Comparative transcriptome analysis reveals defense responses against soft rot in Chinese cabbage

Pectobacterium carotovorum ssp. carotovorum (Pcc) is a necrotrophic bacterial species that causes soft rot disease in Chinese cabbage. In this study, plants harboring the resistant mutant sr gene, which confers resistance against Pcc, were screened from an 800 M2 population mutated by ethyl methane sulfonate (EMS) and scored in vitro and in vivo for lesion size. The transcript profiles showed ~512 differentially expressed genes (DEGs) between sr and WT plants occurring between 6 and 12 h postinoculation (hpi), which corresponded to the important defense regulation period (resistance) to Pcc in Chinese cabbage. The downstream defense genes (CPK, CML, RBOH MPK3, and MPK4) of pathogen pattern-triggered immunity (PTI) were strongly activated during infection at 12 hpi in resistant mutant sr; PTI appears to be central to plant defense against Pcc via recognition by three putative pattern recognition receptors (PRRs; BrLYM1-BrCERK1, BrBKK1/SERK4-PEPR1, BrWAKs). Pcc triggered the upregulation of the jasmonic acid (JA) and ethylene (ET) biosynthesis genes in mutant sr, but auxins and other hormones may have affected some negative signals. Endogenous hormones (auxins, JAs, and SA), as well as exogenous auxins (MEJA and BTH), were also verified as functioning in the immune system. Concurrently, the expression of glucosinolate and lignin biosynthesis genes was increased at 12 hpi in resistant mutant sr, and the accumulation of glucosinolate and lignin also indicated that these genes have a functional defensive role against Pcc. Our study provides valuable information and elucidates the resistance mechanism of Chinese cabbage against Pcc infection.


Introduction
Chinese cabbage (Brassica rapa ssp. pekinensis) originated in central China and is the most widely grown, important vegetable crop in Asia. Soft rot disease caused by the pathogen Pectobacterium carotovorum ssp. carotovorum (Pcc), also known as Erwinia carotovora ssp. carotovora (Ecc), can result in severe losses and is one of the three most economically important diseases of Chinese cabbage. The narrow genetic background of the core collections of Chinese cabbage and the little information available about the molecular mechanism of resistance against Pcc have resulted in very limited breeding material exhibiting resistance to the disease. Pcc is a necrotrophic bacterium with a wide host range 1 and can survive in the soil for several months without the host. It infects the host through natural pores on the plant surface or wounds, and when environmental conditions such as moisture, oxygen, and temperature are conducive, it exists in the vascular tissue, including parenchyma cells 2,3 .
When Pcc invades the host plant, plant cell walldegrading enzymes (PCWDEs) such as polygalacturonase (PGs), pectate lyase (Pel), and cellulase (Cel) are synthesized and secreted from the bacterial cytosol into intercellular spaces of the plant tissue 4 . Pcc employs the Type II secretion system (T2SS), which is the main way that proteins are delivered to host cells and cause soft rot disease 5,6 . The type III secretion system (T3SS) has significant roles by contributing to virulence in hemibiotrophic phytopathogenic bacteria for secreting effectors and transporting virulence factors, but few factors are required for Pcc to attack the host plant. Except for DspE, no T3SS effectors have been identified that elicit plant cell death to promote plant tissue maceration but not to suppress basal defense responses 7,8 . Therefore, the pathogenicity of Pcc does not rely on T3SS to infect host plants 9 .
No resistance genes (R genes) have been identified for Pectobacterium; these genes encode the proteins that can directly and indirectly recognize effectors and elicit defensive reactions against effectors 10 . The R-genemediated immune response, also known as effectortriggered immunity (ETI) 11 , is pathogen specific; intense, inducing programmed cell death (PCD); and causes the hypersensitive response (HR) so that pathogens cannot obtain nutrition from infection plant parts 12 . Except for ETI in the host-pathogen interaction system, the plant's pattern recognition receptors (PRRs) on the surface of cell membranes recognize conserved microbe-or pathogen-associated molecular patterns (MAMPs/ PAMPs). Pathogen pattern-triggered immunity (PTI) 11 is a consequence, and recognition of the pathogen causes a series of host responses, which include eliciting production of reactive oxygen species (ROS), activating the Ca 2+ -mediated and hypersensitive responses, and stimulating the mitogen-activated protein kinase (MAPK) cascade reaction. In addition, the molecular fragments from degradation of the cell wall can act as dangerassociated molecular patterns (DAMPs) and are recognized by PRRs to activate PTI 13 . Specifically, plant cell wall fragments released by the action of the hydrolytic enzymes secreted by Pcc are major elicitors in enhanced immunity toward these pathogens 14 .
Plant hormones have an important role in the regulation of plant growth and development, and they mediate defense responses as signals to pathogens and phytophagous insects 15 . Salicylic acid (SA), jasmonic acid (JA), and ethylene (ET) are primary signals that activate and facilitate immune responses in plants 16 . SA signaling commonly regulates plant defense against biotrophic pathogens, and JA/ET-dependent signaling pathways are required for resistance to necrotrophic pathogens 17 . JA/ ET-dependent signaling pathways have an essential role in resistance to Pcc, but it is unknown whether the SAdependent pathway is required for plant resistance [18][19][20] . Additionally, other hormones, such as auxins, abscisic acid (ABA), gibberellins and cytokinins, are considered modulators of plant-pathogen interactions 21 .
WRKYs are one of the largest families of plant transcription factors, with the conserved WRKY domain regulating plant responses to pathogens. WRKY70 is the key factor in balancing SA-dependent signaling and JAdependent signaling for defense against Pcc 20 . WRKY75 positively regulates JA-or SA-dependent defense 22 , and WRKY33 is a positive regulator of JA-dependent genes but represses the SA-dependent pathway 23,24 . In Arabidopsis, WRKY7 activated the expression of the JA-dependent signaling gene PDF1.2, indicating that WRKY7 is a positive regulatory factor in the JA pathway 25 . Overexpression of encoding pineapple bromelain (BAA1), rice leucine-rich repeat-protein (OsLRP) and polygalacturonase-inhibiting protein 2 (PGIP2) [26][27][28] was reported to improve resistance to Pcc infection.
Plant resistance to Pcc is complex, and little is known about the molecular basis of resistance to this soft rot phytopathogen. The completely sequenced B. rapa genome furnishes exceptional amounts of genetic data 29 that can be used for mutant library research in Chinese cabbage. In our previous research, 5396 mutant plants (M 1 ) were obtained from seeds after treatment with ethyl methane sulfonate (EMS) mutagenesis. All plants were self-pollinated, and 4253 plants produced between 10 and 300 seeds each, which represented the mutant population 30,31 . RNA-Seq is a transcriptome analysis approach using deep-sequencing technology and has replaced previous technologies such as microarrays 32 . RNA-Seq is a more robust method to reveal global gene expression patterns of plant immunity in response to wild-type (WT) and resistant mutant sr soft rot bacterial infection over time. Therefore, the specific objectives in our research were as follows: 1. to create a reliable identification inoculation method for Pcc and obtain resistant mutants against soft rot from our population mutated by EMS; 2. to determine the seminal period corresponding to defense regulation (resistance) to Pcc; 3. to compare the transcript profiles of resistant mutant sr plants to the susceptible WT plants at 0, 6, 12 and 24 hpi (hours postinoculation) in response to Pcc using RNA-Seq to elucidate the putative resistance molecular mechanism operating against Pcc, including the infection process and recognition of the pathogen, signal transduction and synthesized secondary metabolites functioning in the immune system.

Plant materials and bacterial pathogen inoculation
The soft rot-resistant mutant sr was screened from an EMS-mutagenized M 2 population of Chinese cabbage 30,31 and controlled self-pollinated to obtain M 4 generation. All seeds were sown in pots in the greenhouse at 26-28 ℃ with 16 h daytime/15 ℃ with 8 h nighttime and 90% humidity. All samples were collected one week after transplanting.
Pcc pathogen BC1 33 was cultured in LB broth medium overnight in an incubator set at 28 ℃ with continuous shaking (150 rpm). Bacteria were diluted with LB medium to 10 5 cfu/mL for inoculation of plants.
Petioles of the third leaves (from inside to outside) of 7-to-8 leaf plants were lightly scored (through the epidermis) with a sterile scalpel and inoculated with 5-10 μL of a uniform bacterium suspension made from cultures, which were labeled "in vivo" 34 ( Supplementary Fig. 1a). Similarly, the third leaves were cut into 5.5-cm-diameter disks with a homemade tool ( Supplementary Fig. 1) and placed in closed 9-cm-diam petri dishes with two layers of moist filter paper to maintain high humidity. The leaf circles were scored as before, inoculated with 5-10 μL of bacterium suspension, and placed in an incubator (28 ℃, 90% humidity). These cultures were designated as "in vitro" 28 ( Supplementary Fig. 1b).

Harvesting samples and observing disease severity
For RNA-Seq analyses, the leaves that were to be inoculated with WT and sr lines in vivo were harvested 0 hpi (control) and after inoculation (6, 12 and 24 hpi) with three biological replicates. Samples at 0 and 12 hpi in WT and sr were used to determine the concentrations of glucosinolate, lignin and hormones. All samples were frozen immediately in liquid nitrogen and stored at −80 ℃ before analyses.
To accurately evaluate the visible symptoms of Pcc, leaves were inoculated in vivo and in vitro. Disease severity in vivo was scored at 48 hpi because of lower humidity and subsequent disease development compared to in vitro trials. Disease ratings are illustrated in Fig. 1a: 0 (no symptoms), 1 (lesions discrete and <0.5 cm in diam, lignified inoculation spots), 3 (lesions discrete and 0.5-2 cm in diam, lignified inoculation spots), 5 (macerated lesions occupied less than 60% of the petiole), 7 (macerated lesions occupied more than 60% of the petiole), and 9 (macerated lesions occupied the entire petiole and extended to the leaf blade). Plants with disease severity scores of 0, 1, and 3 were categorized as resistant (0 was fully resistant, whereas 1 and 3 were partially resistant); scores of 5 were categorized as partially susceptible; and scores of between 7 and 9 were categorized as susceptible. Lesion diameters on in vitro plants were measured using ImageJ software (National Institutes of Health, USA). The macerated lesions on the leaf disks were scored based on a modification of Park et al. 28 at 24 hpi as disease severity (Fig. 1b): 0 (no symptoms), 1 (discrete lesions <0.5 cm in diam: lignified inoculation spots), 3 (discrete lesions 0.5-1.5 cm in b Fig. 1 The disease grading standard (0, 1, 3,5,7,9) for soft rot resistance of Chinese cabbage seedlings caused by Pcc. a The macerated lesions on the leaves were cored in vivo. b The macerated lesions on the leaves were cored in vitro diameter, lignified inoculation spots), 5 (macerated lesions occupied 25-35% of the entire leaf disk), 7 (macerated lesions occupied 35-50% of the entire leaf disk), and 9 (macerated lesions occupied more than 50% of the entire leaf disk).
cDNA library construction and sequencing data analysis The RNA from three biological replicates of each mutant sr and WT from 0, 6, 12, and 24 hpi was extracted according to the manufacturer's instructions using Trizol reagent (Invitrogen, USA). RNA purity was assessed, and the cDNA library was prepared as previously described 32 .
Raw data (raw reads) in the fastq format were processed and cleaned (clean reads). The clean data were mapped to the B. rapa reference genome (v1.5) from the Brassica database (BRAD) (http://brassicadb.org/brad/) 29 . After filtering the reads, 179.17 Gb of high-quality sequences (more than 96% of the raw reads) of 24 samples (WT and sr at 0, 6, 12, 24 hpi with three replicates) were obtained, ranging from 6.16 to 9.16 Gb per sample, with error rates < 0.1% and 67.60-75.31%; 66.71-74.36% of these sequences were mapped to unique locations, whereas 0.89-1.55% were mapped to multiple genome locations (Supplementary Table S1). A total of 44248 predicted B. rapa genes were annotated.
HTSeq v0.6.1 was used to count the read numbers mapped to each gene, and the FPKM (Fragments Per Kilobase of transcript sequence per Million base pairs sequenced) of each gene was calculated based on the length of the gene and read counts mapped to this gene 35 . Differential expression analyses of two groups were performed using the DESeq R package (1.18.0). The resulting P-values were adjusted to control the false discovery rate (FDR). Genes with an adjusted P-value ≤ 0.05 found by DESeq were considered differentially expressed genes (DEGs). We used KOBAS software to test the statistical enrichment of DEGs in KEGG pathways.

Quantitative real-time PCR (qRT-PCR) analyses
Total RNA was extracted from the same plant samples as those used for RNA-Seq, and first-strand cDNA was synthesized using a ReverTra Ace qPCR RT Master Mix (TOYOBO, Japan) according to the manufacturer's instructions. Bractin was used as an internal reference control, and gene primers were designed by Primer Premier 5.0 software. qRT-PCR analysis was performed on a Lightcycler 96 real-time PCR detection system (Roche, USA) using THUNDERBIRD SYBR qPCR Mix as a fluorescent detection dye (TOYOBO, Japan). The qRT-PCR program was performed in 96-well plates under the following protocol: initial activation at 95 ℃ for 10 min, followed by 45 cycles of 95 ℃ for 10 s, 58 ℃ for 10 s, and 68 ℃ for 10 s. This procedure was followed by melting curve analysis from 95 ℃ for 10 s, 65 ℃ for 60 s, and 97 ℃ for 1 s. The 2 −△△Ct method was used to calculate the relative expression levels of the target genes 36 . All reactions were performed with three biological and technical replicates.

Glucosinolate determination
Glucosinolate was extracted according to the method described by Liao et al. 37 , and compounds were detected using HPLC 38 . Each sample was analyzed with three biological replicates.

Lignin content determination
Lignin was extracted according to the method described by Johnson et al. 39 . Three biological replicates of each of the mutant sr and WT at 0 and 12 hpi were freeze dried and ground into powder. Samples (1.5 mg of DW (dry weight)) were added to 1.5 mL of 20-40% acetyl bromide and 0.2 mL of perchloric acid and maintained at 70 ℃ for 1 h. Afterward, 3 mL of 2 M NaOH and 3 mL of glacial acetic acid were added, and then the entire reaction was diluted to 25 mL with 100% glacial acetic acid. The absorbance of the reactions was measured at 280 nm with a UV-1800 spectrophotometer (Shimadzu, Japan), and the mean amount of lignin was calculated for each sample from five biological replicates.

Hormone treatment in vitro
The third leaves of WT and sr plants were harvested at the same time that in vitro inoculations were completed. Additionally, samples from soft rot-tolerant pak choi ('Huaguan') were collected. Aqueous solutions of the phytohormones (IAA (200 µM), IBA (200 µM), Me-JA (1 mM), and BTH (0.1 mM) 41,42 ) were sprayed onto plants, which were wrapped with a layer of plastic film for 12 h. The film was removed, and the plants were inoculated with Pcc via the previously described protocol for in vitro studies. Controls were treated with sterile, distilled water. After the plants had 7-8 leaves, the petioles of the third leaf (from inside to outside) were wounded, inoculated with 5-10 μL of fresh bacterial suspension as before and identified as in vivo 34 ( Supplementary Fig. 1a). Three biological replicates for sr and WT inoculations were made.

Results
Screening the mutants resistant to Pcc and scoring the disease severity in the M 2 population We randomly chose 800 M 2 plants from 400 different M 1 families to be inoculated with Pcc for in vitro and in vivo studies of Chinese cabbage (Supplementary Table 2). Disease severity was observed at 24 hpi in vitro 28 and at 48 hpi in vivo 34 ( Supplementary Fig. 1). In the M 2 population, the greatest disease grade was 9, and most plants were susceptible to Pcc. The disease severity of WT plants inoculated with Pcc was scored as 9 by both inoculation methods, and all were susceptible to Pcc (Fig. 2a). Only one plant from the M 2 population was evaluated as resistant (disease grade 1) in both in vivo and in vitro methods and thereafter was referred to as sr (Fig. 2b). After 7 days of inoculation with Pcc, the resistant mutant sr plants were still alive; in contrast, WT plants were dead ( Supplementary Fig. 2).

Differentially expressed genes (DEGs) between WT and sr at four time points
A total of 44,248 genes were detected, and their expression was compared between sr and WT. Among these, 616 DEGs were identified at different time points during the plant response to Pcc after inoculation (Fig. 3). The number of DEGs between sr and WT increased from 0 to 12 hpi (36 DEGs at 0 h, 60 DEGs at 6 hpi, 512 DEGs at 12 hpi) and then began to decrease after 12 hpi (23 DEGs at 24 hpi). At 12 hpi, the number of DEGs was the largest, the number of upregulated genes (412) was greater than that of downregulated genes (91), and the expression of defense responses was greater than that at all other time points.

KEGG pathway functional enrichment analysis of the DEGs at 12 hpi
Based on the previous analysis, 12 hpi was the most important defense regulation time point to Pcc in Chinese cabbage. KEGG enrichment analysis was performed between sr and WT at 12 hpi. A total of 391 DEGs were mapped to 72 KEGG pathways and included those KEGG pathways most significantly identified, including several pathways related to immune response against pathogens (Fig. 4, Supplementary Table S3).
To validate the reliability of the resistance-responsive gene expression from RNA-Seq, 16 genes were confirmed based on previous analyses by quantitative realtime PCR using gene-specific primers (Supplementary Table S4).
The expression patterns of the selected resistanceresponsive genes identified by RT-qPCR were largely consistent with the RNA-Seq data ( Supplementary Fig. 3) and indicated that there was a high degree of agreement in the expression patterns between qPCR and RNA-Seq. Glucosinolate in sr and WT was measured at 0 and 12 hpi (Fig. 5). Eight types of glucosinolate were detected, including three aliphatic glucosinolates: 2-hydroxy-3butenyl (PRO), 3-butenyl (NAP) and 4-pentenyl (GBN); four indolic glucosinolates: 3-indolmethyl (GBC), 1-methoxy-3-indolylmethyl (NEO), 4-hydroxy-3indolylmethyl (4OH) and 4-methoxy-3-indolylmethyl (4ME); and one benzenic glucosinolate: 2-phenylethyl (NAS). No significant difference in the concentration of NAS between sr and WT was observed between noninoculated plants and those inoculated with Pcc. However, significant differences in the amounts of aliphatic glucosinolate and indolic glucosinolate between sr and WT were observed. The total content of the two compounds increased in sr and WT when inoculated with Pcc but was significantly greater in sr compared to WT. PRO was the main component of aliphatic glucosinolate and represented the greatest change in sr at 12 hpi. NAP and GBN expression were very low in Chinese cabbage and was reported to be low in B. napus 43 . However, the content of NAP and GBN significantly increased after 12 hpi with Pcc, and GBN significantly increased in sr compared to WT. In contrast, there was no difference between sr and WT at 12 hpi because sr contained more NAP than WT at 0 hpi. The absolute increase was larger in WT, which may be due to PRO generated by the hydroxylation of side chains from NAP in the biosynthesis process 44 . There were no significant differences in four types of indolic glucosinolate before inoculation between sr and WT, but they were induced to increase in sr and WT 12 hpi with Pcc. Among these, GBC and NEO were not significantly different between WT and sr. The 4OH and 4ME forms of indolic glucosinolate increased significantly in sr but not in WT at 12 hpi. Therefore, PRO, GBN, 4OH, and 4ME were determined to be "defense glucosinolate."

Quantitative analysis of lignin after infected Pcc
The acetyl bromide reaction method was used to detect lignin in the proximal petiole (including the infected wound) and in the leaf (excluding the infected wound) in sr and WT. The analyses of sr and WT were carried out at 0 and 12 hpi (Fig. 6). Because the degree of lignification varies in specific tissues, the lignin content in petioles was higher than in leaf blade. The mean lignin content in the blades and petioles significantly increased in both sr and WT at 12 hpi with Pcc, but the rate of increase in sr blades and petioles was 76% and 67%, respectively, and greater than that in WT blades and petioles, at 48% and 47%, respectively.

Comparison of endogenous auxins, JAs and SA in sr and WT
Indole-3-acetic acid (IAA) and its derivatives (methyl indole-3-acetate (ME-IAA), 3-indolebutyric acid (IBA), and indole-3-carboxaldehyde (ICA)) were detected in both sr and WT plants (Fig. 7a). ME-IAA level was not affected in sr or WT after inoculation with Pcc. However, 12 hpi with Pcc, the primary auxin, IAA, decreased in both plant types. Compared to sr, IAA level in WT was higher at 0 hpi and decreased to the same level as sr after 12 hpi. IBA and ICA levels increased under pathogen stimulation in WT but decreased in sr when the resistant host plant was invaded by pathogen.
Four JAs were present in the host plants (Fig. 7b). MEJA and H2JA were constant during the course of disease development, but JA and JA-ILE increased significantly in sr and WT at 12 hpi with Pcc. The JA level significantly increased in the resistant genotype compared to the susceptible genotype. JA-Ile had similar patterns of JA-Ile to the JA patterns in response to Pcc (Fig. 7b). Pathogens triggered the host plant to increase JA biosynthesis in either susceptible or resistant plants during early infection; however, there was a significantly higher expression level in resistant plants 45 . SA levels showed opposite patterns in sr and WT after 12 hpi with Pcc (Fig. 7c). SA levels significantly increased in sr and significantly reduced in WT, although the SA basal level was higher in WT without the pathogen. The SA level was lower in WT compared to sr at 12 hpi and was similar to the level in sr at 0 hpi. In Arabidopsis, the IAA-dependent pathway may have an antagonistic effect on the SA-dependent defense pathway-pathogen interaction 46 . In our study, the SAdependent and IAA-dependent pathways did not show any obvious antagonistic interactions and were opposite to the IBA and ICA patterns.

Effects of exogenous hormone on resistance against Pcc
After the application of exogenous hormones, resistance against Pcc significantly changed in sr, WT and 'Huaguan' (Fig. 8). IBA application significantly enhanced susceptibility of sr and 'Huaguan' compared to application of IAA. JA retarded disease development in WT and 'Huaguan' but did not completely relieve the disease symptoms. The effect of BTH application inhibited symptom development on leaves regardless of the disease grade of plants. IAA and IBA negatively regulated the immune response against Pcc, and IBA significantly promoted disease development and enhanced susceptibility. MEJA and BTH positively affected resistance against Pcc.

The putative resistance mechanism to Pcc in Chinese cabbage
A previous analysis demonstrated that 6 to 12 hpi was the most important defense regulation period against Pcc in Chinese cabbage (Fig. 3), and KEGG enrichment analysis in sr at 0 and 12 hpi revealed the putative mechanism of response to Pcc. In sr, 7747 DEGs (3579 upregulated genes, 4168 downregulated genes) were mapped to 121 KEGG pathways at 12 hpi. Four pathways (glucosinolate biosynthesis, plant-pathogen interaction, plant hormone signal transduction and phenylpropanoid biosynthesis) and related pathways were selected to explain defense mechanisms against Pcc (Supplementary Table S5).
We verified that glucosinolate has an important role in defense against Pcc, as 15 DEGs were enriched in the glucosinolate biosynthesis pathway (Ath00966, 15/19), which is probably a part of defense against pathogen and insect infection in Brassicaceae plants 47 . Thirty-eight genes were involved in the glucosinolate biosynthesis pathway in Arabidopsis, and 87 genes were described in our study. Some of these genes may be homologous to those in Arabidopsis and combined with glucosinolate for defense against Pcc 29,48-50 . In our study, 46 of 87 genes were expressed to synthesize "defensive glucosinolate" in aliphatic, indolic and benzenic glucosinolate pathways through the following three phases: side-chain elongation, core structure formation, and secondary modification. The genes in these phases were regulated by transcription factors (Supplementary  Table S6, and Fig. 9a). Three types of aliphatic glucosinolates (PRO, NAP and GBN) that were classified by side carbon chain length were detected in our study (Supplementary Table S7, and Fig. 9b). n = 4 and n = 5 represented aliphatic glucosinolate with 4 and 5 carbon chains, respectively, in their core structure. The concentrations of PRO, NAP and GBN were stimulated by Pcc, and those of PRO and GBN were significantly higher in sr compared to WT (Fig. 5). GS-OH is responsible for converting NAP to PRO and was upregulated in sr at 12 hpi (Fig. 9c). However, the production of NAP was dependent on AOP2, but three of the AOP2 homologous genes (BrAOP2) were not expressed in our study, and the AOP3 gene was not found in B. rapa. Nevertheless, two of three BrAOP1 genes showed significant changes when plants were challenged with Pcc, and only one BrAOP1 gene (Bra000847) was upregulated by challenge with Pcc. All genes involved in the indolic and benzenic glucosinolate synthesis pathways were significantly upregulated, except for two MYB transcription factors (BrMYB34-Bra029349, BrMYB51-Bra025666, Fig.  9a). The key genes for core structure formation and secondary modification were upregulated, and gene expression level increases were greater in the resistant mutant sr. Limited by the sensitivity of detection technology, only one benzenic glucosinolate (NAS) was formed (Fig. 5). Whether benzenic glucosinolate was produced as a defensive compound is difficult to ascertain. Pfalz et al. 51 demonstrated that multiple genes control secondary modification to form various indolic glucosinolates. However, 1OH-I3M was not detected in our study, and GBC, NEO, 4OH, and 4ME increased only when the plant was infected (Fig. 5). The CYP81F family of enzymes catalyzed GBC in the first step of modification, and CYP81F2, CYP81F3, and CYP81F1 catalyzed GBC to 4OH. CYP81F4 was responsible for the conversion of GBC to 1OH-I3M. 4OH and 1OH-I3M were converted to 4ME and NEO through the function of IGMT1 and IGMT2 (Fig. 9b). Therefore, because most of these genes were upregulated, it was verified that 4OH.  11 . Interestingly, there were no DEGs encoding putative PRRs found at other time points (6 and 24 hpi). Hence, five putative receptors recognized as M/PAMPs or DAMPs to Pcc triggered the defensive response in our study. Chitin elicitor receptor kinase 1 (CERK1, Brcerk1-Bra031293), chitin receptor (LYM1, Brlym1-Bra016402), and leucine-rich repeat receptor-like protein kinase (PEPR1, Brpepr1-Bra003858) are recognition receptors and had significantly higher expression in sr at 12 hpi than in WT (Figs. 10a and 11). Although the other BAK1-LIKE1/SERK4 (BKK1/SERK4, Brbkk1/serk4-Bra040899) genes were not annotated in the pathway, their function may be part of a receptor complex for different D/PAMPs 52 , whose expression was also higher at 12 hpi. The other receptors were WAKs (wallassociated receptor kinases; Brwak2-Bra012273, Brwak4-Bra012272) and had been identified as oligogalacturonide (OG) receptors 53  Ultimately, basic chitinases (PR-3 and PRB1; Brpr3-Bra011464 and Brprb1-Bra013123) were upregulated at 12 hpi and were part of the immune response to Pcc (Fig. 10b).
JAs are produced by a series of enzymatic reactions that begin with α-linolenic acid as the initial substrate, and the expression of several genes was changed at the transcription level in alpha-linolenic acid metabolism downstream regulated genes 54 , which increased in sr but not in WT. However, PDF1.2, HEL, and CHIB are required in the JA/ET signaling pathway to respond to Pcc but were not identified. In our study, SA levels significantly increased in sr but were significantly reduced in WT (Fig. 8c). Unlike JA/ET, there were no SA biosynthesis-related genes found in DEGs. However, WRKY70 (Brwrky70-Bra014692), as a central component in SA signaling, was upregulated to promote the expression of downstream genes in sr but not in WT. IBA and ICA increased in the susceptible WT when inoculated with the pathogen (Fig. 8a). Other evidence showed that exogenous auxin (IAA and IBA) significantly enhanced susceptibility in WT to Pcc (Fig. 9). After inoculation with Pcc, Aux/IAA genes (such as Briaa7-Bra033886, Bra001900; Briaa19-Bra027232, Bra021117) and TIR1 (Brtir1-Bra014378, Bra003518) were inhibited in sr, but seven GH3 family genes (Brgh3.1-Bra039832; Brgh3.2- Fig. 9 The major glucosinolate biosynthetic pathways in Chinese cabbage. a The aliphatic indolic and benzenic glucosinolate biosynthesis pathways by three separate phases (side-chain elongation, core structure formation and secondary modification), which were regulated by transcription factors. b The different types of indolic glucosinolate biosynthesis pathways. c The different types of aliphatic glucosinolate biosynthesis pathways Bra041046; Brgh3.10-Bra034205; Brgh3.12-Bra023403, Bra006194) were upregulated (Fig. 11d). Four GH3 family genes (BrGH3.1, BrGH3.2, BrGH3.10, BrGH3.12) were also upregulated in sr but not in WT at 12 hpi. In contrast, primary auxin (IAA) shared a common biosynthetic pathway with indolic glucosinolate and camalexin, making IAOx a regulatory branch point. CYP79B2 (Brcyp79b2-Bra010644, Bra011821, Bra017871) and CYP79B3 (Brcyp79b3-Bra030246) were upregulated in sr, which promoted the biosynthesis of IAOs in the indole glucosinolate, auxin and camalexin biosynthesis pathways. Indole glucosinolate was synthesized directly from IAOx by CYP83B1 (Brcyp83b1-Bra034941) and was also upregulated. However, there was no significant difference in the expression of Brcyp1a13 and Brcyp71b15, which regulate the synthesis of camalexin. The genes that control the generation of auxin from IAOx are not known. Lignin synthesis pathway genes were enriched in the term biosynthesis of secondary metabolites (Ath01110, 505/995) at 12 hpi. Our study indicated that the expression of genes encoding PAL1 (Brpal1-Bra005221, Bra017210), CCR1 (Brccr1-Bra002236, Bra017580, Bra020021, Bra026090), COMT1 (Brcomt1-Bra012270, Bra015719, Bra016432, Bra025874, Bra025875, Bra035481), and CCoAOMT (Brccoaomt-Bra017624, Bra034600) was upregulated in sr but not in WT (Fig. 10e), and the lignin content of the cell wall of sr was increased after 12 hpi with Pcc (Fig. 6). The accumulation of lignin could provide a positive defense effect against Pcc.

Discussion
Fitness of disease severity scoring method and resistance period in the immune system In this study, the disease severity of soft rot was evaluated in vitro and in vivo for lesion size in Chinese cabbage ( Fig. 1 and Supplementary Fig. 1). Because of low humidity and the speed of disease development, plants could be scored in vivo at 48 hpi 34 and in vitro at 24 hpi 28 , which made the disease severity accurate but also met the requirements of harvesting samples for RNA-Seq analysis. The transcript profiles were investigated with sr and WT at 0 h, 6, 12, and 24 hpi in  response to Pcc using RNA-Seq. The petiole, not the leaf blade, was inoculated and used for in vivo samples. Leaf blades had not been in contact with the pathogen during 6 hpi, and the mutant sr showed the strongest resistance at 12 hpi and remained resistant at 24 hpi. In contrast, WT did not incite protection against Pcc at 12 hpi, and macerated lesions appeared at 24 hpi. Therefore, 6-12 hpi was the initial defense regulation period to Pcc in our study.
The putative immune mechanism of the Chinese cabbage-Pcc interaction Mutated genes for soft rot resistance traits were identified from the F 2 population (two parents: resistant mutant sr and WT) by the MutMap method 55 (data are unpublished). Considering that the F1 plants showed susceptibility to Pcc and that disease severity segregated into susceptibility and resistance at a segregation ratio of 3:1 in the F 2 population, the resistant mutant trait may be controlled by a single recessive locus. A subset of 5 genes having nonsynonymous SNPs was chosen in resistant mutant sr (Supplementary Table S8).
There are three separate modes of action in plant innate immunity responses: ETI, PTI and systemic acquired resistance (SAR), and they are obviously different and closely correlated to interact with pathogens 11 . BTH, as a substitute for SA, maintained a longer chemical effect than did SA and was repeatedly shown to be effective against pathogens by activating the SAR pathway 56 . The effects of BTH application on enhanced resistance were significant against Pcc in susceptible Chinese cabbage and tolerant pak choi (Fig. 8). These results demonstrated that BTH treatment could trigger SAR in the host plant to enhance immunity.
In our study, ETI was not the primary defensive strategy of the host plant against Pcc. However, PTI appears to have a central role in plant defense against Pcc, which is consistent with the review of Davidsson et al. 9 . We found three putative R-structure genes (Bra013144, Bra027047, Bra037141) from DEGs at 12 hpi, but these genes did not occur at other time points, and their expression also increased in the susceptible WT from 0 to 24 hpi (Fig. 5). ETI triggered immune responses with PCD to cause HR and enabled necrotrophic pathogens to acquire more nutrients from dead plant tissues and promote advancement of the infection. However, PCD has the opposite effect on resistance in biotrophic pathogens because it can restrict the growth and colonization of pathogens 9,14 . In our study, the expression of key genes of PTI 57,58 , mitogen-activated protein kinase (MPK), calcium-binding protein (CML), calcium-dependent protein kinase (CPK), respiratory burst oxidase homolog (RBOH), and WRKY33, increased at 12 hpi in the resistant mutant sr but not in the susceptible WT. PTI was triggered by three different PPRs: BrLYM1-BrCERK1 may comprise PGN recognition, BrBKK1/ Fig. 11 Simple schematic diagram of the interactions between Pcc and Chinese cabbage in our study. The major bacterial M/D/PAMPs (Peps, PGNs, and OGs) were recognized by different PRRs (BrBKK1/SERK4-PEPR1, BrLYM1-BrCERK1, and BrWAKs) to activate immune responses. MAPK activation is an important component of PTI signaling. Two major WRKY transcription factors (WRKY25 and WRKY33) are also targets of MAPK phosphorylation, which regulates PR protein activity. JA, ET, and SA were induced to accumulate transduction signals, and auxins were affected as some negative signals. Glucosinolate and lignin, as secondary metabolites, were synthesized and had functional roles in defense against Pcc SERK4-PEPR1 was a receptor complex recognized by BrPeps, and BrWAK2, and BrWAK4 were involved in an immune response against Pcc by recognizing DAMPs such as OGs.
Well-known PAMPs are bacterial flagellin (flg22) and elongation factor Tu (EF-Tu), which are recognized by plant PRRs, such as flagellin-sensitive 2 (FLS2) and EF-Tu receptor (EFR), and trigger plant defenses to induce PTI against different pathogens 59,60 . Interestingly, the expression of PRRs, such as FLS2 and EFR, did not change over the course of the experiments. Chitin is the main wall compound in fungal cell walls that can be hydrolyzed into chitin fragments by plant chitinases as a defensive mechanism. Chitin elicitor receptor kinase 1 (CERK1) recognizes chitin from the fungal cell wall as a PAMP leading to the expression of PTI 61 . Peptidoglycans (PGNs) are gram-positive and gram-negative bacterial cell walls whose structures are similar to chitin found in fungi. PGNs are recognized by AtLYM1 and AtLYM3 combined with AtCERK1 in Arabidopsis to trigger PTI 62 . In Chinese cabbage, Pcc may release PGNs that were recognized by BrLYM1-BrCERK1 and activated genes to protect the host plant from being infected (Fig. 11).
Endogenous small peptides (Pep1-8) act as M/DAMPs and are recognized by PEPR1 and its homolog PEPR2 to activate PTI to pathogens ROS, and ET is also involved in PEPR signaling 63,64 . BRI1-associated receptor kinase 1 (BAK1/SERK3) and its closest paralogue BAK1-Like1/ SERK4 (BKK1/SERK4) are ligands within other PRRs and form complexes contributing to PTI signaling 52 . Similar to FLS2 and EFR, BAK1/SERK3, the closest paralogue to BKK1/SERK4, is also required to elicit PTI to associate with the PEPR-mediated response signaling system in response to AtPeps 64,65 . Therefore, BrBKK1/SERK4-PEPR1 function in a direct role to elicit PTI as part of a receptor complex for some Peps or MAMPs in Chinese cabbage (Fig. 11).
In our study, BrWAK2 and BrWAK4 were identified at 12 hpi and participated in defense against Pcc. WAKs can distinguish and respond to OGs inducing a defense response 53 and are degraded products from pectinderived homogalacturonan released from plant cell walls by PCWDs (such as PGs) and function as DAMPs 66 . WAKs bind to two types of pectin: native pectin regulates cell expansion, and one OG activates the response pathway by the pathogen. The binding of WAKs depends on the affinity for the esterified polymers 67 . One assumption was that different WAKs can distinguish types of pectin or OGs formed by different pathogens, and these two types tend to be recognized by different WAKs. Furthermore, eight genes encoding putative polygalacturonaseinhibiting proteins (PGIPs) were upregulated in sr but not in WT (Fig. 10a). One BrPGIPs gene (Bra005918) was considered a candidate gene harboring one nonsynonymous SNP (leucine to glutamine in an exon) in resistant mutant sr. PGIPs are PG inhibitor proteins of cell wall-degrading enzymes located in plant cell walls 68 . They combine with PG to inhibit the degradation and maintain the integrity of the plant cell wall. The role of PGIPs is defense against fungal pathogens 69 . However, it was also indicated that it may have a potentially important defense role in Chinese cabbage against Pcc 26 . Regardless of which PRRs recognized M/PAMPs or DAMPs to trigger PTI, downstream defense responses (CPK, CML, RBOH MPK3, and MPK4) were strongly activated during infection at 12 hpi in sr (Fig. 10a). In Arabidopsis, PEPR1 and PEPR2 recognized AtPeps to produce ROS 63 and OGs and induced a very strong AtRBOHD-dependent apoplastic ROS burst 70 . These related genes were upregulated in sr, but not in WT, and suggested that PTI had a major role in resistance against Pcc in the mutant. In our study, three copies of WRKY33 were upregulated in sr but not in WT. The WRKY33 transcription factor is a downstream gene for plant resistance to necrotrophic pathogens 24 . Knockout wrky33 mutant plants are highly susceptible to necrotrophic pathogens, but overexpression of WRKY33 increases resistance to Botrytis and Alternaria brassicicola in Arabodopsis 23,24 . WRKY33 is also a specific regulator of the autophagy gene ATG18a, which enables the formation of the degradation autophagosome of cytoplasmic components 57,71 . However, ATG18a, which impacts immune responses significantly against Pcc through PTI immunity, may not be related to autophagy.
Glucosinolate and auxin shared the same branch point but had the opposite effect on the immune response Glucosinolates (GSs) are the products of Brassicaceae species, which are involved in plant defense against insects and pathogens and whose regulatory networks are affected by the plant hormones JA, SA, and ET and by protein kinase and oxidation reduction 72,73 . Indolic glucosinolate is involved in plant growth and defensive responses to pathogens 72,74 . Regardless of the class of glucosinolate, the formation of glucosinolate can be included in the following three separated phases: sidechain elongation, core structure formation, and secondary modification, and the genes in these phases are regulated by transcription factors (Fig. 9a). However, no methylthioalkylmalate synthase family (MAM) genes related to defense after inoculating Pcc at the seedling stage were expressed in our study. This family of genes controls the side-chain length of aliphatic glucosinolate and originates from methionine 75 . Only one gene, BCAT-3 (Brcat-3-Bra029966), was expressed during the side-chain elongation phase and upregulated significantly in sr at 12 hpi, but not in WT. The formation core structure is catalyzed by the CYP79 and CYP83 families that belong to cytochrome P450 enzymes. Our results showed that most CYP79 family genes were upregulated in sr induced by Pcc (Fig. 10a), which is consistent with research on Arabidopsis 74,76 . The gene CYP79F1, which converted the substrate of phenylalanine and methionine to aldoxime 77 , was not expressed, whereas CYP79B2 and CYP79B3 converted tryptophan to indole-acetaldoxime (IAOx), and CYP79A2 participated in the formation of benzenic glucosinolate, which increased in sr at 12 hpi. In our study, the expression of CYP83B1 was upregulated in WT and downregulated in sr. CYP83B1 preferentially uses indole-3-acetaldoxime and aromatic aldoximes as substrates, whereas CYP83A1 acts on aliphatic aldoximes 78,79 . In the side-chain elongation phase, some genes acted on two types of glucosinolate. These genes were also upregulated in sr (except one, Brsur1-Bra036703). In our study, all genes were involved in core structure formation and significantly upregulated in the indolic and benzenic glucosinolate synthesis pathways (Fig. 9a), which implied that indolic and benzenic glucosinolates accumulated and had a functional role in defense against Pcc in Chinese cabbage.
In our study, GS-OH was upregulated in sr at 12 hpi, was responsible for the conversion of NAP to PRO, and explained the accumulated PRO (Fig. 9c). The AOP family has three copies (AOP1, AOP2, and AOP3), and AOP2 and AOP3 were identified as potential genes in the stage of aliphatic glucosinolate modification 80 . However, three AOP2 homologous genes (BrAOP2) were not expressed in our study, and there was no AOP3 gene in B. rapa. Nevertheless, two of the three BrAOP1 genes had significant changes, but only one BrAOP1 gene (Bra000847) had increased expression stimulated by Pcc. The production of NAP and GBN was dependent on AOP2. AOP1 was considered to be ancestral by tandem repeat production to have AOP2 and AOP3, the biological function in the synthesis of NAP and GBN was not clear 81 . CYP81F family genes were responsible for the conversion of indolic glucosinolate (Fig. 9b). Most of these genes that were upregulated verified that indolic glucosinolate increased significantly in sr at 12 hpi but not in WT. No conclusions can be made as to whether benzenic glucosinolate production participated in this pathogen defense. From this evidence, we explicitly suggest that glucosinolate, especially indolic glucosinolate as a secondary metabolite in B. rapa, has a functional role in defense against Pcc.
Demonstrating that resistance to Pcc is due to indolic glucosinolate is difficult because indolic glucosinolates share a common biosynthetic pathway with camalexin and IAA. IAOx is a regulatory branch point that can be degraded into indole acetonitrile (IAN) by CYP1A13, which in turn can be hydrolyzed by nitrilases into IAA and oxidatively decarboxylated into camalexin 82 . Camalexin is a phytoalexin generated by plants under biological or abiotic stress and regulated by cytochrome P450 enzymes CYP79B2, CYP79B3, CYP1A13, and CYP71B15 83 . CYP79B2 and CYP79B3 were upregulated in sr, which promoted the biosynthesis of IAOs in the indole glucosinolate pathway (Fig. 9). There was no significant difference in the expression of Brcyp1a13 and Brcyp71b15, which may suggest that camalexin may not be the reason for induction of defense against Pcc. IAA not only negatively inhibits the response to pathogens but also shares biosynthetic pathways with defense compounds and is elevated after pathogen infection 84 . Whether the homeostasis of IAOx, which IAA and indole glucosinolate shared, was broken, more IAOx flowed to the indole glucosinolate biosynthesis pathway to produce more indole glucosinolate for defense against the pathogen.
After inoculating plants with Pcc, Aux/IAA and TIR1 were inhibited, but some GH3 family genes were upregulated, and the expression pattern was similar to the molecular mechanism of auxin-dependent signaling for defense responses to pathogenesis 46 (Fig. 10d). In contrast to other IAA genes, not all members of the GH3 gene family inactivate IAA, whereas synthetases modify the action of IAA, SA, or JA by conjugating them to amino acids 85 . Endogenous auxins (IAA, IBA, and ICA) and applied exogenous auxins (IAA and IBA) enhanced the susceptibility of plants to Pcc (Figs. 7a and 8). Interestingly, IBA and ICA patterns are opposite to the SA pattern (Fig. 7c). One question proposed is whether the auxin-dependent pathway exerts an antagonistic effect on the SA-dependent defense pathway in plant-pathogen interactions. Four genes (GH3.1, GH3.2, GH3.10, and GH3.12) were upregulated in sr compared with WT at 12 hpi. In Arabidopsis, GH3-12 acted directly on SA or on a competitive inhibitor of SA 86 . However, GH3.2 is suppressed by auxin signaling and does not require activation of the SA or JA signaling pathway in rice 87 . The mechanism of GH3.1 and GH3.10 does not clearly affect the response to any hormone signal 88 . Maybe the difference between rice and Arabidopsis results in different mechanisms of inhibition of auxin-dependent defense or the different members of the GH3 family influence a different response pathway. Our results suggest that disease resistance conferred by the suppression of auxin signaling is involved in the SAdependent pathway to activate the defense against Pcc, but more research is necessary to confirm this hypothesis.
The accumulation of SA, JA, and ET as transduction signals in the defense response SA, JA, and ET signaling pathways are independent but also have complex cross-talk interactions among them and are utilized accurately by different mechanisms in different plant-pathogen interactions to activate immune responses in plants 15,16 . JAs, including jasmonic acid and methyl jasmonate (MeJA), are lipid-derived hormones that regulate plant development, respond to biological and abiotic stresses and have significant roles in disease resistance against necrotrophic pathogens 89 . Pathogens trigger the host plant to increase JA biosynthesis, and there is a significantly higher level of JA expressed in resistant plants 45,90 . JAs are synthesized with a series of enzymatic reactions that begin with α-linolenic acid. LOX, AOS, and AOC are key enzymes involved in the synthesis of JAs, whose expression increased in sr but not in WT (Fig. 10c). The accumulation of JAs after inoculation with Pcc demonstrated their involvement in the immune responses of the host plant (Fig. 7b).
JA/ET signaling pathways interact positively with defense responses against necrotrophic pathogens 14,17 . Although the process of ET biosynthesis involves various regulated enzymes, ACS is largely attributed to the control of ET synthesis via transcriptional regulation and protein expression. In our study, ACS was upregulated in the resistant mutant sr, but not WT, and is similar to other studies 91 . ERF is a common point of the JA and ET pathways and activates JA/ET downstream regulated genes 54 , which were increased in sr but not in WT. The PDF1.2, HEL, and CHIB genes are required in the JA/ET signal pathway to respond to Pcc 18 but were not identified in WT and sr. Because induction of defense gene expression appears to be achieved by a very complicated combination of signals not only from JA/ET but also from some negative pathway effectors such as IAA, it is not possible to discern which hormone signal systemcontrolled defense response is controlled by these genes. In addition, we suspect that the time points chosen in our experiment were earlier than the hormone signal transduction and that the accumulation of JA and ET were synergistically associated with immunity to Pcc.
Resistance against Pcc can be enhanced by the induction of JA/ET-mediated genes, as demonstrated in our study. Interestingly, SA-mediation was also revealed to be an efficient defense against Pcc 19,20 . SA-dependent responses are commonly required for defense against biotrophs 17 . SA increased in the plants following initial infection by pathogens and established SAR with several pathogenesisrelated (PR) genes expressed 92 . In our study, SA levels were significantly increased in sr and concomitantly significantly reduced in WT after Pcc inoculation (Fig. 7c). Furthermore, applications of BTH enhanced resistance significantly against Pcc (Fig. 8). However, cross talk between SA and JA/ET signaling is repressed in the resistant response. WRKY70 is a central component in SA signaling, followed by increased SA and decreased JA signaling, which result in enhanced resistance 20 . In this study, WRKY70 was upregulated in sr to promote downstream genes expressed, but not in WT.
There is apparent controversy regarding whether SA is involved in the response to the Pcc-depressed JAdependent pathway. This could be explained by the different efficacies induced by SA-and JA/ET-dependent pathways. The network of the signaling pathway is extremely complex, and as we expected, a gene could have several roles to defend against pathogens. Furthermore, consistently overlapping the various defensive pathways could be triggered by several genes in different pathways. It is nearly impossible to analyze one gene or signal pathway independently. We suggest that the mechanisms of hormone signals are a joint defense against Pcc and include an induced resistance response that requires JA/ ET-dependent signaling pathways. We further hypothesize that SA-dependent pathways participate in resistance to Pcc and that auxin-dependent pathways interact with JA/ET and SA pathways to inhibit defensive responses.

Lignin protects against further infection in the immune response
Lignin is the natural product for the structural integrity of the cell wall, which has a role in mechanical support and water transportation during the development of plants. In plant defense against damage and disease, lignin is formed to prevent nutrient and water loss and the spread of pathogens from the initial point of attack 93 . Lignin is closely associated with the resistance of plants to pathogens, and increased lignin in plants can enhance this resistance 94 (Fig. 6). The phenylalanine ammonia-lyase (PAL) gene was upregulated in sr but not in WT. PAL is the first enzyme in the phenylpropanoid pathway and is located at the beginning of primary metabolism that leads to secondary metabolism in lignin synthesis. The other genes, including cinnamoyl CoA reductase (CCR), caffeoyl-CoA O-methyltransferase (CCoAOMT), and cinnamyl alcohol dehydrogenase (CAD), were more highly expressed in sr than in WT (Fig. 10e). Our findings are consistent with Zhang et al. 34 and demonstrated that lignin protected the host plant from further infection by Pcc.