Comparative transcriptome analysis between a resistant and a susceptible Chinese cabbage in response to Hyaloperonospora brassicae

ABSTRACT

Downy mildew caused by Hyaloperonosporabrassicae (H. brassicae) leads to up to 90% of the crop yield loss in Chinese cabbage in China. A transcriptome analysis was carried out between a resistant line (13-13, R) and a susceptible line (15-14, S) of Chinese cabbage in response to H. brassicae. The NOISeq method was used to find differentially expressed genes (DEGs) between these two groups and GO and KEGG were carried out to find R genes related to downy mildew response of Chinese cabbage. qRT-PCR was carried out to verify the reliability of RNA-seq expression data. A total of 3,055 DEGs were screened out from 41,020 genes and clustered into 6 groups with distinct expression patterns. A total of 87 candidate DEGs were identified by functional annotation based on GO and KEGG analysis. These candidate genes are involved in plant-pathogen interaction pathway, among which 54 and 33 DEGs were categorized into plant-pathogen interaction proteins and transcription factors, respectively. Proteins encoded by these genes have been reported to play an important role in the pattern-triggered immunity (PTI) and effector-triggered immunity (ETI) processes of disease responses in some model plants, such as Arabidopsis, rice, tobacco, and tomato. However, little is known about the mechanisms of these genes in resistance to downy mildew in Chinese cabbage. Our findings are useful for further characterization of these candidate genes and helpful in breeding resistant strains.

Introduction

Chinese cabbage (Brassica rapa ssp. pekinensis), a member of the Cruciferae family, is one of the most important leafy head vegetables cultivated worldwide especially in China, Korea, and Japan. Because of the abundant vital mineral nutrients, crude fiber, and vitamins, B. rapa has become an integral part of healthy diets. However, B. rapais extremely sensitive to numerous microbes, insects, and herbivores, especially during the reproductive stage, resulting in serious yield loss.

Downy mildew caused by H.brassicae is an oomycete disease specific to the Brassica species, which tends topropagate and spread rapidly in warm and high humid climates, especially in spring and fall. In China, it is one of the devastating diseases to Chinesecabbageleading to up to 90% of the crop yield loss¹. Chinese cabbage can be damaged during all growth periods by H.brassicae. Frequent or heavy application of fungicide is necessary to control the development and spread of the disease. However, fungicide is also a great threat to food safety due to the unacceptable high pesticide residue. Therefore, identification of resistant genes (R genes) and breeding disease-resistant varieties are vital for agriculture production.

Plants have evolved an innate immune system to defend the vast majority of potential pathogens. In the classical “zigzag model,” there are two modes of plant resistance mechanisms, namely PTI and ETI leading to the development of systemic acquired resistance (SAR). According to these two mechanisms, immune receptors are classified into two categories,^2,3based on which plants can recognize molecules derived from pathogens. The molecules known as microbe-associated patterns (MAMPs) or pathogen-associated molecular patterns (PAMPs) are recognized by the first class of immune receptors – cell-surface associated pattern recognition receptors (PRRs) mainly containing receptor-like kinases (RLKs) and receptor-like proteins (RLPs). The perception of MAMPs and PAMPs by PRRs triggers the PTI signaling pathway leading to transcriptional reprograming and biosynthesis of specific defense molecules.^4,5 PTI, as the first barrier, enables plants to respond rapidly and fight efficiently against the invasion of a large range of pathogens.⁶However, pathogens that cause disease in plants may acquire special abilities – injecting pathogenic effectors into plant cells – to escape the recognition of receptors and even suppress PTI.⁷Oomycete effectors target different sites in host plant cells.⁸ Some effectors act outside of plant cells interfering with apoplastic plant proteins involved in pathogen defense, such as inhibitors of plant hydrolases,⁹ arginine-glycine-aspartic acid-containing proteins¹⁰ and plant toxins that enhance microbial virulence leading to necrosis of host cells.¹¹ Another class of Oomycete effectors includes members of the RXLR (Arg-Xaa-Leu-Arg) and Crinkler (CRN) families acting inside host plant cells.⁸ They are recognized by the resistance (R) proteins,¹² which are intracellular immune receptors. As a result, the ETI¹³is activated to prevent pathogen invasion. Recent studies have revealed that many RxLR effectors ofHyaloperonosporaarabidopsis(H.arabidopsis), leading to downy mildew of Arabidopsis specifically, are capable of suppressing plant immunity. For example, HaRxL96 as an evolutionarily conserved effector in different oomycete species suppresses immunity in different plant species such as soybean and nicotiana;¹⁴however, HaRxL23, which is expressed early during infection, triggers an ecotype-specific defense response in Arabidopsis.¹⁵ The perception of signaling molecules induced by ETI activates MAP kinase cascades in a similar way as PTI and leads to hypersensitive responses (HR), including rapid cell death, synthesis of reactive oxygen species (ROS) and salicylic acid (SA), and expression of defense-related genes.¹⁶ At last, plants defend themselves by activating the broad-spectrum antimicrobial and long-lasting systemic acquired resistance (SAR) system,^17,18 which contains related hormone signaling pathways such as jasmonates (JA), salicylates (SA), ethylene (ET), abscisic acid (ABA), and brassinosteroids (BR).¹⁹

Based on the B. rapa reference genome sequence,²⁰ molecular marker technologies have been rapidly developed, such as random amplified polymorphic DNA (RAPDs),²¹ restriction fragment length polymorphisms (RFLPs),²² amplified fragment length polymorphisms (AFLPs),²³ simple sequence repeats (SSRs),²⁴ sequence-tagged sites (STS),²⁵ and specific-locus amplified fragment sequencing (SLAF-seq),²⁶ which have greatly promoted the identification of the R genes. Genetic analysis of downy mildew resistance in Chinese cabbage has shown that the resistance to downy mildew is determined by a single dominant gene.²⁷ Recently, it has been reported that the genetic resistance mechanisms are different in different growth stages of Chinese cabbage.²⁸ For example, at the heading stage of Chinese cabbage, a single dominant locus BrRHP1 located on the A01 chromosome is active,²⁹ while four major QTLs mapped to A08 are active at every developmental stage and two minor QTLs located on linkage groups A06 and A04 are effective on both the rosette stage and the heading stage.²⁸However, the identification of R genes related to downy mildew in B.rapais very limited. Only two R genes,BrRHP1²⁹ and BrRLP48³⁰have been identified to contribute to downy mildew resistance.

In recent field investigation, we had discovered two lines of Chinese cabbage named 13–13 (the resistant line, R) and 15–14 (the susceptible line, S) which showed more significant phenotypic differences in response to H.brassicae compared with our previous study.³¹ Considering the species of Chinese cabbage are enormous and multiple, to further understand and explain the mechanism of the response to H.brassicae in Chinese cabbage, eight samples of these two lines of Brassicarapawere sequenced using the RNA-Seq technology. Finally, 87 candidate DEGs were identified, which are involved in the plant-pathogen interaction pathway. Among these DEGs54 and 33 were categorized into plant-pathogen interaction proteins and transcription factors, respectively. These findings are useful for further characterization of these candidate genes and helpful in breeding resistant strains.

Results

Symptoms of these two genotypes

Leaves from twenty-day-old plants were inoculated with the H. brassicae zoospore suspension, the same strain used in our previous article³¹ and harvested after 7 days. Noticeably, there were more HR lesions on the surface of the leaves of the S genotype (Figure 1, S-T, T means treatment group with inoculation) than that of the R genotype (Figure 1, R-T, T means treatment group with inoculation). The phenotypes of these two species indicated that they were reliable to be used for the transcriptome analysis.

Figure 1.

Symptoms on the leaves of the R (13–13) and the S (15–14) genotype 7 days after inoculation with H. brassicae. Cell death in leaves was examined by trypan blue staining.

We collected two biological replicates of each H.brassicae treated R and Sgeno type sample and the control samples at 168 h (7 days) after inoculation for RNA-seq analysis.

Transcriptome analysis in response to H.brassicae

RNA was extracted using two biological replicates of leaves from R-CK, T,S-CK, R-T and S-T (CK means control group without inoculation, T means treatment group with inoculation), and eight libraries were generated. Eight samples of Brassicarapawere sequenced using the RNA-Seq technology,^32,33 by which 12,935,779 raw sequencing reads were generated. A total of 12,855,513 clean reads were generated after filtering adaptor sequences and low quality reads as well as reads containing unknown nucleotides larger than 10%. Then, clean reads were mapped to the reference gene/genome using the BWA³⁴/Bowtie 2³⁵ tool. The average mapping ratio with the reference gene is 67.29% and the average genome mapping ratio is 81.72% as shown in Table 1. The expression levels of the genes were quantified by RSEM.³⁶ We counted the expressed genes and calculated its proportion to total genes for each sample as shown in Supplementary Figure S1. Of the 41,020 predicted genes in the Brassica rapa genome, the expression levels of 32,483 mapped genes were quantified based on reads per kilobases per million reads (RPKM), which were converted from the mapped read counts of differentially expressed genes (DEGs).

Table 1.

Summary statistics of RNA-seq data and map results. 13–13-168: R genotype. 15–14-168: S genotype. CK: control check without inoculation. T: treatment group with inoculation.

				Map to genes						Map to genome
Sample	Raw Reads Number	Clean Reads Number	Total Bases	Total Mapped Reads (%)	Perfect Match (%)	Mismatch (%)	Unique Match (%)	Multi-position Match (%)	Total Unmapped Reads (%)	Total Mapped Reads (%)	Perfect Match (%)	Mismatch (%)	Unique Match (%)	Multi-position Match (%)	Total Unmapped Reads (%)
13–13-168CK-1	13045738	12972038	636539155	68.54	48.69	19.86	45.16	23.38	31.46	82.04	60.38	21.66	73.28	8.75	17.96
13–13-168CK-2	13045624	12969962	591538976	60.71	42.13	18.58	39.58	21.13	39.29	74.72	53.24	21.48	66.97	7.75	25.28
13–13-168 T-1	13045918	12990595	635629862	68.36	47.84	20.52	43.42	24.93	31.64	84.15	61.36	22.79	73.81	10.34	15.85
13–13-168 T-2	12165094	12072224	635528138	69.1	48.52	20.58	43.59	25.5	30.9	84.03	61.41	22.62	73.66	10.38	15.97
15–14-168CK-1	13046001	12949537	634527313	69.92	48.59	21.34	45.65	24.27	30.08	84.11	61.1	23.01	75.51	8.59	15.89
15–14-168CK-2	13046098	12970648	635561752	68.82	48.09	20.73	45.15	23.67	31.18	83.86	61.4	22.46	75.38	8.47	16.14
15–14-168 T-1	13045950	12962500	635162500	66.18	47.28	18.9	47.4	18.78	33.82	80.12	58.95	21.17	73.77	6.36	19.88
15–14-168 T-2	13045815	12956601	634873449	66.67	46.96	19.71	47.73	18.94	33.33	80.73	58.62	22.11	74.23	6.49	19.27

Screening ofDEGsand clustering analysis

The NOISeq method³⁷ was used to find differentially expressed genes between these two groups, which maintained good True Positive and False Positive rates when increasing sequencing depth. Finally, the DEGs were selected according to the following default criteria: log₂ (fold change) > 2 or < −2 and diverge probability ≥0.8. A total of 3,055 DEGs were screened out from all the 41,020 genes, and 21 DEGs were found in all four paired comparisons (Figure 2a).

Figure 2.

Analysis of differentially expressed genes (DEGs) identified by pairwise comparisons. DEGs from experiment vs control check of both R and S genotypes. (a) Venn diagram of the number of DEGs in the four paired comparison. The number of genes up- or down-regulated in the inoculated and non-inoculated samples with a diverge probability ≥0.8 and log, fold change >2.0 or <−2.0 were shown in the diagram. (b) Statistic of differentially expressed genes. X-axis represents pairwise and Y-axis means number of screened DEGs. Black bar denotes down-regulated genes and red bar for the up-regulated. (c) Cluster analysis of gene-expression patterns of DEGs. The 3055 DEGs were grouped into 20 sub-clusters based on the common expression patterns by the OmicShare online tools (https://www.omicshare.com/tools). Six sub-clusters reflected the significant expression trends of DEGs. The gene number and p-value of each sub-cluster were marked on the top of each profile.

In comparison of the gene expression level in the control groups (S-CK vs.R-CK), there were 387 and 341 DEGs expressed higher and lower in S-CK than in R-CK, respectively, (Figure 2b), showing genotypic differences between the resistant and susceptible genome. In the experiment groups, the expression levels of 422 genes were up-regulated and 1,079 genes were down-regulated in S-Tcompared with that in R-T, which might be specifically involved in the process of Chinese cabbage downy mildew (Figure 2b).

Additionally, paired comparisons were carried out to analyze up- and down-regulated DEGs of the same genotype inoculated or non-inoculated. It was found that in the S genotype, out of 1,950 DEGs, 1,529 DEGs were up-regulated and 421 were down-regulated, while in the R genotype, up to 39% of DEGs were up-regulated with the remaining DEGs down-regulated (Figure 2b).

The 3,055 DEGs were grouped into sub-clusters based on the common expression patterns by the OmicShare tools, a free online platform for data analysis (http://www.omicshare.com/tools), including six distinct sub-clusters reflecting the general expression trends of DEGs in the R and S genotypes after H.brassicae infection (Figure 2c). The numbers of genes within each cluster were marked on each profile.

GO and KEGG pathway analysis

Annotation analysis of GO was performed by mapping screened DEGs to GO terms in the database (http://www.geneontology.org/), and then the WEGO software³⁸was used to perform GO function classification to understand the distribution of gene functions of Chinese cabbage from the macro level (Figure 3).

Figure 3.

GO terms and KEGG pathways enriched in the two lines of Chinese cabbage. (a) The GO terms and KEGG pathways enriched in DEGs of the R genotype (13–13) screened out from the samples before and after inoculation. (b) The GO terms and KEGG pathways.

The most enriched GO terms were annotated in the same paired-comparisons with significant difference at a level of corrected p-value ≤ 0.05 (Table 2). In terms of the number of GO terms, the DEGs of the S genotype were enriched to six terms in biological process, six terms in cellular component and four terms in molecular function while the R genotype DEGs were enriched to three terms in biological process, four terms in cellular component and two terms in molecular function. Most of the common enriched GO terms of the two genotypes (S genotype vs. R genotype) were mainly distributed in the following categories: ‘Response to stress’ (8 GO IDs vs. 10 GO IDs), ‘Response to stimulus’ (12 GO IDs vs. 11 GO IDs), ‘Metabolic process’ (13 GO IDs vs. 7 GO IDs) in biological process, ‘Cell’ (5 GO IDs vs. 6 GO IDs), ‘Plastid’ (8 GO IDs vs. 10 GO IDs) in cellular component and ‘Binding’ (8 GO IDs vs. 5 GO IDs) in molecular function. However, the S genotype had more GO terms than the R genotype including‘Multi-organism process,’ ‘Immune system process,’ and ‘biological regulation’ in biological process, ‘Photosynthetic membrane’ and‘Membrane part’ in cellular component and ‘Enzyme regulation activity,’ ‘Oxidoreductase activity’ and ‘Transmembrane transporter activity’ in molecular function. It is notable that six DEGs of the R genotype were enriched to the GO term ‘Carbon-nitrogenlyase activity’ while the S genotype did was not enriched to this term. In terms of the numbers of DEGs involved in each GO term, the S genotype mobilized more genes to participate in almost every GO term, while the gene number in ‘Cell’ in cellular component was less than that of the R genotype. Under almost every GO term, the expression levels of DEGs showed either an up- or down-regulated trend. However, 43 DEGs involved in ‘Photosynthetic membrane’ of the S genotype were all down-regulated. In addition, in the R genotype, the number of the down-regulated DEGs was more than that of the up-regulated DEGs in 3 GO terms, the number of the up-regulated genes was more than that of the down-regulated genes in 5 GO terms, and in the last one term, the number of the up-and down-regulated DEGs was the same. However, there were more up-regulated DEGs than down-regulated DEGs in 12 out of 16 GO terms of the S genotype except ‘Organelle part,’ ‘Membrane part,’ and ‘Photosynthetic membrane’ (all down-regulated) in cellular component and ‘Transmembrane transporter activity’ in molecular function.

Table 2.

The most enriched GO terms with significant difference at a level of corrected p-value < 0.05 in R Genotype (13–13-168CK vs. 13–13-168 T) and S Genotype (15–14-168CK vs. 15–14-168 T), respectively. The different GO IDs between the two genotypes were highlighted in red.

	R Genotype (13–13-168CK vs.13–13-168 T)		S Genotype (15–14-168CK vs.15–14-168 T)
GO terms	GO ID	Number of DEGs (up-/down-regulated)	GO ID	Number of DEGs (up-/down-regulated)
Biological process
Response to stress	GO:0006950, GO:0006952, GO:0009814, GO:0009408, GO:0006979, GO:1901700, GO:0000302, GO:0009415, GO:0009642, GO:0006978	115 (43/72)	GO:0006950, GO:0006952, GO:0009814, GO:0006979, GO:1901700, GO:0000302, GO:0006978, GO:0009743	381 (314/67)
Response to stimulus	GO:0050896, GO:0009628, GO:0009266, GO:0009314, GO:0009416, GO:0009411, GO:0009642, GO:0009607, GO:0042221, GO:0010035, GO:0010038	162 (63/99)	GO:0050896, GO:0009628, GO:0009314, GO:0009416, GO:0009642, GO:0009607, GO:0009605, GO:0043207, GO:0042221, GO:0010033, GO:0010035, GO:0010038	651 (519/132)
Metabolic process	GO:0019748, GO:0009698, GO:0044550, GO:0009699, GO:1902221, GO:0006558, GO:0006091	40 (22/18)	GO:0030162, GO:0045861, GO:0019748, GO:0009404, GO:0044550, GO:1901565, GO:0006022, GO:0006026, GO:0042430, GO:1901136, GO:0006790, GO:0015979, GO:0006091	161 (113/48)
Multi-organism process			GO:0051704, GO:0051707, GO:0098542	97 (87/10)
Immune system process			GO:0002376, GO:0006955, GO:0045087	40 (37/3)
biological regulation			GO:0031647, GO:0043086, GO:0051336, GO:0051346, GO:0052547, GO:0010466	18 (13/5)
Cellular component
Cell	GO:0005623, GO:0044464, GO:0071944, GO:0030312, GO:0031975, GO:0009579	172 (77/95)	GO:0071944, GO:0030312, GO:0009579, GO:0005618, GO:0044436	144 (105/39)
Extracellular region	GO:0005576	31 (24/7)	GO:0005576	104 (73/31)
Plastid	GO:0009536, GO:0044435, GO:0044434, GO:0009526, GO:0009532, GO:0031976, GO:0009507, GO:0009941, GO:0009570, GO:0009534	63 (32/31)	GO:0009536, GO:0044435, GO:0044434, GO:0009532, GO:0031976, GO:0009507, GO:0009534, GO:0009570	258 (145/113)
Organelle part	GO:0044422, GO:0031984, GO:0031976	56 (29/27)	GO:0031984	77 (10/67)
Photosynthetic membrane			GO:0034357, GO:0009521, GO:0009522, GO:0009523	43 (0/43)
Membrane part			GO:0031225, GO:0098796	61 (10/51)
Molecular function
Binding	GO:0033218, GO:0042277, GO:0003727, GO:0008187, GO:0046906	14 (7/7)	GO:0043167, GO:0043169, GO:0046872, GO:0046906, GO:0005506, GO:0030246, GO:0030247, GO:0001871	301 (229/72)
Carbon-nitrogenlyase activity	GO:0016840, GO:0016841	6 (5/1)
Enzyme regulator activity			GO:0030234, GO:0004857, GO:0061134, GO:0061135	36 (28/8)
Oxidoreductase activity			GO:0016491	145 (120/25)
Transmembrane transporter activity			GO:0022857, GO:0005372, GO:0042887	12 (1/11)

There were 61 DEGs showing a down-regulated trend both in the R genotype and the S genotype samples after infection, among which 13 DEGs were mapped to no GO terms and 19 DEGs were mapped to the same GO terms in the two genotypes (Supplementary File 2, Table S1). Five DEGs had different GO terms under Function Ontology, either mapped to GO terms in the S genotype only or mapped to more GO terms in the S genotype than that in the R genotype. For example, three of these five DEGs were mapped to “ion binding,” “cation binding,” and “metal ion binding” in the S genotype but mapped to no GO terms in the R genotype. Bra005425 was mapped to 3 GO terms in the S genotype while only 1 GO term in the R genotype. However, it was the opposite in cellular component and biological process. Among 21 DEGs with different GO terms referring to cellular component in these two genotypes, 19 DEGs were mapped to more GO terms in the R genotype than that in the S genotype while 4 out of 5 DEGs were mapped to different GO terms relating to biological process in the two genotypes. Six out of these 61 DEGs were mapped to the terms only present in the R genotype, such as ‘Multi-organism process,’ ‘Immune system process,’ and ‘biological regulation’ in biological process, ‘Photosynthetic membrane,’ and ‘Membrane part’ in cellular component. Proteins encoded by six DEGs were related to photosynthesis. For example, Bra005424, Bra026745, and Bra005425 encoded chlorophyll a-b binding proteins andBra001507 encoded a photosystem II oxygen-evolving complex 23 K protein.

In order to understand the biological functions of the DEGs, pathway-based analysis was carried out byKEGG³⁹ (the major public pathway-related database) using Brassica rapa (Chromosome v1.5) as the background reference and significantly enriched metabolic pathways and signal transduction pathways were identified in the DEGs (Figure 3). Firstly, the top 20 of the enriched pathway terms were analyzed (Supplementary File 3: Table S2). The result showed that there was significant a difference in the enriched pathways between these two comparison-pairs. A total of 32 enriched pathway terms were found in these two comparison-pairs with 8 terms in both comparison-pairs. Among these eight pathway terms, the top 2 enriched pathway terms, “Metabolic pathways” and “Biosynthesis of secondary metabolites,” contained 28.22% vs. 26.34% and 18.89% vs. 14.29% (S Genotype vs. R Genotype) of the DEGs of each comparison-pair, respectively. However, the third enriched pathway term in the R Genotype pair was “Protein processing in endoplasmic reticulum,” while in the S Genotype pair, the third enriched pathway term was “Plant-pathogen interaction.”

Then, we analyzed the enrichment pathways in the S Genotype (S-CK vs. S-T) and the R Genotype (R-CK vs. R-T) at a level of Q value≤0.05 (Supplementary file4: Table S3). The results showed that in the R Genotype pairs, among a total of 66 DEGs, there were 26 up-regulated DEGs and 40 down-regulated DEGs which participate in 7 KEGG pathways. In the S Genotype pairs, 471 out of 1,265 DEGs were involved in 20 KEGG pathways, among which, 392 were up-regulated and 79 were down-regulated. Although 4 KEGG pathways were both presented in the two comparison-pairs, DEGs involved in these pathways were obviously different. For example, in “Protein processing in endoplasmic reticulum” pathway, there were 25 and 51 DEGs in the R Genotype pair and S Genotype pair, respectively. Among these genes, 18 were the same. In “Flavonoid biosynthesis” pathway, among 8 vs. 15 DEGs of the two comparison-pairs, only one DEG was found in both pairs. In the R Genotype pair, all DEGs in “Flavonoid biosynthesis” and “Phenylalanine metabolism” were up-regulated while all DEGs in “Photosynthesis-antenna proteins” were down-regulated. In the S Genotype pairs, all DEGs in “Photosynthesis-antenna proteins” were also down-regulated. However, there were 8 KEGG pathways containing only up-regulated DEGs. The most interesting KEGG pathway was the “Plant-pathogen interaction” which has 82 up-regulated and 5 down-regulated DEGs. Then, we explored these 87 DEGs as shown in Table 3. In the plant-pathogen interaction pathway, 54 genes were detected which encode proteins belonging to 8 protein families, while 33 genes were detected which encode proteins belonging to 3 transcription factor families.

Table 3.

Defense response proteins in Chinese cabbage in response to H.brassicae stress. The pathway “Plant-pathogen interaction” was selected out from Supplementary file 4:Table S3 with a Q-value < 0.05. Numbers in column “Gene ID” represent the total number of DEGs in different terms. DEGs with diverge probability ≥0.9 were highlighted in red.

Terms	Gene ID	FPKM				Log2 fold change
		13–13-168CK	13–13-168 T	15–14-168CK	15–14-168 T	13–13-168 T/ 13–13-168CK	13–13-168CK/ 15–14-168CK	13–13-168 T/ 15–14-168 T	15–14-168 T/ 15–14-168CK
Plant-pathogen interaction	87
CML	17
Group III	3
	Bra002984	10.615	20.235	5.98	26.54	0.930748478	0.827886982	−0.391315522	2.149950981
	Bra000430	82.64	50.23	31.485	235.765	−0.718290979	1.392175587	−2.230728382	2.904612989
	Bra004524	0.74	2.595	0.37	6.885	1.810137363	1	−1.407722116	4.217859479
Group Ⅳ	14
	Bra027981	3.33	4.55	1.86	12.765	0.450344368	0.840219556	−1.488255087	2.778819011
	Bra017927	9.405	6.135	2.755	98.365	−0.6163646	1.771375625	−4.003009822	5.158020847
	Bra003712	0.855	1.3	0.215	5.35	0.604515298	1.99158776	−2.041027268	4.637130327
	Bra012889	7.3	3.67	0.935	73.435	−0.992116401	2.964858194	−4.322615865	6.295357657
	Bra025654	21.925	57.54	21.595	267.92	1.391988378	0.021879526	−2.219165155	3.633033059
	Bra028438	2.15	4.255	1.755	12.595	0.984822472	0.292865629	−1.565620085	2.843308186
	Bra030349	0.21	0.43	0.01	7.03	1.033947332	4.392317423	−4.031116124	9.457380879
	Bra033745	2.52	0.64	0.42	17.535	−1.977279923	2.584962501	−4.776021715	5.383704292
	Bra019503	3.715	4	0.62	31.73	0.106637789	2.58302209	−2.987775616	5.677435496
	Bra033467	46.08	37.315	31.955	204.48	−0.304385031	0.528099029	−2.45413215	2.677846149
	Bra018143	13.59	18.855	2.91	68.87	0.472401694	2.223454398	−1.868928527	4.564784619
	Bra021379	8.65	12.105	6.795	50.05	0.484831041	0.348226582	−2.04776699	2.880824613
	Bra039511	2.94	3.22	0.01	6.14	0.131244533	8.199672345	−0.931177967	9.262094845
	Bra013470	6.905	3.14	5.73	32.445	−1.136876855	0.269106276	−3.369159702	2.501389122
CALM	1
	Bra027848	0.35	2.13	0.18	4.685	2.605427	0.959358016	−1.137195617	4.70198
Calcium dependent Protein Kinases (CDPK)	3
	Bra037181	2.14	5.345	3.37	14.92	1.320579151	−0.655137795	−1.480985683	2.146427039
	Bra012058	1.95	7.385	2.04	13.69	1.921123797	−0.065095028	−0.89045262	2.746481389
	Bra038149	2.875	11.75	5.665	32.345	2.031026896	−0.978514	−1.460881956	2.513394852
CNGCs	2
	Bra021266	3.17	5.715	2.785	17.225	0.850270658	0.186805513	−1.591678579	2.62875475
	Bra004537	2.24	5.75	3.075	26.505	1.360063224	−0.457087678	−2.204630679	3.107606225
MAPK	4
MAPK3	Bra038281	10.58	26.825	13.205	64.745	1.342238544	−0.319744674	−1.271190612	2.293684482
MAPKK1	Bra030430	1.1	3.69	4.93	61.625	1.746117293	−2.164084123	−4.06182302	3.64385619
MAPKK7	Bra008099	65.015	144.735	60.895	277.605	1.154569324	0.094448834	−0.939619714	2.188637872
MAPKKK4	Bra039693	0.23	0.125	0.12	5.53	−0.879705766	0.938599455	−5.46727948	5.52617317
HSP 90 s	3
	Bra028952	36.905	121.05	31.805	137.905	1.713714884	0.214562704	−0.188071686	2.116349274
	Bra028948	56.83	182.435	35.395	156.005	1.682657915	0.68310714	0.225790267	2.139974788
	Bra028924	71.87	184.75	49.17	224.225	1.362112773	0.547611331	−0.279372776	2.18909688
ALMP	1
3-ketoacyl-CoA synthase 21	Bra009445	0.01	0.275	0.01	15.02	4.781359714	0	−5.771309384	10.5526691
LRR proteins	23
TAO1	Bra027047	2.67	3.83	1.28	15.525	0.52050465	1.060695932	−2.019176971	3.600377553
TMM	Bra003549	5.98	9.545	39.335	7.435	0.674599713	−2.717596194	0.360412455	−2.403408936
SERK	Bra041057	6.06	6.075	5.47	22.305	0.003566615	0.147776961	−1.876410834	2.02775441
RNI-like	Bra021980	5.91	6.895	2.75	13.865	0.222392421	1.103726512	−1.00782516	2.333944093
IRAK	Bra002544	0.16	1.11	0.31	6.845	2.794415866	−0.95419631	−2.624490865	4.464710421
GSO2	Bra039532	0.81	0.08	0.04	4.18	−3.339850003	4.339850003	−5.707359132	6.707359132
PR1-1	Bra013123	69.025	2291.79	40.49	6097.97	5.053212061	0.76955334	−1.411854199	7.2346196
PR1-2	Bra010447	23.555	25.36	48.545	3.22	0.106521413	−1.043289378	2.977422152	−3.914190116
STS14	Bra037162	33.86	15.605	34.17	8.285	−1.117573616	−0.013148275	0.913434755	−2.044156645
PGIP1	Bra009234	4.74	4.46	3.33	15.915	−0.087843349	0.509364882	−1.835271542	2.256793075
PGIP1	Bra009235	3.03	0.88	1.505	9.89	−1.783742365	1.009554307	−3.490395092	2.716207034
DRT100	Bra038737	1.49	2.885	5.25	0.215	0.953258988	−1.817005092	3.746162754	−4.609908858
PTI 1-like	Bra001662	6.445	9.105	5.215	22.295	0.498478659	0.305513106	−1.291989277	2.095981042
CEBip	Bra009660	3.29	5.28	3.915	19.92	0.682450346	−0.250924724	−1.915607813	2.347133435
PAD4	Bra033431	3.175	9.895	1.62	26.655	1.639943115	0.970762779	−1.42963457	4.040340464
EDS1	Bra033774	4.805	12.745	6.585	116.37	1.407323037	−0.454647009	−3.190715903	4.143391931
EDS1	Bra018085	7.575	11.01	5.005	27.065	0.539496675	0.59787582	−1.297613919	2.434986414
CysP RD-19 C-related	Bra033545	85.87	39.375	82.445	356.155	−1.124874268	0.05872219	−3.177153416	2.111001337
TIR-NBS-LRR	Bra008056	11.29	9.525	8.785	35.54	−0.245254488	0.361931295	−1.899652684	2.01632949
Unknown LRR	Bra034594	0.71	0.565	0.325	5.975	−0.329568157	1.127379306	−3.40261594	4.20042709
Unknown LRR	Bra009434	11.92	14.885	13.215	63.665	0.320474986	−0.148792189	−2.096641243	2.26832404
Unknown LRR	Bra034404	3.875	11.295	5.28	67.985	1.543416055	−0.446341619	−2.589532199	3.686606635
Unknown LRR	Bra000657	2.08	3.425	3.93	16.515	0.71952046	−0.917945784	−2.269601076	2.071175752
Transcription factors	33
ERF TF	3
CRF2	Bra019270	6.13	5.515	5.14	0.25	−0.152526188	0.254118714	4.463360886	−4.361768359
CRF4	Bra040839	1.375	3.215	0.23	9.19	1.225387119	2.579725852	−1.515246124	5.320359095
ERF119	Bra036360	1.735	2.77	2.65	16.085	0.674950313	−0.611056697	−2.537758055	2.601651671
MYB TF	8
51–2	Bra016553	7.83	14.07	3.72	15.34	0.845538116	1.073709686	−0.124676154	2.043923956
57	Bra024760	0.01	0.24	0.01	5.2	4.584962501	0	−4.437405312	9.022367813
44	Bra032176	0.515	2.54	0.13	4.4	2.30218416	1.986060809	−0.792675027	5.080919995
108	Bra004473	0.115	1.73	0.44	11.94	3.911066272	−1.935869663	−2.786958894	4.762155503
24	Bra012482	1.315	2.245	0.49	6.695	0.771652645	1.424209145	−1.576368611	3.772230401
8	Bra017685	6.115	6.45	4.95	22.06	0.076946662	0.304923974	−1.774061725	2.155932361
	Bra008610	13.09	17.82	3.89	18.13	0.44503224	1.750623037	−0.024881588	2.220536865
88	Bra012337	9.96	9.17	7.155	29.26	−0.119224008	0.477193975	−1.673936131	2.031906097
WRKY TF	22
	Bra020546	0.885	2.26	0.115	4.13	1.352573412	2.944043594	−0.869819009	5.166436015
	Bra019265	1.14	13.35	0.66	20.34	3.549734012	0.788495895	−0.607479937	4.945709844
	Bra013732	2.075	24.455	3.15	74.885	3.558946223	−0.602240492	−1.614547301	4.571253032
	Bra032174	14.89	10.295	12.78	58.55	−0.532399924	0.220455918	−2.507725341	2.195781334
	Bra028707	2.48	3.27	0.365	6.61	0.398950515	2.764371752	−1.015359636	4.178681903
	Bra023998	4.25	4.55	4.88	21.07	0.098403704	−0.199418307	−2.211251864	2.110237261
	Bra040557	4.87	0.675	0.44	13.86	−2.850962365	3.468346343	−4.359895945	4.977279923
	Bra003239	0.935	3.76	1.145	12.325	2.007694392	−0.292309328	−1.71278308	3.428168143
	Bra010231	10.5	13.865	5.89	66.125	0.401058288	0.834049789	−2.253748201	3.488856278
	Bra024328	0.01	1.515	0.01	3.15	7.243173983	0	−1.056034035	8.299208018
	Bra031900	2.935	4.925	0.85	23.425	0.746763221	1.787825757	−2.249853418	4.784442396
	Bra000064	0.235	1.68	0.915	9.545	2.837728571	−1.961110987	−2.506283965	3.382901549
	Bra040926	2.21	0.39	0.385	13.145	−2.502500341	2.521116019	−5.074896208	5.093511886
	Bra017117	0.265	0.555	0.07	3.9	1.066495412	1.920565533	−2.812914447	5.799975392
	Bra037796	0.01	1.33	0.115	10.63	7.055282436	−3.523561956	−2.998643446	6.530363925
	Bra010220	0.065	0.455	0.01	3.975	2.807354922	2.700439718	−3.12701641	8.63481105
	Bra011299	3.005	7.4	1.995	18.185	1.30016028	0.590976244	−1.29715175	3.188288274
	Bra014692	2.015	6.195	2.09	15.87	1.620324444	−0.052723104	−1.357125941	2.924727281
	Bra007243	2.835	14.71	3.21	45.5	2.375376606	−0.179224562	−1.629069299	3.825221343
	Bra032084	4.22	8.535	2.95	17.84	1.016148154	0.516528044	−1.063652557	2.596328756
	Bra027057	2.955	3.235	1.175	13.6	0.130607582	1.330497374	−2.071769034	3.53287399
	Bra033158	0.275	3.815	0.275	10.405	3.794179533	0	−1.447522003	5.241701536

Defense responsive genes

In the search of defense responsive genes to downy mildew in Chinese cabbage, a total of 87 up- or down-regulated DEGs were identified (Table 3). Of these DEGs, 54 and 33 were categorized into plant-pathogen interaction and transcription factors, respectively. In the category of plant-pathogen interaction, 23 R proteins were involved while3ethylene responsive factors (ERFs), 8 MYBs and 22 WRKYs were identified as transcription factors.

When comparing the CK groups of the R Genotype and the S Genotype (R-CK vs.S-CK), we found that more than 71%of the total DEGs (including TFs) involved in the plant-pathogen interaction pathway had a higher FPKM (including 4 pairs of equal FPKMs) in the R genotype than that in the S genotype. It is worth mentioning that there were 16 DEGs (including 6 DEGs encoded Calmodulin-like protein (CML), 1DEG encoded Mitogen-Activated Protein Kinase (MAPK), and 2 Resistance(R) genes), 1 ERF TFand 6 WRKY TFsparticipated in the plant-pathogen interaction pathway with an absolute log₂(fold change) value greater than 2, indicating that these genes might be considered as basal defense responsive genes. Additionally, when comparing the T (treatment) groups of the R Genotype and the S Genotype (R-Tvs.S-T), there were only five DEGs (1Heat Shock Protein 90 gene (HSP90), 4Rgenes, and 1ERF TF) showing a higher FPKM in the R genotype. A total of, 82 DEGs were up-regulated by more than 4 folds in the S Genotype samples after inoculation, accounting for 94% of the total DEGs. However, in the R genotype, there were 68 DEGs up-regulated after the inoculation of H.brassicae, among which only 17 DEGs had a log₂(fold change) value greater than 2. These results showed that the S Genotype had a more sensitive and stronger response to the infection of H.brassicae.

In the S genotype, two cyclic nucleotide-gated channels genes (CNGCs) with a number of calcium signaling genes (17 CMLs, 1 calmodulin (CALM) and 3 Calcium Dependent Protein Kinase genes (CDPKs)) were activated indicating that Ca²⁺signaling may play an important role in response to H.brassicae of Chinese cabbage. For example, the CALM gene, Bra027848, was up-regulated by more than 16 folds, and the log₂(fold change) value of the CML family genes, including Bra030349, Bra033745, and Bra019503, was increased by more than 9, 5, and 5 folds, respectively. We also found that after the inoculation of H.brassicae, all four DEGs involved in the MAPK family were dramatically up-regulated in the S genotype with the fold changes of the expression level ranging from 4 to 32, indicating that the MAPK pathway was specifically activated in the S genotype.

A number of candidate LRR proteins were identified in this study. For example, genes encoding target of AvrB Operation 1(TAO1), Peroxin (PEX), Interleukin 1 receptor-associated kinase (IRAK), and GASSHO (GSO2) were expressed by more than 8, 16, 16, and 64 folds in the S genotype after inoculation, while the expression levels of Tao1andPEX showed no significant difference in R genotype. The expression of IRAKwas up-regulated by 8 folds, and the expression of GSO2was down-regulated by 8 folds in the R genotype after inoculation. Interestingly, two Pathogenesis-Related 1 (PR1) genes, namelyPR1-1 andPR1-2showed different patterns in these two genotypes. The expression of PR1-1was increased by 32and 128 folds after inoculation in the R and S genotype, respectively. However, the expression of PR1-2did not change in the R genotype, but was dramatically down-regulated in the S genotype after inoculation.

Among 33 TFs,3 MYB TFs had a log₂ fold change of more than 2, while the rest had no significant changes in the R genotype after inoculation. However, all the MYB TFs were markedly up-regulated in the S genotype. In the R genotype, 9 WRKYs were up-regulated, and 2 WRKYs were down-regulated. The expression of ERF TFs had no obvious change in the R genotype but showed significant changes in the S genotype after inoculation.

Validation of the RNA-Seqresults by qRT-PCR

To validate the RNA-seq data, 18 DEGs with a diverge probability ≥0.9 in the RNA-seq data, which belonged to 4 gene families and WRKY TFs, were selected for real-time quantitative PCR (qRT-PCR) analysis. These 18 DEGs included13 defense responsive genes and 5 WRKYs. The relative expression measurements from qRT-PCR are shown in Figure 4, and the qRT-PCR primers designed for DEGs are listed in Supplementary file 5: Table S4. These two methods showed that the DEGs had similar expression patterns, indicating that RNA-seq could be used for gene expression profiling of Chinese cabbage in response to the infection of H. brassicae.

Figure 4.

Validation of RNA-seq data by qRT-PCR. Relative expression levels represented by gray columns were calculated by the comparative method. FPKMs represented by black lines were from RNA-seq data meaning reads per kb per million reads.

Discussion

Downy mildew caused by H.brassicae is one of the most severe oomycete diseases of Chinese cabbage leading to serious crop yield losses in Chinese cabbage. Although great progress has been achieved in locating the R genes responding to downy mildew in Chinese cabbage, the mechanism is still not clearly understood. Transcriptome analysis has been successfully performed in two grapevines lines (a resistance line and a susceptible line) infection by Plasmoparaviticola (Pv), the pathogen causing downy mildew in grapevine,⁴⁰ which makes it possible to understand the molecular mechanism via identification of important candidate R genes and pathways related to downy mildew responses. In order to have a further insight into the resistance mechanism of Chinese cabbage to downy mildew, a genome-wide comparative transcriptome analysis of two Chinese cabbage species was carried out after inoculated by H. brassicae. The results indicated that there were different numbers of DEGs identified in different lines after inoculation probably due to basic responsive mechanisms. Compared with the R phenotype line, the S phenotype line has a larger number of DEGs indicating that there were more biological processes disturbed by the pathogen infection in the susceptible Chinese cabbage.

GO analysis showed that ‘Carbon-nitrogenlyase activity’ was only presented in the R genotype and ‘Multi-organism process,’ ‘Immune system process,’ and ‘biological regulation’ in biological process, ‘Photosynthetic membrane,’ and ‘Membrane part’ in cellular component and ‘Enzyme regulation activity,’ ‘Oxidoreductase activity,’ and ‘Transmembrane transporter activity’ in molecular function were specific to the S genotype. In the 61 down-regulated DEGs found in both the R genotype and the S genotype, the R genotype had more GO terms than the S genotype. Among the 61 DEGs, 6 were mapped to GO terms related to photosynthesis in the S genotype.KEGG analysis also showed that all the DEGs involved in photosynthesis were down-regulated in both genotypes, indicating that the photosystem in Chinese cabbage was inhibited upon H.brassicaeinfection. A number of DEGs among these 61 DEGs were mapped to “ion binding,” “cation binding,” and “metal ion binding” in the S genotype while mapped to no GO terms in the R genotype, indicating that diverse ion channels might play a role in defense responses of Chinese cabbage after infection.

In our previous study³¹ on the resistance to downy mildew between two lines of Chinese cabbage named 318 R and 361 S, DEGsof the two lines were significantly enriched in the top three pathways: plant-pathogen interaction pathway, plant hormone signal transduction pathway, biosynthesis of secondary metabolites pathway. However, in our current study, DEGs of R (13–13) and S (15–14) were enriched in metabolic pathway, biosynthesis of secondary metabolites pathway, and protein processing in endoplasmic reticulum. These differences should be related to different species and external environment. We also found that three pathways: protein processing in endoplasmic reticulum, fatty acid metabolism and propanoate metabolism were involved in the both R genotypes while two pathways: sulfur metabolism and metabolic pathways were involved in the both S genotypes, which could be the common resistant mechanism of downy mildew in Chinese cabbage. Additionally, in our current study, two pathways: endocytosis pathway and circadian rhythm pathway were found to be specific to R genotype (13–13), indicating that seedlings responded to H.brassicae via enhancing the ability of cellular trafficking⁴¹and regulating its circadian rhythm behavior⁴² as part of its innate immunity. Another two pathways: plant-pathogen pathway and carbon metabolism pathway were also found to be specific to S genotype (15–14), suggesting that there are different tolerance mechanisms of downy mildew in different Chinese cabbage species.

A total of87 candidate DEGs involved in the plant-pathogen interaction pathway were identified, among which 54 and 33 DEGs were categorized into plant-pathogen interaction proteins and transcription factors, respectively. Among these 54 plant-pathogen interaction proteins, 23LRRs, 20 CMLs, 4 MAPKs, 3CDPKs, 3 HSP 90 s, 2 CNGCs, and 13-ketoacyl-CoA synthase (KCS) were involved while 3 ERFs, 8 MYBs, and 22 WRKYs were found as transcription factors (Table 3).

The first layer of defense: defense genes in PTI

In plants, there are a series of innate immune receptors namely pattern recognition receptors (PRRs), which can detect microbe-associated molecular patterns (MAMPs) or damage-associated molecular patterns (DAMPs) on cell surfaces.⁴³ Plant PRRs include receptor-like kinases (RLKs) and receptor-like proteins (RLPs).^44,45The perception of MAMPs and PAMPs by PRRs triggers the PTI signaling pathway.

It has been clarified that chitin elicitor receptor kinase 1 (CERK1) directly binds with chitin. Interacting with three chitin-elicitor binding protein (CEBiP) homologs are indispensable for chitin recognition and signaling,^46–48 which triggers immune responses in many plants, such as rice and Arabidopsis.^49–51 Additionally, it has been reported that the interaction of OsCERK1 with HSP90 and its cochaperone Hop/Sti1 in the endoplasmic reticulum (ER) plays an important role in the transport of PRRs.⁵² A chitin synthase gene has been found in grapevine downy mildew oomycete, Plasmoparaviticola.⁵³ However, there is no evidence indicating whether H.brassicae has chitin-containing cell walls. In our study, the expression level of a PRR gene CEBip was significantly increased in the S genotype after H.brassicae infection, which may activate the chitin signaling and enhance the immune responses of Chinese cabbage.

Polygalacturonase-inhibitor proteins (PGIPs), as an important plant defense family, influence the accumulation of elicitors by regulating the activity of microbial polygalacturonases (PGs). It has been shown that the heterologous expression of a common bean PGIP gene confers a strong resistance against two oomycetes,Phytophthoraparasiticavar.nicotianae and Peronosporahyoscyamif. sp.tabacina, on the transgenic plants.⁵⁴Another study was carried out and identified a pearl millet PGIP1, which plays an important role in host defense.⁵⁵ Through RNA-Seq analysis, two PGIP genes were identified to express more than 4 fold higher in the S genotype after the infection suggesting that they may function in the resistance of Chinese cabbage to downy mildew.

The somatic embryogenesis receptor kinase (SERK) family can form complexes with various immune receptors, such as Flagellin-Sensitive 2 (FLS2), EFR, Elicitor Peptide1 receptor 1 and 2 (PEPR1 and PEPR2), in a ligand-dependent manner tomediatePTI signaling in Arabidopsis.^56–59 Additionally, SERKs associate with ERECTA (ER) and Too Many Mouths (TMM) to activate the MAP kinase cascade and to define the stomatal patterning.^60,61Considering that TMM differentially modulates stomatal development in different organs with a negative role in cotyledons and a positive role in hypocotyls and stems,^62,63 it is not astonishing thatSERKwas up-regulated while TMMwas down-regulated in the S genotype after inoculation.

The stimulation of PRRs by PAMPsand DAMPs induces a series of intracellular signaling events including MAPK cascades and calcium signaling, eventually resulting in the PTI. Typically, a MAPK cascade contains three sequentially activated kinases, MAPKKK, MAPKK, and MAPK, which transduce signals through phosphorylation.⁶⁴ The expression of MAPK-signaling genes, mitogen-activated protein kinase 3 (MAPK3), mitogen-activated protein kinase kinase1/7 (MAPKK1/7) and mitogen-activated protein kinase kinase kinase 4 (MAPKKK4), was increased by at least 2 fold in the S genotype-T compared with that in the S-genotype-CK, upon H.brassicae infection. Based on the expression pattern of these genes in our study, we suspect that the MAPKKK4-MAPKK1/7-MAPK3 cascade positively regulates the defense responses of Chinese cabbage to downy mildew.

It is reported that cytosolic free Ca²⁺ (Ca²⁺_cyt) is increased rapidly by MAMPs in plant cells, and different MAMPsinduce different spatio-temporal Ca²⁺ patterns.^65–67A total of 23 DEGs involved in the calcium signaling were identified by the RNA-Seq analysis, which encode1CALM, 17 CMLs, 2 CNGCs, and 3CDPKs, respectively. The results indicate that the calcium signaling pathway maybe a positive regulator for downy mildew resistance.

As mentioned above, from the transcriptome of the two accessions of Chinese cabbage, one CEBiP, one PGIP, and one SERK were identified as PRRs and exhibited higher expression levels in response to H.brassicae in the S genotype (Table 3). These proteins have been reported to be effective in response to downy mildew in Arabidopsisand rice as the first layer of plant defense. Unfortunately, little has been reported on the mechanisms of these genes in the progress of resistance to downy mildew in Chinese cabbage. Due to the broad-spectrum antibacterial activity of PTI and the conserved function of homologous proteins, we believe that these proteins play a similar role in the PTI process of Chinese cabbage.

The second layer of defense: the defense genes in ETI

In the ETI pathway, the nucleotide-binding site and leucine-rich repeat domain containing (NB-LRR) proteins recognize effectors and trigger the ETI response,⁶⁸ which lead to hypersensitive response (HR) and cause programmed cell death.⁶⁹ Although effectors can be directly recognized by protein–protein interaction with NB-LRRs, the recognition may be interrupted by new effectors obtained by the evolution of pathogens. Thus, indirect effector recognition assists the detection of diverse effectors, in which a sole immune receptor is activated without maintaining the R protein collection in plant defense against pathogens.⁷⁰ Plant NB-LRRs have been proven to resist against various pathogens.^71,72

In the early 1990 s, a series of phytoalexin-deficient (pad) mutants were isolated from the accession Columbia (Col-0) of Arabidopsis in order to test their response to pathogens, including Peronosporaparasitica.⁷³ The results show that PAD1, PAD2, PAD3, and PAD4 are required for the resistance to P.parasitica and pad4-1 exhibits full susceptibility to four of six Col-incompatible parasite isolates. Enhanced disease susceptibility 1 (EDS1), as a Nucleo-cytoplasmic basal resistance regulator, relates to host cell death and transcriptional mobilization of defense pathways by transferring signals to downstream TIR-NB-LRR receptors.⁷⁴Recently, a grapevine downy mildew related candidate RxLR effector derived from Plasmoparaviticola, PvAvh74, has been found in the induction of cell death in Nicotianabenthamiana (N. benthamiana) leaves.⁷⁵Using virus-induced gene silencing (VIGS), it is found that EDS1, Non-race specific disease resistance protein 1(NDR1), Suppressor of G-2 alleleofskip 1(SGT1), Required for Mla12 Resistance 1 (RAR1) and HSP90are required to PvAvh74-induced cell death in N. benthamiana. Notably, PR1has been commonly used as a molecular marker functioning inplant pathogen resistance, but their biological activity has remained elusive.^76,77 The correlation between PR-1 proteins and enhanced resistance against various oomycetes in many plants has been noted.^78,79 A high-throughput cDNA tag sequencing method was carried out to uncover the expression patterns of H.parasitica predicted effectors and Arabidopsis genes related to downy mildew.⁸⁰ As a result, Arabidopsis salicylic acid (SA)-responsive genes including PR1, EDS1, and PAD4 were activated not only at one daypost-inoculation (dpi) with H.parasiticaEmoy2 but also at 3 and 5 dpi with H. parasitica Waco9. The results are consistent with previous reports of the expression profiles of Arabidopsis and H.parasitica.^81–84

In our study, we found that 2 PR1, 2 EDS1, and 1 PAD4 were up-regulated dramatically in the S genotype after inoculation, especially PR1-1. The expression level of PR1-1was increased by more than 149 and 32 times in the S genotype and the R genotype, respectively. Interestingly, PR1-2 was down-regulated in the S genotype after inoculation but had no significant change in the R genotype, indicating that these 2 PR1 s have different mechanisms and functions in different pathways related to disease resistance. Similar results have been reported. For example, tobacco PR-1gexhibits the highest fungicidal activity when encountered with an oomycetePhytophthorainfestans, whereas tobaccoPR-1a and −1b only show a marginal effect.⁸⁵ We also noticed that PR1, EDS1, and PAD4 were involved in the SA pathway, indicating that SA signaling was activated in the S genotype after H.brassicae infection, which is consistent with previous studies.^73,75,80

Additionally, 5 genes encoding 1 cysteine (Cys) protease and four unknown LRR-proteins were found up-regulated while 1 gene encoding a DNA-damage repair/toleration 100 protein (DRT100) was down-regulated in the S genotype after infection. However, we did not find any evidence proved that they were involved in the responses of Chinese cabbage to downy mildew.

TFs response to downy mildew

TFs rapidly respond to the defense signals, and related transcripts are accumulated to withstand pathogens. Some transcriptional factors have been found to play an important role in signaling pathways related to resistant responses. For example, AtWRKY70 was found to be indispensable for both full RPP4-mediated disease resistance and basal defense against the oomycete Hyaloperonosporaparasitica.⁸⁶ It also reported that WRKY proteins are responsible for the basal expression of an Arabidopsis downy mildew resistance gene, namely RPP8, and that H. arabidopsidis-induced up-regulation of WRKY is driven by SA signaling.⁸⁷ As revealed in our study, 22 WRKYs in Chinese cabbage were identified in response to the attack of H.brassicae.Three ERF TFs, as positive regulators in the ET-signaling pathway,^88,89 were dramatically changed in expression in the S genotype after inoculation, indicating the activated ET signaling in response to downy mildew. Six ethylene response factor (ERF) transcription factor genes, AtERF24, AtERF25, AtERF108, AtERF109, AtERF112,and AtERF113,play an important role during downy mildew infection of Arabidopsis.⁹⁰ Numerous MYBs have been characterized by genetic approaches, which function in the control of plant-specific processes, such as primary and secondary metabolism, cell fate and identity, developmental processes, and responses to biotic and abiotic stresses.⁸⁴ In the responses of grapevine to downy mildew, MYBs regulate the transcriptional activation of distinct stilbene synthase genes(STS). The expression level of STSwas found to be inversely proportional to the incidence and severity of disease symptoms.⁹¹STS14 was identified in this study with a lower FPKM in the S genotype after infection compared with that of the S genotype CK group. However, STS14 had the highest FPKM in the R genotype CK group but had no significant change after infection. It is consistent with the results mentioned above. In addition, 22 MYBswere activated in response to downy mildew. However, the interaction mechanism between MYBs and STS14 was still unknown in Chinese cabbage.

Conclusions

A transcriptome analysis was carried out between a resistant line (13–13, R) and a susceptible line (15–14, S) of Chinese cabbage in response to H.brassicae. A large number of DEGs were identified in both the R and S lines.DEGs of R and S were enriched in metabolic pathway, biosynthesis of secondary metabolites pathway, and protein processing in endoplasmic reticulum. Two pathways named endocytosis pathway and circadian rhythm pathway were found specific to R genotype while another two pathways named plant-pathogen pathway and carbon metabolism pathway were specific to S genotype, which may relate to the basal and species-specific resistance to downy mildew of Chinese cabbage, respectively. Plant-pathogen interaction pathway was found to be active only in S genotype, among which some LRRs, R proteins, TFs were identified, indicating that these proteins may be important in regulating species-specific resistance to downy mildew of Chinese cabbage. A total of 87 candidate DEGs were identified from Chinese cabbage with different phenotypes after infection by H.brassicae. These candidate genes were involved in the plant-pathogen interaction pathway, among which 54 and 33 DEGs were categorized into plant-pathogen interaction proteins and transcription factors, respectively. Proteins encoded by some of these DEGs were related to Ca²⁺signaling, MAPK pathway meaning the two pathways specifically activated in the S genotype. Some other DEGs encoded LRRs including R proteins, the function of which has been studied thoroughly in model plants, such as Arabidopsis, rice, tomato, and tobacco. However, little is known about the mechanisms of these genes in resistant response to downy mildew in Chinese cabbage. Our findings are critical to further reveal the tolerance mechanism of H.brassicae in Chinese cabbage and helpful in breeding resistant strains.

Materials and methods

Plant materials and growth conditions

Two Chinese cabbage inbred lines,13–13 and 15–14, were used in this study for 13–13 is has a highly resistant phenotype to downy mildew, while 15–14 has a susceptible phenotype. Both lines were acquired from Shandong Academy of Agricultural Sciences, China. The seedlings were sown in soil in a plastics greenhouse with a 14 h light/10 h dark, 25°C to10°C (day to night) and relative humidity at 60% to 95% (day to night) regime.

Inoculation with H.brassicae

The diseased leaves of Chinese cabbage infected byH. brassicae were collected from the field. The leaves were cultivated at 20°C and 95% relative humidity in an incubator (Dongnan, Ningbo, China) overnight for proliferation of conidia after distilled water washed and brushed to clean the mildew layer and other attachment. The leaves were then brushed using a sterile hairbrush to dislodge the conidia into sterilized distilled water and diluted to a final concentration of 1 × 10⁶ spores per milliliter.

Plants at the two-leaf stage (20 days after sowing) were used to inoculate by spraying the bacterium solution evenly onto the leave surfaces. In the control groups, the leaves were sprayed with sterilized water. Then, the treatment groups and control groups were covered with a plastic film at 20°C and 95% relative humidity in the incubator for 24 h.

Finally, the plants were transferred into a plastics greenhouse with a 14 h light/10 h dark, 25°C to 10°C (day to night) and relative humidity at 60% to 95% (day to night) regime. The leaves were harvested on the 7th day after inoculation for RNA isolation. The experiments were repeated three times. At the same time, leaves were taken for trypan blue staining. The leaves were soaked in boiling trypan blue solution (10 mL lactic acid, 10 mL glycerol, 10 mL ddH₂O, 10 g phenol, and 20 mg trypan blue for 5 min and incubated for 3 h. Samples were then decolorized overnight in 2.5 g/mL chloral hydrate solution to clear the background and photographed.

We collected

RNA isolation

Two biological replicates of each H.brassicae treated R and S genotype sample and the control samples at 168 h (7 days) after inoculation were collected for RNA-seq analyses. Samples were immediately frozen in liquid nitrogen and stored at −80°C until use. Total RNA was extracted from each sample using Trizol reagent (Invitrogen, Carlsbad, CA, USA) and treated with RNase-free DNase Ⅰ (Takara, Dalian, China) for 45 min according to the manufacturer’s protocol. The quality of the RNA was measured by agarose gel electrophoresis with a 1% agarose gel and on a 2100 Bioanalyzer RNA Nanochip (Agilent, Santa Clara, CA, USA). Equal quantities of total RNA from three replicates were mixed for cDNA library construction.

cDNA library construction, sequencing, and data processing

cDNA libraries were constructed at Beijing Genomics Institute (BGI, Shenzhen, China) according to the manufacturer’s protocol (Illumina Inc., San Diego, CA, USA). During the QC step, Agilent 2100 Bioanalyzer (Agilent Technologies Inc., Santa Clara, CA, USA) and ABI StepOnePlus Real-Time PCR System (Applied Biosystems, Inc., Foster City, CA, USA) are used to qualify and quantify of the sample library. The library products are ready for sequencing via Illumina HiSeqTM 2000 (Illumina Inc). Raw reads after quality control (QC) are filtered into clean reads which will be aligned to the reference sequences. Clean reads after QC are named alignment data, which is used to calculate distribution of reads on reference genes and mapping ratio. Bowtie2³¹ was used to map clean reads to reference gene and BWA³²was used to map to reference genome.

Gene quantification, screening DEGs, GO, and KEGG analysis

FPKM (Reads Per kb per Million reads) method³⁰ is used in calculated expression level of genes, which is able to eliminate the influence of different gene length and sequencing discrepancy on the calculation of gene expression. NOISeq method³⁴ was used to identify differentially expressed genes (DEGs) according to the following default criteria: Fold change ≥2 or ≤-2and diverge probability ≥0.8. All DEGs were mapped to GO terms in the database (http://www.geneontology.org/). GO terms with a p-value ≤ 0.05 are defined as significantly enriched GO terms in DEGs. KEGG (http://www.genome.jp/kegg/) was used to perform pathway enrichment analysis of DEGs.

Quantitative RT-PCR analysis

First-strand cDNA was synthesized using PrimeScript 1^st Strand cDNA synthesis Kit (Takara). Quantitative real-time PCR (qRT-PCR) was carried out using a SYBR Green Master mix (Takara, Dalian, China) on an IQ5 Real-Time PCR Detection System (Bio-Rad, Hercules, CA, USA). The qRT-PCR primers designed for the candidate DEGs and actin gene (BrACT1) are listed in Supplementary file 5. and the BrACT1 was proved to be succeed in being the internal control in the research published by Jeongyeo Lee et al.⁸⁸ in 2103. The PCR cycling conditions comprised an initial polymerase activation step of 95°C for 1 min, followed by 40 cycles of 95°C for 10 s, and 60°C for 30 s. After each PCR run, a dissociation curve was designed to confirm the specificity of the product and to avoid the production of primer dimers. Three replicates of each sample were conducted to calculate the average Ct. Relative expression level was calculated by the comparative 2^‐ΔΔCt method.⁹²

Funding Statement

This work was supported by the Modern Agricultural Industrial Technology System Funding of Shandong Province, China [SDAIT-05-04]; Agricultural Science and Technology Innovation Project of SAAS [CAAS-XTCX2018021]; National Key Research and Development Program of China [2017YFD0101801].

Author contributions

Jingjuan Li, JianweiGao and Yuping Bi conceived and designed the experiments. Han Zheng, Yihui Zhang performed the experiments. Han Zheng, Lilong He, and Fengde Wang analyzed the data. Han Zheng, Lilong He, Jingjuan Li, and JianweiGao drafted the manuscript. All authors read and approved the final manuscript.

Conflicts of interest

The authors declare that there is no conflict of interests regarding the publication of this paper.

Data availability

All sequence data were deposited in the NCBI Sequence Read Archive (SRA, http://www.ncbi.nlm.nih.gov/Traces/sra) under accession numberSRR11128169, SRR11128168, SRR11128167, and SRR11128166 for R-CK, R-T, S-CK, and S-T, respectively.

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website.