Label-Free Quantitative Proteomic Analysis of Systemic Responses to

Article pubs.acs.org/jpr

Label-Free Quantitative Proteomic Analysis of Systemic Responses to Local Wounding and Virus Infection in Arabidopsis thaliana Annette Niehl,#,†,‡ Zhe Jenny Zhang,#,∇,§,⊥ Martin Kuiper,∥ Scott C. Peck,*,§,⊥ and Manfred Heinlein*,†,‡ †

Institut de Biologie Moléculaire des Plantes du CNRS (UPR 2357), Université de Strasbourg, 67084 Strasbourg, France Zürich-Basel Plant Science Center, Botany, Department of Environmental Sciences, University of Basel, 4051 Basel, Switzerland § Department of Biochemistry, University of MissouriColumbia, Columbia, Missouri 65211, United States ⊥ Bond Life Sciences Center and Interdisciplinary Plant Group, University of MissouriColumbia, Columbia, Missouri 65211, United States ∥ Department of Biology, Norwegian University of Science and Technology, 7491 Trondheim, Norway ‡

S Supporting Information *

ABSTRACT: Plants are continuously exposed to changing environmental conditions and must, as sessile organisms, possess sophisticated acclimative mechanisms. To gain insight into systemic responses to local virus infection or wounding, we performed comparative LC−MS/MS protein profiling of distal, virus-free leaves four and five days after local inoculation of Arabidopsis thaliana plants with either Oilseed rape mosaic virus (ORMV) or inoculation buffer alone. Our study revealed biomarkers for systemic signaling in response to wounding and compatible virus infection in Arabidopsis, which should prove useful in further addressing the trigger-specific systemic response network and the elusive systemic signals. We observed responses common to ORMV and mock treatment as well as protein profile changes that are specific to local virus infection or mechanical wounding (mock treatment) alone, which provides evidence for the existence of more than one systemic signal to induce these distinct changes. Comparison of the systemic responses between time points indicated that the responses build up over time. Our data indicate stress-specific changes in proteins involved in jasmonic and abscisic acid signaling, intracellular transport, compartmentalization of enzyme activities, protein folding and synthesis, and energy and carbohydrate metabolism. In addition, a virus-triggered systemic signal appears to suppress antiviral host defense. KEYWORDS: Arabidopsis, systemic signaling, virus, wound, label-free quantification

■

transcripts.4,6,9 A number of hypotheses have been proposed for the mechanistic basis of the primed state. Priming has been linked with modifications in chromatin states, leading to enhanced transcription initiation and gene expression after the primed plant encounters a new stress.10−13 However, defense priming may also be conferred through the modification of primary metabolism,14 accumulation of inactive signaling proteins,15 or hormone action.9,16,17 SAR, the more active systemic response, has been shown in genetic studies to involve salicylic acid (SA), jasmonic acid (JA) and lipid signaling.18 Our knowledge about systemic priming against compatible pathogens is still limited. Both wounding and compatible virus treatments have been reported to induce local and systemic responses at the transcript level,19−29 but only wounding is also known to induce local and systemic changes at the protein level.30−39 Whether a local compatible virus infection causes specific systemic responses at the proteome level is yet unknown.

INTRODUCTION As sessile organisms, plants are exposed to a plethora of environmental stresses to which they must respond to maintain efficient growth and survival. For instance, plants have evolved sophisticated multilayered mechanisms to sense and restrict the spread of invading pathogens.1,2 Whereas many plant responses to pathogens are studied in relation to the actual infected tissue, referred to as local defenses, plants also transmit their “knowledge” about local infection to more distant, uninfected tissues.3−5 One of the functions of this systemic communication leads to the establishment of a “primed state”, a condition in which plants are able to elicit faster or more robust defense responses to subsequent biotic stresses.6 In contrast to systemic acquired resistance (SAR), in which pathogen-associated molecular patterns (PAMPs) or resistance gene-mediated recognition of an elicitor or effector molecule lead to active systemic defense responses involving the systemic induction of pathogenesis-related (PR) genes,7,8 priming of defense can also be induced by compatible pathogens and confers a more dormant response not associated with strong induction of defense gene © 2013 American Chemical Society

Received: November 13, 2012 Published: April 17, 2013 2491

dx.doi.org/10.1021/pr3010698 | J. Proteome Res. 2013, 12, 2491−2503

Journal of Proteome Research

Article

not-treated plants were removed. At 4 and 5 dpt, the three youngest leaves of each of 50 plants per treatment were collected and shock-frozen in liquid nitrogen. The experiment was performed temporally three times. Thus, samples from three independent experiments were analyzed (three experiments × three treatments × two time points = 18 samples). To ensure that distant leaf samples of virus-treated plants were free of virus, the plants were kept for additional three weeks and examined for symptom development. To control for efficiency of infection, another set of 20 plants was inoculated with the virus but the inoculated leaves were not removed. All of these 20 control plants developed ORMV symptoms at 10 to 14 dpt, indicating an infection efficiency of 100%.

To determine whether a local compatible virus infection in lower leaves causes a systemic response in upper, noninfected leaves, we investigated systemic responses of the Arabidopsis proteome to compatible virus infection in a time-resolved manner. Rosette leaves of Arabidopsis thaliana plants were inoculated with Oilseed rape mosaic virus (ORMV), which causes a systemic somatic recombination response in Arabidopsis and tobacco,13,40,41 thus confirming that a systemically translocated signal is generated during this plant−virus interaction. We compared the proteome responses in upper, virus-free leaves of the virus-treated plants with the responses in upper leaves of not-treated plants and of plants treated with abrasive and buffer alone (mock treatment). Comparative LC−MS/MS protein profiling of the distal leaves of virus-treated, mock-treated and not-treated plants revealed a large overlap between proteins changing in response to both the mock and virus treatment. However, the responses also revealed proteins showing differential protein abundance in response to mock versus virus treatment, thus identifying potential biomarkers specific for these respective treatments and suggesting the presence of multiple systemic signals. The qualitative comparison of the responses at four and five days post treatment (dpt) indicates that the responses build up over time. Together, these results ̈ untreated tissues add to our understanding of how naive, respond to systemic signals arising from locally applied biotic and abiotic stresses.

■

Protein Extraction

In each experiment, pooled leaf samples of 50 not-treated, mocktreated, or virus-treated plants were homogenized in liquid N2 using a Sartorius (Aubagne, France) micro-dismembrator. Proteins were extracted using TRI-reagent (Sigma, St. Louis, MO) according to the manufacturer’s instructions with minor modifications. 0.75 mL of TRI-reagent was added to the frozen leaf material, which was then ground and vortexed. After centrifugation at 12000g for 10 min, the supernatant was collected and nucleic acids were extracted with 0.2 mL of chloroform. After removing the upper aqueous phase, DNA was extracted from the organic phase by ethanol precipitation at 2000g. Subsequently, protein was precipitated from the phenol−ethanol supernatant with three volumes of ice-cold acetone and centrifugation at 12000g for 10 min. The proteincontaining pellet was washed three times in 0.3 M guanidine hydrochloride/95% ethanol solution. After washing the protein pellet in ethanol, the protein was maintained as a precipitated pellet in acetone.

MATERIALS AND METHODS

Chemicals

All chemicals were ultrapure grade (Sigma Chemical or Fisher Scientific). HPLC-grade water was produced by a Millipore Synthesis system (Millipore Corp. Billerica, MA). Sequencinggrade modified trypsin was purchased from Promega.

1D Gel Analysis and Tryptic Digestion

Virion Preparation

From the acetone pellet, proteins were solubilized in sample loading buffer heated to 75 °C and separated by 8% SDSPAGE. After staining with colloidal Coomassie G-250 overnight,43 gels were destained in water before dividing each lane into 8 gel strips with a razor blade. Proteins in resulting gel pieces (1 mm2 each) were reduced with 50 mM (2-carboxyethyl) phosphine hydrochloride salt (TCEP-HCl) and alkylated with 50 mM iodoacetamide before in-gel digestion with trypsin overnight in 100 mM ammonium bicarbonate buffer (pH 8.3) at 37 °C. After digestion, peptides were eluted twice with 1% trifluoroacetic acid in 60% acetonitrile. The extracted peptide mixture was dried by lyophilization overnight. Dried peptides were stored at −80 °C until LC−MS/MS analysis.

To prepare ORMV virions for inoculation, ORMV-infected Nicotiana benthamiana leaves were homogenized to fine powder in liquid N2. After addition of 1 mL of 0.5 M sodiumphosphate buffer pH 7.4 and 0.1% 2-mercaptoethanol per gram of leaf material, virions were extracted with 1 volume butanol/ chloroform (1:1 (v/v)) and the phases separated by centrifugation (two times 15 min at 12000g). Virions in the upper, aqueous phase were precipitated with 4% PEG-8000 at 20000g. The pellet was resuspended in 10 mM sodiumphosphate buffer (pH 7.4) and cleared by centrifugation at 5000g for 10 min. The supernatant was precipitated again with 4% PEG-8000 and 1% NaCl and resuspended in 10 mM sodium-phosphate buffer, pH 7.4. Virion concentration was estimated from absorbance values at 260 nm.

LC−MS/MS Analysis

Lyophilized peptides were dissolved in 0.1% formic acid for MS analysis. Peptides were applied to a 10 cm prepacked capillary column (Picotip, 75-μm inner diameter, 15-μm tip, New Objective, Woburn, MA) and eluted into the nanoelectrospray ion source of a LTQ-Orbitrap LC−MS/MS mass spectrometer (Thermo Scientific) controlled by XCalibur version 2.2.1. A fully automated chromatography run was carried out with the mass spectrometer operating in data-dependent mode. The buffer solutions used for chromatography were 0.1% formic acid in 100% acetonitrile (buffer A), 0.1% formic acid (buffer B). The elution gradient was with a 1% per min incremental gradient of buffer A for the first 45 min and 11% per min of buffer A for the final 5 min. Survey scans at resolution of 30 000 covered the range 300−2000 m/z with a threshold of

Plant Growth, Mock Treatment, Viral Infection, and Tissue Collection

Arabdidopsis thaliana Columbia-0 plants were grown from seeds on soil at 12/12 h light/dark cycles and 21/18 °C for 34 days. At the time of inoculation, the plants were in early developmental stages 1.07−1.10.42 The plants were inoculated at the adaxial side of two rosette leaves with 10 mM phosphate buffer pH 7.4 containing 150 ng of purified ORMV (virus treatment) or with buffer alone (mock-treatment), both in the presence of Celite as an abrasive. Not-treated plants served as controls for the wounding response triggered by the mechanical treatment using an abrasive. One day (24 h) after mock or virus treatments, the treated leaves and the corresponding leaves of 2492



Article

analysis tool (http://metagenealyse.mpimp-golm.mpg.de/) was used. Principal component analysis (PCA) was applied to reduce the dimensionality of the data to a set of six and seven principal components (PCs) for 4 and 5 dpt data, respectively. ICA of this reduced data set led to the identification of independent components (ICs) that were ranked by the kurtosis measure.46 The algorithm used for ICA was CuBICA4.47 Functional classification of the proteins with consistent changes at 4and 5 dpt was performed using the classification super viewer tool48 (http://bar.utoronto.ca/welcome.htm) and MAPMAN.49

3 microscan/s and a scan time of 1 s. For MS/MS spectra, the m/z range was 150−2500 with CID activation and a scan time of 30 ms per MS2 acquisition. Charge state rejection was enabled for +1 with a peak detection window of 10 ppm. Dynamic exclusion was enabled with one count for 30 s after acquisition with a tolerance of 10 ppm. Exclusion duration was 45 s with list size of 300 precursor ions with no expiration. Rejection mass list contains 81 major peptides from Arabidopsis RuBisCO small and large subunits, porcine trypsin, and human keratin.

■

Peptide and Protein Identification

Tandem mass spectra were extracted and charge state deconvoluted by Mascot Distiller version 2.0. Deisotoping was not performed. All MS/MS samples were analyzed using Mascot (Matrix Science, London, U.K., server version 2.3) and X! Tandem independently (www.thegpm.org; version CYCLONE (2010.12.01.1)) searching the TAIR10 database. A total of 35 386 sequence entries within TAIR10 database were searched. Iodoacetamide derivatives of cysteine and oxidized methionine were specified as variable modifications. Scaffold (version Scaffold 3.6.0, Proteome Software, Inc., Portland, OR) was used to validate MS/MS-based peptide and protein identifications. Peptide and protein identifications44,45 were accepted by filtering for at least 95% peptide probability, at least two peptides and at least 99% protein probability. Proteins that contained similar peptides and could not be differentiated on the basis of MS/MS analysis were grouped.

RESULTS

Experimental Design and Protein Identification

Our objective was to investigate changes in protein abundance in distant leaves of Arabidopsis plants (i.e., leaves that never experienced the initial treatment) reacting to a systemic signal triggered by local wounding or virus infection. Rosette leaves of Arabidopsis plants were either not treated or treated by mechanical inoculation with buffer (mock treatment) or with virus-containing buffer (virus treatment). Both mock and virus treatments caused the wounding of the leaves since the inoculation procedure involved the rubbing of the leaves in the presence of abrasive. We expected that following these treatments, a systemic signal would move into upper, not-treated leaves either as a volatile compound, as a soluble factor along the source-sink connections, as electrical or hydraulic stimuli, or as some combination of these. To increase the likelihood that the systemic signal reached the distant, not-treated leaves, two leaves at opposite sides of the plant were treated. For analysis of the proteome in response to the systemic signal the three youngest, distal leaves were harvested at 4 and 5 days post treatment. To prevent the systemic spread of the virus before harvest, the locally mock- and virus-treated leaves (also the control leaves of not-treated plants) were removed at 24 h posttreatment (see experimental design summary in Figure 1). The

Data Analysis and Bioinformatics

The number of identified proteins differed between the independently performed experiments (Table S1, Supporting Information). To restrict the influence of the technical factors that caused these differences in total spectral counts (SCs), the fold-level changes of proteins occurring in response to the different treatments were calculated separately for each of the three experiments. To normalize between the treatments within each experiment the unweighted SCs for each protein were multiplied with adjust ratios (R) that were obtained by dividing the mean value of all SCs within the experiment by the mean value of all SCs within the treatment-specific sample. The resulting normalized SC (NSC) values were used for further analysis. Proteins robustly identified upon at least one treatment condition in all three experiments were divided into three categories according to their NSC: (a) proteins with a NSC no less than three (NSC ≥ 3); (b) proteins with a NSC greater than zero (NSC > 0); (c) proteins with a NSC no less than zero (NSC ≥ 0). For the category c proteins, zero NSC values were substituted with 0.5 to facilitate further calculation. Log2 values were calculated on the basis of NSC values for samples of virus-treated (infected) (I) and mock-treated (M) plants using samples of not-treated plants (N) as baseline, i.e., log2(I/N) and log2(M/N), respectively. Normality tests and tolerance interval calculations were then performed to create the final candidate protein list (MiniTab v.16.0). Briefly, two-sided tolerance limits were calculated for the log2(I/N) and log2(M/N) NSC values for each experiment. To calculate the upper tolerance limit (UTL) and lower tolerance limit (LTL), we chose a tolerance interval that covers a proportion of 90% of the proteins with a probability of γ = 0.95. Proteins falling outside these limits in at least two of the three experiments were considered to display significantly altered abundance upon the respective treatment. For independent component analysis (ICA) of the NSC data in Table S2 (Supporting Information) the MetaGeneAlyse v1.7.1

Figure 1. Experimental procedure. Treated leaves were removed at 1 dpt. Distant leaves were harvested at 4 and 5 days after treatment. The total protein extracts were loaded on a 10% SDS-PAGE gel to separate the complex proteome before staining with Colloidal Coomassie Blue (G-250). Gel bands were then trypsin-digested. Resulting peptides were extracted and lyophilized. The resolubilized peptide mixture was analyzed by LC−MS/MS.

absence of any systemic spread of the virus into untreated tissue was confirmed by the lack of ORMV symptoms in distal leaves remaining on the plants three weeks after sampling (data not shown). A second set of plants from which the inoculated leaves 2493



Article

were not removed showed typical, systemic ORMV symptoms (stunting and leaf curling) at this time, thus demonstrating that plants had been efficiently inoculated (100%, data not shown). Three temporal experiments, each with nottreated, mock-treated and virus-treated plants, were conducted. Total proteins isolated from the sampled young, distal leaves were separated by SDS-PAGE. Subsequently, the gel-separated proteins were cut into multiple bands, trypsin-digested, and subjected to LC−MS/MS analysis.

Figure 2. Number of proteins identified in distant leaves at 4 and 5 dpt. The Venn diagram depicts (A) all proteins identified at 4 and 5 dpt and (B) proteins reproducibly identified in all three experiments for at least one treatment (i.e., virus-treated, mock-treated, or nottreated).

Analysis of Protein Profiles

The proteins identified by mass spectrometry were viewed in Scaffold (version 3.6.0) and the selected filter settings (95% probability for peptide identification, SC for at least two peptides per protein identified, and 99% probability for protein identification) displayed 1474 and 1441 identified proteins for 4 and 5 dpt, respectively. We noticed that the total number of proteins identified for each of the three independently performed biological replicate experiments differed, which we assume is due to independent sample preparation, processing and analysis (Table S1, Supporting Information). On the other hand, the number of proteins obtained for each specific treatment within the respective experiment was comparable. Given these conditions, we decided to normalize and compare the protein data within each specific experiment rather than to average the data across all three experiments. Thus, for each experiment, the raw, unweighted SC data were exported from Scaffold and normalized (Table S2, Supporting Information). The unweighted SC values for each protein in each sample were adjusted for relative protein abundance by normalizing them to the average of all SC values in the experiment, giving NSC. To gain an overview about the biological relevance of the measured data, the normalized spectral counts (NSC) for all proteins identified at 4 and 5 dpt (Table S2, Supporting Information) were subjected to independent component analysis (ICA). ICA is an unsupervised method to visualize parameters in a given data set. Thus, ICA can assist in the interpretation of data by revealing experimental and biological parameters independently of experimental knowledge.46 ICA separated the samples of mock- and virus-treated plants from samples of not-treated plants for 4 dpt and all the different treatments for 5 dpt confirming the existence of biologically relevant information in the data set (Figure S1, Supporting Information). Moreover, these results demonstrate the development of a virus-specific response over time. Comparison of the proteins identified at 4 and 5 dpt revealed that the majority of the proteins (1289, 79%, Figure 2A) were present in samples from both time points. For quantitative proteome comparisons, only proteins that were found in each of the three experiments were selected. Thus, the subsequent analysis was restricted to the 598 and 566 proteins that were reproducibly identified at 4 dpt and 5 dpt, respectively (Figure 2B). To quantify the systemic response of Arabidopsis leaves to local virus- or mock-treatment, fold-changes in protein abundance were calculated from NSC values. Zero-NSC values were substituted with 0.5 to facilitate log calculation. Because undetected proteins (i.e., NSC = 0) can produce large ratio differences, we categorized and analyzed the data by considering specific protein abundance spectra. Thus, category a contained protein candidates with an NSC ≥ 3, category b contained protein candidates with an NSC > 0, and category c contained protein candidates with an NSC ≥ 0 (Figure 3A).

Figure 3. Overview about the data analysis procedure. (A) The data analysis scheme. Unweighted spectral counts (SC) were normalized by calculation of adjust ratios for individual proteins for each biological replicate experiment using the formula: R = (mean of SC [all treatments])/(mean of SC [specific treatment]). Multiplication of R with each individual SC value resulted in a normalized SC (NSC) value for each protein. Subsequently, three subdata sets were created to cover three distinct data populations: proteins with an NSC no less than three (NSC ≥ 3) were grouped in category a; proteins with an NSC greater than zero (NSC > 0) were grouped in category b; and proteins with an NSC no less than zero (NSC ≥ 0) were grouped in category c. Proteins with increased or decreased abundance upon virus treatment or mock-treatment were obtained by calculating the ratio between NSC values for treated versus not-treated samples and log2 transformation. Normality test and tolerance interval calculation were performed to identify significant changes in protein abundance. (B) Demonstration of the overlapping coverage of the three protein categories described in (A). (C) An example of a normal distributed data population with upper and lower tolerance limits. This example is shown to illustrate our approach to calculate significant differences in protein changes upon virus or mock treatment compared to nottreated controls. We calculated tolerance limits for tolerance intervals covering 90% of the log2 NSC(I/N) and NSC(M/N) values with 95% confidence. We then selected those protein fold changes that fell outside the limits, thus displaying higher or lower fold changes than the average as significant.

As would be expected, the more restrictive NSC ≥ 3 data retrieves a smaller subset of protein candidates compared to the total NSC ≥ 0 set (Figure 3B). However, all three NSC categorized 2494



Article

Table 1. Proteins with Significant Fold Abundance Changes in Distant Leaves of Virus-Treated and Mock-Treated Plants at 4 dpta fold change (log2) log2(I/N) protein name ATHDH, HISN8, HDH | histidinol dehydrogenase c ROC4 | rotamase CYP 4 a FLA1 | FASCICLIN-like arabinogalactan 1 a LOX2, lipoxygenase 2 a TGG2, BGLU37 | glucoside glucohydrolase 2 c

a Ribosomal protein S5 family protein a HSP60, HSP60−3B | heat shock protein 60 c Ribosomal protein S7e family protein c EMB2719, HAP15 | PAM domain (PCI/PINT associated module) protein c vacuolar ATP synthase subunit D c EMB1241 | Co-chaperone GrpE family protein c Amidase family protein c BOU | Mitochondrial substrate carrier family protein a BGL1, BGLU18, ATBG1 | beta glucosidase 18 c ADL6, DRP2A | dynamin-like protein 6 b APA1, ATAPA1 | aspartic proteinase A1

accession number

functional category

molecular weight (kDa)

3

responsive to

49

2.070

2.510

2.388

2.577

2.036

2.358

virus/mock

28 45

−0.243 −0.286

4.510 −0.603

4.710 −1.483

0.176 0.372

3.358 −0.814

0.000 −0.776

virus/mock virus/mock

102 63

1.729 0.497

0.132 0.764

1.540 0.292

1.080 0.754

0.725 0.723

1.748 0.834

virus/mock virus/mock

33

0.245

0.145

0.839

0.220

1.195

1.546

virus/mock

AT3G23990 protein

61

0.075

0.757

1.254

0.250

−0.270

1.098

virus/mock

AT3G02560 protein

22

−0.549

−2.042

−3.134

−0.950

−2.042

−0.361

virus/mock

AT1G20200 protein

56

−4.111

−0.703

2.388

−2.119

−2.627

0.000

virus/mock

AT3G58730 transport AT5G17710 protein

29 35

−3.526 0.129

−0.440 0.204

−3.134 −0.746

−3.526 −2.526

−0.744 −0.007

−1.361 −3.134

virus/mock mock

AT3G25660 tetrapyrrole synthesis AT5G46800 transport

57 31

−0.456 0.000

0.882 −0.703

−0.746 1.803

−2.526 1.992

−2.627 −0.270

−0.191 3.358

mock mock

AT1G52400 misc. gluco-, galacto-, mannosidases AT1G10290 misc. dynamin

60

0.677

0.690

0.477

0.994

0.993

0.224

mock

99

−1.008

3.247

3.611

−0.239

0.000

3.095

virus

AT1G11910 protein

55

0.866

0.882

−0.331

0.050

0.408

−1.361

virus

AT5G63890 amino acid metabolism AT3G62030 cell AT5G55730 cell wall AT3G45140 hormone metabolism AT5G25980 hormone metabolism/secondary metabolism AT2G33800 protein

1

2

log2(M/N) 3

1

2

a

Bold and underlined numbers are log2 values that exceed the upper and lower limit of the 90% tolerance interval for each experiment, respectively. I: virus-treated plants; M: mock-treated plants; N: not-treated plants; a, b, c: proteins found outside of the 90% tolerance interval (p-value 0; and (c) NSC ≥ 0. Functional categories were assigned according to MAPMAN49 annotation.

Gfeller et al.,32 we analyzed leaves distant from the treated leaf. Wounding responses, for example, have been demonstrated to diminish with increasing distance from the damaged site.32,50

subdata sets (a, b and c in Figure 3A,B) composed of log2 values passed the normality test with a p-value of

Label-Free Quantitative Proteomic Analysis of Systemic Responses to

Recommend Documents