The transcriptome analysis of the Arabidopsis thaliana in response to the Vibrio vulnificus by RNA-sequencing

Because of the recent increase in the demand for fresh produce, contamination of raw food products has become an issue. Foodborne diseases are frequently caused by the infection of leguminous plants by human bacterial pathogens. Moreover, contamination by Vibrio cholerae, closely related with Vibrio vulnificus, has been reported in plants and vegetables. Here, we investigated the possibility of Vibrio vulnificus 96-11-17M, an opportunistic human pathogen, to infect and colonize Arabidopsis thaliana plants, resulting in typical disease symptoms at 5 and 7 days post-inoculation in vitro and in planta under artificial and favorable conditions, respectively. RNA-Seq analysis revealed 5,360, 4,204, 4,916 and 3,741 differentially expressed genes (DEGs) at 12, 24, 48 and 72 h post-inoculation, respectively, compared with the 0 h time point. Gene Ontology analysis revealed that these DEGs act in pathways responsive to chemical and hormone stimuli and plant defense. The expression of genes involved in salicylic acid (SA)-, jasmonic acid (JA)- and ethylene (ET)-dependent pathways was altered following V. vulnificus inoculation. Genetic analyses of Arabidopsis mutant lines verified that common pathogen-associated molecular pattern (PAMP) receptors perceive the V. vulnificus infection, thus activating JA and ET signaling pathways. Our data indicate that the human bacterial pathogen V. vulnificus 96-11-17M modulates defense-related genes and host defense machinery in Arabidopsis thaliana under favorable conditions.


Introduction
Over the past few decades, the availability of fresh foods for optimum human health has received considerable attention. Although suppliers have put tremendous effort into reducing PLOS  the contamination of fresh foods, the number of cases related to contaminated food products has been increasing [1,2]. Contamination of raw food products is mainly caused by bacteria, viruses and parasites. Bacteria are prokaryotic microorganisms that inhabit and adapt to soil, water and acidic hot springs and function as symbionts or pathogens in animals and plants [1]. Human bacterial pathogens including Salmonella [3] and Escherichia coli O157 contaminate fresh fruits and vegetables [4]. Between 2000 and 2008, a number of cases of foodborne diseases caused by Salmonella spp. and norovirus were reported in the US, and Salmonella has been ranked as the major bacterial pathogen responsible for the hospitalization and death of humans [5]. In addition to the direct impact on human health, these pathogens also cause substantial economic losses, as contaminated produce has to be recalled by the supplier. For example, the recall of contaminated spinach resulted in the loss of $350 million in the USA in 2006 [6]. Plants may be contaminated at any stage of growth or during post-harvest processing. Human pathogen population increases both before the harvest of plants [7,8] as well as postharvest [7]. Sterilization and sanitation are used to eliminate the contamination of fresh produce by decreasing the population of human pathogens in plants. However, despite extensive efforts, internalized human pathogens cannot be completely eliminated from plants [9,10]. Thus, understanding the interaction between human pathogens and plants is critical for controlling the outbreak of foodborne illnesses.
To date, V. vulnificus is a Gram-negative halophilic bacterium and a lactose-positive opportunistic human pathogen [11]. V. vulnificus grows in aquatic environments worldwide, especially in warm waters in tropical and subtropical regions [12,13]. V. vulnificus has been classified into three distinct biotypes, based on its lipopolysaccharide (LPS) antigens, expression of capsule and host range [14,15]. Biotype 1 has been predominantly associated with shellfish colonization and human diseases and was initially divided into five LPS groups [16]. Biotype 2 mostly infects marine vertebrates (e.g., eels) and possesses a single type of LPS antigen known as serogroup E [17]. Biotype 3 has been relatively recently updated in Israel fish farms by wound infection with live fish [15]. Infection by V. vulnificus occurs via two main routes: 1) septicaemia resulting from the consumption of V. vulnificus-contaminated raw seafood; and 2) wound infection [18]. Recently, V. vulnificus infection has been reported to cause gastroenteritis [18].
It is known that V. cholerae shares many similarities with V. vulnificus [19]. Moreover, contamination of V. cholerae has been reported in diverse plants and vegetables for a long time [20,21,22,23]. However, despite its ability to cause severe disease, V. vulnificus has not yet been reported to cause any plant disease. This notion prompts us to test the possibility of V. vulnificus causes disease in plants and how it modulates the plant defense responses under artificial and favorable conditions. To investigate further experiments, we first identified the strains of V. vulnificus that clearly contribute to the development of pathogenicity in Arabidopsis in vitro. Among the tested strains, V. vulnificus 96-11-17M was selected for further analyses. Next, we examined the disease severity and pathogen population density in Arabidopsis plants inoculated with this strain in planta. To obtain molecular evidence, we performed RNAsequencing (RNA-Seq) analysis of Arabidopsis plants infiltrated with V. vulnificus 96-11-17M in planta. The results showed that V. vulnificus 96-11-17M infection altered the transcript levels of genes involved in salicylic acid (SA)-, jasmonic acid (JA)-and ethylene (ET)-dependent pathways at the tested time points relative to time zero (control). Furthermore, several differentially expressed genes (DEGs) identified in Arabidopsis plants inoculated with V. vulnificus 96-11-17M significantly overlapped with those identified in Arabidopsis plants inoculated with Pseudomonas syringae pv. maculicola ES4326, a common plant bacterial pathogen. Taken together, our results provide that V. vulnificus modulates defense-related genes and host defense machinery in Arabidopsis thaliana under favorable conditions used in this study.

Plant materials
In vitro. Seeds of Arabidopsis ecotype Col-0 were surface-sterilized with 6% sodium hypochlorite for 5 min and then washed five times with sterile distilled water. The sterilized seeds were plated on half-strength Murashige and Skoog (0.5× MS) medium supplemented with 3% sucrose and 0.6% plant agar. Seeds were germinated and incubated in a growth chamber at 24 ± 2˚C under a 12 h light/12 h dark photoperiod and fluorescent light (7,000 lux light intensity) for 3 days. The germinated seeds were transferred to fresh 0.5× MS plates and incubated for 2 weeks under the same conditions for further experiments.

Disease severity and population in vitro
To perform in vitro assays, 2-week-old Arabidopsis seedlings were separately drop-inoculated with bacterial suspensions of four strains of V. vulnificus including 96-11-17M, A1402, CMCP6 and YJ016 at an optical density at 600 nm (OD 600 ) of 2.0 and with Pseudomonas syringae pv. tomato DC3000 (positive control; OD 600 = 1). Briefly, bacterial strains were picked from each glycerol stock at -70˚C, streaked on LB agar medium containing 2% NaCl (Sigma, CA, USA) and 5 μg/ml rifampicin (Yuhan Corporation, Seoul, South Korea) and incubated at 30˚C for 2 days. P. syringae pv. tomato DC3000 was also steaked on LB agar medium containing 5 μg/ml rifampicin and incubated at 30˚C for 2 days. After incubation, colonies were scripted by plastic bar using sterilized 10% PBS (Bioneer, Daejeon, South Korea). The 10% PBS buffer contained NaCl (13.7 mM), KCl (0.27 mM), Na 2 HPO 4 (1 mM) and KH 2 PO 4 (0.18 mM) at pH 7.4. The inoculated plants were incubated in a growth chamber for 5 days and then scored for disease severity. Plants inoculated with P. syringae pv. tomato DC3000 re-suspended in 10% PBS were scored on a scale from 0 to 5, where 0 indicated no symptoms, 1 indicated the development of yellowish color, 2 indicated chlorosis only, 3 indicated necrosis and chlorosis, 4 indicated partial necrosis of the inoculated area and 5 indicated complete necrosis of the inoculated area [24]. The 10% PBS used a negative control because all strains of bacteria were resuspened in 10% PBS.
For population quantification in vitro assays, leaves at the same stage and age were collected at 5 days after drop-inoculation of V. vulnificus 96-11-17M and P. syringae pv. tomato DC3000, homogenized, spread on LB agar medium containing 2% NaCl and 5 μg/ml rifampicin and LB agar medium containing 5 μg/ml rifampicin, respectively, and finally incubated at 30˚C for 2 days. P. syringae pv. tomato DC3000 was also steaked on and incubated at 30˚C for 2 days. The number of V. vulnificus 96-11-17M and P. syringae pv. tomato DC3000 colonies was counted at 2 days after incubation.

Total RNA extraction, cDNA synthesis and qRT-PCR
Leaves of 3-week-old wild-type plants infiltrated with V. vulnificus 96-11-17M (OD 600 = 1 in 10% PBS) were harvested at 0, 12, 24, 48 and 72 h post-infiltration and immediately frozen in liquid nitrogen. Subsequently, the leaf tissues were ground to a fine powder using a mortar and pestle. Total RNA was isolated from 100 mg ground tissue using RNeasy 1 Plus Mini kit, according to the manufacturer's protocol (Qiagen, CA, USA). First-strand cDNA was synthesized using 2 μg total RNA, oligo-dT primer, dNTPs and Moloney murine leukaemia virus reverse transcriptase (M-MLV RT; Enzynomics, Daejeon, South Korea). To validate the RNA-Seq data, qRT-PCR was performed using a Chromo4 Real-time PCR System (Bio-RAD, CA, USA). Each reaction mixture contained 2× Brilliant SYBR Green Supermix (Bio-RAD, CA, USA), cDNA template and 0.5 μM gene-specific primers. Sequences were amplified using the following conditions: initial denaturation at 95˚C for 10 min, followed by 44 cycles of denaturation at 95˚C for 30 sec, annealing at 60˚C for 30 sec and extension at 72˚C for 42 sec. Gene expression levels were normalized relative to the expression of AtACT2 (GenBank accession no. Q96292). Primer sets used in this study are listed in S1 Table.

Construction of RNA-Seq library
Leaves of 3-week-old wild-type Arabidopsis seedlings were infiltrated with V. vulnificus 96-11-17M (OD 600 = 1 in 10% PBS) and collected at 0, 12, 24, 48 and 72 h post-infiltration; two independent replicates were performed in planta for this experiment. Intact total RNA was extracted from the infiltrated leaves, as described above. The quality and integrity of RNA were confirmed by agarose gel electrophoresis, and unwanted contamination with proteins and solvents was assessed using a Nanodrop spectrophotometer (Nanodrop Technologies, Inc., DE, USA). An RNA-Seq library was constructed using TruSeq RNA Sample Preparation kit v2 (Illumina, CA, USA), according to the manufacturer's guidelines. Briefly, mRNA was purified from total RNA using magnetic beads attached to specific poly oligo-dT primers. The purified mRNA was then fragmented and used as template for cDNA synthesis. The cDNA was ligated to adapters, and the fragments were amplified by PCR. The RNA-Seq library was sequenced to generate paired-end (2 × 100 bp) reads using the Illumina Hiseq-2000 system (Illumina, CA, USA) at the Human Derived Material Center of KRIBB, Daejeon, South Korea.

RNA-Seq data processing and statistical analysis
To analyze the RNA-Seq data, the reference genome sequence of A. thaliana was first obtained from The Arabidopsis Information Resource (TAIR, version 10, assembly ID: TAIR10). The bowtie2-build component of Bowtie2 (ver. 2.0) and SAMtools (ver. 0.1.18) were used for reference genome indexing. Reads from V. vulnificus-infiltrated samples at different time points were mapped onto the reference genome (ver. 2.0) using the Tophat2 reference genome mapper. The RNA-Seq data generated in this study were deposited at the NCBI Gene Expression Omnibus (GEO) public database under the accession number GSE61418 and data were publicly available.
To analyze the transcriptome, a hierarchical clustering algorithm that uses a centered correlation coefficient as the measure of similarity and complete linkage clustering was applied, as described previously [25]. When performing cluster analysis, the value of fragments per kilobase of exon per million reads (FPKM) was calculated for each sample to estimate gene expression levels. The FPKM data were normalized using quantile normalization in R (version 3.0.1). The FPKM values were log2-transformed and median-centered across genes and samples. To assess differences in gene expression patterns between samples, scatterplot matrices were produced using Pearson correlation tests. Gene set enrichment analysis was carried out to identify the most significant gene sets associated with biological processes. The significance of over-represented gene sets was estimated using Fisher's exact test. Gene set enrichment analysis was performed using DAVID Bioinformatics Resources (ver. 6.7).

Statistical analysis
Analysis of variance (ANOVA) of experimental data sets was performed using JMP software version 5.0 (SAS Institute, Cary, NC, USA). Significant effects of treatment were determined based on the magnitude of the F-value (P = 0.05). When a significant F-test was obtained, separation of means was accomplished using Fisher's protected least significant difference test at P = 0.05.

V. vulnificus causes disease in Arabidopsis
To assess whether the human pathogenic marine bacterium V. vulnificus causes disease in A. thaliana, we drop-inoculated Arabidopsis leaves with suspensions of four V. vulnificus strains in vitro. All of the tested strains produced typical disease symptoms in Arabidopsis leaves similar to those caused by P. syringae pv. tomato (Pto) DC3000 (positive control), while no disease symptoms were observed in Arabidopsis leaves inoculated with 10% phosphate buffer saline (PBS) (negative control) (Fig 1A). After selection of candidate strain for further experiments, we inoculated V. vulnificus 96-11-17M onto A. thaliana leaves, and incubated 5 days in the chamber. Inoculated leaves were collected, homogenized and spread on LB agar medium supplemented with 2% NaCl and 5 μg/ml rifampicin. The plates were incubated at 30˚C for 2 days. A colony selected from the plates after incubation was used for preparing bacterial suspension (OD 600 = 2.0). We drop-inoculated with this bacterial suspension onto 2-week-old-Arabidopsis leaves and confirmed the same disease symptom at 5 days after inoculation. To confirm that V. vulnificus colonized Arabidopsis plants, we monitored the total bacterial population density after drop-inoculation with V. vulnificus 96-11-17M and Pto DC3000 in the same age and stage leaves. The population density of V. vulnificus 96-11-17M and Pto DC3000 was approximately 10-fold higher on day 5 than on day 0, whereas the negative control showed no V. vulnificus colonies (Fig 1B, in vitro). For consistent results, we infiltrated with V. vulnificus 96-11-17M and Pto DC3000 in Arabidopsis plants grown in soil; this was accompanied by a 15-fold increase in the total population density of V. vulnificus 96-11-17M and Pto DC3000 (Fig 1B, in planta). These results indicate that V. vulnificus causes disease in A. thaliana.

V. vulnificus modulates the expression levels of Arabidopsis genes
To investigate whether V. vulnificus modulates Arabidopsis plants at the molecular level, we investigated gene expression levels in Arabidopsis leaves infiltrated with V. vulnificus 96-11-17M at 0, 12, 24, 48 and 72 h post-inoculation in planta analysis. The correlation coefficients of all samples were highly similar, suggesting that none of the samples exhibited abnormal or artificial variation in gene expression levels (S1 Fig). Unsupervised hierarchical clustering analysis revealed genes showing significantly different expression patterns at each time point compared with the 0 h time point (control) and classified the genes into four clusters (Fig 2A).
According to the results of heat map and line plot analyses, cluster 1 genes were specifically up-regulated at 48 h post-inoculation, cluster 2 genes were up-regulated at both 12 and 24 h post-inoculation (relatively early time points), cluster 3 genes were down-regulated at tested time points compared with 0 h control and cluster 4 genes were down-regulated from 12 to 48 h post-inoculation (Fig 2B). To validate the RNA-Seq data, we examined the expression of five cluster 1 and cluster 2 genes using quantitative real-time PCR (qRT-PCR). The expression of these genes was relatively high at 48 h or from 12 to 24 h after V. vulnificus inoculation (Fig 2C  and 2D).
Next, we aimed to identify genes showing differential expression at specific time points  Table. We validated the expression levels of the top 10 up-regulated genes using qRT-PCR. The results showed that all 10 genes were significantly induced by V. vulnificus infiltration, and the fold change ratios were clearly comparable to the RNA-Seq data (Fig 3A and 3B).

V. vulnificus is likely perceived by common receptors of Arabidopsis and activates JA and ET signaling
To investigate how V. vulnificus affects the plant defense mechanism for its propagation in planta, we monitored the population density of V. vulnificus in wild-type (Col-0) and mutant plants of Arabidopsis; these knockout mutants lacked genes that encode pattern recognition receptors (PRRs), such as Flagellin Sensing2 (FLS2) and Elongation Factor 2 Related 1 (EFR1), that recognize pathogen/microbe-associated molecular patterns (PAMPs/MAMPs). Higher V. vulnificus population densities were observed in the knockout mutant lines than in wild-type plants at 3 days post-infiltration (Fig 3C), indicating that PAMPs/MAMPs of V. vulnificus such as flagellin and elongation factor are recognized by common PRRs to activate plant innate immunity.
In addition, analyses of RNA-Seq data revealed that genes involved in plant defense hormone-related pathways were up-regulated in V. vulnificus-infiltrated plants (S2 Fig and S6  Table). These results suggested that defense-related hormone pathways should be required for V. vulnificus infiltration. This phenomenon needed to be confirmed by genetic analyses of Arabidopsis; we quantified the bacterial population density in wild-type and mutant Arabidopsis plants at 4 days post-infiltration in planta. The population density of V. vulnificus was the same in the SA signaling pathways NahG transgenic, SA-deficient mutant (sid2) and Nonexpressor of Pathogenesis-Related gene 1 mutant (npr1) lines compared with the Col-0 (control).  By contrast, the bacterial population density was at least 10-fold higher in the JA response mutant (jar1) and Ethylene Receptor 1 mutant (etr1) than in the Col-0 (control) (Fig 3D).

V. vulnificus shares common features with Pseudomonas syringae pv. maculicola ES4326
Because our results suggested V. vulnificus modulated Arabidopsis defense-related genes, we compared the transcriptome data obtained from Arabidopsis plants inoculated with V. vulnificus with transcriptome data obtained from Arabidopsis plants inoculated with P. syringae pv. The results of GO analysis showed that, under the biological process category, genes overlapping between the two transcriptome data sets were classified under response to JA, response to ET, response to stimulus, immune system process, flavonoid biosynthesis, flavonoid metabolism, phenylpropanoid metabolism, developmental process, and wax metabolism and biosynthesis, regardless of the sampling time point (Fig 4B). The results of Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway analysis showed that, in plants sampled at 24 and 48 h, many of these overlapping genes were involved in photosynthesis, plant hormone signal transduction and starch and sucrose metabolism (Fig 4C). The 324 and 323 Arabidopsis genes were significantly up-and down-regulated by both V. vulnificus 96-11-17M and P. syringae pv. maculicola ES4326 at 24 and 48 h post-infection, respectively. Between 324 and 323 DEGs at two time points, approximately 80% of genes were overlapped (S7 Table).

Discussion
Despite the increasing interest in fresh food contamination and rising sea level, there has yet been no evidence suggesting that the marine bacterial pathogen V. vulnificus regulates the transcriptome and physiology of plants. In the current study, we showed that V. vulnificus infection caused disease symptoms in Arabidopsis plants, consistent with the increase in its population density under artificial and favorable conditions. Moreover, V. vulnificus modulated the transcriptome of Arabidopsis plants at certain time points, and a considerable overlap in DEGs was observed between Arabidopsis plants infiltrated with V. vulnificus and those infected with P. syringae pv. maculicola ES4326. These overlapping DEGs were further characterized to obtain potential functional clues that indicate how V. vulnificus modulates Arabidopsis.
In this study, we first asked how V. vulnificus could survive in plants and develop pathogenicity and virulence against plants under natural conditions. To maintain the pathogenicity and virulence of V. vulnificus in Arabidopsis, we artificially maintained the bacterial colonies under favorable conditions on LB agar medium containing 2% sodium chloride, with salinity ranging from 15 to 25 parts per thousand [26,27]. To inoculate and infiltrate plants, bacterial colonies were resuspended in 10% PBS, thus supplying additional NaCl. The resuspended bacterial colonies were subsequently infiltrated into the apoplast, which is the space between the plasma membrane and cell wall, to enable easy penetration into the plant cell by avoiding the cell wall. The results showed typical disease symptoms on Arabidopsis leaves both in vitro and in planta, accompanied by an increase in the population density of V. vulnificus in infected plants (Fig 1B). These results indicate that V. vulnificus acts normally in the plant host under favorable conditions.
To complete its life cycle under unfavorable conditions, V. vulnificus must overcome the immune response of the human body [28]. The plant cell apoplast represents a favorable habitat for pathogens, as it mediates the transport of water and solutes between tissues or organs and provides a carbon dioxide (CO 2 )-rich environment [29]. V. vulnificus uptakes nutrients and converts amino acids to amines and CO 2 under acidic conditions [28,30]. The apoplast undergoes acidification in plant roots, coleoptiles, internodes and leaves [29]. Intriguingly, in this study, parameters such as acidification, salinity and CO 2 levels of the plant apoplast were appropriate for facilitating the infection of Arabidopsis plants by V. vulnificus.
Next, it is important to explain how plants respond to the infection by V. vulnificus. The notion that V. vulnificus acts as a plant pathogen is based on the hypothesis that plants have evolved an elaborate immune system that activates defense pathways in response to V. vulnificus infection. The first line of plant defense is the activation of PRRs localized in the plasma membrane upon the recognition of PAMPs/MAMPs [31]. To investigate whether the common PAMP receptors FLS2 and EFR1in Arabidopsis are involved in the perception of flagellin and elongation factor, respectively, and thus in the activation of the plant defense response, we quantified the population density of V. vulnificus in wild-type and PAMP receptor mutant plants of Arabidopsis. The bacterial population density was 10-fold higher in the mutants than in wild-type plants (Fig 3C). These results indicate that the human pathogen V. vulnificus is recognized by plant receptors and acts as a pathogen in Arabidopsis under favorable conditions.
The plant hormones SA, JA and ET play a pivotal role in plant defense against pathogens. Biotrophic pathogens that activate programmed cell death in plants are dependent on SA signal transduction, whereas necrotrophic pathogens rely on JA and ET signaling pathways [32]. In agreement with the above notion, genes involved in SA, JA and ET signaling pathways were highly up-regulated in V. vulnificus-infiltrated plants at all time points (Fig 3A and 3B and Fig  4). This suggests that V. vulnificus modulates plant defense-related hormone signaling pathways. This phenomenon was confirmed by genetic analyses of Arabidopsis jar1 and etr1 mutants showing that the bacterial population density was higher in these mutants than in the wild type (Col-0; Fig 3D). Thus, gene expression analysis and genetic analyses indicate that V. vulnificus modulates plant defense response through JA-and ET-dependent signaling pathways.
In conclusion, V. vulnificus clearly produced disease symptoms in Arabidopsis leaves both in vitro and in planta under artificial and favorable conditions. We propose that V. vulnificus infects plants by activating common PRRs and regulates defense-related gene expression and plant defense responses (Fig 5). Our results revealed several intriguing aspects of how V. vulnificus to regulate Arabidopsis under favorable conditions. Here, we describe for the first time the possible mechanism of infection of Arabidopsis by V. vulnificus, thus expanding the current knowledge on the pathogenicity of V. vulnificus. If we verify the communication between V. vulnificus and A. thaliana under natural conditions in further investigation it should prompt public health officials to pay more attention to the contamination of fresh vegetables and crops by V. vulnificus, which may be as important as infection with Salmonella and E. coli O157. In addition, the evidence provided in this study contributes to the screening and elucidation of virulence factors of V. vulnificus alternatively using plant systems, which can ultimately be used to eradicate animal and human diseases caused by V. vulnificus. Taken together, our results reveal a new paradigm that a human pathogenic marine bacterium modulates defense responses in Arabidopsis under artificial and favorable conditions.