcultured on X1.5 medium under light at 26°C, during which time the cotyledons turned green (Fig. 2h). Each resistant shoot was transferred to woody plant medium (WPM) to induce seedling formation (Fig. 2i, j). Finally, seedlings were transplanted into nutrient substrate for further growth (Fig. 2k, l). In this study, the duration of the transformation cycle, from Agrobacterium inoculation to identification of mutant transgenic lines, was~15 months.
To confirm the presence of the transgene in the regenerated plantlets, we extracted genomic DNA from leaves and performed PCR using oligonucleotide primers specific for vector sequences corresponding to RTcas9 and Hygjc2 of the 14 regenerated plants initially tested; 10 were found to contain both sequences (Fig. 3a).
For further study, we focused on transgenic lines targeting a site near the carboxyl-terminal end of the Barwin domain (T3 target site; Fig. 1a). We found that all 72 samples sequenced showed mutations within an~5-base sequence at the T3 target sequence immediately upstream of the PAM site (Fig. 3b). The most frequently observed mutation (47 samples) was a dinucleotide deletion (CGCCA > CCA), resulting in a reading frame shift after Tyr127 in the encoded protein and in termination after an additional 16 amino acids (Fig. 3c, d). A complex mutation, CGCCA > TAACC, was also frequently observed (19 samples). This resulted in a nonconservative substitution of Asn-Gln for Ala-Lys129 in the encoded protein. In addition, all 19 samples containing the CGCCA > TAACC mutation additionally contained a C > A transversion 11 bases upstream of the 5′ end of the target sequence, converting Gln119 to Lys (Fig. 3b, c). Finally, we identified a small number of samples (6) containing a single nucleotide insertion of cytosine (CGCCA > CGCCCA), resulting in a reading frame switch after Ala128 in the encoded protein and termination after an additional 15 amino acids (Fig. 3c, d). The Cas9 protein encoded by the CRISPR/Cas9 vector contains a 3×-FLAG epitope tag for immunological detection (Fig. 1b). Western blotting of protein extracts from regenerated plants revealed an immunoreactive species of the size expected for the Cas9-3×-FLAG protein (163 kDa), which was absent in nontransgenic plants (Fig. 3e).  (Table 1). We amplified each sequence from 10 independent samples using PCR and specific oligonucleotide primers (Table S1) and subjected the amplicons to sequencing. The results showed that there were no off-target events in this gene editing study ( Table 1).
Knockout of VvPR4b in "Thompson Seedless" decreases resistance to P. viticola Three representative transgenic lines (#1-3) were chosen to assess the impact of loss of function of VvPR4b on the resistance to P. viticola. Lines #1 and #2 represented the dinucleotide deletion (CGCCA > CCA), whereas line #3 represented the single-nucleotide insertion (CGCCA > CGCCCA). Under optimal growth conditions and in the absence of the pathogen, there were no visible differences in phenotype between VvPR4b knockout lines and wildtype plants (Fig. 4a). We evaluated resistance to P. viticola using a detached leaf disc assay. Leaf discs were excised from sterilized leaves of transgenic or wild-type plants, inoculated with a P. viticola sporangium suspension, and maintained in petri dishes under 100% relative humidity in a controlled-environment chamber. After 72 h (72 h post-inoculation or hpi), discs from two of the transgenic lines (#1, #2) began to show abundant sporangia, while discs from transgenic line #3 and wild-type plants were relatively unaffected (data not shown). After four days (96 hpi), leaf discs from transgenic lines #1 and #2 were covered with sporangia, while the symptoms on the discs from transgenic line #3 and wild-type plants were much less severe (Fig. 4b). To quantify the relative extent of downy mildew disease progression, quantitative real-time PCR was conducted with oligonucleotide primers specific for ACTIN sequences in the grapevine (VvACTIN1) or P. viticola (PvACTIN) genomes. The relative genomic abundance of P. viticola in transgenic lines #1, #2, and #3 was 1.2, 2.5, and 1.4 times that in the wild type, respectively, which was consistent with the observed disease phenotype (Fig. 4c).

Observation of P. viticola colonization and reactive oxygen species (ROS)
To assess downy mildew disease progression within the leaf tissues, we used epifluorescence UV microscopy to examine P. viticola colonization. Detailed observations were made at 24 hpi and 72 hpi in both WT and VvPR4b knockout lines (Fig. 5). The first sign of infection was the appearance of substomatal vesicles around stomata, which was evident in both WT and VvPR4b knockout lines at 12 hpi and 24 hpi (Fig. 5a-d and Fig. S2A1-D1). In addition, germ tubes were observed in the wild type, but not in the VvPR4b knockout lines, at 24 hpi (Fig. 5a1). In addition, the mycelia in the WT cells were shorter than those in the three VvPR4b knockout lines at 24 hpi (Fig. 5a1-d1 and Fig. S2A2-D2). Finally, staining of cells with 3,3-diaminobenzidine (DAB) failed to detect H 2 O 2 in either the WT or the VvPR4b knockout lines (Fig. 5a2-d2 and Fig. S2a1-d1, a2-d2). By 48 hpi, the mycelia branched rapidly, completely covering the infected area (Fig. S2A3-D3). At 72 hpi, hyphae were apparent in the inner tissues in the wild type and transgenic line #3, but there were no obvious sporangiophores or sporangia (Fig. 5e1, h1). In contrast, we observed abundant sporangiophores and sporangia in VvPR4b knockout line #1 (Fig. 5f1). Finally, both the WT and VvPR4b knockout lines developed sporangiophores at the inoculation sites at 96 hpi ( Fig. S2A5-D5, A6-D6). DAB staining revealed H 2 O 2 accumulation in the mesophyll tissue of the WT, but staining was generally absent in the mesophyll and was restricted to the periphery of the stomata in the VvPR4b knockout lines (Fig. 5e2-g2). Specifically, in the leaves of wild-type plants, DAB staining appeared at 48 hpi around the infection sites, as the mycelia extended and sporangiophores differentiated in the mesophyll tissue with increasing infection time (Fig. S2a1-a6). In comparison, DAB staining at 48 hpi was restricted to the periphery of only a few stomata of transgenic lines #1 and #2 ( Fig. S2b1-b6, c1-c6). For transgenic line #3, there was almost no accumulation of H 2 O 2 throughout the infection process (Figs. 5h2 and S2).

Discussion
Grapevine is one of the most economically valuable horticultural crops in the world, but increasingly variable biotic and abiotic pressures pose serious threats to grape production. One of the most formidable and widespread diseases of grapevine is downy mildew. Although there have been many studies on the pathology of the causative agent, P. viticola, few studies related to functional analyses of resistance genes in grapevine have been investigated. One reason for this is the recalcitrance of grapevine to Agrobacterium-mediated transformation, and subsequent regeneration. However, recent advances in transformation technologies using embryogenic callus cells have now overcome this roadblock 28 . Previous approaches to gene targeting in plants have used transcription activator-like effector nucleases and zinc-finger nucleases. These approaches have often been successful but are complicated in target design 29 . The CRISPR/Cas9 system thus represents a tremendous advancement for studies of gene function in grapevine.
In the present study, we used sgRNA to target three independent sites in VvPR4b exons. Analyses of mutation frequency and off-target sequences showed that the Cas9/ sgRNA expression vector could effectively induce mutations at precise loci in grapevine. Interestingly, we found that the Cas9/sgRNA construct resulted in both deletions Fig. 4 Downy mildew disease progression on leaf discs from wild-type (WT) and VvPR4b knockout (#1-3) lines. a Phenotype of wild-type (WT) and VvPR4b knockout (#1-3) lines in the absence of pathogen; (b) phenotype of WT and VvPR4b knockout transgenic lines infected with P. viticola. Leaf discs were photographed at 96 hpi. Lines #1 and #2 represent the dinucleotide deletion (−2 bp), and line #3 represents the singlenucleotide insertion mutation (+1bp); (c) relative genomic abundance of P. viticola at 120 hpi. Error bars represent standard errors. An asterisks indicate statistically significant differences between WT and transgenic lines (*P < 0.05, **P < 0.01) and insertions 3 bp upstream of the PAM sequence (Fig. 3b), which was consistent with previous studies 30, 31 . Most CRISPR-induced mutations identified in previous studies in grapevine were short deletions and insertions [23][24][25]27 . However, long deletions (>5 bp) have also been observed, possibly resulting from multiple, closely spaced targets 32 . In our study, the VvPR4b knockout lines showed only two mutations that changed the length of VvPR4b, namely, a dinucleotide deletion (CGCCA > CCA) and a single C nucleotide insertion (CGCCA > CGCCCA), all of which were located 3 bp upstream of the PAM. Importantly, the base insertion mutations occurred only once in the transgenic lines, whereas the remaining mutations were all two-base deletions (Fig. 3b). We speculate that reason for the low number of mutation types may be related to there being a single target and to the different materials used for genetic transformation ("Chardonnay"/"Thompson Seedless"), but additional experiments are needed to verify this hypothesis. Previous studies have shown that editing efficiency is impacted by GC content in the sgRNA. For example, a sgRNA sequence with >50% GC resulted in higher editing efficiency as compared with a sgRNA with <35% GC 18,23 . In our study, 129 regenerated grape seedlings were identified, of which 26 lines were VvPR4b knockout lines, and the gene editing efficiency was 20.16%, with 60% GC content in the target sequence. Furthermore, no off-target mutation was detected in this study (Table 1), which could be due to the use of only a single sgRNA. This suggested that the CRISPR/Cas9 system can be effective in precisely editing genes in grapevine.
However, the mutations found in our study indicated that the allele had undergone distinct editing. For example, (1) a dinucleotide deletion (CGCCA > CCA) was present in allele I, while allele II exhibited base replacement (CGCCA > TAACC), and (2) a single C nucleotide was inserted into allele I (CGCCA > CGCCCA) while allele II exhibited base substitution (CGCCA > TAACC) (Fig. 3c). This gene knockout pattern has been widely observed in previous studies and is expressed as +n/-n (or −n/+n) 23,24 . The protein encoded by these mutations is expected to lack a complete Barwin domain and thus is likely to be dysfunctional (Fig. 3d).
Both the PEM callus and SEs have been found to be suitable tissues for transformation via cocultivation with Agrobacterium. Many previous studies have been conducted to improve both transformation efficiency and regeneration in grapevine 33 . In this study, we used PEM from the cultivar 'Thompson Seedless', developed through recurrent cycles of secondary embryogenesis (Fig. 2a). In addition, we evaluated the expression of the Cas9 protein in transgenic lines by western blotting. It has been reported that Cas9 remains bound to target DNA after cleavage 25,34 .
It is well known that PR proteins play pivotal roles in plant defense against pathogen challenge. PR4, which is a chitinase and chitin-binding protein, plays an important role in pathogen responses in many plants. In this study, we demonstrated that VvPR4b encodes a chitinase II-like protein containing a amino-terminal signal peptide and carboxyl-terminal Barwin domain (Fig. 1).
A previous study showed that the expression of PR4 was induced after inoculation in 'Riesling', suggesting that PR4 may play a role in the resistance to P. viticola 35 . In our study, at both 72 hpi and 96 hpi there was more extensive colonization on leaf discs from two VvPR4b knockout lines (#1 and #2), than on discs from the WT (Fig. 4b). We also found that at 120 hpi, the relative genomic abundance of P. viticola in all three VvPR4b knockout lines was obviously higher than that in the WT (Fig. 4c), which demonstrated that VvPR4b is required for full resistance to P. viticola.
Leaf discs stained with aniline blue 12, 24, 48, 72, 96, and 120 hpi (Fig. S2) were used to further evaluate colonization by P. viticola. The infection cycle of P. viticola includes the attachment of the zoospore to a stoma and generation of a germ tube, which enters the substomatal cavity and expands into a substomatal vesicle. Subsequently, hyphae are differentiated and spread into the intercellular space of the parenchyma 36 . At 96 hpi, we observed that the inoculation sites of the WT and VvPR4b knockout lines were full of sporangiophores ( Fig. S2A5-D5, A6-D6), which because that "Thompson Seedless" has relatively low resistance to grapevine downy mildew. Meanwhile, the growth of P. viticola was consistent with the phenotype of grape leaf discs infected with downy mildew (Fig. 4), reflecting the difference in resistance between wild-type and VvPR4b knockout lines.
The mechanism underlying the plant response to fungal infection, and how this affects the interaction between plants and pathogens, involves a very complex regulatory network in which the generation of ROS plays an important role, including in the regulation of gene expression of PR1, chitinase and β-1, 3-glucanase 37 . It has been reported that the PR4b protein has antifungal activity in pepper (Capsicum annuum), wheat (Triticum aestivum) and rice (Oryza sativa L.) 38 . In tobacco, the PR4 gene was induced by overexpression of SWPA4 (a major stress-induced gene), resulting in increased H 2 O 2 production 39 . PR1, PR4, and PR10 were significantly upregulated in OsWRKY67-overexpressing transgenic lines, with rapidly accumulation of ROS to increase the resistance to two rice pathogens (Magnaporthe oryzae and Xanthomonas oryzae pv. oryzae) 40 . Similarly, the relative transcriptional level of PR genes was upregulated in OsMAPK15 (a negative regulatory gene for disease resistance in rice) knockout mutants, while chitin promoted an increase in the accumulation of ROS and significantly enhanced the resistance of rice to the two abovementioned diseases, and the opposite results were obtained in these gene overexpression lines 41 . In our study, we used DAB staining to visualize the accumulation of H 2 O 2 inside the leaf tissues (Fig. S2). H 2 O 2 is produced in cells as a defensive response to pathogens that infect via stomata 42,43 . Previous studies have shown that both susceptible and resistant cultivars of grapevine produce ROS when infected with downy mildew 44 . Our DAB staining results suggested that P. viticola developed rapidly in the VvPR4b knockout lines compared with the WT.
Overall, our research demonstrated that the CRISPR/ Cas9 system can be a high-efficiency tool for generating targeted mutations in grapevine and will be beneficial for gene functional analyses in this plant. In addition, this work showed that grapevine VvPR4b plays an important role in downy mildew resistance.

Amplification of the VvPR4b genomic clone
The PR4 gene in "Pinot Noir" is located on chromosome 14, has a total length of 527 bp, and contains a 95-bp intron (Grape Genome Browser 12x; http://www. genoscope.cns.fr/externe/GenomeBrowser/Vitis/). This sequence was used to design oligonucleotide primers for the amplification of VvPR4b from "Thompson Seedless" (Table S1). For amplification, we employed a high-fidelity DNA polymerase (KOD-plus Neo, KOD-401(856600), TOYOBO, Japan) in a total reaction volume of 50 μL. The amplification parameters were 94°C for 3 min; 40 cycles of 98°C for 10 s, 59°C for 30 s, and 68°C for 25 s; and a final extension at 68°C for 7 min. The PCR product was cloned into the pLB-Simple vector and introduced into the E. coli strain Top10 for propagation and subsequent sequencing.

Target selection and CRISPR/Cas9 knockout vector construction
To identify potential target sites in VvPR4b, we analyzed the entire VvPR4b genomic sequence using CRISPR-P (http://cbi.hzau.edu.cn/cgi-bin/CRISPR). sgRNAs were selected on the basis of their position in the gene and the potential for off-target mutations. To generate the binary vector, the pBR322-based pCACRISPR/Cas9 vector was subjected to recombination with the VvPR4b-AtU6-sgRNA cassette.

Grapevine transformation and regeneration
The PEM was cultured for 7 days prior to transformation to improve activity on culture medium X6. The VvPR4b CRISPR/Cas9 binary vector was transformed into Agrobacterium tumefaciens strain GV3101 through the freeze-thaw method 45 . Agrobacterium was grown in LB liquid medium containing antibiotics [gentamicin (Gent), rifampicin (Rif), and kanamycin (Kan)]. Cells were collected by centrifugation and resuspended in resuspension solution (½-strength MS liquid medium containing 100 μmol/L acetosyringone). The infection procedure was carried out as described by Su 46 with minor changes. Briefly, the resuspended culture was incubated on a shaker at 28°C for 30 min, the density was adjusted to OD600 = 0.75, and the PEM callus was added. The combined Agrobacterium/PEM culture was then incubated for 10 min with shaking. Then, PEM cells were collected on sterile filter paper soaked in resuspension solution and cultured in darkness at 24°C for 48 h. Then, the PEM was washed with sterile water containing 500 mg/L cefomycin (Cef) and 500 mg/L carbenicillin (Carb), transferred to delayed screening medium (Murashige-Skoog basal medium containing 60 g/L sucrose, 3.0 g/L phytagel, 1.5 g/L activated carbon, 250 mg/L Cef, and 250 mg/L Carb; pH 5.8) and cultured in darkness at 24°C for 1 month. Next, the PEM was subcultured on X6 screening medium (Murashige-Skoog basal medium containing 60 g/L sucrose, 3.0 g/L phytagel, 1.5 g/L activated carbon, 250 mg/L Cef, 250 mg/L Carb, and 7.5 mg/L hygromycin (Hyg); pH 5.8). The nontransformed tissue gradually died and turned black. The resistant embryos were transferred to X3 screening medium (Murashige-Skoog basal medium containing 30 g/L sucrose, 3.0 g/L phytagel, 1.5 g/L activated carbon, 250 mg/L Cef, 250 mg/L Carb, and 7.5 mg/L Hyg; pH 5.8) to promote development into plantlets. After cotyledon formation, plantlets were transferred to X1.5 (Murashige-Skoog basal medium supplemented with 15 g/L sucrose, 3.0 g/L phytagel, and 1.5 g/L activated carbon; pH 5.8) and maintained under a 16-h light/8-h dark photoperiod to induce germination. Then, the long embryonic axis was transferred to WPM medium (WPM basal medium containing 30 g/L sucrose, 7.0 g/L agar, and 0.2 mg/L 6benzyladenine (6-BA)) to regenerate the plant 46 . Finally, the regenerated seedlings were transplanted into an artificial soil mixture and cultured in a controlledenvironment chamber.

Identification of transgenic lines
To identify regenerated plants containing the VvPR4b-CRISPR binary vector, young-leaf genomic DNA was extracted through a CTAB-based method and subjected to amplification using oligonucleotide primers specific for the RTcas9 and Hygjc2 sequences (Table S1). The VvPR4b target site sequence was amplified by PCR using a high-fidelity DNA polymerase (KOD-plus Neo, KOD-401 (856600), TOYOBO, Japan) using the cri-pr4b-F/R primers (Table S1) and sequenced with the specific primer cri-pr4b-cx (Table S1). In addition, the PCR products were extracted from the gel and cloned into the pMD19-T vector (code no.6013, TAKARA). For each transgenic line, 10 randomly selected clones were sequenced. Sequences were aligned by DNAMAN (version 6.0.40; Lynnon Biosoft).

Off-target analysis
Putative off-target sites were identified through CRISPR-P. Oligonucleotide primers specific for the sites (Table S1) were used for PCR amplification with highfidelity DNA polymerase (KOD-plus Neo, KOD-401 (856600), TOYOBO, Japan). Amplified sequences were cloned into the pMD19-T vector (code no.6013, TAKARA) for sequencing.
Pathogen culture and inoculation P. viticola was isolated from an infected leaf of a grapevine plant 48 in the Grape Repository of Northwest A&F University, Yangling, Shaanxi, China. The leaf was washed three times aseptically, placed with the abaxial side facing upwards on wet filter paper, and cultured overnight in darkness at 22-25°C. The following day, zoospores were collected in sterile water with a sterilized soft brush, and the suspension was agitated in the dark for 10 min and then filtered through three layers of gauze. Sporangia were counted using a hemocytometer, and the concentration was adjusted to 5 × 10 4 per mL in sterile water. For leaf disc assays, leaves were detached from plants maintained in a controlled-environment chamber, washed three times with sterile water, and allowed to dry at room temperature. Discs were excised using a 10-mmdiameter, stainless steel hole punch and placed with the abaxial surface facing upwards on three layers of sterile wet filter paper in a petri dish. Then, 20 μL of the sporangial suspension was applied, and samples were incubated for 24 h in darkness at 22 ± 2°C with 100% RH, followed by incubation at the same temperature under a 16-h light/8-h photoperiod. Experiments were carried out with three biological replicates for each line (WT or different transgenic lines), with each replicate using at least nine leaf discs 49,50 .
Statistics on the relative genomic abundance of P. viticola To determine the relative genomic abundance of P. viticola infecting leaf discs, the inoculated discs were collected. Genomic DNA was extracted using a CTABbased method, and the DNA concentration was adjusted to 50 μg/L. For PCR amplification, 4 μL of genomic DNA was used in combination with oligonucleotide primers specific for the VvACTIN1 and PvACTIN genes (Table  S1). The relative genomic abundance of P. viticola was calculated as 2 −▵▵Cq 51,52 .

DAB and aniline blue staining and analyses
For detection of H 2 O 2 using DAB, leaf discs were immersed in 1 mg/mL 3, 3-diaminobenzidine solution (DAB, Sigma, dissolved in distilled water at pH 3.8, acidified by hydrochloric acid) and then cultured under light for 8 h. The DAB staining solution was then replaced with 95% alcohol, and the samples were decolorized in saturated chloral hydrate solution (250 mg/100 mL) until translucent, washed three times with 0.05% aniline blue (dissolved in 0.067 M K 2 HPO 4 , pH 9-10) and incubated overnight in the same solution. The leaf discs were placed on a glass slide with the abaxial side facing up, and hyphal growth was observed under blue/purple light with a fluorescence microscope (Olympus bx-51) 49,53 . The accumulation of H 2 O 2 was observed under a bright field.

Data analysis
Data were analyzed by SPSS 16.0 software, and significant differences were identified using the Tukey test for one-way ANOVA.