Introduction

Grape downy mildew is a common disease in most parts of the world where grapes are grown, caused by Plasmopara viticola (Berk. et Curtis) Berl. et De Toni, which is a biotrophic oomycete with a narrow host range (Dick 2002). The pathogen is able to infect all green plant tissues. Grape downy mildew has drawn much attention because of the great economic losses it causes (Gessler et al. 2011). To protect against this pathogen, vinyards require frequent fungicide applications to produce healthy fruit, a practice that is not only costly for the grower but also contributes to environmental pollution and may negatively impact human health (Dubos 2002). Breeding resistant cultivars are thus the most economical and effective way to control downy mildew (Bisson et al. 2002). Although European Vitis vinifera cultivars are highly susceptible to P. viticola, American and wild Vitis species, e.g., V. amurensis, V. cinerea, V. riparia, V. rupestris, on the other hand, exhibit varying levels of resistance to the pathogen (Cadle-Davidson 2008) and are commonly used as parents in resistance breeding efforts and in genetic studies.

Genome sequencing of the Eurasian grape, Pinot Noir, has provided important material for discovering disease-resistant genes of grape (Jaillon et al. 2007). The research on resistance to downy mildew of grapevine mainly focused on resistance screening of grapevine species, marker-assisted selection, gene mapping, and gene cloning. Genetic linkage map of grape resistance to downy mildew was constructed by RAPD molecular marker technology (Fischer et al. 2004; Welter et al. 2007). The resistance gene analogs (RGAs) associated with downy mildew resistance were cloned (Seehalak et al. 2011). The resistance gene MrRPV1 was located on chromosome 12 of round leaf grape and European grape (Merdinoglu et al. 2003). It was reported that plants carrying the MrRPV1 gene have obvious resistance to downy mildew, and exhibit a hypersensitive response at the site of infection (Feechan et al. 2013).

Mechanisms of post-infection resistance in resistant grapevines include stomatal closure (Allègre et al. 2009), callose deposition (Yu et al. 2012), cell wall-associated defence processes (Diez-Navajas et al. 2008; Jürges et al. 2009), the expression of defence genes (Hamiduzzaman et al. 2005; Perazzolli et al. 2011), the accumulation of reactive oxygen species (ROS) (Dubreuil-Maurizi et al. 2010), increased enzymatic activity (Godard et al. 2009; Harm et al. 2011) and the activation of hypersensitive response (HR) (Bellin et al. 2009; Nascimento et al. 2019) in reaction to inoculation with P. viticola. Resistant accessions are able to activate defense responses upon pathogen infection, which culminate in localized necrosis, resulting into lower rates of sporangia release compared to susceptible accessions (Bellin et al. 2009; Polesani et al. 2010).

Gene expression profiling has been used widely to study the molecular mechanism of grapevine and P. viticola interaction. Solexa sequencing has been used for deep sequencing the transcriptomes of P. viticola infected leaves of Vitis amurensis Rupr. cv. ‘Zuoshan-1’, and the differentially expressed genes (DEGs) associated with ribosome structure, photosynthesis, amino acid and sugar metabolism were also identified (Wu et al. 2010). Early transcriptional changes associated with P. viticola infection in susceptible V. vinifera and resistant V. riparia plants were analyzed using the combimatrix microarray platform; measurements of jasmonic acid and methyl jasmonate in infected leaves suggested that this hormone may also be involved in V. riparia resistance to P. viticola (Polesani et al. 2010). Transcriptional changes associated with early stages of P. viticola infection indicate the presence of a weak defence response in susceptible grapevines (Polesani et al. 2010).

The transcriptome sequencing technology and bioinformatics analysis packages have greatly facilitated studies on the plant and pathogens (Han et al. 2015; Yuan et al. 2018). Transcriptomic analysis mainly highlights defense-related genes, providing clues for subsequent molecular-level studies. In this study, we used Illumina RNA-Seq analysis to characterize the transcriptional dynamics associated with the highly susceptible cultivar, ‘Centennial Seedless’, and the highly resistant cultivar, ‘Beta’, during early response to P. viticola inoculation. The changes of gene expression in resistant and susceptible cultivars were analyzed and multiple disease-resistant candidate genes were screened. This study provides insight into the molecular mechanism of resistance to downy mildew of grape, which contributes to our understanding of the grape downy mildew pathosystem.

Materials and methods

Plant materials and pathogen infection

‘Beta’ (V. labrusca×V. riparia) is a grape cultivar that is highly resistant grape to downy mildew, while ‘Centennial Seedless’ (Vitis vinifera) is a grape cultivar that is highly susceptible to downy mildew. The grapevine cultivars, ‘Beta’ and ‘Centennial Seedless’ were grown in 15 cm pots filled with a mixture of 60% vermiculite and 40% meadow soil and cultured in growth chambers under a light regime of 16 h light/8 h dark cycles at 25–26 °C. In order to inoculate plants, P. viticola sporangia were collected from ‘Centennial Seedless’ at the experimental research station of the Liaoning Academy of Agricultural Sciences (China). A detached leaf with ‘oil spots’ (a diagnostic symptom of downy mildew) was thoroughly washed under tap water to remove the sporangia and then incubated in a humid chamber at room temperature overnight to induce sporulation; the following day, freshly formed sporangia were suspended in sterile distilled water by gently washing the leaves (Liang et al. 2015). The sporangia suspension was filtered with three-layers of sterilization gauzes. Then, the concentration was adjusted to 1 × 106 sporangia ml−1 by using a hemocytometer under a light microscope to count sporangia. Inoculation with sporangia was performed on the abaxial leaf surface of the third to fourth fully expanded leaves from the top. Zoospore infection initiation could occur within two hours of wetting at 25 °C and the incidence of infection reached 100% when the dorsal side of the leaves were wetted for 6–7 h (Sha et al. 2011). Genes showed greatly fold change in susceptible and resistant Vitis vinifera cultivars at 6 hpi after P. viticola inoculation (Monteiro et al. 2013).Therefore, the leaves from each grape cultivar were collected at 6 h after inoculation. Leaves treated with water under the same conditions served as a control. Each treatment was obtained from a pool of nine independent leaves, with three replications for RNA extraction. The collected samples were then frozen immediately in liquid nitrogen and stored at −80 °C until further use.

RNA extraction and quality determination

Total RNA was isolated from the collected leaves using a modified SDS /Phenol method (Zhang et al. 2003). The RNA samples were treated with DNase I (TaKaRa, Japan) for 4 h to remove any genomic DNA contamination. RNA purity and concentration were measured at 260/280 nm using a spectrophotometer (NanoDrop-1000, Thermo Scientific) and RNA integrity was verified by agarose gel electrophoresis. The integrity of the RNA samples was examined with an Agilent 2100 Bio analyzer.

cDNA library preparation and Illumina sequencing

Equal quantities of high-quality RNA from three replications were combined into a single large pool for cDNA synthesis. Sequencing libraries were generated using the TruSeq RNA Sample Preparation Kit (Illumina, San Diego, CA, USA). The mRNA was purified from total RNA using poly-T oligo-attached magnetic beads. Fragmentation was carried out using divalent cations under elevated temperature in a proprietary Illumina fragmentation buffer. First strand cDNA was synthesized using random oligonucleotides and SuperScript II. Second strand cDNA synthesis was performed using DNA Polymerase I and RNase H. Remaining overhangs were converted into blunt ends via exonuclease/polymerase activities and the enzymes were removed. After adenylation of the 3′ ends of the DNA fragments, Illumina PE adapter oligonucleotides were ligated to prepare for hybridization. To select cDNA fragments of at least 200 bp in length, the library fragments were purified using the AMPure XP system (Beckman Coulter, Beverly, CA, USA). DNA fragments with ligated adaptor molecules on both ends were selectively enriched using an Illumina PCR Primer Cocktail in a 15 cycle PCR reaction. Products were purified (AMPure XP system) and quantified using the Agilent high sensitivity DNA assay on a Bioanalyzer 2100 system (Agilent). The sequencing library was then sequenced on an Illumina Hiseq 2000 platform, and 126 bp paired-end reads were generated by Shanghai Personal Biotechnology Cp. Ltd. The raw sequencing data of four samples in this test have been submitted to NCBI (https://www.ncbi.nlm.nih.gov/sra/PRJNA560305).

Read mapping to the reference genome

To obtain high-quality clean data for comparing with the reference genome, the raw reads were filtered by removing the adapter sequences, low quality sequences (reads with ambiguous bases `N′), and reads in which more than 20% of bases had a Q-value <30. Quality checks were performed with fastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The reads were mapped to the grape (V. vinifera cv. Pinot Noir) reference genome (Vitis_vinifera.IGGP_12x.24.dna.toplevel.fa; ftp://ftp.ensemblgenomes.org/pub/release-24/plants/fasta/vitis_vinifera/dna) using bowtie2/tophat2 with default parameters (Radakovits et al. 2012).

Functional annotation

We annotated unigenes based on a set of sequential BLAST searches (Altschul et al. 1997) designed to find the most descriptive annotation for each sequence. The unigenes were compared with sequences in the Nonsupervised Orthologous Groups (eggNOG) database (http://eggnog.embl.de/version_3.0/) (Powell et al. 2011), and the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway database (Minoru et al. 2004). The Blast2GO software (Conesa et al. 2005) was used to obtain GO annotation of the unigenes. The software was then used to perform GO functional classification of unigenes to view the distribution of gene functions (Ashburner et al. 2000). Unigenes were also queried against the PlnTFDB (http://plntfdb.bio.uni-potsdam.de/) database using BLASTx to identify putative transcription factors.

Analysis of differentially expressed genes (DEGs)

Gene expression levels were measured in RNA-Seq analysis as the number of reads and were normalized with RPKM (Reads per kilobase per million mapped reads) (Mortazavi et al. 2008). The RPKM measure of read density reflects the molar concentration of a transcript for RNA length and for the total read number in the measurement. RPKM was used to quantify gene expression with Bowtie 2 (2.2.4, default setting) (Langmead and Salzberg 2012) and the calculation method of RPKM is described below (Ammar et al. 2012).

$$ \mathrm{RPKM}={\displaystyle \begin{array}{c}\mathrm{Total}\ \mathrm{exon}\ \mathrm{reads}\div \\ {}\mathrm{Mapped}\ \mathrm{reads}\ \left(\mathrm{millions}\right)\times \mathrm{Exon}\ \mathrm{length}\ \left(\mathrm{KB}\right)\end{array}} $$

HTSeq v0.6.1 software was used to analyze the gene expression level in each sample, and then we used the DESeq R software package (1.18.0) to analyze DEGs of the four libraries (Wang et al. 2009). RPKM values were normalized, and the differentially expressed genes (DEGs) were considered to be significant if a false discovery rate (FDR) ≤ 0.01 and the absolute of log2 fold change (R) ≥ 1.

Screening of candidate resistance genes

The encoded proteins of the candidate resistant genes were searched for pfam domains using HMMER 3.0 program (Eddy 2011). A protein sequence similarity search was done using BLAST (ref) against proteins in the Plant Resistance Gene Database. Cluster 3.0 was used to perform Cluster analysis between genes and samples. Clustering results were plotted with TreeView.

RT-qPCR

To confirm the results of transcriptome analysis, we determined the expression levels of 18 DEGs by RT-qPCR, which were up or down regulated in the two cultivars and largerly changed in differentially expressed multiples. Total RNAs from each sample were extracted using SDS/Phenol and treated with DNase I to remove any genomic DNA contamination. The concentration of each RNA sample was adjusted to 1 mg/ml with nuclease-free water and total RNA was reverse transcribed in a 20 μl reaction system using the TaKaRa Super RT Kit. Reactions were carried out with Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen) and specific primers using the Light Cycler 480 (Roche Diagnostics, Germany). PCR condition was: 95 °C for 4 min as initial steps, followed by 35 cycles of 95 °C for 15 s, 57 °C for 15 s and 72 °C for 25 s. Each sample was examined in three technical replicates, and dissociation curves were analyzed to verify the specificity of each amplification reaction. The relative expression levels were normalized to the EF1a gene (Monteiro et al. 2013) (Supplementary Table 1) and calculated using the 2-ΔΔCT method (Livak and Schmittgen 2001).

Results

Illumina sequencing and alignment with the reference genome

We sequenced four cDNA libraries, R0 (‘Beta’ water control), R1 (‘Beta’ inoculated treatment), S0 (‘Centennial Seedless’ water control) and S1 (‘Centennial Seedless’ inoculated treatment). In this study, 84,831,516 clean reads were obtained, each of which were 126 bp in length, comprising 20.74 Gb of sequence data. The number of clean reads of four samples accounted for more than 90% of the total reads. GC content of sequence data from the four libraries was all approximately 50%, and Cycle Q20% was above 94%, which suggested that the accuracy and quality of the sequencing data were sufficient for further biological information analysis (Table 1).

Table 1 Summary of ‘Centennial seedless’ and ‘Beta’ RNA-seq reads against the ‘Pinot Noir’ genomic sequence. S0 = ‘Centennial seedless’ CK, S1 = ‘Centennial seedless’ inoculated treatment, R0 = ‘Beta’ CK, R1 = ‘Beta’ inoculated treatment
Full size table

Differentially expressed genes (DEGs) among the four grape samples

RPKM were used to estimate gene expression levels in the four libraries. There were 959 differentially expressed genes in R0 vs S0, of which 547 genes were up-regulated and 412 genes were down-regulated. These results show that there are wide discrepancies in gene expression in leaf of the two cultivars. After inoculation with P. viticola, gene expression of the two grapevine cultivars changed significantly. 1091 differentially expressed genes were detected in R1 vs R0, of which expression levels of 222 genes were up-regulated; 869 genes were down-regulated. In S1 vs S0, 849 differentially expressed genes were detected, of which 250 genes were up-regulated and 599 genes down-regulated. There were 961 differentially expressed genes between R1 and S1 (Supplementary Table 2). In general most of the DEGs were down regulated in the resistant and susceptible cultivar following P. viticola inoculation. In this process, the two cultivars shared 313 (227 + 39 + 19 + 28) DEGs in response to downy mildew according to the Venn diagram (Fig. 1). In addition, the number of DEGs belonging to ‘Beta’-specific genes in R0 vs R1 and S0 vs S1 was 778 (480 + 184 + 24 + 90). There were 545 (365 + 62 + 28 + 90) DEGs belonging to specific genes after treatment between S0 vs R0 and S1 vs R1 (Fig. 1). These genes were predicted to be important in interactions between grape and P. viticola. The data illustrate the difference between the resistant and susceptible cultivars, and deep analysis of these genes may shed light on resistance mechanisms.

Fig. 1
figure 1

Venn diagrams of gene expression in four samples. S0 = ‘Centennial Seedless’ CK, S1 = ‘Centennial Seedless’ inoculated treatment, R0 = ‘Beta’ CK, R1 = ‘Beta’ inoculated treatment

Full size image

eggNOG functional category analysis

All unigenes were aligned to the eggNOG database for further functional prediction and classification. 491 genes were categorized into 23 functional groups with an eggNOG classification. Among these eggNOG categories, the cluster of ‘general function prediction’ had the highest number, 62 (12.63%); followed by ‘posttranslational modification, protein turnover, chaperones’, 61 (12.42%); ‘translation, ribosomal structure and biogenesis’, 50 (10.18%) and ‘signal transduction mechanisms’, 48 (9.78%) (Fig. 2). Analysis of these genes will contribute to our understanding of the underlying mechanism of resistance to downy mildew.

Fig. 2
figure 2

eggNOG Function Classification of Consensus Sequence in all unigenes

Full size image

GO annotation analysis

Gene Ontology (GO) analysis was used for functional classification of the assembled transcripts. Under the GO terms, the DEGs were found to be involved in biological process, cellular component, and molecular function. In the R1 vs R0 group, 4074 DEGs were assigned to one or more GO terms, with 52.2% assigned to biological processes, 31.9% to cellular components, and 15.9% to molecular functions. GO enrichment analysis showed that for biological processes the high number of DEGs were associated with metabolic processes, cellular processes, and transport. For cellular components the three top classifications were cell, intracellular and membrane. For molecular functions, binding was the highest. In the S1 vs S0 group, 3099 DEGs were assigned to one or more GO terms, with 49.6% assigned to biological processes, 35.6% to cellular components, and 14.8% to molecular functions. GO enrichment analysis showed that for biological processes, the DEGs were primarily associated with metabolic processes, transport and response to abiotic stimulus. For cellular components, the three top classifications were membrane, plastid and thylakoid. GO enrichment analysis showed no significant enrichment associated with molecular functions. Interestingly, the number of DEGs associated with metabolic processes was particularly high in the two comparisons (Fig. 3).

Fig. 3
figure 3

Gene Ontology (GO) analysis of differentially expression genes in two grapevine species. GO analysis was summarized into three categories: biological process, cellular component, and molecular function. Go terms with FDR < 1 in S0 vs S1 was used as a condition for screening

Full size image

Pathway annotation analysis

KEGG is a public database for networks of molecular interactions in cells and their variants specific to particular organisms. To further examine the usefulness of the grape genes generated in the present study, the DEGs were compared with the KEGG database using BLASTX and the corresponding pathways were established. Out of 547 (5.79%) DEGs, 175 were assigned pathways in the R1 vs R0 group and 413 (4.37%) were assigned to 135 pathways in the S1 vs S0 group. Six significantly enriched pathways associated with the DEGs were identified, which were also associated with downy mildew-related gene expression (Table 2), plant-pathogen interactions, phenylalanine metabolism, phenylpropanoid biosynthesis, plant hormone signal transduction, flavonoid biosynthesis, and terpenoid backbone biosynthesis. In plant hormone signal transduction pathways, genes in the R1 vs R0 group were associated with the SA, JA, ABA, BRs, CTK and auxins response, and exhibited differential expression. Biosynthesis of other secondary metabolites (phenylpropanoid biosynthesis and flavonoid biosynthesis) was stimulated in response to downy mildew in two groups. In the plant–pathogen interaction pathway, genes in MAPK signal transduction pathway mediating resistance to pathogens in the anti-infective cultivar plants, including five mitogen-activated protein kinases (MAPK) genes, two WRKY29 genes, and three PR-1 genes, showed differential expression. This indicated that the MAPK signal transduction pathway was involved in the response of grape to downy mildew. After P. viticola infection, the expression of a HSP90 protein-related gene was up-regulated 3.4-fold in the strongly resistant cultivar ‘Beta’. RPM1 interacting protein 4 (RIN4), a negative regulator of plant disease resistance, was found in the highly susceptible cultivar, ‘Centennial Seedless’, and its expression was down-regulated upon P. viticola infection, thereby activating the up-regulated expression of the disease resistance gene RPM1.

Table 2 KEGG enrichment analysis of metabolic pathways
Full size table

Transcription factor analysis

Analysis of transcription factors of resistant and susceptible grapevine cultivars induced by P. viticola resulted in 201 transcription factors (66 up-regulated and 135 down-regulated) in two grapevine cultivars being identified. The expression of transcription factors showed strong cultivar-specificity in two grapevine cultivars inoculation with P. viticola. There were 29 overlapping transcription factors, 82 transcription factors were unique to ‘Beta’ and 61 transcription factors were unique to ‘Centennial Seedless’ after inoculation. The largest transcription gene family present in the DEGs was the AP2-EREBP family, followed by the MADS family, the MYB-related family, the Orphans family, and the bHLH family (Fig. 4). Transcription factors ERF, MYB, WRKY, and bHLH associated with disease resistance were screened, which indicated that grape leaves could induce pathogenesis related proteins and signal pathways by regulating transcription factors to improve disease resistance against downy mildew.

Fig. 4
figure 4

Distribution of up-and down regulated transcription factors in gene families in two grapevine species. The distribution of transcription factor gene-family members in two grapevine species

Full size image

Resistance candidate genes screening

The differentially expressed genes in resistant and susceptible cultivars after exposure to downy mildew were compared, to identify candidate resistant genes which were defined as genes which responded to pathogen infection in the resistant strain but not in the susceptible strain. The results showed that a total of 196 genes were identified as candidate resistance genes (Fig. 5). Among them, 174 genes were down-regulated in response to pathogen infection and the remaining 22 genes were up-regulated (Supplementary Table 3). There were no functional annotations for the protein products of 172 of these genes. Among the candidate resistant genes, five encoded proteins with the leucine rich repeat (LRR_8) domain that is well characterized as a domain for plant resistant genes (Supplementary Table 2). Three of these 5 proteins also had the leucine rich repeat N-terminal domain (LRRNT_2). These genes included VIT_13s0064g01260, VIT_08s0007g02110, and VIT_08s0058g00540. Other highly represented domains were UDP-glucoronosyl, UDP-glucosyl transferase (UDPGT, 7 proteins), the thaumatin family (Thaumatin, 7 proteins), the core histone H2A/H2B/H3/H4 (Histone, 6 proteins), the protein kinase domain (Pkinase, 6 proteins), and protein tyrosine kinase (Pkinase_Tyr, 6 proteins) (Supplementary Table 3).

Fig. 5
figure 5

Expression patterns of some important candidate resistant genes related to downy mildew resistance. In this figure, from left to right, the four columns show the expression of S0 (‘Centennial Seedless’ CK), R0 (‘Beta’ CK), S1 (‘Centennial Seedless’ inoculated treatment) and R1 (‘Beta’ inoculated treatment), respectively. Each row represents a different gene. Blue, black and yellow indicate low, medium and high expression levels of genes respectively. Genes were selected with the absolute of log2 fold change (R) ≥ 1and FDR < 0.01

Full size image

Comparison of DGE tag data with qRT-PCR expression patterns

To verify the reliability of the RNA-seq results, a total of 18 genes and their expression patterns in leaves following inoculation with P. viticola were examined by qRT-PCR (Supplementary Table 1). The results indicated that the relative expression trend of the 18 genes in qRT-PCR was consistent with that of the sequencing data, although there were differences in the relative expression among multiple genes for the two methods. In general, the relative expression of the 18 genes in qRT-PCR were the same as that of the sequencing data, indicating that the information obtained by transcriptome sequencing was reliable. For example, both qRT-PCR and RNA-seq analyses indicated that genes encoding LRR receptor-like serine(VIT_10s0003g02910), abscisic acid receptor PYL4 (VIT_08s0058g00470), WRKY transcription factor (VIT_01s0010g03930), 14 kDa proline-rich protein (VIT_02s0154g00320), LRR receptor-like serine (VIT_06s0004g06430), ethylene-responsive transcription factor (VIT_16s0013g01000), cytochrome P450 (VIT_17s0000g05110), beta-amylase (VIT_02s0012g00170), xyloglucan endotransglucosylase/hydrolase (VIT_02s0012g02220), cellulose synthase (VIT_02s0025g01750), and receptor kinase-like (VIT_17s0000g04400) were more highly expressed in the two inoculated grape samples (R1 and S1) than in the the two non-inoculated samples (R0 and S0) (Fig. 6). Likewise, expression patterns of two genes encoding receptor-like protein kinas (VIT_06s0004g02880), G-type lectin S-receptor-like serine (VIT_00s0286g00130), probable inactive receptor kinase (VIT_17s0000g02010), MLP-like protein (VIT_01s0026g00570), metallothionein-like protein (VIT_06s0004g04640), putative cell wall protein (VIT_12s0059g01830), and GDSL esterase/lipaseGDSL (VIT_10s0003g00620) were more highly down-regulated in P. viticola-infected leaves of two grape samples (R1 and S1) than the two non-infected samples (R0 and S0) (Fig. 6).

Fig. 6
figure 6figure 6

Q-PCR validation of up / down DEGs of grape leaves infected with P. viticola. EF1α was used as an internal control. The y-axis represents Relative fold express compared to R0 and S0. Date represent means ± SD of three replicates

Full size image

Discussion

Downy mildew caused by P. viticola is a devastating disease for grapevine, involved in many biological processes and metabolic pathways changes. In the current study, RNA-seq technology was applied to the transcriptome of different cultivars of grape leaves infected with P. viticola using Illumina HiSeq™ 2500 platform. We have provided important, comprehensive, and rich transcription information from RNA-Seq high-throughput sequencing technology. In this experiment, the measured samples were compared with the known grape genome. The ratio of reads to total reads in the grape genome was approximately 60%, with 40% of the unigenes not matching any known gene, which could be due to differences in grapevine species, or due to incomplete grape genome annotation information. In our transcriptome sequencing analysis, we found that most genes and pathways were similar in resistant and susceptible grapevine cultivars inoculated with P. viticola. Resistance in V. riparia was induced after infection, and was not based on differences in basal gene expression between the susceptible and resistant grapevine species (Polesani et al. 2010). In general, we found more DEGs in metabolic pathways in the resistant cultivar than in the susceptible cultivar. The functional mining of differential genes in metabolic pathways will lay the foundation for mechanistic studies of stress responses to downy mildew of grape.

Secondary metabolites like viniferins, lignin, anthocyanins, flavonoid and phenolic compounds may play important roles in the defense response of grapevine to downy mildew (Doster and Schnathorst 1985; Malacarne et al. 2011; Wang et al. 2018; Fröbel et al. 2019). Resistant grapevines react to P. viticola inoculation by rapidly up-regulating genes involved in metabolic processes, including the phenylpropanoid pathway (Kortekamp 2006; Polesani et al. 2010; Wu et al. 2010). In our study, the genes in the metabolic pathway of flavonoids biosynthesis and phenylalanine metabolism were more highly expressed. Our results suggested that secondary metabolites protected plants against P. viticola infection, specifically by flavonoids and phenylpropanoid biosynthesis. In the metabolic pathway of flavonoids biosynthesis, chalcone synthase (CHS), flavonol synthase (FLS), shikimate O-hydroxycinnamoyltransferase (HCT) and leucoanthocyanidin reductase (LAR) genes were up-regulated in the resistant cultivar but were not affected in the susceptible cultivar. Likewise, in the metabolic pathway of phenylalanine metabolism, Shikimate O-hydroxycinnamoyl transferase (HCT), phenylalanine ammonialyase (PAL) and 4-coumarate CoA ligase (4CL) genes were up-regulated in the resistant cultivar but were down-regulated or unchanged in the susceptible cultivar. It has been suggested that the resistant cultivar may enhance resistance to downy mildew by synthesis of lignin (which is used to reinforce the cell wall) or by producing polyphenols (flavones, flavonols, isoflavones and anthocyanins) with antibacterial activity.

Constitutive resistance is also suggested by the high levels of expression of some genes related to stress and defence in resistant grapevines, as compared to the lower expression levels observed in susceptible uninoculated grapevines (Kortekamp and Zyprian 2003; Kortekamp 2006; Fung et al. 2007; Figueiredo et al. 2008). In our study, there were more differentially expressed genes (DEGs) in the resistant cultivar than the susceptible cultivar in the plant hormone signal transduction pathway (including significant enrichment). Recently studies have showed jasmonates have been implicated in resistance against downy mildews in grapevine (Polesani et al. 2010) and in resistance induced by BABA and by β-1,3-glucan sulfate against P. viticola (Hamiduzzaman et al. 2005; Trouvelot et al. 2008). This study showed that many key genes in the JA biosynthesis pathway and JA signal transduction pathway were differentially expressed as a result of P.viticola infection, including two allene oxide synthase (AOS) genes, two 12-oxophytodienoate reductase (OPR) genes and three SAM-dependent methyltransferase genes. Resistant grapevines react to P. viticola inoculation by rapidly up-regulating genes coding for pathogenesis-related (PR) proteins (Busam et al. 1997; Kortekamp 2006) and genes involved in defence-related signal transduction. Our data supported the downstream regulatory gene PR1 in ‘Centennial Seedless’ cultivar was down-regulated 2.4-fold with P. viticola infection and was down-regulated 3.65-fold in the ‘Beta’ cultivar. These results indicate that P. viticola can stimulate the JA and SA signal pathways in response to downy mildew. Multiple studies have shown that ABA plays a significant role in plant-pathogen affinity interaction (Stec et al. 2016). The current study has shown that the key genes of the ABA signal transduction pathway were induced by P. viticola. For example, two genes encoding the ABA receptor protein PYR/PYL were up-regulated (VIT_13s0067g01940 and VIT_08s0058g00470), and three genes encoding PP2C protein phosphatase were up-regulated which in turn negatively regulated the ABA signal pathway. However, defense responses were not activated in grapevines without pathogens, major accumulation of defence gene products were observed in elicited plants after they were inoculated with P. viticola as compared to uninoculated plants.

Transcriptional regulation is a crucial step in plant responses to pathogen infection. Transcription factors play a vital role in this regulation process. Our results showed that the number of coding genes of AP2-EREBP, MADS, WRKY, MYB, MYB-related, and Orphans accounted for the primary proportion of expression levels which were significantly different in each control group. These results suggested that these types of transcription factors were important in response to downy mildew. Studies have shown that overexpression of the VvWRKY1 gene of ‘Cabernet Sauvignon’ grapes in tobacco enhanced resistance to downy mildew and powdery mildew (Marchive et al. 2007). WRKY transcription factors are expressed at higher levels in resistant grapevine specie than susceptible specie (Polesani et al. 2010). An SSH library of powdery mildew-infected grapes was constructed and the differential expression of the transcription factor MYB was investigated, indicating MYB could be involved in the defense process of grapevine against powdery mildew (Fekete et al. 2009). In our study, we confirmed that the transcription factor WRKY and MYB were expressed at higher levels in resistant grapevine cultivar than susceptible cultivar following inoculation with P. viticola and were the factor driving different responses to downy mildew on two grapevine cultivars.

In this study, RNA-Seq technology was used to construct a transcriptome library of grapevine leaves with P. viticola inoculation. The differential expression of genes in resistant and susceptible grapevine cultivars to downy mildew were analyzed at the transcriptome level to identify candidate genes involved in downy mildew resistance. We plan to investigate the roles of these candidate genes in grapevine biotic stress response and analyze the gene expression regulation network that may have disease resistance candidate genes involved. The specific roles of candidate genes in plant disease resistance and the molecular mechanisms involved in disease resistance can be further clarified through their stable expression profiles in plants.