Arabidopsis phospholipase Dβ1 modulates defense responses to bacterial and fungal pathogens

Authors


Authors for correspondence:

Xuemin Wang

Tel: +1 314 587 1419

Email: swang@danforthcenter.org

Jian Zhao

Tel: +86 027 8738 5199

Email: jianzhao@mail.hzau.edu.cn

Summary

  • Pathogen infection of higher plants often induces rapid production of phosphatidic acid (PA) and changes in lipid profiles, but the enzymatic basis and the function of the lipid change in pathogen–plant interactions are not well understood.
  • Infection of phospholipase D β1 (PLDβ1)-deficient plants by Pseudomonas syringae tomato pv DC3000 (Pst DC30000) resulted in less bacterial growth than in wild-type plants, and the effect was more profound in virulent Pst DC3000 than avirulent Pst DC3000 (carrying the avirulence gene avrRpt2) infection. The expression levels of salicylic acid (SA)-inducible genes were higher, but those inducible by jasmonic acid (JA) showed lower expression in PLDβ1 mutants than in wild-type plants.
  • However, PLDβ1-deficient plants were more susceptible than wild-type plants to the fungus Botrytis cinerea. The PLDβ1-deficient plants had lower levels of PA, JA and JA-related defense gene expression after B. cinerea inoculation.
  • PLDβ1 plays a positive role in pathogen-induced JA production and plant resistance to the necrotrophic fungal pathogen B. cinerea, but a negative role in the SA-dependent signaling pathway and plant tolerance to infection with biotrophic Pst DC3000. PLDβ1 is responsible for most of the increase in PA production in response to necrotrophic B. cinerea and virulent Pst DC3000 infection, but contributes less to avirulent Pst DC3000 (avrRpt2)-induced PA production.

Introduction

Plants have developed different but interrelated defense systems to battle against invasion by pathogenic microorganisms (Durrant & Dong, 2004; Abramovitch et al., 2006; Jones & Dangl, 2006). One system recognizes and responds to molecules common to many classes of microbes, including nonpathogens. This defense response is initiated by recognition of a microbe- or pathogen-associated molecular pattern (MAMP or PAMP, respectively) by the corresponding pattern recognition receptor, which is typically an integral plasma membrane protein. Another system of defense responds to pathogen virulence effectors, either directly or through their effects on host targets. The effector-triggered defense response involves resistance (R) proteins, usually polymorphic NB-LRR proteins that specifically recognize particular pathogen effectors and launch a specific defense response, such as the Arabidopsis R protein RPS2, which recognizes the Pseudomonas syringae effector AvrRpt2, a type III effector from the Pseudomonas syringae (Glazebrook, 2005; Jones & Dangl, 2006). PAMP-triggered immunity (PTI) and effector-triggered immunity (ETI) are closely related and interact in an array of defense responses, such as the hypersensitive response, the biosynthesis of the signaling molecules salicylic acid (SA), jasmonate (JA) and ethylene, and the production of pathogenesis-related (PR) proteins and phytoalexins (Durrant & Dong, 2004; Abramovitch et al., 2006). Several parallel, yet cross-talking signaling pathways and defense strategies exist, such as SA-dependent systemic acquired resistance in avirulent bacterial infections and JA/ethylene-dependent inducible systemic resistance during necrotrophic fungal pathogen infections (Glazebrook 2005; Spoel et al., 2003; Truman et al., 2007). However, the biochemical pathways mediating the cross-talk of different defense responses are not well understood (Glazebrook, 2005; Jones & Dangl, 2006).

Different aspects of lipid metabolism and signaling play important roles in disease resistance and susceptibility. For example, suppression of α-Dioxygenase1 (α-DOX1), a 16- and 18-C fatty acid-dioxygenase, confers enhanced susceptibility to P. syringae (De León et al., 2002). Mutation of DEFECTIVE in INDUCED RESISTANCE 1 (DIR1), a putative lipid transfer protein, renders plants deficient in systematic acquired resistance (SAR; Maldonado et al., 2002). Mutation of a plastidic stearoyl acyl-carrier protein desaturase, SUPPRESSOR OF SALICYLIC ACID INSENSITIVITY2 (SSI2), results in an increased SA-mediated SAR (Nandi et al., 2004). Mutation of phospholipase A (PLA) suppressor of AvrBst-elicited resistance1 (SOBER1) increases the resistance of Arabidopsis ecotype Pi-0 to Pseudomonas syringae tomato (Pst) DC3000 (avrBst, a type III effector from Xanthomonas campestris pv vesicatoria). This study also indicates a potential cross-talk between the lysophospholipid-producing PLA and the phosphatidic acid (PA)-producing phospholipase D (PLD) pathways in the plant–bacterial pathogen interaction (Kirik & Mudgett, 2009).

PLD, which hydrolyzes membrane phospholipids to generate PA and a free-head group, is involved in cellular processes including reactive oxygen species (ROS) generation (Sang et al., 2001; Park et al., 2004), hormone signaling (Zhang et al., 2004), and disease resistance (den Hartog et al., 2003; Andersson et al., 2006; Bargmann et al., 2006; Yamaguchi et al., 2009). Plant PLD is composed of a family of heterogeneous enzymes with distinguishable catalytic and regulatory properties (Wang et al., 2006). PLDβ1 in Arabidopsis binds to Ca2+ and phosphatidylinositol-4,5-bisphosphate (PIP2) and hydrolyzes phosphatidylethanolamine (PE) preferentially over phosphatidylcholine (PC) (Pappan et al., 1997; Zheng et al., 2000). The expression of Arabidopsis PLDβ1 was induced by both bacterial and fungal pathogen infections (Zabela et al., 2002). Tomato (Lycopersicon esculentum) LePLDβ1 was induced by fungal elicitors, and RNAi knockdown of LePLDβ1 resulted in an increased defense response to fungal elicitors (Laxalt et al., 2001; Bargmann et al., 2006). Knockdown of PLDβ1 in rice (Oryza sativa) increased resistance to Pyricularia grisea and Xanthomonas oryzae pv oryzae. These results indicate that PLDβ1 plays an important role in plant–pathogen interactions and also raise further questions. As a phospholipid-hydrolyzing enzyme, what effect would PLDβ1 have on membrane glycerolipid species without and with pathogen infection? How would a gene-knockout of PLDβ1 affect SA- and JA-mediated defense responses? In this study, we used PLDβ1-knockout and RNAi-suppressed Arabidopsis plants to investigate the effect of PLDβ1 on the interaction of Arabidopsis with the bacterial pathogen P. syringae and the fungal pathogen Botrytis cinerea. The results indicate that PLDβ1 plays a role in promoting PA production after pathogen infection. PLDβ1 plays a negative role in plant resistance to bacterial pathogens but a positive role in the plant response to the necrotrophic fungal pathogen B. cinerea.

Materials and Methods

PLDβ1 T-DNA-insertion knockout isolation and complementation

A T-DNA insertion mutant in PLDβ1 (At2 g42010), designated pldβ1-1, was identified from the Salk Arabidopsis thaliana (L.) Heynh T-DNA knockout collection (Salk_079133) and seeds were obtained from the Arabidopsis Biological Resource Center (ABRC) at Ohio State University. A PLDβ1 homozygous T-DNA insert mutant was isolated by PCR screening using PLDβ1 gene-specific primers 5′-ATA CCC TCC ACC TGA AAC TAA ACC GCA-3′ (forward primer) and 5′-TGTGGCTTATTGTGGAATGCATTACGA-3′ (reverse primer), and the T-DNA left border primer 5′-GCG TGG ACCGCT TGC TGCAACT-3′. Positive T-DNA insertion lines were confirmed by sequencing. The F3 generation shows a homozygous mutation, and deficiency of PLDβ1 transcripts was confirmed by northern blotting. pldβ1-1 showed co-segregation with kanamycin resistance in a 3:1 ratio, indicating that the mutant had a single T-DNA insertion. For complementation of the PLDβ1 knockout mutant, the native PLDβ1 gene, including its own promoter region, was amplified from 1000 bp upstream of the start codon and 300 bp after the stop codon and then was cloned into the pEC291 vector. The primers for PLDβ1 complementation were 5′-ATGGCGCGCCAGATTCTCGTCCACTGAGGA-3′ (forward) and 5′-ATGGCGCGCCTAGAGATGGGCTCTGGAGAT-3′ (reverse). The construct, pEC291-PLDgγ3, was transformed into Agrobacterium tumafaciens strain GV3101 via electroporation. Homozygous pldβ1-1 plants were transformed (Clough & Bent, 1998), the seeds were collected, and transformants were selected on medium containing half-strength Murashige and Skoog (MS) medium, 50 μg−1 ml−1 kanamycin, and 1% agar.

PLDβ1 RNAi construct and generation of RNAi suppression lines

The sense cDNA fragment (492 bp; the last exon plus part of the 3′ untranslated region (UTR), from 2784 to 3276 bp in PLDβ1 mRNA; GenBank accession number U84568) followed by an intron (101 bp; the last intron, from 29 898 to 29 999 bp in Arabidopsis BAC clone T6D20; GenBank accession number U9292382) was amplified using PCR with the forward primer BIR51, 5′-CCCAAGCTTATTTAGAGTGATAATATATC-3′ (HindIII site underlined), and the reverse primer BIR31, 5′-CCGGAATTCAGATCTATGGATACAG AAT-3′ (EcoRI site underlined). Antisense cDNA (492 bp; the last exon plus the 3′-UTR from 2784 to 3276 bp in PLDβ1 mRNA; GenBank accession number U84568) was amplified with the forward primer BIR52, 5′-CCGGAATTCTGGTCAGGTAAATCCCGCAAAC-3′ (EcoR1 site underlined), and the reverse primer BIR32, 5′-CCGCTCGAGATTTAGAGTGATAATATATC-3′ (XhoI site underlined). Two PCR products were digested with EcoRI and then ligated into a fragment of 1100 bp containing two inverse cDNA repeats separated by an intron. This fragment was further digested with the XhoI and HandIII restriction enzymes, and the resulting fragment was purified and then ligated into the pKYLX71-35S2 vector at XhoI and HandIII sites. The resulting RNAi vector was confirmed by complete sequencing and transformed into Arabidopsis by the floral dipping method. F1 and F2 seeds were screened using both kanamycin plates and PCR. Two homozygous lines, RNAi1 and RNAi2, were finally obtained with dramatically decreased PLDβ1 transcripts in leaves, as confirmed by northern blotting.

Pathogen growth and inoculation

Arabidopsis thaliana (L.) Heynh ecotype Col-0, pldβ1-1, RNAi mutants, and pldβ1-1 complemented with PLDβ1 (PLDβ1-complemented) plants were grown in soil in growth chambers at 23 : 21°C and 60–80% relative humidity under 8 : 16 h day : night photoperiods (85 μmol m−2 s−1) for pathogen inoculation or 16 : 8 h day : night photoperiods (120 μmol m−2 s−1) for physiological and genetic analysis. The virulent strain Pseudomonas syringae tomato (Pst) DC3000 and an avirulent strain carrying the avirulence gene avrRpt2, Pst DC3000 (avrRpt2), were grown in King's medium B at 28°C with appropriate antibiotics (50 mg l−1 of rifampicin for Pst DC3000 and 25 mg l−1 of rifampicin plus 50 mg l−1 of kanamycin for Pst DC3000 (avrRpt2)). Five-week-old, soil-grown plants were inoculated with Pst DC3000 and Pst DC3000 (avrRpt2). Bacterial inoculations were performed as previously described (Nandi et al., 2004). The fully expanded leaves of each plant were infiltrated with a suspension containing 107 colony-forming units (CFU) per milliliter of Pst DC3000 or Pst DC3000 (avrRpt2). In parallel, plants similarly infiltrated with 10 mM MgCl2 served as the controls. At 0, 3 and 5 d post inoculation, six infected leaves were collected per genotype to measure the growth of the pathogen. Bacterial counts are expressed as colony-forming units per leaf disc. The rate of increase in the lesion area was determined according to methods described previously (Nandi et al., 2004). Each data point represents three replicates, with three leaf discs per replicate. Approximately 50 plants of each genotype and > 300 leaves were treated.

Botrytis cinerea IMI169558 was cultivated in potato dextrose broth medium at 22°C in a growth chamber. Spores were harvested and infections performed as described previously (Nandi et al., 2004). Briefly, a 10-μl drop of freshly harvested B. cinerea conidial spore suspension (5 × 107 spores ml−1) was placed on three leaves of 4-wk-old soil-grown plants pricked with a 23-gauge needle. The inoculum was allowed to air-dry; plants were covered with a transparent plastic dome and cultivated at 18°C (80–100% humidity) in a growth chamber programmed for a 14 : 10 h light : dark cycle. Control plants were inoculated with potato dextrose broth medium alone. The number of leaves showing different levels of lesion was scored at 5 d post inoculation (Nandi et al., 2004). Approximately 50 plants of each genotype and > 300 leaves were tested.

Extraction and quantification of SA

About 0.3 g of fresh leaves was harvested and quickly frozen in liquid N2. SA extraction and determination were performed according to a method described by Nandi et al. (2004). Briefly, frozen samples were ground and underwent extraction once with 3 ml of 90% methanol and once with 3 ml of 100% methanol. The combined extracts were dried under N2 gas and suspended in 2.5 ml of 5% trichloroacetic acid. The samples were acid-hydrolyzed by adding 200 μl of HCl and incubating at 95°C for 30 min. SA was extracted with 5 ml of a mixture containing cyclohexane : ethylacetate : isopropanol (50 : 50 : 1). The sample was dried under N2 gas and dissolved in 0.5 ml of the mobile phase (69 : 27 : 4 mix of water : methanol : glacial acetic acid). Samples were filtered through a 0.22-μm filter, and 7.5 to 20 μl was used for high-performance liquid chromatography. Samples were passed over a 4.6 × 250-mm C18 reverse-phase column, and SA was eluted with the mobile phase at a flow rate of 0.8 ml min−1. Absorbance of the eluted samples was recorded at 310 nm, and SA concentrations were determined by comparison with SA standards.

RNA blotting and immunoblotting

Five-week-old Arabidopsis seedlings that were treated with pathogens or a control solution (water or 10 mM MgSO4) were used for gene expression analysis. Whole plants were taken for extraction of RNA (Zhao et al., 2011). Twenty micrograms of RNA was loaded and separated by formaldehyde-agrose gel electrophoresis and transferred to a hybridization membrane. 32P-labeled DNA probes for the pathogenesis-related genes PR-1 (At2 g14610), plant defensin 1.2 (PDF1.2) (At5 g44420), PLDα1 (At3 g15730), PLDβ1 (At2 g42010), lipoxygenase 2 (LOX2) (At3 g45140), allene oxide synthase (AOS) (At5 g42650) and PR-5 (At1 g75040) were generated using full-length cDNA of these genes. The immunoblotting for PLDβ1 was performed as described previously (Zheng et al., 2002). Proteins were extracted from Arabidopsis leaves and equal amounts of proteins were subjected to 8% SDS-PAGE, followed by immunoblotting using antibodies against PLDβ1 and PLDα1. The PLD antibodies were raised against the C-terminal 13-amino acid residue peptide (Zheng et al., 2002).

Real-time qRT-PCR

Rosette leaves of appropriate age were inoculated with pathogens. Leaves were harvested before infection and mock treatments (time-point 0) and at the indicated time-points after treatment. Total RNA was isolated with the RNeasy Plant Mini kit (Qiagen, USA). RNA was reverse-transcribed using an M-MLV Reverse Transcriptase kit (Life Technologies, Invitrogen, Grand Island, NY, USA) according to the manufacturer's instructions. Primer pairs listed in Supporting Information Table S1 online were used for real-time quantification by an iQ®5 multicolor real-time PCR detection system (Bio-Rad). Individual PCR reaction mixtures contained 1 μl of diluted cDNA, 10 μl of SYBR Green Mastermix (Thermo Fisher Scientific Inc., St Louis, MO, USA), and 250 μM of each primer in a final volume of 10 μl. In all experiments, three biological replicates of each sample and two technical (PCR) replicates were performed. Data were analyzed with iQ®5 Optical System software by the comparative ΔΔCT method. The amount of target genes was normalized to the abundance of the constitutive ubiquitin 5 (UBQ5) and Tubulin8 genes. Three biological replicates were analyzed, each consisting of eight individually infected leaves.

Detection of inline image and measurement of H2O2

After inoculation with bacterial or fungal pathogens for the time indicated in the figures, inoculated leaves were detached and infiltrated with nitroblue tetrazolium (NBT) for 2 h. The purple formazan precipitate indicates the location and extent of inline image accumulation. For quantitative measurement of H2O2, frozen leaves (c. 0.2 g) were ground to a powder under liquid nitrogen and homogenized with 0.5 ml of 0.2 M HClO4 in a pre-cooled mortar and pestle. The extract was centrifuged at 10 000 g for 10 min at 4°C. The supernatant was collected and neutralized to pH 7.0–8.0 with 0.8 M NH4OH, and briefly centrifuged at 3000 g for 5 min to sediment the insoluble material. After eight-fold dilution, H2O2 concentration was measured with an Amplex Red hydrogen peroxide/peroxidase assay kit (Cat. no. A22188; Molecular Probes, Eugene, OR, USA).

Lipid profiling and JA measurement

The process of lipid extraction, analysis, and quantification was performed as described previously (Welti et al., 2002). Briefly, both bacterial and fungal inoculated leaves were collected at the sampling time and immersed immediately in 3 ml of isopropanol with 0.01% butylated hydroxytoluene (BHT) for 15 min at 75°C to inhibit lipolytic activities. This was followed by the addition of 1.5 ml of chloroform and 0.6 ml of water. After shaking for 1 h, the extracting solvent was transferred to a clean tube. The tissues were extracted with chloroform-methanol five times with 30-min agitation each time. The extracts were combined and washed with 1 M KCl, followed by another wash with water. The solvent was evaporated with a stream of nitrogen. The remaining plant tissues were dried in an oven at 105°C overnight and then weighed for ‘dry weight’, which here refers to the dry weight minus the lipid content. Lipid samples were analyzed with an electrospray ionization triple quadruple mass spectrometer (API 4000; Applied Biosystems, Foster City, CA, USA). The molecular species of each lipid class were quantified in comparison to two internal standards, as previously described (Welti et al., 2002; Devaiah et al., 2006). Five replicates of each treatment for each phenotype were processed and analyzed. The Q-test for discordant data was performed on the replicates of total lipid data or samples. Paired values were subjected to a t-test to determine statistical significance.

JA analysis in B. cinerea-inoculated and control plants was carried out according to the method described previously (Yang et al., 2007; Pan et al., 2008). Leaf samples were harvested at 0, 12 and 24 h post inoculation. About 100 mg of fresh Arabidopsis leaves was sealed in 1.5-ml snap-cap vials and quickly frozen in liquid N2. The leaf tissues were ground into powder and 500 μl of 1-propanol/H2O/concentrated HCl (2 : 1 : 0.002, v/v) with internal standards (50 ng) was added, and the samples were agitated for 30 min at 4°C. One ml of CH2Cl2 was added, and the samples were agitated for another 30 min and then centrifuged at 13 000 g for 5 min. The lower phase (25 μl) was directly infused into a hybrid triple quadrupole/linear ion trap mass spectrometer (ABI 4000 Q-TRAP®; Applied Biosystems) outfitted with an electrospray (ESI) ion source.

Statistical analysis

For most experimental data, including SA and H2O2 measurements, three independent experiments were performed; lipid profiling was performed with five repeats. The data were analyzed using Student's t-test. The differences between two tails of data with the error bars represent 95% confidence limits.

Results

PLDβ1 deletion increased defense response and resistance to Pst DC3000

To characterize the function of PLDβ1, we generated RNAi mutants; two lines, RNAi1 and RNAi2, in which PLDβ1 expression was suppressed to different degrees, were isolated and used in this study (Fig. 1a). RNA blotting indicated that RNAi1 had c. 20–25% of the wild-type level of PLDβ1 transcript, whereas RNAi2 had < 10% of the wild-type level of PLDβ1 transcript (Fig. 1b). Immunoblotting using PLDβ1 antibody also confirmed that both RNAi1 and RNAi2 had a decreased level of PLDβ1 protein (Fig. 1c). We also isolated one knockout line (pldβ1-1) with the T-DNA inserted in the first exon of PLDβ1, 877 bp downstream of the start codon (Fig. 1d). Elimination of the PLDβ1 transcript in pldβ1-1 was confirmed by RNA blotting (Fig. 1d). The three PLDβ1 mutant lines with varying decreases in PLDβ1 transcripts were used for further study. Under normal growth conditions, the PLDβ1-deficient plants displayed no apparent morphological alterations.

Figure 1.

Production and confirmation of phospholipase D β1 (PLDβ1) RNAi knockdown and T-DNA insertion knockout Arabidopsis lines. (a) Construction of the PLDβ RNAi vector. Inverted repeats of exons (gray boxes) and an intron (empty boxes) with restriction sites were cloned in tandem, and their expression was driven by double CaMV 35S promoters and terminated by the nopaline synthase (NOS) terminator. (b) RNA blotting of PLDβ1 and PLDα1 expression in RNAi lines. Equal amounts of total RNA from Arabidopsis leaves were loaded and rRNA detected with ethidium bromide (EtBr) was used as a loading control. PLDα1 protein levels were also used as loading controls. The density of the PLDβ1 RNA band was quantified from three experiments. Data bars represent the mean (± SD) of three repeats. (c) Immunoblotting of PLDβ1 and PLDα1 protein levels in RNAi lines. Equal amounts of Arabidopsis leaf proteins were subjected to SDS-PAGE, followed by immunoblotting using antibodies against PLDβ1 and PLDα1. The band density of PLDβ1 was quantified from three experiments. Data bars represent the mean (± SD) of three repeats. (d) T-DNA insertional knockout of PLDβ1. Left panel, T-DNA insertion position in PLDβ1 with exons (gray boxes) and introns (bars between exons). Right panel, northern blot showing the absence of the PLDβ1 transcript in T-DNA insertion knockout. PLDα1 expression and 28S RNA (detected with ethidium bromide) were used as loading controls. EtBr, ethidium bromide.

We examined the expression of PLDβ1 in leaves treated with several compounds associated with stresses to gain insights into its function. PLDβ1 transcript level was increased in leaves with 0.1 mM methyl jasmonate (MeJA), but decreased with 0.1 mM SA or methyl salicylate (MeSA) (Fig. S1a,b). When leaves were treated with 2 mM H2O2, the expression of PLDβ1 was suppressed in the early phase (3 h) but increased later (20 h). In contrast, abscisic acid (ABA) treatments induced a transient increase in PLDβ1 expression at 6 h. When Arabidopsis leaves were inoculated with pathogens, the expression of PLDβ1 was induced by both virulent (Pst DC3000) and avirulent (Pst DC3000 (avrRpt2)) bacterial pathogens (Fig. 2a, left panel). Infection with the fungal pathogen B. cinerea also strongly induced PLDβ1 expression (Fig. 2a, right panel). In contrast, PLDβ2 was not induced by the bacterial or fugal pathogens, When Col-0, RNAi1 and RNAi2 plants were infected with the pathogen B. cinerea, the PLDβ1 transcript was increased in Col-0 plants, but only small amounts of PLDβ1 transcripts were detected in the RNAi lines (Fig. S2).

Figure 2.

Responses of phospholipase D β1 (PLDβ1) Arabidopsis mutants to bacterial pathogens. Fully expanded leaves of Columbia (Col-0), pldβ1-1, RNAi mutants, and pldβ1-1 complementation (COM) plants were infiltrated with a suspension containing Pseudomonas syringae pv tomato DC3000 (Pst DC30000) with or without avrRpt2. (a) Expression of PLDβ1 and PLDβ2 in Col-0 plants infected with Pst DC3000 pathogens or Pst DC3000 (avrRpt2, a type III effector gene from the Pseudomonas syringae) pathogens (2 × 107 CFU ml−1) for 12 h. Leaves treated with 10 mM MgSO4 were used as a control (mock). The relative expression level of PLDβ1 and PLDβ2 was analyzed by quantitative PCR. Expression levels were normalized with respect to the housekeeping genes UBQ (ubiquitin 5) and Tubulin. Data bars represent the mean (± SD) of three repeats. (b) Defensive response of Arabidopsis plants with altered PLDβ1 expression to virulent and avirulent Pst DC3000 (2 × 107 CFU ml−1). The photographs show representative leaves inoculated with bacterial pathogen for 12 h. (c) Proliferation of the virulent Pst DC3000 (left panel) and avirulent Pst DC3000 (avrRpt2) (2 × 105 CFU ml−1) (right panel) pathogens in Col-0, pldβ1-1 and RNAi plants. Bacterial growth in five leaf discs was measured. Bacterial numbers are expressed as the increase in colony-forming units (CFU) per leaf disc after bacterial inoculation (i.e. log CFU post inoculation – log CFU at time zero). Data represent the average of three samples ± SD. Different letters indicate sample groups with significant differences (< 0.05) from each other.

To investigate the role of PLDβ1 in the plant response to pathogens, we inoculated PLDβ1 mutant and wild-type Arabidopsis plants with virulent Pst DC3000 or Pst DC3000 (avrRpt2). Most plants displayed leaf cell collapse in the pathogen-inoculated leaves within 24 h (Fig. 2b). However, the leaf cell collapse was more evident in the two RNAi lines and pldβ1-1 plants than in wild-type plants, with PLDβ1 mutants displaying both earlier development of cell collapse and larger collapse area (Fig. 2b). The accelerated cell collapse in PLDβ1 mutants occurred with both Pst DC3000 and Pst DC3000 (avrRpt2) (Fig. 2b). Genetic complementation of pldβ1-1 with native PLDβ1 restored the mutant's collapses to the wild-type phenotype, suggesting that ablation of PLDβ1 results in an enhanced resistance to the biotrophic pathogen Pst DC3000 (Fig. 2b). No symptoms occurred in control plants treated only with 10 mM MgSO4. A bacterial growth assay showed that PLDβ1 mutant plants were suppressed in bacterial pathogen growth, and the difference in pathogen growth between pldβ1-1 and wild-type after virulent Pst DC3000 infection was greater than that after avirulent Pst DC3000 (avrRpt2) infection (Fig. 2c).

Both RNAi mutants and pldβ1-1 produced higher levels than in wild-type of total SA (free SA and SA glycoside) in response to Pst DC3000 infections (Fig. 3a) and Pst DC3000 (avrRpt2) infections (Fig. S3). These differences were evident in both inoculated (local) leaves and neighboring (distal) leaves, suggesting a systematic induction of SA production (Fig. 3b). Both quantitative RT-PCR and RNA blotting were used to evaluate defense-related gene expression in treated/nontreated samples. Consistent with the higher level of SA, the SA-inducible pathogenesis-related (PR) gene PR1 was up-regulated in PLDβ1 mutants (Figs 3c, S4). However, PLDβ1 mutants displayed lower levels of expression of the JA biosynthetic genes AOS and LOX2, and the JA/ethylene-regulated defensive gene PDF1.2, than the wild-type (Figs 3, S4).

Figure 3.

Salicylic acid (SA) in pathogen-inoculated phospholipase D β1 (PLDβ1) Arabidopsis mutants. Total SA (free SA + SA glycosides) in the wild type, pldβ1-1, RNAi1, and RNAi2 were determined by HPLC. Data are presented as mean ± SD (n = 3). Different letters indicate sample groups with significant differences (< 0.05) from each other. (a) SA concentrations in inoculated leaves as a function of time post Pseudomonas syringae pv tomato DC3000 (Pst DC30000) inoculation (2 × 107 CFU ml−1). (b) SA concentrations in local and distal leaves from plants infected with Pst DC3000 for 12 h. (c) Expression of defense-related genes in Col-0, pldβ1-1, RNAi1, RNAi2, and pldβ1-1 complementation (COM) plants infected with Pst DC3000 pathogens (pst) or Pst DC3000 (avrRpt2) (Rpt) pathogens (2 × 107 CFU ml−1) for 12 h. Leaves treated with 10 mM MgSO4 (m) were used as a control. The relative expression level of PR-1 was analyzed by quantitative PCR. Expression levels were normalized with respect to the housekeeping genes. Data bars represent the mean (± SD) of three repeats. AOS, oxide synthase; LOX, lipoxygenase; PR, pathogenesis-related; PDF, plant defensin.

PLDβ1 mutants increased susceptibility to Botrytis cinerea

When plants were inoculated with spores of B. cinerea, PLDβ1 mutants displayed accelerated and larger necrotic lesions compared with wild-type plants, as indicated by lesion areas measured at day 5 post inoculation (Fig. 4a). In addition, more infected leaves had necrotic lesions with chlorotic halos (Fig. 4b). qRT-PCR and RNA blotting analyses showed that PLDβ1 mutants had lower levels of expression of LOX2, AOS, and PDF1.2, but a higher level of PR1 expression than wild-type in response to B. cinerea (Figs 4c, S5).

Figure 4.

Responses of phospholipase D β1 (PLDβ1) Arabidopsis mutants to Botrytis cinerea. (a) Leaves of 3-wk-old Columbia (Col-0), pldβ1-1, RNAi, and pldβ1-1 complementation (COM) plants were inoculated with B. cinerea spores (5 × 107 spores ml−1) or water (control). The photographs show representative leaves 5 d post inoculation. (b) Lesions scored at 5 d post inoculation. The lesion development rate was the number of leaves with a necrotic area among the inoculated leaves in each line. Data are presented as mean ± SD (n = 25). Different letters indicate sample groups with significant differences (< 0.05) from each other. (c) Expression of defense-related genes in Col-0 and PLDβ1 mutants infected with fungal pathogens determined by quantitative PCR. Expression levels were normalized with respect to ubiquitin 5 (UBQ5). Data bars represent the mean (± SD) of three repeats. Leaves treated with water were used as a control. (d) JA concentrations in wild-type (Col-0), pldβ1-1, and pldβ1 complementation (COM) plants at various hours post inoculation (hpi) with B. cinerea. C indicates control (i.e. mock-inoculated) and T indicates treated (i.e. pathogen-inoculated). Three-week-old soil-grown seedlings were treated with B. cinerea spores. Values are mean ± SE of three independent replicate experiments. (e) Salicylic acid (SA) in B. cinerea spore-inoculated PLDβ1 mutants. Samples were harvested after inoculation with B. cinerea spores (5 × 107 CFU ml−1) for different times. Total SA (free SA + SA glycosides) in the wild-type, pldβ1-1, RNAi1, RNAi2, and pldβ1-1 complementation plants was determined by HPLC. Data are presented as mean ± SD (n = 3). AOS, oxide synthase; LOX, lipoxygenase; PR, pathogenesis-related; PDF, plant defensin.

We then measured endogenous JA and SA concentrations in PLDβ1-knockout (KO), wild-type, and PLDβ1-complemented plants with or without pathogen inoculation. JA concentrations in these plants increased greatly 24 h after B. cinerea spore inoculation, but the JA concentration of pldβ1-1 was significantly lower than that of wild-type and PLDβ1-complemented plants (Fig. 4d). Meanwhile, SA measurement results suggest that PLDβ1-KO and knockdown mutants all had slightly higher concentrations of total SA after 3 d of spores-inoculation (Fig. 4e), which is consistent with higher PR1 expression. These results suggest that abolishing PLDβ1 impairs fungal pathogen-induced JA production, rendering plants more sensitive to the pathogen.

PLDβ1 mutation increased ROS production in response to pathogens

We examined whether there was a difference between PLDβ1 mutants and wild-type in cellular ROS accumulation during the defense response. When inline image reacts with NBT, a precipitate of purple formazan forms (Rao et al., 2000) (Fig. 5a). Pathogen-treated leaves of the PLDβ1-deficient mutants displayed more intense blue staining than those of the wild-type. The intensity of blue staining was inversely associated with the level of PLDβ1 transcript in the two RNAi mutants and pldβ1-1 (Fig. 5a,b). To verify the association of PLDβ1 deficiency with ROS production, we measured H2O2 concentrations in infected leaves. PLDβ1 mutants generated more H2O2 than wild-type plants in response to infections with the bacterial pathogens Pst DC3000 and Pst DC3000 (avrRpt2) and the fungal pathogen B. cinerea (Figs 5c,d, S6). The enhanced ROS production in PLDβ1-deficient plants was correlated with a stronger defensive response and more severe necrotic responses than observed in wild-type plants.

Figure 5.

Reactive oxygen species (ROS) generation in the Arabidopsis response to pathogen infection. The wild-type (Columbia (Col-0)), phospholipase D β1-1 (pldβ1-1), RNAi1, and RNAi2 plants were inoculated with Pseudomonas syringae pv tomato DC3000 (Pst DC30000) (2 × 107 CFU ml−1) or Botrytis cinerea spores. Data are presented as mean ± SD (n = 5). (a) Nitroblue tetrazolium staining of 5-wk-old Arabidopsis leaves 12 h post inoculation with Pst DC3000 bacterial pathogen (2 × 107 CFU ml−1). (b) H2O2 quantification in Pst DC3000-infected leaves 12 h post inoculation. (c) Nitroblue tetrazolium staining of 5-wk-old Arabidopsis leaves 12 h post inoculation with B. cinerea. (d) H2O2 quantification in B. cinerea spore-inoculated leaves. Different letters indicate sample groups with significant differences (< 0.05) from each other.

PLDβ1 ablation decreased virulent pathogen-induced production of phosphatidic acid

To determine whether ablation of PLDβ1 changed the cellular concentration of PA, we quantitatively profiled phospholipids and galactolipids in Col-0, pldβ1-1, and PLDβ1-complemented plants with or without pathogen treatments. There were no differences in PA concentration among Col-0, pldβ1-1, and PLDβ1-complemented plants without treatment. Inoculation with B. cinerea spores induced significant increases in PA in all three genotypes, but the PA increase in pldβ1-1 was significantly lower than that in Col-0 and PLDβ1-complemented plants (Fig. 6). Concentrations of the major PA species, 34:2-, 34:3-, 36:4-, 36:5-, and 36:6-PA, in pldβ1-1 were significantly lower than those in wild-type plants after the fungal pathogen treatment (Fig. S7).

Figure 6.

Phospholipid and galactolipid contents in leaves of wild-type and phospholipase D β1-1 (pldβ1-1) Arabidopsis plants as affected by Botrytis cinerea, Pseudomonas syringae pv tomato DC3000 (Pst DC30000), or avirulent Pst DC3000 (avrRpt2) infection. Columbia (Col-0), pldβ1-1, and pldβ1 complementation (COM) plants were inoculated with control solutions (-C) or with pathogens (-T). Control solutions were water for B. cinerea spores (5 × 107 spores ml−1) and 10 mM MgSO4 for the virulent strain Pst DC3000 and the avirulent strain Pst DC3000 (avrRpt2) (2 × 107 CFU ml−1). Leaf samples were collected for lipid profiling 24 h post inoculation. Values are mean ± SE (n = 5). Different letters indicate sample groups with significant differences (< 0.05) from each other. PA, phosphatidic acid; PC, phosphatidylcholine; PE, phosphatidylethanolamine, PI, phosphatidylinositol; PG, phosphatidylglycerol; PS, phosphatidylserine, MGDG, monogalactosyldiacylglycerol; DGDG, digalactosyldiacylgly.

The PA increase was also observed when the plants were treated with the virulent pathogen Pst DC3000 (Fig. 6). Pst DC3000 infection significantly increased PA production 24 h post inoculation, but pldβ1-1 had lower PA concentrations than wild-type and PLDβ1-complemented plants after Pst DC3000 infection (Fig. 6). The major PA species, such as 34:2-, 34:3-, 36:4-, 36:5- and 36:6-PA, were significantly lower in pldβ1-1 than in wild-type and PLDβ1-complemented plants treated with virulent bacterial pathogens (Fig. S7). Inoculation with the avirulent pathogen Pst DC3000 (avrRpt2) induced more PA production than did Pst DC3000 and B. cinerea infections, but pldβ1-1 did not show much difference from wild-type or PLDβ1-complemented plants in PA concentrations (Fig. 6). These results suggest that PLDβ1 plays a different role in virulent and virulent bacterial pathogen-induced PAPA production: whereas it contributes significantly to the virulent pathogen-induced PA production, but less to the avirulent pathogen-induced PA under this assay condition.

The concentrations of total phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylinositol (PI); phosphatidylglycerol (PG), and digalactosyldiacylglycerol (DGDG) were comparable in wild-type and pldβ1-1 with or without pathogen infection (Fig. 6). However, infections resulted in different changes in wild-type and pldβ1-1 when lipid molecular species were analyzed (Fig. S8). Treatment with B. cinerea tended to result in higher concentrations of 34:3, 36:5 and 36:6 PC and 34:3 PE, and treatment with Pst DC3000 resulted in higher concentrations of 36:5 and 36:6 PC in the PLDβ1 mutant relative to wild-type (Fig. S8). These higher concentrations of PC and PE molecules in PLDβ1 mutants versus wild-type plants after pathogen treatments were associated with the lower levels of PA production in PLDβ1 than in wild-type. These data might suggest that PC and PE are potential in vivo substrates of PLDβ1.

PLDβ1-KO had elevated lysophospholipid concentrations in virulent pathogen-infected leaves

In contrast to the smaller increase in PA in pldβ1-1, higher concentrations of total lysophosphatidylcholine (LPC), lysophosphatidylethanolamine (LPE), and lysophosphatidylglycerol (LPG) were found in pldβ1-1 than in wild-type and complemented plants after B. cinerea infections, whereas without infection, pldβ1-1 and wild-type displayed no significant difference (Fig. 7, top panel). After virulent Pst DC3000 infection, pldβ1-1 plants also had higher concentrations of LPE and LPG than wild type (Fig. 7, middle panel). However, after infection with the avirulent pathogen Pst DC3000 (avrRpt2), only LPG was significantly higher in pldβ1-1 than in wild-type, and the difference in LPG between wild type and pldβ1-1 was also smaller than that for virulent Pst DC3000-infected plants (Fig. 7, bottom panel). In addition, the avirulent Pst DC3000 (avrRpt2) infection showed an increase in lysophospholipid concentrations in wild-type, pldβ1-1, and complemented plants (Fig. 7, bottom panel). However, no such increase was observed in wild-type and complemented plants after virulent Pst DC3000 or B. cinerea infection (Fig. 7, top and middle panels).

Figure 7.

Concentrations of total LPC, LPE, and LPG in leaves of wild-type and phospho-lipase D β1-1 (pldβ1-1) Arabidopsis plants with and without inoculation with Botrytis cinerea, Pseudomonas syringae pv tomato DC3000 (Pst DC30000), or avirulent Pst DC3000 (avrRpt2). Columbia (Col-0), pldβ1-1, and pldβ1 complementation (COM) plants were inoculated with control solutions (-C) or with pathogens (-T). Control solutions were water for B. cinerea and 10 mM MgSO4 for the virulent strain Pst DC3000 and the avirulent strain Pst DC3000 (avrRpt2). Leaf samples were collected for lipid profiling 24 h post inoculation. Values are mean ± SE (n = 5). Different letters indicate values with significant differences (< 0.05) from each other LPC, lysophosphatidylcholine; LPE, lysophosphatidylethanolamine; LPG, lysophosphatidylglycerol.

The major LPC and LPE species were 18:3, 18:2, and 16:0 species (Fig. S9a,b), whereas 18:3, 16:1, and 16:0 were the major LPG species (Fig. 7d). At 24 h post inoculation with Pst DC3000, both 18:3-LPC and 18:2-LPC were higher in pldβ1-1 than in wild-type and complemented plants (Fig. S9c). In the pldβ1-1 plants, the formation of 16:1-LPG after pathogen inoculation was particularly apparent (Fig. S9c).

Discussion

Our results show that PLDβ1 is a negative regulator of SA-dependent resistance to Pst DC3000, but a positive regulator of the JA-dependent pathway and resistance to B. cinerea. The negative role of PLDβ1 was indicated by the increased production of SA, SA-inducible genes, and decreased damage of virulent bacterial infection. The positive effect of PLDβ1 on the JA-dependent signaling pathway was indicated by decreased expression of JA-biosynthetic and -responsive genes, decreased JA production, and increased damage in response to B. cinerea inoculation. Botrytis cinerea inoculation has been reported to induce a substantial increase in JA (Truman et al., 2007; Yang et al., 2007). However, the expression of PLDβ1 is induced by JA and also by wounding (Wang et al., 2001). Of the four PLDs (α1, γ1, γ2, and β1) examined, the greatest increase for PLDβ1 mRNA occurred 30 min after wounding, whereas that for PLDγ1 and PLDγ2 transcripts occurred at 60 min after wounding and that for PLDα mRNA occurred at 3 to 6 hr after wounding (Wang et al., 2001). However, the wounding-induced JA production in Arabidopsis peaked 1 h after wounding (Wang et al., 2001). These results could mean that PLDβ1 is involved in pathogen- and wounding-induced JA production. The SA-dependent and JA/ethylene-dependent signaling pathways in Arabidopsis contribute to resistance to distinct microbial pathogens (Spoel et al., 2003; Durrant & Dong, 2004; Truman et al., 2007). Yet, a close interplay exists between the two pathways. For example, the NON-EXPRESSOR OF PR GENES1 (NPR1) mutant, which is impaired in SA signaling, produces a higher concentration of JA and shows an increased susceptibility to pathogens relative to wild type (Spoel et al., 2003). The effects of PLDβ1 on plant–pathogen interactions indicate that PLDβ1 and associated lipid changes are involved in the SA-dependent and JA/ethylene-dependent plant defenses against bacterial and fungal pathogens.

Treatments of plants with P. syringae with or without AvrRpm1 or AvrRpt2, rhizobium, fungal pathogens, or elicitors have all been reported to induce PA production (Van der Luit et al., 2000; de Jong et al., 2004; Andersson et al., 2006; Bargmann et al., 2006). However, the specific PLD responsible for the pathogen-induced PA production was unknown. The current study showed that knockout of PLDβ1 decreased the PA production induced by B. cinerea and virulent Pst DC3000 infection, but the loss of PLDβ1 had no major impact on the PA production induced by the avirulent pathogen Pst DC3000 (avrRpt2). The results indicate that PLDβ1 is responsible for a major portion of the PA generated during virulent bacterial and fungal pathogen attack, but it contributed less to avirulent pathogen-induced PA production under the present assay conditions. It is possible that other PLDs, such as PLDα1 and PLDδ, besides PLDβ1, are activated in the plant–avirulent DC3000 (avrRpt2) interaction, thus masking the difference in PA production between pldβ1-1 and wild-type plants. Specifically, PLDδ has been shown to be responsible for most H2O2-induced PA production (Zhang et al., 2003). The avirulent infection induced more H2O2 production than the virulent infection (Fig. S6 versus Fig. 5d), which may result in increased PLDδ activity during the DC3000 (avrRpt2) infection. Thus, these data indicate that other PLDs, besides PLDβ1, are involved in pathogen-induced PA production, and during avirulent infection the other PLDs contribute more, thus overwhelming the effect of PLDβ1 abrogation. However, pldβ1-1 still displayed increased resistance to avirulent DC3000 (avrRpt2) infection, even though no difference in PA concentration was detected between pldβ1-1 and wild-type plants. One possible explanation for this is that the PA produced by other PLDs may play a less important (or a different) role in heightening disease sensitivity than PLDβ1-derived PA. Different PLDs and their derived PA have been shown to play distinguishable roles in the plant response to different stresses (Zhang et al., 2003, 2009).

Unlike the attenuated increase in PA, the concentrations of the lysophospholipids LPC, LPE, and LPG in pathogen-infected PLDβ1-deficient plants were higher than those in wild-type plants. This inverse relationship between PA and lysophospholipids in bacterial infection was also observed with another mutant, SOBER1, which affected Arabidopsis–Pst DC3000 (avrBst) interactions (Kirik & Mudgett, 2009). SOBER1 is a PLA2 that negatively regulates pathogen-induced PA accumulation. Both SOBER1 and PLDβ1 are negative regulators in Pst DC3000 (avrBst) and virulent Pst DC3000 defense, respectively. The data could indicate a cross-talk between the PLD and PLA pathways in plant–pathogen interactions. Such an interaction has also been proposed to occur in the plant response to wounding and senescence (Ryu & Wang, 1996; Ryu et al., 1997). LPE was reported to inhibit PLDα to decrease PA production and delay senescence (Ryu et al., 1997). It is conceivable that lysophospholipids may inhibit PLDβ and other PLDs to suppress PA production (Kirik & Mudgett, 2009). The present findings suggest that the ablation of PLDβ1 results in an increase in PLA activity. It would be of interest in future studies to determine how PLDβ and its derived PA suppress PLA and the production of lysolipids.

PLD and PA are both involved in ROS production and response (Sang et al., 2001; Park et al., 2004; Andersson et al., 2006). PLDα1-derived PA has been shown to directly interact with NADPH oxidase on the plasma membrane and promote ROS production (Zhang et al., 2009). However, the present study showed that PLDβ1-deficient plants produced a lower concentration of PA, but a higher concentration of ROS in response to virulent bacterial and fungal pathogen infection. The same trend has also been observed in PLDβ1-deficient rice and tomato (Bargmann et al., 2006; Yamaguchi et al., 2009). These results suggest that PLDβ1 and its derived PA are unlikely to promote ROS production, and instead they may play a role in mediating the ROS response, similar to that of PLDδ, where ablation of PLDδ rendered plants more sensitive to ROS-promoted cell death (Zhang et al., 2003). PLDβ1 mutants showed an increased hypersensitive response phenotype in response to avirulent or virulent Pst DC3000 infections. The effects may result from increased ROS damage in PLDβ1 mutants compared with wild-type plants. The increased concentration of ROS in PLDβ1-deficient mutants may increase SA biosynthesis, which in turn increases the ROS concentration because SA inhibits catalase activity (Durner & Klessig, 1995; Martinez et al., 2000). Rapid biosynthesis of SA is important for plants to establish systematic resistance to Pst DC3000 (Durrant & Dong, 2004).

Conclusion

The present data indicate that PLDβ1 plays a positive role in pathogen-induced PA production, JA accumulation, JA-dependent defense gene expression, and plant resistance to pathogenesis of the necrotrophic fungal pathogen B. cinerea. In contrast, PLDβ1 suppresses the SA-dependent signaling pathway and defense gene expression, thus negatively affecting plant resistance to infection with biotrophic Pst DC3000. In addition, the results suggest that PLDβ1 may suppress the production of lysolipids, implying a cross-talk between PLDβ1- and PLA-mediated signaling in the plant response to bacterial and fungal pathogens. Further study is needed to determine whether the inverse relationship between PA and lysophospholipid production may underlie a chemical mechanism that modulates plant–pathogen interactions.

Acknowledgements

We thank Mary Roth for lipid analysis. The work was supported by grants from the National Science Foundation (IOS-0818740; MCB-0922879) and the US Department of Agriculture to X.W. Equipment acquisition and method development at the Kansas Lipidomics Research Center were funded by the National Science Foundation (EPS 0236913, MCB 0455318 and 0920663, DBI 0521587), Kansas Technology Enterprise Corporation, K-IDeA Networks of Biomedical Research Excellence (INBRE) of National Institute of Health (P20RR16475), and Kansas State University.

Ancillary

Advertisement