Proteome analysis of the Albugo candida-Brassica juncea pathosystem reveals that the timing of the expression of defence-related genes is a crucial determinant of pathogenesis.

White rust, caused by Albugo candida, is a serious pathogen of Brassica juncea (Indian mustard) and poses a potential hazard to the presently developing canola-quality B. juncea industry worldwide. A comparative proteomic study was undertaken to explore the molecular mechanisms that underlie the defence responses of Brassica juncea to white rust disease caused by the biotrophic oomycete Albugo candida. Nineteen proteins showed reproducible differences in abundance between a susceptible (RH 819) and a resistant variety (CBJ 001) of B. juncea following inoculation with A. candida. The identities of all 19 proteins were successfully established through Q-TOF MS/MS. Five of these proteins were only detected in the resistant variety and showed significant differences in their abundance at various times following pathogen inoculation in comparison to mock-inoculated plants. Among these was a thaumatin-like protein (PR-5), a protein not previously associated with the resistance of B. juncea towards A. candida. One protein, peptidyl-prolyl cis/trans isomerase (PPIase) isoform CYP20-3, was only detected in the susceptible variety and increased in abundance in response to the pathogen. PPIases have recently been discovered to play an important role in pathogenesis by suppressing the host cell's immune response. For a subset of seven proteins examined in more detail, an increase in transcript abundance always preceded their induction at the proteome level. These findings are discussed within the context of the A. candida-Brassica juncea pathosystem, especially in relation to host resistance to this pathogen.

Introduction

The oomycete pathogen Albugo candida causes the globally devastating white rust disease of mustard (Brassica juncea) and some types of rapeseed (e.g. Brassica rapa, Brassica campestris) (Sokhi and Khangura, 1992). Filamentous oomycetes are considered to have descended from an aquatic non-fungal ancestor and belong to the kingdom Stramenopiles (along with brown algae and diatoms) (Sogin and Silberman, 1998). Despite the separate phylogeny of oomycetes and true fungi (kingdom Eumycota), the biotroph A. candida causes plant disease in a manner similar to the biotrophic rust fungi, including penetration by one or more germtubes arising from asexual inoculum via stomata, mesophyll colonization of a compatible host tissue, and sporulation within spore-bearing pustules. Hence, the disease caused by A. candida has been referred to as white blister rust (Holub ). The disease is characterized by the formation of white to cream-coloured zoosporangial pustules on cotyledons, leaves, stems, and inflorescences, with staghead galls formed as the result of inflorescence infection (Verma and Petrie, 1980). The combined infection of leaves and inflorescences causes yield losses of 20–60% (Barbetti, 1981; Bernier, 1972; Lakra and Saharan, 1989). As deployment of host resistance offers the most cost-effective and reliable means of managing this disease, studies have been undertaken to identify sources of resistance (Li , 2008). In particular, the recent identification of race 2V of A. candida in Australia (Kaur ), has opened the way for investigation into the mechanism(s) associated with host resistance against this highly virulent pathogen race.

Plants exhibit numerous responses when challenged by pathogens. Some of these involve the activation of host defence genes that bring about physical and biochemical changes in the host (Jones and Dangl, 2006). For example, during incompatible host–pathogen interactions, damage caused by the pathogen remains restricted as a result of the plant's defensive response. This response is associated with a co-ordinated and integrated set of metabolic alterations (Van Loon, 1997; Epple ). During this process various proteins are induced which are collectively referred to as ‘pathogenesis-related (PR) proteins’. These PR proteins are defined as proteins coded for by the host plant but induced specifically in pathological or related stress situations (Antoniw and Pierpoint, 1978; Van Loon ). They not only accumulate locally in the infected tissues, but are also induced systemically and are associated with the development of systemic acquired resistance (SAR) against further invasion by plant pathogens. The details of the biochemical and molecular events occurring in host plant tissues during disease progression are not clear in the A. candida–Brassica juncea pathosystem, with no study describing the molecular changes in the host that accompany infection as a result of either a compatible or incompatible interaction.

The identification and characterization of host proteins whose abundance is affected by pathogen challenge is important in elucidating their roles in mediating the host response. In this study, proteome-level differences between a white-rust susceptible (RH-819) and a resistant (CBJ-001) variety of B. juncea, selected after screening of germplasm from Australia, India, and China, were examined at different time points following inoculation with A. candida. Two-dimensional gel-electrophoresis and mass spectrometry were utilized to identify 19 differentially expressed proteins followed by quantitative real-time reverse transcriptase (qRT)-PCR to confirm the observed proteome changes for seven of those at the transcriptome level. The present study describes for the first time the possible involvement of a thaumatin-like protein (PR-5) in the resistance mechanism in B. juncea towards A. candida. A differential peptide match to the peptidyl-prolyl cis/trans isomerase (PPIase) isoform CYP20-3 that belongs to the cyclophilin C-Cyp subfamily was found only in the susceptible variety and showed no detectable expression at the protein level in the resistant variety. Its abundance was up-regulated in the susceptible variety in the presence of the pathogen as compared to mock-inoculated plants. While cyclophilins have been reported to play an important role in pathogenesis by suppressing the host cell's immune response (Kromina ), to date they have not been reported to be involved in the B. juncea-Albugo candida pathosystem.

Materials and methods

Plant growth

Five B. juncea varieties highly resistant to A. candida, three of which showed some low level of hypersensitive response (HR), and six highly susceptible varieties were selected from the screening trials of Li ) (see Supplementary Table. S1 at JXB online). Seed was sown into steam-treated potting mix (composted pine bark:cocoa peat:river sand, 2:1:1 by vol.) in 8-cell plastic trays. Plants were grown at 13/18 oC night/day temperatures with a 16 h photoperiod and a light intensity of 520 μmol m−2 s−1.

Pathogen culture and inoculation method

Cotyledons of B. juncea variety RH-819 were infected with a clonal isolate of A. candida race 2V isolated from B. juncea (Kaur ). The white rust pustules were collected 10 dpi and stored at –80 °C. For use as an inoculum, zoosporangia were dispersed from infected cotyledons into deionized water and filtered through cheesecloth to remove plant debris. The concentration of the zoosporangia was determined using a haemocytometer and adjusted to 105 zoosporangia ml−1. Fully expanded cotyledons from seedlings 10 d after sowing were inoculated by spotting 10 μl of the zoosporangial suspension onto the adaxial surface of each of the two lobes of each cotyledon. Plants were then subjected to 4 d of enhanced humidity (=95% RH). As a control, cotyledons were mock-inoculated with 10 μl deionized water and seedlings treated as those inoculated with A. candida. Disease was assessed on a 0 (resistant) to 9 (highly susceptible) scale, based on the percentage of cotyledon surface area covered with pustules (Williams, 1985). Inoculated cotyledons were collected for further analysis, immediately frozen in liquid nitrogen, and stored at –80 °C. The cotyledon stage was specifically chosen as the disease severity on cotyledons has been significantly and positively correlated with disease progression in mature plants in work by Li ).

Protein extraction

Cotyledon tissue (1 g) was ground to a fine powder in liquid nitrogen before homogenizing in 10 ml cold 20% (w/v) trichloroacetic acid in acetone containing 0.2% (w/v) dithiothreitol (DTT). Supernatants were incubated overnight at –20 °C to precipitate the protein completely before collecting by centrifugation at 30 000 g for 20 min at 4 °C. The pellet was washed three times with ice-cold acetone containing 0.2% (w/v) DTT, dried at room temperature for 30 min, and resuspended in 1 ml of rehydration buffer [2 M thio-urea, 7 M urea, 2% (w/v) 3-[(3-cholamidopropyl) dimethyl-ammonia]-1-propanesulphonate hydrate (CHAPS), 10% (w/v) phenylmethylsulphonyl fluoride (PMSF), and 0.2% (w/v) DTT]. The sample was mixed using a vortex mixer and incubated with orbital shaking for 16 h at 4 °C to solubulize the protein completely. The insoluble material was removed by centrifugation as above. The concentration of the protein in the supernatant was determined (Peterson, 1983) and samples were stored at –80 °C. Extracts from pooled cotyledon tissue were prepared from at least three independent inoculation experiments.

Two-dimensional gel electrophoresis

The method described by Yajima for 2D separation of proteins was used with some modifications. A solution containing 500 μg of total protein in 0.04% (w/v) ampholytes in a total volume of 200 μl was used for passive rehydration of 11 cm immobilized pH gradient strips (IPG), (ReadyStrip IPG strips pH 4–7, Bio-Rad, USA). The IPG strips were exposed to a total of 30 kVh using the IEF program (PROTEAN IEF unit, Bio-Rad): 300 V in linear mode for 1 h, 8 kV in linear mode for 3 h, and 8 kV in rapid mode until 30 kVh were reached, before storing at –20 °C overnight. Proteins in the IPG strips were reduced by gently agitating the strips in 6 M urea, 2% (w/v) SDS, 0.375 M TRIS-HCl, pH 8.8, 20% (v/v) glycerol, and 130 mM DTT at ambient temperature for 30 min, replacing with fresh solution once. Proteins were alkylated in 6 M urea, 2% SDS, 0.375 M TRIS-HCl, pH 8.8, 20% (v/v) glycerol, and 135 mM iodoacetamide (Aldrich, USA), with gentle agitation at ambient temperature for 30 min, replacing the solution once. Each IPG strip was embedded at the top of an SDS-PAGE in 0.5% (w/v) low-melting point agarose in 1× TRIS-Glycine-SDS (TGS) buffer (Aldrich, USA). Precision wide-range standard proteins (Sigma, USA) were used for molecular mass determinations. Proteins were separated by electrophoresis in 1× TGS buffer (Protean II XI, Bio-Rad) at 150–200 V constant current. Gels were stained for at least 3 h with Coomassie Brilliant Blue R250 (Aldrich, USA) and destained according to the manufacturer's instructions.

2D gel analyses and protein identification

Images of the 2D gels were acquired (GS-800 calibrated densitometer, Bio-Rad, USA) and recorded using a red filter (595–750 nm) and a resolution of 36.3×36.3 μm before analysis (PD-Quest software, Bio-Rad). The software allowed the creation of match-sets where replicate gels could be compared with corresponding gels from other treatments. The analysis was re-evaluated by manual validation to include missing spots and to eliminate artefacts. The mean value for the normalized spot volume and the standard deviation (SD) were determined for each spot. Each set of 2D gels was analysed using the Student's t test tool of the software to determine the statistically significant differences (P =0.05) in protein amount between pathogen-inoculated and mock-inoculated cotyledons. Among the differentially staining spots, only those protein spots that were reproducibly unique to only one variety (susceptible or resistant) or that showed reproducibly altered staining intensities compared with the control gel images (obtained from cotyledons mock-inoculated with DI water) in response to pathogen infection were quantified (Spot quantification tool, PD-Quest software, Bio-Rad, USA).

Protein spots were removed from the gels, destained, trypsin digested, and analysed by mass spectrometry (MS) (4800 MALDI TOF/TOF Analyzer, Proteomics International, Perth, Australia) according to the method of Bringans . Mass spectra were analysed using peptide sequence matching software (Mascot, Matrix Science Ltd.). For the PR-5 protein, MS de novo sequence analysis was performed by further analysis of the spectra (PEAKS Studio, Version 4.5, SP2 program, Bioinformatics Solutions, Waterloo, ON, Canada) (see Supplementary Table. S3 at JXB online). The ion score automatically calculated by the software was –10×log (P) where P was the probability that the observed match was a random event. Ion scores exceeding the threshold value for a random event indicated sequence identity or extensive homology (P <0.05) (Sharma ). The identity of the spot was selected as the protein that produced the highest score and, consequently, the best match with its peptide sequence. For the MS peptides identified, database searches (BLAST, National Center for Biotechnology Information, USA) were used to create protein sequence alignments. Searches were performed over the full molecular weight and pI range and no species restriction was applied. Since the full genome sequence of B. juncea is not yet available, only two sets of peptides could be directly matched to B. juncea protein accessions. The rest of the peptide sets were matched to the closest protein sequence in another species (see Supplementary Table. S3 at JXB online).

Quantitative real-time PCR (qRT-PCR) primer design

Where available, B. juncea cDNA sequences were used for primer design. When B. juncea peptides could only be matched to homologues in related Brassica species, it was assumed that the corresponding cDNA sequences would have sufficient identity with those from B. juncea to be useful in qRT-PCR (Sharma ). To ensure maximum sequence complementarity, primers were placed in those regions of the cDNA that matched the B. juncea peptide sequences. Primers were designed (Primer Express 3 software, Applied Biosystems) to generate amplicons of approximately 60 bp (see Supplementary Table. S2 at JXB online) using the default programme settings. Primer sequences were cross-checked by database searches (National Center for Biotechnology Information) to ensure that only the target of interest would be amplified in PCR reactions. All primers were tested on B. juncea cDNA and the products analysed by melt-curve analysis and agarose gel electrophoresis to ensure that a single PCR product was formed. The PCR efficiency of each primer pair was calculated (see Supplementary Table. S2 at JXB online) (LinRegPCR 7.0 program; Ramakers ).

qRT-PCR

Total RNA was extracted (RNeasy Plant Mini Kit, Qiagen, Australia) from pooled cotyledon samples of the inoculated and mock-inoculated plants and treated with DNase I (RNase-free DNase I, Qiagen) to ensure the absence of contaminating genomic DNA according to the manufacturer's instructions. First strand cDNA was synthesized using oligo(dT) primers (iScript cDNA synthesis kit, Bio-Rad) according to the manufacturer's directions. qRT-PCR was performed (ABI 7500 FAST cycler, Applied Biosystems) using SYBR green (Power SYBR Green master mix, Applied Biosystems) following the manufacturer's standard protocol. The relative abundance of mRNAs was expressed as 40–ΔCt, with ΔCt being the difference in the threshold cycle (Ct) number of the gene-of-interest and a reference gene with 40 being randomly chosen since it represents the highest possible Ct value (Bari ). The actin gene (B. napus) (GenBank accession no. AF111812) (Yang ) was used as the reference gene in all experiments.

Statistical analyses

For statistical analyses of germplasm screening data a single factor analysis of variance was conducted (8th edn, Lawes Agricultural Trust, Hertfordshire, UK). Fisher's least significant difference (l.s.d) at 95% significance level was used to test the differences among genotypes. The correlation between the two screening trials was tested by regression analysis. For statistical analyses of proteomic data the Student's t test was used and a significance level of (P <0.05) applied to test for differences.

Results

Identification of variable resistance to A. candida in B. juncea germplasm

Inoculation of A. candida onto the cotyledons of 14 B. juncea genotypes (see Supplementary Table. S1 at JXB online) caused disease symptoms to occur 10 days post-inoculation (dpi). White rust pustules formed on the abaxial surface of the cotyledons, with chlorosis appearing on the corresponding adaxial surface. Highly susceptible varieties showed smaller numerous pustules on both sides of the cotyledon that ultimately coalesced into larger pustules. By contrast, the resistant varieties showed only pinhead-sized pustules. First visible symptoms appeared at 10–11 dpi, but the assessments were conducted at 14 dpi to be able to distinguish clearly between different degrees of resistance. The tests were repeated and the results pooled for analysis. Regression analysis demonstrated a strong correlation between the two tests (y=0.4039x+0.285; P <0.05; r=0.61; n=14). Significant differences (P <0.001; Least Significant Difference 2.44) were found among the 14 genotypes in relation to cotyledon infestation by A. candida by 14 dpi (see Supplementary Table. S1 at JXB online). Based on the disease scoring system used, different levels of host response were identified among the genotypes, with Genotype RH-819 from India and CBJ-001 from China being the most susceptible and resistant cultivar, respectively.

Pathogen-induced changes occurring in CBJ-001 and RH-819 plant proteomes

To understand the biochemical and molecular events occurring in the host tissues in the A. candida–B. juncea pathosystem, proteome-level differences between white rust-susceptible and -resistant varieties of B. juncea were examined as the A. candida infestation progressed up to 72 hours post-inoculation (hpi). The CHAPS soluble protein fractions of cotyledons of susceptible (RH-819) and resistant (CBJ-001) varieties of B. juncea were characterized in response to A. candida using two-dimensional gel electrophoresis (2DE). Plants from both cultivars were either inoculated with the pathogen or mock-inoculated in triplicate. Tissues were harvested and the protein samples from each biological replicate for both the mock- and the pathogen-inoculated cotyledons separately subjected to isoelectric focusing (IEF) followed by SDS-PAGE. Each of the six 2DE gels for each treatment pair showed hundreds of proteins separated according to their molecular weight (MW) and isoelectric point (pI) (Fig. 1A, B), thus providing an overall picture of pathogen-induced changes that occurred in CBJ-001 and RH-819 plant proteomes.

Fig. 1.

2D gel analyses of Brassica juncea cotyledon samples from a resistant (CBJ 001) and susceptible (RH-819) variety in response to inoculation with Albugo candida. (A) representative image of the resistant variety of B. juncea cotyledon total proteins at 24 hpi separated by 2-DE and visualized with Coomassie Blue (proteins spots, selected for MS/MS analyses, having intensities that were altered significantly (P <0.05) as a result of pathogen challenge are indicated by arrows and number). (B) Representative image of the susceptible variety of B. juncea cotyledon proteins at 24 hpi separated by 2-DE and visualized with Coomassie Blue (proteins spots, selected for MS/MS analyses, having intensities that were altered significantly (P <0.05) as a result of pathogen challenge are indicated by arrows and number). (C) Closer view of six spots either present in the resistant or susceptible variety and showing significant changes between the mock-inoculated control (C) and the pathogen-challenged seedlings (I).

Identification of differentially expressed proteins following pathogen infection

After normalization of spot intensities, comparison of gels showed a total of 19 proteins with reproducible differences in abundance between the pathogen-inoculated and mock-inoculated control samples from RH-819 and CBJ-001 B. juncea varieties. Some of the differences in abundance were due to the presence or absence of a protein in either the resistant or the susceptible variety, while other differences were quantitative (Fig. 1A, B; Table 1). In the CBJ-001 resistant variety, the abundance of five proteins increased during the infection, while the amounts of three proteins decreased. Of these eight proteins, five were not detectable in the susceptible variety over the time course (Table 1). On the other hand, in the susceptible variety, the abundance of 11 proteins changed significantly in amount with seven increasing and four decreasing in their steady state levels with infection. Protein spot 7, which increased in intensity with infection, was not detected in the resistant variety during the course of the experiment (Table 1).

Table 1.

Fold changes in intensity levels of 2D gel spots produced by Brassica juncea CBJ-001 (resistant) and RH-819 (susceptible) varieties inoculated with Albugo candida, as compared with the mock-inoculated plants at 2, 4, 8, 24 and 72 hours post inoculation (hpi)

Spot no.	Protein	hpi	Statusb
			Resistant variety		Susceptible variety
			Fold change	±SE	Fold change	±SE
1	Plant-Thaumatin-like (PR-5) Protein [Brassica juncea]a	2	+1.99	±0.21	*	*
		4	+1.17	±0.02	*	*
		8	+2.12	±0.10	*	*
		24	+4.68	±0.46	*	*
		72	+2.50	±0.45	*	*
2	Glutathione S-transferase GSTF9 [B. juncea]a	2	*	*	*	*
		4	*	*	*	*
		8	+4.44	±0.50	*	*
		24	+4.35	±0.72	*	*
		72	+11.23	±0.70	*	*
3	Cysteine synthase OAS-TL B [Arabidopsis thaliana]a	2	*	*	*	*
		4	*	*	*	*
		8	+0.80	±0.11	*	*
		24	+1.01	±0.22	*	*
		72	+1.67	±0.13	*	*
4	Superoxide dismutase CSD2 (Cu/Zn) [A. thaliana]	2	+1.34	±0.30	*	*
		4	+2.38	±0.06	+1.09	±0.04
		8	+2.11	±0.24	+1.02	±0.02
		24	+1.16	±0.04	*	*
		72	+1.25	±0.08	*	*
5	Superoxide dismutase FSD1 (Fe/Mn) [Raphanus sativus]a	2	+2.03	±0.20	*	*
		4	+3.82	±0.01	*	*
		8	+6.67	±0.48	*	*
		24	+19.07	±1.00	*	*
		72	+37.33	±2.63	*	*
6	Red chlorophyll catabolite reductase ACD2 [B. napus]a	2	*	*	*	*
		4	*	*	*	*
		8	–2.23	±0.88	*	*
		24	–1.47	±0.10	*	*
		72	–1.43	±0.12	*	*
7	Peptidyl-prolyl cis-trans isomerase CYP20-3 [A. thaliana]a	2	*	*	+1.15	±0.03
		4	*	*	+0.97	±0.16
		8	*	*	+1.48	±0.02
		24	*	*	+1.10	±0.03
		72	*	*	+1.04	±0.06
8	Carbonic anhydrase [A. thaliana]	2	*	*	*	*
		4	–1.28	±0.09	*	*
		8	–1.18	±0.03	*	*
		24	–1.27	±0.42	*	*
		72	–1.26	±0.08	–1.17	±0.14
9	Cytochrome b6-f complex iron-sulphur subunit [A. thaliana]	2	–0.54	±0.01	–0.72	±0.05
		4	–0.73	±0.21	-0.69	±0.09
		8	–1.96	±0.20	*	*
		24	–2.64	±0.55	*	*
		72	–1.72	±0.06	*	*
10	Triosephosphate isomerase [A. thaliana]	2	*	*	*	*
		4	*	*	+1.65	±0.09
		8	+1.03	±0.09	+0.97	±0.16
		24	+0.90	±0.00	+1.28	±0.04
		72	+0.99	±0.02	+1.16	±0.12
11	Triosephosphate isomerase [A. thaliana]	2	*	*	*	*
		4	*	*	+1.55	±0.18
		8	+1.03	±0.15	+1.62	±0.08
		24	+1.18	±0.03	+1.45	±0.10
		72	+1.19	±0.14	+1.81	±0.30
12	Triosephosphate isomerase [A. thaliana]	2	*	*	*	*
		4	*	*	+1.78	±0.51
		8	+1.87	±0.10	+1.70	±0.19
		24	*	*	+1.51	±0.06
		72	+1.36	±0.12	+1.25	±0.09
13	Malate dehydrogenase [A. thaliana]	2	*	*	*	*
		4	*	*	+2.11	±1.53
		8	*	*	+1.64	±0.01
		24	+1.44	±0.30	+1.34	±0.01
		72	+1.95	±0.85	+1.53	±0.14
14	Chlorophyll a/b binding protein [B. oleracea]	2	+1.56	±0.17	*	*
		4	+1.58	±0.37	*	*
		8	*	*	–1.28	±0.13
		24	*	*	–1.08	±0.12
		72	*	*	–2.66	±0.76
15	AT4g24770/F22K18_30 [A. thaliana]	2	*	*	*	*
		4	*	*	*	*
		8	*	*	*	*
		24	*	*	+1.21	±0.10
		72	–0.75	±0.01	+1.55	±0.13
16	Phosphoribulokinase [A. thaliana]	2	*	*	*	*
		4	*	*	*	*
		8	*	*	*	*
		24	*	*	–2.04	±0.83
		72	-1.58	±0.37	–1.67	±0.10
17	Phosphoglycerate kinase [A. thaliana]	2	*	*	*	*
		4	*	*	*	*
		8	*	*	*	*
		24	*	*	–1.93	±0.11
		72	–1.89	±0.35	–1.23	±0.21
18	Glutamine synthetase [B. rapa]	2	*	*	*	*
		4	*	*	*	*
		8	*	*	*	*
		24	*	*	+2.42	±0.19
		72	+1.36	±0.57	+3.03	±0.53
19	ATP synthase subunit alpha [Arabis hirsuta]	2	*	*	*	*
		4	*	*	*	*
		8	–2.14	±1.11	–1.51	±0.77
		24	*	*	–2.65	±1.77
		72	–1.88	±0.82	–1.79	±0.85

Protein spots only detected either in resistant variety or susceptible variety and not detectable in the other variety.

* Spot not detected.

The identities of 18 of the 19 differentially abundant proteins were successfully established by ESI-Q-TOF-MS/MS, with the final spot (spot 1) being identified through peptide de novo-sequence analysis (Table 1; see Supplementary Table S3 at JXB online). These 19 proteins were classified as ‘significant hits’ based on individual peptide ion scores. Of the 19 proteins identified, two could be directly matched to known B. juncea proteins, while others could only be matched to putative orthologues in other Brassicaceae due to the limited coverage of the B. juncea genome database. Three proteins were matched to proteins from other Brassica spp., 13 proteins were matched to Arabidopsis thaliana proteins, and one protein was matched to a Raphanus sativus protein sequence (Table 1; see Supplementary Table. S3 at JXB online).

A rigorous criterion was applied to the selection of differentially expressed candidate protein spots for further analysis. A protein was only considered differentially expressed if its abundance differed by at least 1.5-fold in all three replicates of at least one time point between the inoculated and mock-inoculated plants. Surprisingly, many candidate proteins were only detected either in the susceptible or resistant variety. To keep the focus on the host–pathogen interaction this study concentrated on the downstream analysis of the seven proteins that had statistically significant changes in abundance (P <0.05) (Fig. 1C; Table 1). All of these proteins belong to protein families that have already been identified in other pathosystems as important in mediating resistance or susceptibility to plant pathogens.

Proteins detected only in the resistant variety that are differentially expressed upon pathogen infection

Pathogen responsive protein 5 (PR-5) or plant-thaumatin-like protein

This protein was found to be expressed only in the resistant variety with a 1.9±0.21, 1.17±0.02, 2.12±0.10, 4.68±0.46, and 2.50±0.45-fold increased abundance in the pathogen challenged as compared to the mock-inoculated plants at 2, 4, 8, 24, and 72 hpi, respectively (Table 1). There was no detectable expression of this protein recorded at any of the time points in the susceptible variety, either in the mock-inoculated or the pathogen-inoculated plants. The peptide sequence obtained was found to match a salicylic acid-induced thaumatin (PR-5) protein sequence of B. juncea (Table 1) (Genbank accession no. ABX10753).

Superoxide dismutase

Two isoforms of superoxide dismutase (SOD), one with closest similarity to Cu/Zn-binding SODs, and the other with closest similarity to Fe/Mn-binding SODs, were identified. Both enzymes catalyse the conversion of superoxide radicals to molecular oxygen and H2O2. Both isoenzymes showed elevated expression in the resistant variety challenged with A. candida compared with the mock-inoculated plants, but the close homologue to the plastid-localized Cu/Zn SOD CSD2 in Arabidopsis (At2g28190, Kliebenstein ) showed a similar induction at 4 hpi and 8 hpi in the susceptible variety. By contrast, the Fe/Mn-SOD with the highest homology to a protein in Raphanus sativus that shares 93% identity to the plastid-localized, diurnally regulated, Cu-binding FSD1 protein in Arabidopsis (At4g25100, Kliebenstein ) showed a steady increase in expression following the pathogen challenge of the resistant cultivar from 2 hpi onwards steadily increasing to 37.33±2.63-fold at 72 hpi in the pathogen-inoculated compared with the mock-inoculated plant with no detectable expression in the susceptible variety in the presence or absence of the pathogen at any of the analysed time points (Table 1).

Glutathione S-transferase

A jasmonic acid-inducible glutathione S-transferase (GST) in B. juncea with 96% sequence identity to the A. thaliana GSTF9 protein (At2g30860) from the plant-specific Phi (π) class of the GST C-terminal domain superfamily was identified as being expressed only in the resistant variety. Its abundance was 4.44±0.50, 4.35±0.72, and 11.23±0.70-fold up-regulated in plants inoculated with A. candida as compared to the mock-inoculated seedlings at 8, 24, and 72 hpi, respectively (Table 1), with no detectable expression at the protein level during the early hours, i.e. 2 hpi and 4 hpi. The protein could not be detected in either the pathogen-inoculated or in the mock-inoculated samples of the susceptible variety.

Cysteine synthase

Expression of the B. juncea orthologue of the plastid localized isoform of cysteine synthase in Arabidopsis, also called O-acetyl-L-serine(thiol)lyase B (OAS-TL B, At2g43750) was only detected in the resistant variety and changed 0.80±0.11, 1.01±0.22, and 1.67±0.13-fold in the pathogen-challenged compared with the mock-inoculated control plants at 8, 24, and 72 hpi, respectively (Table 1). This enzyme belongs to the β-substituted-alanine-synthase branch of the tryptophane synthase β II superfamily. Similar to the expression profile of the GSTF9 homologue, there was no detectable expression in the resistant cultivar at 2 hpi and 4 hpi, neither in the pathogen-inoculated nor in the mock-inoculated treatments.

Red chlorophyll catabolite reductase

The closest match for spot 6 was the B. napus red chlorophyll catabolite reductase (RCCR) protein. RCCR has 84% sequence identity to the plastid-localized ACD2 (ACCELERATED CELL DEATH2, At4g37000) protein of Arabidopsis and was expressed only in the resistant variety. Its expression was 2.23±0.88, 1.47±0.10, and 1.43±0.12-fold down-regulated in the pathogen-inoculated as compared to the mock-inoculated plants at 8, 24, and 72 hpi, respectively (Table 1). Like GST and CSase no expression was recorded during the early hours post-inoculation.

Proteins detected only in the susceptible variety that are differentially expressed upon pathogen infection

Peptidyl-prolyl cis/trans isomerase

Unlike the proteins identified above, a protein closely matching the peptidyl-prolyl cis/trans isomerase (PPIase) isoform CYP20-3/ROC4 (At3g62030) of Arabidopsis thaliana that belongs to the cyclophilin C-CyP subfamily and is the only Arabidopsis cyclophilin that is located in the chloroplast stroma, was only detected in the susceptible variety and showed a transient 1.15±0.03, 0.97±0.16, 1.48±0.02, 1.10±0.03, and 1.04±0.06-fold up-regulation in the A. candida-inoculated compared to the mock-inoculated susceptible variety at 2, 4, 8, 24, and 72 hpi, respectively (Table 1). There was no expression detectable for this protein in the resistant variety.

Analyses of gene expression using quantitative real-time PCR (qRT-PCR)

Given the sensitivity constraints of 2-D gels, the apparent absence of some of the protein spots in one of the cultivars was further investigated by measuring relative abundance of the cognate transcript using qRT-PCR. This allowed further investigation as to whether the observed proteome changes are associated with changes in mRNA levels or whether they are due to post-transcriptional regulation. qRT-PCR analysis was performed using total RNA isolated from cotyledons of mock-inoculated or A. candida challenged plants using the same three biological replicates as for the proteomic analysis. The differential gene expression data for the five candidate proteins that were only detected in the resistant variety but not in the susceptible variety supported our proteomic data (Fig. 2). Transcripts encoding orthologues of GSTF9, PR-5, ACD2, OAS-TLB, and FSD1 mostly showed greater expression in the resistant compared with the susceptible variety in both mock-inoculated and pathogen-inoculated samples. These results suggest a higher constitutive expression of these genes in the resistant over the susceptible variety even in the absence of the pathogen. Furthermore, the transcript abundance for these genes was increased in the pathogen-challenged compared with the mock-inoculated plants. The responses of GSTF9, FSD1, and PR-5 transcripts to the presence of the pathogen were much less intense and relatively slow in the susceptible variety. While GSTF9 and FSD1 transcripts were constitutively induced in all mock-treatments, PR-5 showed a transient induction in the resistant variety at the early time-points of the mock inoculation. However, this early induction is more pronounced and sustained in the presence of the pathogen in the resistant variety. In the susceptible variety, there is no pre-induction of PR-5 by the mock treatment and a delayed weaker response upon pathogen challenge. CSD2 transcript levels on the other hand were elevated to a similar extent in both cultivars upon pathogen challenge. For the CYP20-3 homologue, significantly higher transcript amounts were observed in the susceptible over the resistant variety in both the control and pathogen-inoculated plants. Its expression increased transiently by 4–8-fold within 2–4 hpi in plants of the susceptible variety challenged with A. candida. The very low overall expression level for this gene in the resistant variety is in agreement with the absence of a detectable protein spot in the 2-D gel analysis.

Fig. 2.

qRT-PCR analysis of gene expression of identified candidate proteins. The relative expression of genes encoding homologues of glutathione S-transferase GSTF9 (Brassica juncea) (A); cysteine synthase OAS-TL B (Arabidopsis thaliana) (B); superoxide dismutase FSD1 (Raphanus sativus) (C); superoxide dismutase CSD2 (A. thaliana) (D); plant-thaumatin-like protein PR-5 (B. juncea) (E); red chlorophyll catabolite reductase ACD2 (B. napus) (F); and peptidyl-prolyl cis-trans isomerase CYP20-3 (A. thaliana) (G), at 2, 4, 8, 24, and 72 h following inoculation with Albugo candida in the mock-inoculated and pathogen-inoculated susceptible (RH-819) and resistant (CBJ-001) varieties of B. juncea are shown. Expression levels are given on a log scale expressed as 40-ΔCT, 40 therefore equals the expression level of the actin isoform (accession no. AF111812) used as a reference gene. The fold difference in expression can be deduced as 2ΔΔCT assuming a PCR efficiency of 2 (e.g. an ordinate value of 42 represents 4-fold higher expression than for actin and a value of 38 represents a 4-fold lower expression level than for actin).

Discussion

To gain insight into the molecular mechanisms underlying activation of complex plant defence responses that occur in resistant and susceptible varieties of B. juncea, challenged with A. candida, a time-course protein profiling experiment followed by transcript studies was conducted. The latter studies investigated whether the observed proteome changes were substantiated by changes in mRNA levels. Our results revealed a clear differentiation between resistant and susceptible interactions and led to the identification of six differentially regulated proteins potentially involved in defence reactions in the resistant variety. A seventh differentially expressed protein is potentially involved in suppressing the host defence system, as it was detected only in the susceptible variety.

Plant defence responses often begin with gene-for-gene recognition of the pathogen. Production of virulence effectors by a pathogen leads to its recognition by a host plant (Jones and Dangl, 2006). Recognition results in the rapid activation of defence responses which are usually accompanied by an oxidative burst, that is, the rapid production of reactive oxygen species (ROS). ROS production is also required for the hypersensitive response (HR), a type of programmed cell death thought to limit the access of the pathogen to water and nutrients (Glazebrook, 2005).

In the present study the plastid localized RCCR protein was found to be down-regulated in the pathogen-inoculated compared with the mock-treated resistant cultivar. The protein was only detected from 8 hpi onwards in the mock-treated seedlings and, to a lesser extent, the pathogen-inoculated resistant seedlings. RCCR transcript levels, however, showed a transient up-regulation in both resistant and susceptible varieties under both treatments at 2–8 hpi suggesting a diurnal control of gene expression. This different trend in the 2-D gel-based proteome analysis compared with the observed changes in transcript abundance recorded by qRT-PCR may be due to post-transcriptional or post-translational modifications which may have led to the altered steady-state protein amounts in the two cultivars.

Pheophorbide (Pheide) α oxygenase (PaO) and RCCR catalyse the key reaction of chlorophyll catabolism, the porphyrin macrocycle cleavage of Pheide α to a primary fluorescent catabolite (pFCC) (Wuthrich ). Disruption of RCCR activity is likely to result in the accumulation of RCC and Pheide α which are phototoxic and cause cell death by the light-dependent production of free oxygen radicals (Mach et al., 2001). Alternatively, accumulation of RCC itself may cause a specific signal that triggers cell death (Mach et al., 2001). Cell death and chlorophyll breakdown have been related to plant senescence and disease progression (Rodoni ; Wuthrich ; Mach ; Pruzinska ).

The Arabidopsis ACD2 gene encodes RCCR and was found to suppress the spread of disease symptoms caused by Pseudomonas syringae (Mach ). Inoculated plants that lack RCCR/ACD2 activity show spreading cell death beyond the initial containment zone around the initial lesion. On the other hand, plants expressing high levels of the RCCR/ACD2 protein showed higher tolerance to infection with a virulent Pseudomonas syringae strain. This suggests that an increase in RCCR/ACD2 protein amount may alter the flux of chlorophyll catabolites that normally accumulate during disease and trigger cell death (Mach ). In the present study, the inability to detect RCCR protein at any stage of the infection cycle in the susceptible variety might point towards the inability of these seedlings to initiate a successful HR during the early stages of pathogen attack.

The production of ROS presents a challenge to a plant because of the potential of these compounds to damage cellular components. The plant must therefore maintain a balanced system that produces ROS for defence at the same time as it produces antioxidants to protect against ROS-mediated oxidative damage. One of the critical metabolites for maintaining redox balance in the face of oxidative stress is glutathione, which is a ubiquitous supplier of reducing power for cellular processes (Noctor and Foyer, 1998). Glutathione (GSH) is a conjugate of glutamate, cysteine, and glycine. The last two steps in cysteine biosynthesis are catalysed by a bi-enzyme complex of serine acetyltransferase (SAT) and cysteine synthase, also called O-acetylserine (thiol) lyase (OAS-TL). Our present study found that a B. juncea homologue of the plastid-localized isoform AtOAS-TL B was present in the resistant variety, but not detectable in the susceptible variety. The plastid-localized isoform of OAS-TL B contributes up to 65% of the total OAS-TL enzyme activity in wild-type Arabidopsis plants (Heeg ).

Cysteine is incorporated into GSH by a two-step enzymatic reaction catalysed by γ-glutamylcysteine synthetase, an enzyme localized exclusively to the plastid, and GSH synthetase, an enzyme that shows dual targeting to both the cytosol and plastids (Wachter ). GSH biosynthesis is strongly regulated by cysteine availability (Strohm ). Overexpression of a plastidic OAS-TL isoform or an inactive SAT isoform, caused moderate to very high increases of cysteine and GSH content in tobacco, respectively (Noji ; Wirtz and Hell, 2007). The pool size and redox status of GSH are important for determining a plant's ability to control ROS production and combat microbial pathogens (Noctor and Foyer, 1998).

Both primary and secondary sulphur metabolism were found to be induced at the transcriptional, translational, and metabolite levels upon pathogen attack in Arabidopsis when grown under sufficient sulphate supply, while basal resistance against necrotrophic fungi was compromised under sulphate-limiting conditions (Kruse ). An increase in GSH content was also reported to occur in leaves attacked by avirulent biotrophic pathogens (El-Zahaby ; Fodor ; Vanacker , 1999). Together with our results, this suggests that a boost in cysteine and GSH levels is important for the incompatible interaction between A. candida and B. juncea.

One of the processes consuming GSH could involve the B. juncea GSTF9 homologue. This GST is encoded by one of 13 phi-class GST genes in Arabidopsis. In B. juncea, GSTF9 was detected only in the resistant variety and increased in abundance as the infection progressed. The plant-specific phi-class GSTs are related to fungal and bacterial proteins and little is known about their substrate specificities (Dixon ). The B. juncea peptides identified closely match a jasmonic-acid inducible B. juncea GST, with close homology to the cytosolic dimeric GSTF9 protein from Arabidopsis. Abundance of GSTF6, F9, U5, and U13 transcripts were found to be methyl-jasmonate responsive in Arabidopsis (Jost ). A quadruple knock-out line of Arabidopsis lacking GSTF6, F7, F9, and F10 showed no impairment in growth or de velopment even under numerous stress conditions, but the subtle metabolic changes observed suggest an impaired response to oxidative stress conditions (Sappl ). The endogenous products of oxidative damage, including membrane lipid peroxides and products of oxidative DNA degradation, are highly cytotoxic. GSTs conjugate GSH with such endogenously produced electrophiles, resulting in their detoxification (Pickett and Lu, 1989; Dudler ; Bartling ; Berhane ). Some GSTs also function as glutathione peroxidases to detoxify oxidative products directly (Williamson and Beverley, 1987; Bartling ) and there are many reports that suggest a central role for GSH in plant defence activation (Marrs, 1996; Dixon ).

Our study also detected two plastid localized isoforms of SOD in the resistant B. juncea variety that were absent from the susceptible variety. One of these SOD isoforms, the FSD1 homologue, showed an early and sustained increase in abundance only in the resistant B. juncea variety on both transcript and protein level. Interestingly, the other isoform, the CSD2 homologue, was induced to a similar extent on transcript level in both varieties under pathogen attack, but protein accumulation was only detectable in the resistant cultivar upon pathogen challenge. This again suggests strong post-transcriptional regulation of this particular isoform that is resulting in different outcomes in the resistant and susceptible varieties. SODs are known to play an important role in protecting cells against the toxic effects of ROS produced in various subcellular compartments (Fridovich, 1986; Halliwell and Gutteridge, 1989). An oxidative burst is a key feature underlying successful pathogen recognition at the site of the initial infection (Alvarez ; Van Breusegem and Dat, 2006). Rapid induction of SOD and accumulation of H2O2 are characteristic early features of the HR following perception of avirulence signals from the pathogen (Lamb and Dixon, 1997).

ROS-detoxifying enzymes such as SOD catalyse the dismutation of ROS such as superoxide, O2.−, to hydrogen peroxide, H2O2. H2O2 itself can trigger cell death, but at the same time functions as a messenger in cellular communication demanding a finely tuned adjustment of its cellular concentration (Foyer and Noctor, 2000). SODs therefore play an important role in ameliorating stress during conditions such as pathogen infection where overproduction of ROS is induced (De Gara ).

Various ROS species have similar deleterious effects on plant cells but they each activate different signalling pathways (Laloi ). Moreover, the cellular compartment in which ROS are generated determines the nature of this signal (Miller ). It was quite interesting to find the two plastid-localized SODs responding differently to A. candida in our study. The CSD2 protein seems to accumulate transiently in the resistant variety upon pathogen inoculation, while the FSD1 protein is very strongly induced and keeps increasing throughout the experiment. There was a transient increase in both FSD1 and CSD2 transcript abundances upon pathogen inoculation in the susceptible variety but these are not sustained at the protein level.

Recently, a cyclophilin was found to be necessary for host–pathogen recognition in Arabidopsis (Coaker ). The PPIase subfamily of cyclophilins is known to have chaperone function assisting in protein folding and the assembly of large protein complexes (Kromina ). One of the seven cytosolic cyclophilin isoforms of Arabidopsis, ROC1, is the closest homologue of human cyclophilin A and yeast CPR1 and is required for activation of the Pseudomonas syringae effector protein AvrRpt2 (Coaker ). However, many of the 29 cyclophilins in Arabidopsis are targeted to the plastid. While most of these reside in the thylakoid lumen, one, CYP20-3/ROC4, is found in the chloroplast stroma. In our study, CYP20-3 protein was detected only in the susceptible B. juncea variety, and this was confirmed by a very low CYP20-3 transcript abundance in the resistant variety. This suggests that the B. juncea CYP20-3 protein might play an important role in the compatible interaction between A. candida and B. juncea. But its subcellular localization makes it less likely that it is a direct target for pathogen-derived effector proteins. Earlier studies suggest that its activity might be sensitive to changes in redox balance and that it is induced by light. It was recently demonstrated in Arabidopsis that CYP20-3 may play a role in the regulation of cysteine biosynthesis in the chloroplast stroma through its interaction with the plastid isoform SAT1. This interaction seems to be required for optimal enzyme activity under stress conditions, most likely by stabilizing the assembly of one SAT trimer and three OAS-TL dimers into the cysteine synthase complex. It was shown that cyp20-3 mutants do not show an increase in thiol content in plants subjected to salt stress or high light conditions (Dominguez-Solis ). The authors suggest that ROS elicit the transcriptional activation of SAT1 as well as the oxidation of CYP20-3 in a peroxiredoxin-dependent manner. Thioredoxin would then be needed to regenerate active CYP20-3 (Motohashi ). It is interesting to note that the B. juncea CYP20-3 protein could only be detected in leaves of the susceptible, but not the resistant cultivar. This could indicate that oxidized CYP20-3 accumulates in this cultivar due to a redox imbalance in the chloroplast rendering the cysteine synthase complex inactive but this needs to be investigated further.

Resistance to pathogens is associated with the activation of signalling pathways that lead to the expression of certain pathogenesis-related (PR) proteins (Selitrennikoff, 2001). While the precise mechanism of action of the PR-5 protein is not yet completely understood (Richardson ; Wang ; Coca ; Ibeas ), it is most likely involved in the degradation of the pathogen cell wall. Several PR-5 proteins have been reported to cause cell permeability changes in fungal cells with intact cell walls (Roberts and Selitrennikoff, 1990). The maize PR-5 protein zeamatin inhibits growth of suspension cultures of Candida albicans, Neurospora crassa, and Trichoderma reesei by causing rapid cell lysis, even at low temperatures (Roberts and Selitrennikoff, 1990). PR-5 proteins may have a direct role in cell wall degradation as they bind (1,3)-β-glucans and have in vitro (1,3)-β-glucanase activity (Grenier et al., 1993; Trudel ). In the present study the induction of the PR-5 protein following inoculation with A. candida in the resistant B. juncea variety suggests an anti-pathogenic role. This role is further indicated by its absence from the proteome profiles of the susceptible variety and by the lower transcript abundance in the susceptible variety compared with the resistant variety. The delayed induction of PR-5 transcript accumulation in the susceptible line compared with the resistant line suggests that PR-5 needs to be present in the early stages of infection to exert its anti-pathogenic effect. Thus, in the context of disease progression in the A. candida–B. juncea pathosystem, the timing of the induction of PR-5 in resistant and susceptible varieties appears to be crucial for mounting an effective defence response to this pathogen.

All but two of the candidate proteins identified in this study reside in the chloroplast stroma and are directly or indirectly involved in redox homeostasis. This is quite remarkable given that chloroplasts are increasingly being regarded as sensors of environmental changes and mediators of plant stress responses (Fernandez and Strand, 2008). Defence against pathogens might therefore require the activation of photoinhibitory ROS production combined with the balancing power of the reactive oxygen regulatory network of the chloroplast (Mittler ; Dietz ). This may be important for generating an oxidative burst that limits the uncontrolled spread of cell death once a HR has been triggered by the presence of a pathogen (Belhaj ). This is the first time that the host–pathogen interaction of the B. juncea–A. candida pathosystem has been dissected on a molecular level to differentiate between compatible and incompatible interactions. The results of the present study demonstrate that the timing of the expression of defence-related genes plays a crucial role during pathogenesis and incompatible interactions and that the redox balance within the chloroplast may be of crucial importance for mounting a successful defence response. These findings also indicate that synergistic and conserved strategies are utilized by the resistant host to fight off the A. candida attack.

Determining the changes in protein expression patterns elicited by A. candida in a resistant versus a susceptible variety of B. juncea has not only led to the likely identification of key molecular components of this specific host–pathogen interaction, but now opens the way for developing genetic markers that could be used to screen for resistance within B. juncea germplasm collections. Such studies may also open novel avenues for engineering durable resistance to this pathogen. The present study is the first step in identifying the metabolic processes and defence-related proteins that are required for mounting a successful defence response in B. juncea against A. candida. Looking at functional relationships between different genes-of-interest using transgenic over-expression and RNAi lines will probably be the next step in the characterization of this pathosystem.

Supplementary data

Supplementary data can be found at JXB online.

. Mean disease severity ratings obtained using a scale of 0 to 9 for the different genotypes of Brassica juncea inoculated with Albugo candida at the cotyledon stage.

. Primer sequences for qRT-PCR analyses.

. Summary of the Mascot results of the differentially expressed proteins identified from resistant Brassica juncea (CBJ-001) and susceptible B. juncea (RH-819) upon Albugo candida challenge.