RESEARCH ARTICLE A parallel proteomic and metabolomic analysis of the hydrogen peroxide- and Sty1p-dependent stress response in Schizosaccharomyces pombe Mark E. Weeks 1 , John Sinclair 1 , Amna Butt 2 , Yuen-Li Chung 3 , Jessica L. Worthington 2 , Caroline R. M. Wilkinson 2 , John Griffiths 3 , Nic Jones 2 , Michael D. Waterfield 1 and John F. Timms 1, 4 1 Ludwig Institute for Cancer Research, University College London, Cruciform Building, London, UK 2 Paterson Institute for Cancer Research, Christie Hospital, Manchester, UK 3 Department of Basic Medical Sciences, Medical Biomics Centre, St. George’s Hospital Medical School, Cranmer Terrace, London, UK 4 Department of Biochemistry and Molecular Biology, University College London, Cruciform Building, London, UK Using an integrated approach incorporating proteomics, metabolomics and published mRNA data, we have investigated the effects of hydrogen peroxide on wild type and a Sty1p-deletion mutant of the fission yeast Schizosaccharomyces pombe. Differential protein expression analysis based on the modification of proteins with matched fluorescent labelling reagents (2-D-DIGE) is the foundation of the quantitative proteomics approach. This study identifies 260 differentially expressed protein isoforms from 2-D-DIGE gels using MALDI MS and reveals the complexity of the cellular response to oxidative stress and the dependency on the Sty1p stress-activated protein kinase. We show the relationship between these protein changes and mRNA expression levels identified in a parallel whole genome study, and discuss the regulatory mechanisms involved in protecting cells against hydrogen peroxide and the involvement of Sty1p-dependent stress-acti- vated protein kinase signalling. Metabolomic profiling of 29 intermediates using 1 H NMR was also conducted alongside the protein analysis using the same sample sets, allowing examination of how the protein changes might affect the metabolic pathways and biological processes involved in the oxidative stress response. This combined analysis identifies a number of inter- linked metabolic pathways that exhibit stress- and Sty1-dependent patterns of regulation. Received: October 11, 2005 Revised: November 17, 2005 Accepted: November 23, 2005 Keywords: Hydrogen peroxide / Metabolomics / Redox stress 2772 Proteomics 2006, 6, 2772–2796 1 Introduction The stress responses by which organisms react to environ- mental changes are complex and involve the regulation of many genes [1–4]. Studies of stress responses in human cell lines has led to apparent contradictory results because of their complex regulatory networks, making the use of sim- pler model cell systems more desirable [5]. The fission yeast Schizosaccharomyces pombe is one such model organism for which the complete genome has been sequenced [6] and which can be easily genetically manipulated [7]. In particular, its stress-activation pathways share homology with higher organisms and transduce signals to the nucleus, resulting in Correspondence: Dr. John F. Timms, Ludwig Institute for Cancer Research, University College London, Cruciform Building, Gower Street, London WC1E 6BT, UK E-mail: [email protected]Fax: 144-20-7679-6334 Abbreviations: CESR, core environmental stress response; Cy2, 3- (4-carboxymethyl) phenylmethyl-3’-ethyloxacarbocyanine halide; Cy3, 1-(5-carboxypentyl)-1’-propylindocarbocyanine halide; Cy5, 1-(5-carboxypentyl)-1’-methylindocarbocyanine halide; 2-D-DIGE, 2-D difference gel electrophoresis; GPC, glycerophosphocholine; GSH, glutathione; Hsp, heat shock protein; NHS, N-hydroxysuccin- imidyl; ROS, reactive oxygen species; TMAO, trimethylamine-N- oxide; WT , wild-type DOI 10.1002/pmic.200500741 2006 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
25
Embed
A parallel proteomic and metabolomic analysis of the hydrogen peroxide- and Sty1p-dependent stress response inSchizosaccharomyces pombe
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
RESEARCH ARTICLE
A parallel proteomic and metabolomic analysis of the
hydrogen peroxide- and Sty1p-dependent stress
response in Schizosaccharomyces pombe
Mark E. Weeks1, John Sinclair1, Amna Butt2, Yuen-Li Chung3, Jessica L. Worthington2,Caroline R. M. Wilkinson2, John Griffiths3, Nic Jones2, Michael D. Waterfield1
and John F. Timms1, 4
1 Ludwig Institute for Cancer Research, University College London, Cruciform Building, London, UK2 Paterson Institute for Cancer Research, Christie Hospital, Manchester, UK3 Department of Basic Medical Sciences, Medical Biomics Centre, St. George’s Hospital Medical School,
Cranmer Terrace, London, UK4 Department of Biochemistry and Molecular Biology, University College London, Cruciform Building, London, UK
Using an integrated approach incorporating proteomics, metabolomics and published mRNAdata, we have investigated the effects of hydrogen peroxide on wild type and a Sty1p-deletionmutant of the fission yeast Schizosaccharomyces pombe. Differential protein expression analysisbased on the modification of proteins with matched fluorescent labelling reagents (2-D-DIGE) isthe foundation of the quantitative proteomics approach. This study identifies 260 differentiallyexpressed protein isoforms from 2-D-DIGE gels using MALDI MS and reveals the complexity ofthe cellular response to oxidative stress and the dependency on the Sty1p stress-activated proteinkinase. We show the relationship between these protein changes and mRNA expression levelsidentified in a parallel whole genome study, and discuss the regulatory mechanisms involved inprotecting cells against hydrogen peroxide and the involvement of Sty1p-dependent stress-acti-vated protein kinase signalling. Metabolomic profiling of 29 intermediates using 1H NMR wasalso conducted alongside the protein analysis using the same sample sets, allowing examinationof how the protein changes might affect the metabolic pathways and biological processesinvolved in the oxidative stress response. This combined analysis identifies a number of inter-linked metabolic pathways that exhibit stress- and Sty1-dependent patterns of regulation.
Received: October 11, 2005Revised: November 17, 2005
Accepted: November 23, 2005
Keywords:
Hydrogen peroxide / Metabolomics / Redox stress
2772 Proteomics 2006, 6, 2772–2796
1 Introduction
The stress responses by which organisms react to environ-mental changes are complex and involve the regulation ofmany genes [1–4]. Studies of stress responses in human celllines has led to apparent contradictory results because oftheir complex regulatory networks, making the use of sim-pler model cell systems more desirable [5]. The fission yeastSchizosaccharomyces pombe is one such model organism forwhich the complete genome has been sequenced [6] andwhich can be easily genetically manipulated [7]. In particular,its stress-activation pathways share homology with higherorganisms and transduce signals to the nucleus, resulting in
Correspondence: Dr. John F. Timms, Ludwig Institute for CancerResearch, University College London, Cruciform Building, GowerStreet, London WC1E 6BT, UKE-mail: [email protected]: 144-20-7679-6334
altered patterns of gene expression that are critical for theresponse to environmental stresses such as heat, hyper-osmolarity and oxidative stress. A central element of thestress-activated protein kinase cascade in S. pombe is theprotein kinase Sty1p (Spc1/Phh1p), a homologue of mam-malian p38 kinase, which becomes activated in response tosimilar stresses and whose inactivation results in a pleio-tropic sensitivity to stress [8–10]. Sty1p regulates stress-de-pendent gene transcription, at least in part, through thedirect phosphorylation of the b-ZIP transcription factor,Atf1p which anchors Sty1p in the nucleus [11–14]. Impor-tantly, atf1 mutants show defects in Sty1p-dependent tran-scription, but only display a subset of the phenotypes dis-played by sty1 mutants, suggesting that Sty1p controls otheras yet unidentified proteins and pathways [11, 13].
In the study of oxidative stress, it is particularly relevantthat each human cell metabolizes approximately 1012 mole-cules of oxygen per day with 1% of oxygen metabolismresulting in the production of reactive oxygen species (ROS),such as the superoxide anion (O2
-), the hydroxyl radical(?OH) and hydrogen peroxide (H2O2) [15]. Because of theirhigh reactivity, ROS can bring about cellular damage to var-ious macromolecules. For example, oxidative DNA damagecan alter purine and pyrimidine bases, as well as cleave thephosphodiester DNA backbone leading to genetic mutation[15], whilst oxidation of protein cysteine thiol groups cancause intermolecular protein cross-linking and enzymeinactivation which potentially lead to cell death. Importantly,ROS have been implicated in a number of human diseases[16–18], and for example, it is recognized that many types ofcancer cells exhibit increased production of H2O2 that hasbeen linked to proliferative signalling and tumourigenesis[19–22], although low levels of H2O2 appear to be required fornormal proliferative signalling [23].
Cells have evolved a number of mechanisms to counteractoxidative damage including the direct reversal of mutationsthrough mismatch repair and DNA excision pathways [24], theswift elimination of ROS and the reversal of oxidative damage.The latter are achieved through the induction of antioxidantand redox enzymes such as catalase, superoxide dismutase,peroxidases and thioredoxin, through the maintenance of highlevels of molecular scavengers such as glutathione (GSH) andascorbic acid and by S-thiolation, whereby oxidized thiolgroups form mixed disulphides with GSH which are thenregenerated by glutaredoxins and GSH reductase [25, 26]. Thiolgroups are reported to have numerous roles within the cell andtheir redox states affect the activity and structure of many pro-teins including transcription factors, proteases and phospha-tases. It is not surprising, therefore, that all organisms containregulatory machinery whose purpose is to maintain the redoxstatus of SH groups in both proteins and low-molecular-massthiols [27–29]. In yeast, it has been shown that basal protein S-thiolation is maintained at a low level, but is increased duringoxidative stress where it serves an adaptive function by repro-gramming metabolism and protecting protein synthesisagainst irreversible oxidation [30, 31].
Traditionally, molecular and cellular studies have tendedto concentrate on individual genes and their products, al-though more recently, several microarray studies havedetailed the global genomic responses of yeast to a variety ofstresses including oxidative stress [1–3, 32]. In addition, theglobal analysis of gene products by proteomics has alloweda systematic overview of thousands of proteins at the sametime. 2-DE is one of the most widely used proteomicseparation methods and is often employed for the analysisof differential protein expression across biological samples[33, 34]. A significant improvement came with the intro-duction of 2-D-difference gel electrophoresis (2-D-DIGE), inwhich several samples can be codetected on the same 2-DEgel after differential covalent labelling with matched fluo-rescent tags [35–37]. By alleviating the problems of gel-to-gelvariation, the 2-D-DIGE strategy provides improved accu-racy in quantifying protein differences between samples. Afurther level of global investigation has been realized withthe advent of metabolomics where, for example, NMRspectroscopy can be used to measure changes in the levelsof multiple low-molecular weight metabolites across bio-logical samples. Moreover, by integrating data from metab-olite analysis, gene expression profiling and proteomics, keyperturbed metabolic pathways can be identified providing agreater understanding of the biological processes involvedin the response to environmental changes. Metabolomicshas already been used to study toxicological mechanismsand disease processes and offers great potential as a meansof investigating the complex relationship between stressand metabolism [38, 39].
In this study, we have used 2-D-DIGE to examine prote-omic changes generated by the application of peroxide stressto fission yeast and have compared this data with thatobtained from transcriptional and metabolomic profilingacquired under identical growth conditions. The degree towhich each data set correlates and how the changes relate toaltered biological function is shown and discussed. Further-more, we have examined the effects of the loss of the stress-dependent MAP kinase Sty1 on the peroxide stress responseand have been able to differentiate altered gene expressionstates and metabolic pathways between the unstressed andperoxide-stressed cell types.
2 Materials and methods
2.1 Growth and hydrogen peroxide treatment of
S. pombe
Yeast strains used were wild-type (WT) strain 972 h2 and anisogenic Sty1p-deleted mutant strain sty1D (sty1::ura41 ura4-D18 h2). Two litre cultures of WT or sty1D in yeast extract(YE) medium [40] were grown to mid-exponential phase(56106 cells/mL), and subjected to oxidative stress by addi-tion of 0.5 mM H2O2 for 60 min or left untreated. Cells wereharvested by centrifugation (580g), washed twice in water
2774 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
and re-suspended in 1 mL of grinding/resuspension buffer(10 mM HEPES–HCl pH 7.4, 0.1% IGEPAL CA-630 (pre-viously NP 40), 2 mM EDTA, 4 mg/mL leupeptin, 4 mg/mLaprotinin and 5 mg/mL pepstatin A). Cell suspensions wereimmediately frozen by addition of liquid N2.
2.2 Cell lysis and CyDye-labelling of protein extracts
Frozen cells were broken by grinding for 20 min underliquid N2 in a RM100 mortar grinder (Retsch, Germany) asdescribed [41] and ground lysates stored at 2807C. ForNHS-Cy dye labelling 100 mL aliquots of the lysates weresuspended in 2-D buffer (8 M urea, 2 M thiourea, 4%CHAPS, 0.5% IGEPAL CA-630, 10 mM Tris-HCl pH 8.3)without reductant to give a sample volume of 600 mL. Sam-ples were homogenized by passing through a 25-gaugeneedle six times prior to agitation at room temperature for10–15 min. Insoluble material was removed by centrifuga-tion (12 0006g/10 min/47C). Protein concentrations weredetermined using Coomassie Protein Assay Reagent(Pierce) and a BSA standard curve with four replicate assaysperformed per sample.
A sample amount of 150 mg was aliquoted into tubes forNHS-Cy dye labelling. Equal amounts of protein from eachsample were also mixed to create an internal standard to belabelled with Cy2 and run on each gel. NHS-Cy2 was pur-chased from GE Healthcare (Amersham, UK) whilst NHS-Cy3 and Cy5 were synthesized in-house and stored asdescribed [42]. They are structurally identical to the com-mercially available compounds and highly pure. Sampleswere labelled by addition of 4 pmol of NHS-Cy dye permicrogram of protein (600 pmol/150 mg) and incubation onice in the dark for 30 min. Reactions were quenched with a20-fold molar excess of L-lysine to dye and further incubationon ice in the dark for 10 min. Cy3 and Cy5 were randomlyassigned to biological triplicate samples from separatelygrown and treated cultures for the four conditions underinvestigation. Cy3- and Cy5-labelled samples were mixedappropriately (Table 1) and the same amount (150 mg) of theCy2-labelled pool was added. Samples were reduced withDTT at a final concentration at 65 mM. Carrier ampholines/pharmalyte mixture was added to a final 2% v/v and bromo-phenol blue added. The volume was adjusted to 450 mL withlysis buffer and the samples were agitated prior to cen-trifugation at 12 0006g for 10 min.
Table 1. Typical experimental design showing allocation of Cydye labelling of biological replicates across gels. Cy2was used to label an equal pool (by protein amount) ofall sample replicates
1 2 3 4 5 6
Cy2 Pool Pool Pool Pool Pool PoolCy3 sty1D-60’ WT-60’ WT-0’ sty1D-60’ WT-0’ WT-60’Cy5 WT-0’ Sty1D-0’ WT-60’ Sty1D-0’ Sty1D-0’ Sty1D-60’
2.3 2-DE, fluorescence imaging and image analysis
2-DE was carried out essentially as previously described [43]using 24 cm pH 4–7 L and pH 3–10 NL IPG strips (GEHealthcare) and 12% homogenous SDS-PAGE gels bonded tolow-fluorescence glass plates. Gels were run in Ettan12 geltanks (GE Healthcare) at 2.2 mA per gel and 167C until the dyefront had run off the bottom. All steps were carried out in adedicated clean room. Gels were scanned between glass platesusing a Typhoon 9400 variable mode imager and Image-Quant software (both GE Healthcare). The photomultipliertube voltage was adjusted on each channel (Cy2, Cy3 and Cy5)for preliminary low-resolution scans to give maximum pixelvalues within 5–10% for each Cy-image, but below the satura-tion level. These settings were then used for high-resolution(100 mm) scans of all gels. Images were exported as tiff files forimage analysis. Images were then curated and analysed usingDeCyder software v5.0 (GE Healthcare) essentially as pre-viously described [43]. Here comparison of test spot volumes(Cy3 or Cy5 labelled) with the corresponding standard spotvolume (Cy2 labelled) gave a standardized abundance for eachmatched spot and values were averaged across biological tri-plicates. Only spots displaying a �1.5 average-fold increase ordecrease in abundance between each condition, matchingacross all gel images and having p values ,0.05 were selectedfor identification.
2.4 Spot picking, tryptic digestion and protein
identification
All gels were poststained with colloidal Coomassie Blue G-250 (CCB) and imaged as previously described [43]. CCB-stained images were then matched with the correspondingfluorescent images using DeCyder and a pick list of coordi-nates relative to two reference markers (stuck to plates atcasting) was generated for robotic excision from gels usingan Ettan automated spot picker (GE Healthcare). Spots werecollected in 96-well plates, proteins digested with trypsin andresultant peptides extracted as described previously [43].Extracts were dried, resuspended in 5 mL of 20 mM ammo-nium phosphate and spotted (0.5 mL/sample) onto MALDItarget plates with 1 mL of saturated 2,5-dihydroxybenzoic acid(DHB) in 20 mM ammonium phosphate using the drieddroplet method.
MALDI-TOF MS with peptide mass fingerprinting wasused for protein identification. MALDI mass spectra wereacquired using an Ultraflex mass spectrometer (Bruker Dal-tonics) in the reflector mode externally calibrated using an‘in-house’ mixture of standard peptides. Spectra were ana-lysed in FlexAnalysis (Bruker Daltonics). Where possiblethe spectra were internally calibrated using the trypsin auto-lysis peaks at m/z 842.51 and m/z 2211.10. The SNAP algo-rithm in FlexAnalysis was used to pick up to the 100 mostprominent peaks in the mass range m/z 600–5000. Masslists were extracted from the Ultraflex data files using an in-house Perl script, UltraMassList (http://www.ludwig.ucl.
ac.uk/bachem_html/software.htm). Another Perl script Com-monContam (http://www.ludwig.ucl.ac.uk/bachem_html/software.htm), based on the software of Schimdt et al. [44], wasused to identify masses common to all peak lists (i.e. matrixpeaks, trypsin peaks) above a user-defined threshold. Com-mon masses were removed and the resulting peptide massfingerprints were searched against an up-to-date NCBI andS. pombe database using MASCOT with the automated MAS-COT Daemon (Matrix Sciences v1.9.0 or later v.2.0). A positiveidentification was accepted when a minimum of six peptidemasses matched to a particular protein (mass error of650 ppm allowing one missed cleavage), matched peptidesrepresented�25% of the protein sequence, the MOWSE scorewas over the threshold score (p = 0.05), the hit appeared in thetop five hits in both database searches and the gel-based mo-lecular weight was in general agreement with the predictedmolecular weight of the identified protein.
2.5 Deletion of SPACUNK4.17 and SPAC23H3.15c
S. pombe gene deletions were constructed using establishedmethods [45]. All oligonucleotide sequences used are avail-able upon request. To test for sensitivity to H2O2, 8 mL ofserial five-fold dilutions of cells were plated on YE mediawith and without H2O2 and grown for 4 days at 307C prior toexamination of growth. The starting concentration of cellswas 46106 cells/mL. The sty1D and atf1D mutants, whichhave known sensitivity to H2O2, were used as controls.
2.6 1H NMR analysis of cellular metabolites
Ground yeast lysate (corresponding to 7.5 mg of total protein)from five biological replicates were extracted in 6% perchloricacid as previously described [46]. Neutralized extracts werefreeze-dried and reconstituted in 1 mL of D2O, and the extracts(0.5 mL) were placed in 5 mm NMR tubes. 1H NMR of cellextracts was performed on a Bruker 600 MHz NMR system(pulse angle: 907; repetition time: 5 s). The water resonancewas suppressed by gated irradiation centred on the waterfrequency. Sodium 3-trimethylsilyl-2,2,3,3-tetradeuteropro-pionate (25 mL, 10 mM) was added to the samples for chemicalshift calibration and quantification. All cell extract spectrawere acquired under identical conditions.
3 Results and discussion
3.1 2-D-DIGE and MS analysis of H2O2 and
Sty1p-dependent protein expression
Lysates of control and H2O2-treated WT and sty1D mutantS. pombe cells were compared as biological triplicates using2-D-DIGE allowing the averaging of possible variations ingrowth conditions, sample processing and gel running. Atreatment of 0.5 mM H2O2 for 60 min (60’) was chosen as anoxidative stress since this is known to induce stress genes
while causing minimal cell death and was the concentrationused in a previous parallel global transcriptional analysis [1].It should be noted that the transcriptional responses ofS. pombe vary with H2O2 concentration [47], and here wedescribe the responses to an intermediate level of oxidant.Samples were labelled with either Cy3 or Cy5 and run againsta Cy2-labelled pool of all samples, run on all gels as an inter-nal standard for improved cross-gel matching and quantita-tion [35, 48]. Three experiments were conducted using pH 3–10 nonlinear pH range IPG strips and two with pH 4–7separations. For each experiment DeCyder software was usedto find protein features displaying a .1.5 average fold-change in abundance in WT and sty1D cells in response toH2O2 or between the WT and sty1D cells. Figure 1 shows atypical master gel as an overlay of images (WT-60’ vs. sty1D-60’) with the position of all proteins for which unambiguousidentifications were obtained by MALDI-TOF MS and pep-tide mass fingerprinting.
A total of 260 protein isoforms were identified whichdisplayed significant differences in expression in one ormore of the comparisons (WT-60’ vs. WT-0’, sty1D-60’ vs.sty1D-0’, sty1D-0’ vs. WT-0’ and sty1D-60’ vs. WT-60’). Thecomplete data set is shown in Table S1 of SupplementaryMaterial. Average abundance ratios (based on fluorescenceintensity) for 836 data points out of a possible 1040 wereobtained. Unrecorded data points fell outside the selectionparameters (i.e. not matching across all gels or had p values.0.2). There were 777 data points with significant p values of,0.05. The 260 proteins represented 158 different geneproducts demonstrating a high degree of PTMs or proteo-lysis. Within the entire sample set, 33 spots yielded dataindicating that they contained two proteins with one spotyielding three. In these instances, the quantitative data can-not be assigned to an individual gene product, although thetarget of altered expression could be inferred in some casesfrom correlative changes in mRNA levels (see below). Acomparison of predicted vs. gel-based Mr and pI was con-ducted (Fig. 2). The correlation of Mr was reasonable and wasused as an additional level of confidence for protein identifi-cations, although it was skewed for higher Mr gene productsas these are under-represented on 2-D gels. The pIs corre-lated much less well, again demonstrating a high degree ofPTMs that affect pI (see below).
Figure 3A shows the number and direction of regulationof proteins changing due to H2O2 treatment for each strainand how these overlap, whilst Fig. 3B shows the number anddirection of regulation between cell types in the absence andpresence of stress. The greatest number of differences wasobserved when comparing WT-60’ with sty1D-60’ indicating acritical role of Sty1p in the stress response. Indeed, 47 pro-tein isoforms were up-regulated in WT in response to H2O2
compared to 14 for sty1D, whilst there were a similar numberof down-regulated proteins. Differences were also observedbetween unstressed WT and sty1D with 60 proteins (51 geneproducts) requiring Sty1p for basal expression. A number ofgene products were also de-repressed in sty1D showing that
2776 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
Figure 1. Master gel image showing the positions of S. pombe protein isoforms displaying peroxide and/or Sty1p-dependent changes inexpression. Figure shows an overlay of two pseudo-coloured images of a WT-60’ sample repeat (Cy3, red) and a sty1D-60’ sample repeat(Cy5, blue) created in Photoshop v7. Images were taken from a bonded pH 3–10 NL IPG, 24 cm624 cm61 mm, 12.5% bis/acrylamide gel.Numbers and arrows indicate positions of differentially expressed spots selected by DeCyder software that were identified with high con-fidence by MALDI-TOF MS (see Table S1).
Figure 2. (A) Graphical compar-ison of predicted vs. experimen-tally/gel-determined Mr and (B)pIs for the identified protein iso-forms. Predicted pI and Mr weretaken from databases. Gel-based pI and Mr were calculatedusing DeCyder software basedon selected reference proteinswhere gel-based values andpredicted values were in agree-ment.
Figure 3. Schematic representationof regulated isoform expression. (A)Venn diagram showing the numbersof H2O2-dependent up- and down-regulated isoforms (and gene prod-ucts) in WT and sty1D cells and theiroverlapping patterns of expression.(B) Venn diagram showing the num-bers of Sty1p-dependent up- anddown-regulated isoforms (and geneproducts) in untreated and H2O2-treated cells. Isoforms were takenfrom Table S1 and were included ifthey displayed a .1.5 average-foldchange in abundance (n = 3;p , 0.05).
Sty1p can also repress gene expression in agreement withthe parallel whole-genome transcriptional analysis [1]. How-ever, it is important to note that Sty1p may also alter the iso-form distribution of gene products through induced PTMs.Notably, 48% of all differentially expressed proteins wereregulated by H2O2 in a similar way in the WT and sty1Dstrains, showing that despite the central role of Sty1p instress-activated signalling, many cellular responses do notrequire its function.
The identified proteins were also grouped using k-meansand hierarchical clustering enabling us to rapidly view co-regulated isoforms displaying H2O2- and/or Sty1p-dependencyand to assess whether groups of functionally related proteinswould cocluster (Figs. 4 and S1). Proteins showing potent,Sty1p-dependent H2O2 induction (Fig. 4A) included catalase,GSH S-transferase II, thioredoxin reductase, malate dehy-drogenase, phosphoglycerate dehydrogenase, elongation fac-tor 2, protein tyrosine phosphatase/Pyp3p and three geneproducts of unknown function (an aldo/keto reductase(SPBC215.11c), brefeldin A-resistance protein p20(SPAC3C7.14c) and a hypothetical serine-rich protein(SPAC23H3.15c)). In contrast, Fig. 4B shows a cluster of 20proteins that were de-repressed in the sty1D mutant but rela-tively unaffected by stress. Prominent in this group were mem-
bers of the heat shock protein (Hsp) family (Hsp90 homologueSwo1, Hsp70 family Ssa2, heat shock cognate protein Hsc1,Hsp homologue Pss1 and Hsp70 family Ssc1). This suggeststhat loss of Sty1p may put cells in a permanent state of stress.Another large de-repression (17-fold) was seen for the leucinebiosynthetic enzyme 3-isopropylmalate dehydrogenase, whichwas also down-regulated by H2O2 treatment.
3.2 Functional classification of identified proteins
Each identified protein was assigned a functional classifica-tion based on the gene ontology annotation in the S. pombedatabase (http://www.genedb.org/genedb/pombe/index.jsp)and pathway assignment detailed on the S. pombe KyotoEncyclopaedia of Genes and Genomes database (KEGG,http://www.genome.jp/kegg/kegg2.html). The functionalityof all proteins in our data set is represented in Fig. 5. Themajority (57%) of differentially regulated protein isoformsfell within four functional groups (glycolysis, amino acidmetabolism, molecular chaperones and protein synthesis),while 11% were of unknown function. The remaining 32%were represented in 15 different functional categories. Pro-teins involved in glycolysis (see Table 2) represented thelargest functional group with 50 isoform entries (14 gene
2778 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
Figure 4. k-Means and hierarchicalclustering of protein differences.Upper panel shows a cluster of pro-tein isoforms that were induced byH2O2 in a Sty1-dependent manner.Lower panel shows isoforms whoseexpression was induced by loss ofSty1p (i.e. de-repressed). Clusteringof all identified isoforms was carriedout on the log-transformed 2-D-DIGE ratios (Table S1) using adownloadable version of Multi-Experiment Viewer (MeV) (http://www.tm4.org/mev.html) from TheInstitute for Genome Research(TIGR).
products), indicating a sensitivity of this central metabolicpathway to H2O2 and/or loss of Sty1p. In general in WTcells,the glycolytic enzymes in the upper half of the pathway wereup-regulated while those in the lower half were down-regu-lated. This suggests that glycolytic flux is redirected, perhapsto the pentose phosphate pathway as evidenced by changesin the expression of several pentose phosphate pathwayenzymes (phosphoglucomutase, ribose 5-phosphate isomer-ase, transketolase, transaldolase, 6-phosphogluconate dehy-drogenase and glucose-6-phosphate 1-dehydrogenase/Zwf1p), some of which were induced in a Sty1p-dependent
manner. Previous studies of oxidative stress in S. cerevisiaealso came to this conclusion [30, 49], although the study byShenton et al. showed that H2O2 inhibited the activities ofglyceraldehyde-3-phosphate dehydrogenase, enolase andalcohol dehydrogenase through reversible S-thiolation ratherthan through altered expression and there was little effect onglucose-6-phosphate dehydrogenase/Zwf1 or 6-phosphoglu-conate dehydrogenase activity, enzymes that catalyseNADPH production via the pentose phosphate pathway.Thus, inhibition of glycolytic flux alone was proposed toresult in glucose equivalents entering the pentose phosphate
Figure 5. Functional classification and distribution of all identified protein isoforms. Gene Ontology (GO) annotation terms for biologicalprocess and molecular function were taken from the S. pombe gene database (http://www.genedb.org/genedb/pombe/index.jsp), theKyoto Encyclopaedia of Genes and Genomes (KEGG) Pathways Database (http://www.genome.jp/kegg/kegg2.html) and the Swiss-Prot/TrEMBL protein database (http://us.expasy.org/sprot/). Unknown proteins include those which have no ascribed function, but may fall intoa putative enzyme class based on sequence homology.
pathway for the generation of NADPH and reductive capacity[30]. Phosphoglucomutase (induced in a Sty1-dependentmanner) is also required for the synthesis of the knownstress-protectant sugar trehalose, in agreement with obser-vations made in S. cerevisiae [49]; however, other enzymes inthis pathway were not found.
Amino acid metabolism also featured heavily with 42identified isoforms (25 gene products), as did protein syn-thesis (27 identifications) and protein degradation/proces-sing (9 identifications). These results suggest that globalalterations in protein turnover may occur in response toH2O2 through the regulation of amino acid synthesis, pro-tein synthesis and degradation. Although oxidative stress hasbeen shown to inhibit protein synthesis in S. cerevisiae [30],here, no broad pattern of coregulation could be discernedwhich might suggest a global switch in translational activity.For example, whilst isoforms of five eukaryotic translationinitiation factors were identified (eIF2a, eIF2Bg, eIF3 RNA-binding subunit, eIF3 p39 subunit and eIF4A), they dis-played different patterns of expression in response to H2O2,as did the seven ribosomal proteins identified. Molecularchaperones appeared 29 times, the majority of which wereHsps (Table 3). These changes may be required to aid proteinfolding in a more oxidizing environment or for the removalof aggregated or misfolded proteins; however, there was nocommon pattern of regulation apparent.
Of particular interest to this study were proteins knownto be involved in redox regulation and it is not surprising thatthere were 32 oxidoreductases in the 158 gene productsidentified. The redox enzymes catalase, thioredoxin reduc-tase, GST2 and thioredoxin peroxidase were all up-regulatedin WT cells exposed to H2O2, but were not induced in sty1D
cells (Table 3). This demonstrates the critical role played bySty1p signalling in the induction of antioxidants for removalof cellular H2O2 and protection from oxidative stress. Inter-estingly, one of these antioxidant enzymes, the highly con-served 2 cys-peroxiredoxin thioredoxin peroxidase (Tpx1p),was recently shown to directly activate Sty1p by a mechanisminvolving the formation of a peroxide-induced disulphidecomplex between Tpx1p and Sty1p, with overexpression ofTpx1p resulting in hyperactivation of Sty1p [50]. Thus, thereappears to be a feedback mechanism whereby Sty1 activity isrequired for the induction of one of its own activators. Nota-bly, GSH synthetase large chain/Gsa1p (SPAC3F10.04) wasinduced by H2O2 in both cell types (Table 3) and may berequired to increase GSH production, whilst Sty1p-depend-ent glucose-6-phosphate dehydrogenase induction andinduction of succinate semialdehyde dehydrogenase wouldalso increase NADPH reducing equivalents for reduction ofGSH and protein. Several gene products of unknown func-tion also displayed Sty1p-dependent induction (aldo/ketoreductase SPBC215.11c, brefeldin A-resistance protein p20SPAC3C7.14c, sugar oxidoreductase SPACUNK4.17 and hy-pothetical proteins SPAC23H3.15c, SPCC777.06c,SPAC1002.18), suggesting that they too may be antioxidants.To further investigate the function of two of these unchar-acterized gene products, whose mRNAs were also induced(namely hypothetical serine-rich protein SPAC23H3.15c andsugar oxidoreductase SPACUNK4.17), we constructedmutant strains lacking these genes and tested their sensitiv-ity to H2O2. Figure 6 shows that growth of the sty1D mutant,and to a lesser extent an atf1D mutant, was as expected, sen-sitive to H2O2. However, there was no apparent effect ofdeleting SPAC23H3.15c or SPACUNK4.17. Thus, being
2780 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
Table 2. Differentially regulated enzymes involved in glycolysis. Proteins are organized alphabetically with spot numbers from DeCyderanalysis. Spots containing two gene products are labelled a and b
a) Identifications found in multiple experiments. ORF names and Enzyme Commission (EC) number were taken from the S. pombe genedatabase (http://www.genedb.org/genedb/pombe/index.jsp) and from the Kyoto Encyclopaedia of Genes and Genomes (KEGG) Path-ways Database (http://www.genome.jp/kegg/kegg2.html). Average-fold differences in expression between H2O2-treated and untreatedWT and sty1D cells and between WT and sty1D cells were taken from DeCyder analyses of triplicate samples run in four independent 2-D-DIGE experiments. Only changes are reported where spots matched across all gels within an experiment and fold-changes had pvalues ,0.05 (except those in italics where p values were .0.05 and ,0.2). Isoforms up- and down-regulated by .1.5-fold are high-lighted in dark grey and light grey, respectively. Experimental pI and Mr (gel) were calculated in DeCyder using pI and Mr of knownproteins as references. Multiple isoforms of proteins are listed in order of increasing pI. Predicted pI and Mr were taken from databases.Proteins were denoted with (fragment?) where gel Mr was significantly lower than predicted.
highly induced by H2O2 does not necessarily indicate that agene is essential for a protective response to oxidative stress,and some redundancy must therefore exist.
3.3 Regulated expression, isoform distribution and
post-translational modifications
As well as changes in gene expression, the observed regu-lated protein expression in response to H2O2 or loss of Sty1pis likely to involve that post-transcriptional regulatory mech-anisms such as altered translation, degradation and PTMs.Indeed, this was evidenced by the existence of multiple iso-forms of the same gene products; of 158 gene productsidentified, 44 occurred as multiple isoforms. For example,glyceraldehyde-3-phosphate dehydrogenase was identified as16 different isoforms, consisting of a mixture of the tdh1 andtdh2 gene products (Table 2). Upon H2O2 stress in the sty1Dmutant, several basic isoforms were down-regulated, whilst
the acidic forms were up-regulated indicative of induciblePTMs that affect pI. Indeed, some isoforms (including thoseof enolase and fructose bisphosphate aldolase) had matchedpeptide pairs separated by 79 Da (results not shown) indica-tive of phosphorylation. However, pI shifts may also be dueto the irreversible oxidation of cysteinyl thiols to the R-SO2
2
and R-SO32 forms as reported for peroxiredoxins in redox
stressed cells [51, 52]. This oxidative damage is likely to bemore prevalent in stressed sty1D cells, where redox protectiveenzymes such as catalase and thioredoxin peroxidase are notinduced (Table 3).
Several proteins were also identified as putative frag-ments by virtue of the sequence coverage obtained by MSand their unexpected gel positions. For example, poly(A)-binding protein/Pab1p involved in RNA processing (poly(A)shortening and translation initiation) was identified as threeisoforms; two as the likely whole proteins which were down-regulated in sty1D and one as a lower Mr fragment which was
2788 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
Figure 6. Effect of H2O2 on thegrowth of SPAC23H3.15cD SPA-CUNK4.17D sty1D and atf1Ddeletions mutants comparedwith WT. Cells were seriallydiluted and plated on YE mediawith and without H2O2 (at0.5 mM and 1 mM) and grownfor 4 days at 307C.
up-regulated in sty1D (Table 3). This suggests that Sty1p maynormally inhibit proteolytic cleavage of Pab1p and therebyalter translation initiation to affect protein expression. Sev-eral coregulated proteins could also be grouped according totheir inclusion in multi-subunit complexes. For example,subunits E1 and E2 of pyruvate dehydrogenase were co-regulated (Table 2), and may affect pyruvate levels and hencethe metabolic pathways that utilize pyruvate and acetyl CoA.Similarly, subunits a and b of the vacuolar ATP synthasewere both repressed in sty1D cells (Table 3), suggesting thatSty1p may regulate vacuolar acidification and protein pro-cessing. Three subunits of the chaperonin-containing T-complex were also down-regulated in the sty1D mutant,again implicating Sty1p signalling in the regulation of cel-lular chaperone activity. However, coregulation was not ageneral observation, as in the case of acetolactate synthase(involved in the first step of valine, leucine and isoleucinesynthesis), where the large and small subunits were oppo-sitely regulated (Table 3). Thus, further functional studiesare required to assess the consequence of such subunitchanges on complex activity.
3.4 Correlation of changes in protein and mRNA
abundance
An important resource in interpretation of the protein datahas been the publication of the complete fission yeast genomeand the subsequent publication of a global investigation of thetranscriptional responses of S. pombe to a number of environ-mental stresses [1]. This allowed the parallel comparison ofour quantitative proteomics data with quantitative mRNA datafor the same set of genes under the same conditions. Figure 7displays the overall correlations between published micro-array data and proteomic data for the four comparisons. Theoverall level of correlation was low (e.g. R2 = 0.2 for WT-60’ vs.WT-0’) in agreement with previously published data [53, 54]and did not change when only singly identified isoforms wereconsidered. Of particular interest was the very low correlationswhen comparing sty1D-60’ versus sty1D-0’ (R2 = 0.0033,Fig. 7B) and untreated sty1D versus WT (R2 = 0.045, Fig. 7C).This suggests a significant increase in PTMs (possibly irre-versible oxidation of cysteinyl thiols) and that loss of Sty1p hasa more profound effect on post-transcriptional events than ongene expression per se. The correlation between sty1D and WT
cells under stress (R2 = 0.1768, Fig. 7D) is harder to explain asit would be expected that this correlation would be the lowestgiven the above assumptions. One possibility is that Sty1p-in-dependent, H2O2-inducible gene transcription componentincreases the correlation.
The mRNA analysis showed 44 gene products wererepressed in the absence of Sty1p, while the protein analysisidentified 51 repressed gene products with eight matching tothe mRNA data (Table 4). In most cases, the same direction-ality of mRNA and protein regulation was observed for thisgroup, identifying them as Sty1p-dependent gene targets.Conversely, 95 genes and 57 gene products were de-repressedby loss of Sty1p, with eight matching between the data sets,though with less correlation between transcription and pro-tein expression. Of the 56 peroxide-specific and 12 super-induced genes identified by microarray analysis, only eightgene products were identified by proteomic analysis (Table 4).All of these except a putative aminotransferase(SPAC56E4.03) displayed Sty1p-dependent induction of theprotein that agreed with the mRNA data, although changes inthioredoxin reductase were isoform-dependent. The inductionof thioredoxin reductase seems logical given its role in theregeneration of reduced thioredoxin. However, possible rolesfor brefeldin A-resistance protein, 2-hydroxy acid dehy-drogenase and protein-tyrosine phosphatase/Pyp3p are lessclear. In the transcriptional analysis, 140 genes induced by anumber of different environmental stresses were termed coreenvironmental stress response (CESR) genes [1]. Within thisgroup, 16 were identified by 2-D-DIGE/MS and all were up-regulated with H2O2 in the WT but not sty1D cells, exceptpyridoxal reductase, which was overexpressed in sty1D(Table 4). This group included seven enzymes involved in car-bohydrate metabolism and generation of NADPH reducingequivalents (glucose-6-phosphate dehydrogenase, glycerol-3-phosphate dehydrogenase, malate dehydrogenase, succinatesemialdehyde dehydrogenase and three putative oxido-reductases), the redox enzymes catalase and thioredoxin per-oxidase and five proteins of unknown function, including thetwo that were characterized by assessing the H2O2 sensitivityof their deletion mutants (see above).
Hsp16 was the only Hsp family member identifiedwhere an increase in mRNA levels correlated with increasedprotein expression, identifying it as a stress-inducible gene.Importantly, expression of most S. pombe Hsp family genes
Table 4. Comparison of protein and mRNA ratios. Table shows the fold-change of protein isoform and mRNA across the four experimentalconditions for genes found in [1] that were; H2O2-specific genes, CESR genes super-induced with H2O2, genes that required Sty1pfor basal level expression, genes that were de-repressed in unstressed sty1D, induced CESR genes, repressed CESR genes
ORF Name Name mRNA ratio Protein ratio Function/Pathway
WT-60’vs.WT-0’
sty1D-60’vs.sty1D-0’
sty1D-0’vs.WT-0’
sty1D-60’vs.WT-60’
WT-60’vs.WT-0’
sty1D-60’vs.sty1D-0’
sty1D-0’vs.WT-0’
Sty1D-60’vs.WT-60’
H2O2-specific genes
SPAC3C7.14c Brefeldin A resistance proteinp20 (flavodoxin-like)
7.07 2.36 1.05 0.35 7.26 0.36 10.22 0.50 Unknown
SPAC3C7.14c Brefeldin A resistance proteinp20 (flavodoxin-like)
was transcriptionally down-regulated by H2O2 [1]. Of the 106down-regulated CESR genes, 6 appeared in the protein listand all were down-regulated in sty1D cells with the exceptionof ornithine aminotransferase (up-regulated) and glutaminesynthetase (where four isoforms displayed different expres-sion). In some instances where two proteins were identifiedin the same spot, examination of the mRNA data was used topredict which of the proteins was likely to be regulated. Forexample, hypothetical serine-rich protein (SPAC23H3.15c)and an isoform of glutamine synthetase were found in the
same spot (spot 941) which was strongly induced by H2O2 ina Sty1p-dependent manner. Since the expression of theSPAC23H3.15c mRNA was also strongly induced, the up-regulation of this protein is likely to account for the increasedspot intensity. Catalase and protein tyrosine phosphatase/Pyp3p were found to comigrate (spot 492) and the abun-dance of this spot was increased by H2O2 in a Sty1p-depend-ent manner. The mRNAs of both genes also displayed Sty1p-dependent H2O2 induction (Table 4) and so the observedchange in spot abundance is likely to be an additive effect.
2792 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
Figure 7. Global correlations ofmRNA and protein data. Log10values of the mRNA ratios for allidentified gene products [1]were plotted against the log10protein ratios from the prote-omic analysis for the fourexperimental comparisons; (A)WT-60’ vs. WT-0’, (B) sty1D-60’vs. sty1D-0’, (C) sty1D-60’ vs.WT-0’ and (D) sty1D-60’ vs. WT-60’ and linear regression coeffi-cients (R2) for the data deter-mined.
3.5 Metabolomic analysis
A metabolite analysis was also conducted on the same sam-ple sets using 1H-NMR. Samples from five biological repli-cates were prepared for identical analyses with an internalstandard for calibration and quantitation of metabolitepeaks. In total 31 individual peaks were quantified and ofthese 29 have been previously assigned to a specific cellularmetabolite. Both Sty1p- and H2O2-specific changes wereapparent (Fig. 8). Loss of Sty1p resulted in a significant(p,0.05) reduction of the amino acids leucine, isoleucine,valine, alanine, glutamate, glutamine, tyrosine, histidine,phenylalanine and of the metabolites lactate, U1 (uni-dentified peak), glycerophosphocholine (GPC), trimethyl-amine-N-oxide (TMAO), glycerol and UTP 1 UDP. Con-versely, betaine (involved in glycine metabolism), glycine, b-D-glucose, orotic acid (involved in pyrimidine metabolism)and the ATP 1 ADP pool were increased in unstressed sty1Dcells (Fig. 8A). Peroxide treatment of WT cells resulted in theinduction of acetate, glutamate, lysine, choline, GPC,TMAO, betaine, glycine, glycerol, U2, UTP 1 UDP and ino-sine, and all of these, with the exception of inosine, displayeda lower level of induction or even repression in sty1D(Fig. 8B). Of the metabolites repressed in the WT, leucine,valine and U1 were actually induced in sty1D. The moststriking difference was a six-fold increase in b-D-glucose inthe sty1D cells, but not the WT, in response to H2O2. Togetherthese data show that the Sty1p pathway is a critical regulatorof multiple, diverse metabolic pathways, including aminoacid biosynthesis and metabolism, and is required for thenormal response to oxidative stress. Repression of amino
acids in sty1D may be due to a slower rate of cellular metab-olism, resulting in reduced protein synthesis (lowered levelsof branched amino acids) and membrane turnover(decreased GPC level). With a slower rate of cell metabolism,less energy will be required by these cells, which in turncould lead to reduced glycolysis (reduced level of lactate andelevated glucose level) and amino acid metabolism.
The changing metabolite profiles were compared withthe protein data to identify correlations between the regula-tion of specific enzymes and their substrates or products. Forexample, higher levels of glycine in sty1D may be explainedby the lower expression of the major glycine-degrading en-zyme glycine dehydrogenase (SPAC13G6.06c), which may inturn affect betaine levels and purine metabolism, since gly-cine is essential for purine synthesis. The data also suggestthat an H2O2- and Sty1p-dependent conversion of glutamineto glutamate, possibly involving the regulated expression ofglutamine synthetase (SPAC23H4.06), glutamate dehy-drogenase (SPCC622.12c) and/or succinate semialdehydedehydrogenase (SPAC139.05). Induction of the latter mayalso produce the observed reduction in succinate (Fig. 8B).Sty1p-dependent induction of D-1-pyroline-5-carboxylatedehydrogenase (SPBC24C6.04) could also raise glutamatelevels by conversion from proline, whilst overexpression ofthe glutamine-hydrolysing enzyme GMP synthase(SPBPB2B2.05) in sty1D would reduce glutamine and alterpurine metabolism (Table 3). Enzymatic synthesis of GSHoccurs from the component amino acids (glutamate, cyste-ine, and glycine) via the sequential action of two ATP-de-pendent cytosolic enzymes, so it may be that the observedincreases in glutamate and glycine levels enable increased de
Figure 8. Multiple metaboliteanalysis using high-resolution1H NMR spectroscopy. Five bio-logical sample repeats wereprepared and analysed accord-ing to ‘Section 2’. Percentagechanges in all metabolites areshown for; (A) sty1D vs. WTunder conditions of no stressand stress and (B) H2O2 stressvs. no stress for each strain.Data are expressed as themean 6 SEM with * indicatingsignificant changes (p , 0.05).Peaks labelled U1 and U2 couldnot be assigned.
novo synthesis of GSH for the provision of reducingpotential. The Sty1p-dependent increase in glycerol follow-ing stress is also interesting, since this metabolite isknown to play a cytoprotective role, and its induction maybe mediated via up-regulation of glycerol-3-phosphatedehydrogenase/Gpd1 (SPBC215.05) or alcohol dehy-drogenase (SPAC3A11.12c). In turn, higher glycerol levelsmay increase GPC generation and alter membrane turn-over. Like glycerol, TMAO is thought to act as a chemical
chaperone to protect proteins from misfolding and so itsup-regulation following H2O2 exposure (in a Sty1-depend-ent manner) may be an additional mechanism to protectagainst oxidative protein damage. Glycerol, TMAO, GPC,choline and betaine are also known osmoprotective metab-olites, and their induction suggests that some level ofcross-protection may exist between the peroxide andosmotic stress response, a theory which warrants furtherinvestigation.
2794 M. E. Weeks et al. Proteomics 2006, 6, 2772–2796
Figure 9. Schematic representations ofmetabolic responses to H2O2 in (A) WTand (B) sty1D. Arrows between path-ways indicate known connections takenfrom the Kyoto Encyclopaedia of Genesand Genomes (KEGG) Pathways Data-base. Vertical arrows indicate a predic-tion of the up- or down-regulation of apathway based upon how protein and/ormetabolite levels change in response tostress. Stress response of the WT glyco-lytic pathway is more complex, sincethere was down-regulation of theenzymes in the lower half of pathwayand up-regulation of the enzymes in theupper part of the pathway (see text).Metabolic pathways in bold italic typeindicate that the predicted direction ofregulation of this pathway is supportedby protein data only whilst those in nor-mal type face indicate that the directionof regulation is supported by bothmetabolite and protein data.
4 Concluding remarks
In this study, we have identified a large number of S. pombeprotein isoforms that display H2O2-dependent changes inabundance and have related these changes to alterations inthe mRNA and metabolic profiles of cells treated underidentical conditions. This work shows that a diverse range ofcellular processes are affected by oxidative stress and furtherdemonstrates the role of the Sty1p signalling pathway inregulating some of these processes. Although a number ofthe recorded protein changes correlated well with alteredgene expression (Table 4), there was a low overall correlationbetween protein and mRNA expression in agreement withsome previous studies [53, 54], but not others, where higher
correlations were observed for specific subsets of genes [55,56]. Our work thus highlights the complexity of the cellularresponse to stress and indicates that many levels of regula-tion are at play. Indeed, a large number of proteins appearedto be post-translationally altered through proteolysis, phos-phorylation and/or oxidation with no correlative change intheir mRNA expressions. However, mRNA stability or theefficiency of translation may also be regulated by stress [57]and therefore may contribute to the low correlation observed.This emphasizes the requirement for more detailed prote-omic analyses if we are to fully understand the molecularmechanisms involved in responses to stress and other sti-muli. Moreover, given the relatively poor sensitivity and cov-erage of current proteomic methodologies, it is likely that
there are many more low-abundance targets to be identified,although it is important to note that 2-D-DIGE does provide arobust, quantitative method for global expression profiling.
When interpreted alone, each experimental methodidentifies putative targets of the oxidative stress response andspecifically the Sty1p-responsive elements of numerousmetabolic pathways. However, the real value of the data canonly be discerned when data sets are combined and theresults viewed globally. Such a systems approach allows sig-nificantly more detail to be proposed regarding how variouscellular processes are affected by oxidative stress and Sty1psignalling and how regulation is achieved. It is clear that thestress response is global and has profound effects on multi-ple, diverse metabolic pathways and that the Sty1p stress-activated protein kinase is a critical modulator of these eventsbasally, and in particular, during the stress response. Itinvolves a number of changes including alterations in pro-tein and amino acid synthesis, a switch in energy productionfrom glycolysis to the pentose phosphate pathway for thegeneration of NADPH and the induction of stress-specificantioxidant enzymes and cytoprotectants that would act toreduce cellular ROS, protect against molecular damage andreverse oxidative protein modification. This is in generalagreement with the conclusions of a previous proteomicanalysis of the H2O2 response in the budding yeast S. cerevi-siae [49], and indeed, there was considerable overlap in theprotein changes identified between this study and ours, par-ticularly of enzymes involved in carbohydrate and aminoacid metabolism. Finally, whilst WT cells elicit a robust re-sponse that is required for their survival, the sty1D responseis muted and appears to involve shutting down of proteinsynthesis, a number of metabolic pathways and energy pro-duction possibly in preparation for entering a dormant phaseor prior to cell death (Fig. 9).
We would like to thank Mr. Richard Jacob (UCL) for soft-ware writing and support for MS data analysis and the EMFBiological Research Trust for funding. We would also like toacknowledge The Medical Biomics Centre at St George’s, London,for the use of their 600 MHz NMR Spectrometer.
5 References
[1] Chen, D., Toone, W. M., Mata, J., Lyne, R. et al., Mol. Biol. Cell2003, 14, 214–229.
[2] Gasch, A. P., Spellman, P. T., Kao, C. M., Carmel-Harel, O. etal., Mol. Biol. Cell 2000, 11, 4241–4257.
[3] Causton, H. C., Ren, B., Koh, S. S., Harbison, C. T. et al., Mol.Biol. Cell 2001, 12, 323–337.
[4] Weeks, M. E., James, D. C., Robinson, G. K., Smales, C. M.,Proteomics 2004, 4, 123–135.
[5] Madeo, F., Engelhardt, S., Herker, E., Lehmann, N. et al., Curr.Genet. 2002, 41, 208–216.
[6] Wood, V., Gwilliam, R., Rajandream, M. A., Lyne, M. et al.,Nature 2002, 415, 871–880.