Article pubs.acs.org/jpr
Components of Mitochondrial Oxidative Phosphorylation Vary in Abundance Following Exposure to Cold and Chemical Stresses Yew-Foon Tan,† A. Harvey Millar,†,‡ and Nicolas L. Taylor*,†,‡ †
ARC Centre of Excellence in Plant Energy Biology and ‡Centre for Comparative Analysis of Biomolecular Networks (CABiN), MCS Building M316, The University of Western Australia, 35 Stirling Highway, Crawley WA 6009, Western Australia, Australia S Supporting Information *
ABSTRACT: Plant mitochondria are highly responsive organelles that vary their metabolism in response to a wide range of chemical and environmental conditions. Quantitative proteomics studies have begun to allow the analysis of these large-scale protein changes in mitochondria. However studies of the integral membrane proteome of plant mitochondria, arguably the site responsible for the most fundamental mitochondrial processes of oxidative phosphorylation, protein import and metabolite transport, remain a technical challenge. Here we have investigated the changes in protein abundance in response to a number of chemical stresses and cold. In addition to refining the subcellular localization of 66 proteins, we have been able to characterize 596 protein × treatment combinations following a range of stresses. To date it has been assumed that the main mitochondrial response to stress involved the induction of alternative respiratory proteins such as AOX, UCPs, and alternative NAD(P)H dehydrogenases; we now provide evidence for a number of very specific protein abundance changes that have not been highlighted previously by transcript studies. This includes both previously characterized stress responsive proteins as well as major components of oxidative phosphorylation, protein import/export, and metabolite transport. KEYWORDS: Arabidopsis, proteome, mitochondria, oxidative phosphorylation, cold, hydrogen peroxide, menadione, antimycin A, quantitative proteomics, normalized spectral abundance factor (NSAF)
■
INTRODUCTION Arabidopsis thaliana mitochondria show a highly dynamic response when a plant cell is exposed to chemical or environmental stresses. To date this response has been studied at the transcript level using microarrays to monitor large-scale gene expression during plant stress1−4 and at the protein level in response to environmental stresses5−13 and chemically induced oxidative stress.14,15 Transcript studies have highlighted the lack of transcriptional responses to stress among the classical membrane bound respiratory chain components but large transcription responses in the alternative nonphosphorylating electron transport pathways.1,16 However, the protein studies have generally been limited to studying the soluble mitochondrial proteins and membrane associated hydrophilic proteins, owing to limitations in the traditional gel based techniques to characterize hydrophobic integral membrane proteins,17 and the predominance of hydrophilic peptide identification in shotgun LC−MS/MS experiments.18,19 An exception to this has been the use of BN-PAGE that allows the exploration of large intact protein complexes and their subunit constituents.20 However, BN-PAGE only assesses the protein subunits for these complexes that are assembled, and many of the low abundance and small protein complexes are not resolved by BNPAGE. Thus, the hydrophobic mitochondrial proteome remains a challenge due to low protein abundance of many, the problems with protein insolubility, the predominance of analysis of hydrophilic © 2012 American Chemical Society
peptides, and the characteristics of standard 2D IEF/SDS-PAGE methods. The mitochondrial membranes contain integral proteins that carry out arguably the most fundamental mitochondrial functions including the mitochondrial electron transfer chain, ATP synthesis, protein import and metabolite transport. These membranes provide a barrier to allow the generation of concentration and pH gradients and also contain proteins that allow selective transport of metabolites and precursor proteins within mitochondrial sub compartments. For example, the major mitochondrial process of oxidative phosphorylation is carried out by integral membrane proteins of the electron transfer chain that are imported into the mitochondria as precursors, mostly encoded by the nucleus. These proteins act to generate a proton gradient across the inner mitochondrial membrane that is harnessed by ATP synthase to generate ATP. This ATP is then transported out of the mitochondria to other parts of the cell by another integral membrane protein the ATP/ADP antiporter (adenine nucleotide transporter; ANT). ANT is a member of the mitochondrial carrier substrate family of proteins (MCSF), and previous work has highlighted the difficulty in characterizing integral membrane proteins, as only 6 out of 58 predicted members of the MCSF Received: April 11, 2012 Published: May 11, 2012 3860
dx.doi.org/10.1021/pr3003535 | J. Proteome Res. 2012, 11, 3860−3879
Journal of Proteome Research
Article
protein family in Arabidopsis have been identified.21 The most comprehensive analysis of the Arabidopsis mitochondrial hydrophobic proteome performed by LC−MS/MS analysis to date has yielded 114 nonredundant constitutively expressed proteins.22 In that study membrane proteins were isolated from control mitochondria and hydrophobic proteins isolated using three different techniques; 31 proteins were found by more than one method, while the remainders were found by only one isolation approach. However, given that study occurred some time ago, there were few opportunities for quantitative controls to exclude contaminants from other Arabidopsis membrane systems in the approach. It is likely that integral membrane proteins are differentially expressed under conditions of environmental stress, but this possibility has not been extensively investigated, and little quantitative data exists to substantiate such claims. A few studies have shown changes in abundance in components of oxidative phosphorylation using IEF-SDS PAGE gels including components of complex I and complex V.15,23 However when BNPAGE gels were run of the same samples in one study, no significant changes in protein abundance could be observed.23 This may suggest that along with the known deficiencies of BNPAGE such as the inability to resolve smaller respiratory chain components such as complex II, alternative oxidase and alternative dehydrogenases, it may also lack the resolution to see small but significant changes in protein abundance of the large subunits. We have previously shown that a detailed study of specific peptides was required to show a mitochondrial basic amino acid carrier (BAC) increases in abundance under aerobic conditions in rice, when no transcript response was evident, and this allowed us to highlight its role in mitochondrial arginine metabolism and the urea cycle during growth at low oxygen.24 Also as a response to stress conditions it is likely that new proteins must be imported, leading to changes in the import apparatus, and allowing the possibility of identifying mitochondrial membrane proteins that have not previously been documented. Further, any small molecules that are required in the stress response, which cannot be synthesized internally, must be transported through the outer and inner membranes into mitochondria, and thus significant changes in the MCSF protein profile of mitochondria may occur to facilitate this. In addition, chemical stress leads to protein degradation,15 and as many proteases are membrane-bound, we might expect to observe changes in the protease profile of mitochondria providing new leads on the roles of different protease classes. Here we have investigated the changes in protein abundance of integral membrane proteins in response to chemical and environmental stresses. We have identified 191 mitochondrial integral membrane proteins, 66 of which had not previously been confirmed to be present in Arabidopsis thaliana mitochondria. Some of these novel mitochondrial proteins had previously been claimed to be localized in other organelle preparations (40); however, using a quantitative enrichment approach, or new literature evidence, we can now confidently claim them as mitochondrial, while the localization of other proteins had not been previously determined (26). We have used a quantitative proteomics approach to profile 596 protein × treatment combinations in protein abundance that provide information on 144 distinct integral membrane proteins in response to antimycin A, menadione, hydrogen peroxide, CuCl2, and cold. This study reveals that while it is true the membrane proteome is relatively more stable than the soluble proteome, components of oxidative phosphorylation can be shown to significantly change in abundance in response to chemical and
environmental stresses. Associated with these changes are variations in components of mitochondrial protein import, members of the MCSF and proteases. Together these suggest that even though the key mitochondrial process of oxidative phosphorylation are not transcriptionally responsive to stress conditions, at a protein level their abundance is tuned in response to changing external conditions, likely in an attempt to maintain respiratory homeostasis.
■
EXPERIMENTAL PROCEDURES
Propagation of Arabidopsis thaliana Cell Suspension Cultures
The heterotrophic cell culture used was established from callus of Arabidopsis thaliana (cv. Landsberg erecta). Cells were maintained under continuous light at 22 °C in an orbital shaker (Thermoline Scientific, Australia) at 125 rpm. Cells to be used for the isolation of mitochondria were cultivated in the dark to impair plastid development in order to minimize plastid contamination during isolation of mitochondria. The cultures were maintained under aseptic conditions by removing a 20 mL aliquot of 7-day old culture and adding to 100 mL of fresh medium containing Murashige and Skoog Basal Salt mixture (Phytotechnology Laboratories), 3% (w/v) sucrose, 0.5 mg/L of naphthalene acetic acid and 0.05 mg/L of kinetin, pH 5.8. Induction of Oxidative Stress in Arabidopsis Cell Culture
Prior to the isolation of mitochondria, 7-day old dark-grown cell suspension cultures were treated with either 100 μM CuCl2, 10 mM H2O2, 400 μM menadione (in methanol) or 25 μM antimycin A (in methanol) for 8 h. Equivalent volumes of either water or methanol were added as controls for respective treatments. For cold treatments 5 day old dark-grown cell suspension cultures were placed at 4 °C for 48 h. The concentrations of antimycin A and menadione have previously been optimized.25 Isolation of Mitochondria from Arabidopsis Cell Cultures
Mitochondria were isolated from Arabidopsis thaliana cell suspensions as previously described.26 Briefly, all procedures were performed at 4 °C to preserve mitochondrial integrity and protein function. Cell suspension cultures were cooled on ice to 4 °C for 15 min before collection by filtration through one layer of Miracloth (Calbiochem). The cells were homogenized in batches of 50 g FW by 1 high speed and 2 low speed 15 s bursts in a precooled Warring blender in 150 mL of extraction medium (0.45 M mannitol, 50 mM tetrasodium pyrophosphate, 0.5% (w/v) PVP-40, 2 mM EGTA, 0.5% (w/v) BSA and 20 mM cysteine, pH 8.0). The homogenate was filtered through 3 layers of Miracloth and 1 layer of Muslin to remove cell debris. Superfluous cell debris was pelleted and discarded following centrifugation at 1500g for 5 min. The supernatant was retained and centrifuged at 24000g for 15 min to form a crude organelle pellet. The crude pellet was resuspended in a small volume of wash medium containing BSA (0.3 M mannitol, 10 mM TES, 0.1% (w/v) BSA), which was then layered over an 18/25/40% step Percoll gradient made in wash medium containing BSA. This gradient was centrifuged at 40000g for 30 min, to form a mitochondrially enriched band at the 25 and 40% Percoll interface. This band was carefully aspirated, placed in new centrifuge tubes, and diluted in 5 times the volume with wash medium for centrifugation at 24000g for 15 min. The supernatant was discarded, and the resulting pellet was overlaid onto a self-forming 35% Percoll gradient made in buffer containing 0.3 M sucrose, 10 mM TES-NaOH, pH 7.5, and 0.1% (w/v) BSA and centrifuged at 40000g for 30 min. The purified mitochondrial band near the top of the gradient was retained 3861
dx.doi.org/10.1021/pr3003535 | J. Proteome Res. 2012, 11, 3860−3879
Journal of Proteome Research
Article
Build 65.4, Patches 1,2,3,4, Agilent Technologies). Peptides were loaded onto the trapping column at 4 μL min−1 in 5% (v/v) acetonitrile and 0.1 (v/v) % formic acid with the chip switched to enrichment and using the capillary pump. The chip was then switched to separation, and peptides eluted during a 1 h gradient (5−40% Acetonitrile) directly into the mass spectrometer. The mass spectrometer was run in positive ion mode, and MS scans run over a range of m/z 275−1500 and at 4 spectra s−1. Precursor ions were selected for auto MS/MS at an absolute threshold of 500 and a relative threshold of 0.01, with max 3 precursors per cycle, and active exclusion set at 2 spectra and released after 1 min. Precursor charge-state selection and preference was set to 2+ and 3+, and precursors selected by charge then abundance. Resulting MS/MS spectra were opened in MassHunter Workstation Qualitative Analysis (ver B.01.02, Build 1.2.122.1, Patch 3 Agilent Technologies), and MS/MS compounds detected by “Find Auto MS/MS” using default settings. The resulting compounds were then exported as .mzdata files, which were then searched against an in-house Arabidopsis database comprising ATH1.pep (release 9) from The Arabidopsis Information Resource (TAIR). This sequence database contained a total of 33 621 protein sequences (13 487 170 residues). Searches were conducted using the Mascot search engine version 2.2.03 (Matrix Science) utilizing error tolerances of ±100 ppm for MS and ±0.5 Da for MS/MS, “Max Missed Cleavages” set to 1, the Oxidation (M), Carboxymethyl (C), variable modifications and the Instrument set to ESI-Q-TOF and Peptide charge set at 2+ and 3+. Results were filtered using “Standard scoring”, “Max. number of hits” set to 20, “Significance threshold” at p < 0.05 and “Ions score cutoff” at 0. Ions with a score >32 (P < 0.05) were considered high confidence matches and were used for further analysis. Data files (.mzdata) have been annotated to according to MIAPE (v1.0) and submitted to the Proteome Commons Tranche Repository (https:// proteomecommons.org/tranche/). This data can be retrieved by using the following hash: 3p7v7fFisLOqswtj+w23/LWRiSLPlxWwiXxmooOkuQr4BykUlFmgxpG7NLx8+eVkQj4uwliuAXp25q4Kz2+yqdPRHfQAAAAAAAAT7g==.
by careful aspiration and transferred to new tubes where it was washed in 5 times the volume in wash medium without BSA by centrifugation at 24000g for 15 min. A yield of 10 mg of mitochondrial protein is typically obtained from 100 g of fresh weight of cell suspension culture. A low purity mitochondria fraction was obtained by omitting the second gradient. A second dimension of purification of mitochondria was also achieved by free-flow electrophoresis by the same method outlined in Huang et al.27 resulting in an even higher purity mitochondrial sample (FFE mitochondria). Isolation of the Integral Membrane Protein Fraction
Mitochondria were resuspended in 500 μL of Milli-Q water and subjected to 3 freeze/thaw cycles in liquid nitrogen. The lysed mitochondria were then centrifuged at 20000g for 10 min. The pellet was resuspended in 500 μL of Milli-Q water, and the freeze/thaw cycles and centrifugation were repeated to ensure complete lysis of mitochondria. The resulting pellet, the total membrane fraction, from lysed mitochondria was then treated with 0.1 M Na2CO3 to electrostatically displace peripherally attached proteins. This salt and alkaline extraction of integral membrane proteins was performed according to Fujiki et al.28,29 The pellet was resuspended in 100 μL of 0.1 M Na2CO3 by pipetting. An additional 900 μL of 0.1 M Na2CO3 was added to the solution and vortexed gently to resuspend the pellet. The solution was centrifuged at 20000g for 10 min, and the supernatant discarded. The pellet was again resuspended in 100 μL of 0.1 M Na2CO3 and then made up to 1 mL using 0.1 M Na2CO3. The lysate was vortexed gently to ensure thorough depletion of peripheral proteins. The solution was then centrifuged at 20000g for 10 min, and the supernatant discarded. The pellet was retained as the integral membrane protein fraction. Integral membrane protein pellets were rinsed 5 times in water to remove Na2CO3 that may cause ionic suppression during MS analysis. Delipidation of Integral Membrane Protein Samples
Prior to trypsin digestion, membrane proteins were delipidated and precipitated using chloroform and methanol as described by Wessel et al.30 All solutions used were ice-cold. Briefly, the integral membrane protein pellet was resuspended in 100 μL of 10% (w/v) SDS. At this point, the protein content of small aliquots of solution were estimated by the modified FolinCiocalteu method.31 Then, to the resolubilised pellet, 4 volumes of methanol, 1 volume of chloroform and 2.5 volumes of Milli-Q water were added with vortexing between the additions of each solvent. The solution was centrifuged at 10000g for 5 min creating a proteinaceous layer separating the upper and the lower aqueous phase. The upper layer was carefully aspirated and discarded. To the remaining solution, 1 mL of methanol was added and vortexed to wash the pellet. The washed protein pellet was collected by centrifugation at 9000g for 10 min. The supernatant was discarded, and the pellet was air-dried.
Quantitative Assessment of Protein Abundance by Spectral Counting
A spectral counting method was used as a means to quantitatively assess for the variation in membrane protein expression following exposure to various forms of oxidative stress. To do this, 3 independent biological replicates of mitochondrial integral membrane proteins were extracted from each individual treatment and compared to an appropriate control. The changes in abundance were determined in samples where a protein could be found in all three of the treatment samples. Proteins detected once or twice in a set of replicates were discarded, except when a protein was not found in all three replicates. The mitochondrial origin of each sample was then confirmed by its detection in ≥2 mitochondrial proteomic or GFP studies, by quantitative enrichment/quantitative depletion (QE/QD) or from other published evidence. For a positive mitochondrial call from QE, peptide enrichment was required from either the “dirty” to control sample or from control to “FFE” sample. The presence of predicted membrane spanning regions of proteins were determined by interrogation of the Aramemnon database.32 Proteins that had ≥1 predicted transmembrane domain or other evidence indicating membrane localization were retained, and other proteins were excluded. Calculations of protein abundance changes were made by utilizing the natural log transformation of the normalized spectral abundance factor (NSAF).33 This allowed statistical analysis of the noncontinuous
LC−MS Analysis of Integral Membrane Protein Samples
Integral membrane protein samples were analyzed with a nongel approach, using complex mixture LC−MS/MS analysis. The protein extracts were digested overnight at 37 °C with trypsin, and insoluble components were removed by centrifugation at 20000g for 5 min. The complex lysates were analyzed on an Agilent 6510 Q-TOF mass spectrometer with an HPLC Chip Cube source. The chip consisted of a 160 nL enrichment column (Zorbax 300SB-C18 5 μm) and a 150 mm separation column (Zorbax 300SB-C18 5 μm) driven by Agilent Technologies 1100 series nano/capillary liquid chromatography system. Both systems were controlled by MassHunter Workstation Data Acquisition for Q-TOF (ver B.01.02, 3862
dx.doi.org/10.1021/pr3003535 | J. Proteome Res. 2012, 11, 3860−3879
Journal of Proteome Research
Article
data sets by t-tests carried out using Microsoft Excel. Also to prevent zero errors in the natural log transformation, zero spectral count values were replaced with an empirically derived fractional value of 0.65. This value was obtained in an iterative process by selecting the smallest possible value between 1 and 0 that does not significantly alter the sum and maintains the normal distribution of the ln(NSAF).33 ln(NSAF) distribution was evaluated by a ShapiroWilk test using the Analyze-it (www.analyze-it.com) add on in Microsoft Excel.
combinations that were detected in all three replicates of a stress treatment. In all, 596 protein × treatment combinations met these criteria. To calculate protein abundance using the normalized spectral abundance factor (NSAF) method, spectral counts of 0 were replaced by a fractional spectral value of 0.65 that represented the average value between 0 and 1 that did not significantly alter the normal distribution of each of the individual replicate with all of the treatment and control groups.33 From this, NSAFκ33 and logNSAF33 were calculated for each of the control and treatment replicates and Student’s t test carried out on spectral count data for each protein using the biological replicate data sets. Of the 596 protein × treatment combinations, involving 144 distinct proteins (Supporting Information Table S2), 28 protein × treatment combinations showed statistically significant (p < 0.05) increases (Table 2), and 66 protein × treatment combinations showed statistically significant (p < 0.05) decreases in abundance (Table 3). In total, 503 protein × treatment combinations showed no statistically significant (p < 0.05) change (Supporting Information Table S3). Thus, while the majority of the membrane proteome was relatively homeostatic, clear examples of proteins changing in abundance were observed. Cold stress led to 19 abundance decreases, and we could confirm that a further 86 proteins did not change in abundance. Similarly, hydrogen peroxide led to 3 proteins increasing in abundance, 12 decreases and 85 proteins whose abundance was unchanged. The methanol control for antimycin A and menadione showed increases in 1 protein and decreases in 4 proteins, while 68 proteins remained unchanged compared to the water control. Antimycin A treatment led to 6 protein increases and 4 protein decreases compared to the methanol control, and 88 proteins were stable in abundance. Treatment with menadiaone induced 17 increases in abundance compared to the methanol control, while 111 proteins could be confirmed as not changing in abundance. Finally the copper treatment leads to a single protein abundance increase, but 27 proteins were observed to decrease in abundance, and 65 remained stable in abundance. Of the proteins that showed a statistically significant changes, 43% were proteins that had not previously had information on subcellular location reported or had had their location as mitochondrial confirmed by quantitative enrichment in this study. Most of the proteins observed to statistically significant change, appeared to be relatively low abundance proteins in mitochondria. The only major mitochondrial proteins (greater that 0.5% of total mitochondrial protein17) that decreased in abundance were all ATP synthase subunits (ATP1, ATP2, ATP5 and ATP16), and these proteins represented only 10% of the proteins that showed a statistically significant change in abundance. Hierarchical clustering was carried out on the abundance of the 144 proteins that were found in all three replicates of a treatment or control samples in order to assess any coordinated responses or trends of groups of proteins to the various stress treatments. To do this, the number of observed peptides for each protein in its three replicates were summed and represented as a proportion of the highest number of observed peptides from any treatment; thus, 1 (red) represents the treatment where a particular protein was at it greatest abundance, and 0 (green) represent protein × treatment combinations that had no peptides detected in a treatment, and gray boxes represent treatments where
