Complementary application of ESI and MALDI based mass spectrometry in microbial proteomics
Inauguraldissertation
zur
Erlangung des akademischen Grades
doctor rerum naturalium (Dr. rer. nat.)
an der mathematisch naturwissenschaftlichen Fakultät
der
Ernst-Moritz-Arndt-Universität Greifswald
vorgelegt von
Bernd Heßling
geboren am 24.08.1980
in Bocholt
Greifswald, September 2012
Dekan: Prof. Dr. Klaus Fesser
1. Gutachter: Prof. Dr. Michael Hecker
2. Gutachter: Prof. Dr. Andreas Tholey
Tag der Promotion: 07.01.2013
Table of Contents
List of abbreviations and symbols ...... I Scope and outline of this thesis ...... II Articles and author contribution ...... IV 1 Introduction ...... 1 1.1 Technical requirements for mass spectrometry-driven proteomics ...... 1 1.2 The two prevalent proteomic workflows utilizing mass spectrometry ...... 3 2 The history of mass spectrometry driven microbial proteomics (article I) ...... 6 3 LC-ESI-MS for global proteomics and post-translational modification analysis ...... 9 3.1 Global proteome studies (article II) ...... 9 3.2 Investigation of post-translational modification (article III and IV) ...... 13 4 Proteomics for systems biology ...... 14 4.1 Dynamic adaptations of B. subtilis metabolism analyzed in a systems biology approach (article V) ...... 15 4.2 A newly developed approach to obtain absolute protein quantities (article VI) ...... 16 5 Exchangeability and complementarity of ESI-MS and MALDI-MS ...... 19 5.1 Global relative quantification with LC-MALDI-MS (article VII) ...... 20 5.2 Complementary of MALDI and ESI (article VII) ...... 22 6 Conclusion ...... 23 7 References ...... 25 Article I ...... 31 Article II ...... 43 Article III ...... 91 Article IV ...... 99 Article V ...... 123 Article VI ...... 131 Article VII ...... 141 Affirmation ...... 177 Curicculum vitae ...... 178 Acknowledgement ...... 179
List of abbreviations and symbols
1D one dimensional MS mass spectrometry
2D two dimensional MS/MS tandem mass spectrometry
B. subtilis Bacillus subtilis MSF membrane shaving fraction;
BEF biotinylation-enriched fraction OD optical density
ChIP chromatin immunoprecipitation PAGE polyacrylamide gel electrophoresis DNA deoxyribonucleic acid pH pondus hydrogenii e.g. for example pI isoelectric point EMF enriched membrane protein fraction PMF peptide mass fingerprimt
ESI electrospray ionization PTM post-translational modification
FTICR Fourier-transform ion cyclotron Q quadrupole resonance QconCAT quantification concatemer GFP green fluorescent protein RNA ribonucleic acid HPLC high-performance liquid S. aureus Staphylococcus aureus chromatography SDS sodium dodecyl sulfate i.e. that is SILAC stable isotope labeling by/with IT ion trap amino acids in cell culture ITRAQ isobaric tags for relative and absolute quantitation TOF time of flight VISA vancomycin intermediate LC liquid chromatography Staphylococcus aureus MALDI matrix assisted laser VRSA vancomycin resistant desorption/ionization Staphylococcus aureus MRSA methicillin resistant Furthermore, the usual codes for amino acids Staphylococcus aureus and chemical elements were used.
Page I
Scope and outline of this thesis
Scope and outline of this thesis
This thesis will discuss the different fields of application of the two soft ionization techniques ESI and MALDI in microbial proteomics and their importance for a better understanding of bacteria physiology.
The general development in the past 25 years coming from 2D-gel analysis and protein identification by peptide mass fingerprint analysis via MALDI-TOF to genome wide quantitative LC-ESI-MS experiments with fast and sensitive ESI instruments is exemplary shown for the Gram-positive bacterium Bacillus subtilis in article I.
Even though 2D-PAGE in conjunction with MALDI-MS is still an important tool in proteomic research, the more recently established global quantitative LC-ESI-MS workflows gain more and more relevance as they overcome 2D-PAGE based protein restrictions and enable the acquisition of higher accurate protein quantities. In article II such a workflow was used to analyze the physiological adaptation of Staphylococcus aureus to vancomycin treatment on a global-scale. Also post-translational modifications of proteins, that are important for regulation of their activity and allow rapid adaption to changed environmental conditions, could be analyzed by LC-ESI-MS workflows using special enrichment strategies ( article III and IV ).
Despite the mentioned discrimination and less accurate quantification of proteins, 2D-PAGE analyses are still advantageous when analyzing large-scale time series experiments. To gain highly time resolved data but also very accurate relative quantities on a global-scale, 2D-PAGE-MALDI-MS and LC-ESI-MS techniques have been combined to investigate dynamic proteome adaptations of B. subtilis during nutrition shift as part of a global systems biology approach (article V ).
Also absolute quantities of proteins are of high interest for systems biology, but are still challenging to obtain on large-scale as well as with sufficient accuracy. In article VI a method that again combined 2D- PAGE-MALDI-MS and LC-ESI-MS was introduced to gain absolute protein quantities on global-scale. Utilizing the complementarity of 2D-PAGE and LC-ESI-MS this new workflow enabled fast and cost efficient data acquisition on absolute scale.
In article VII we described for the first time a global quantitative LC-MALDI-MS workflow. Cross validation with an LTQ Orbitrap proofed that LC-MALDI-MS is able to process complex samples and obtain highly reliable quantities. The comparative analysis of data gained with both instrument types
Page II
Scope and outline of this thesis
revealed biases for certain biochemical properties of MALDI as well as ESI instruments, resulting in a general complementarity of both ionization techniques.
Page III
Scope and outline of this thesis
Articles and author contribution
Article I
Becher, D., Büttner, K., Moche, M., Hessling, B., Hecker, M., 2011. From the genome sequence to the protein inventory of Bacillus subtilis . Proteomics 11, 2971–2980.
B. H. assisted in writing the manuscript.
Article II
Hessling, B., Bonn, F., Herbst, F.-A., Rappen, G.-M., Bernhardt, J., Hecker, M. and Becher, D. Global proteome analysis of vancomycin stress in Staphylococcus aureus . Submitted to Mol. Cell Proteomics .
B.H., F.-A. H., F.B. and G.-M.-R. performed the experiments; B.H. analyzed the data; B.H. wrote the manuscript.
Article III
Elsholz, A.K.W., Turgay, K., Michalik, S., Hessling, B., Gronau, K., Oertel, D., Mäder, U., Bernhardt, J., Becher, D., Hecker, M., Gerth, U., 2012. Global impact of protein arginine phosphorylation on the physiology of Bacillus subtilis . Proc. Natl. Acad. Sci. U.S.A . 109, 7451–7456.
B.H. and K.G. performed the mass spectrometry analysis.
Article IV
Chi, B.K., Gronau, K., Mäder, U., Hessling, B., Becher, D., Antelmann, H., 2011. S- bacillithiolation protects against hypochlorite stress in Bacillus subtilis as revealed by transcriptomics and redox proteomics. Mol. Cell Proteomics 10, M111.009506.
B.H. and K.G. performed the mass spectrometry analysis.
Article V
Buescher, J.M., Liebermeister, W., Jules, M., Uhr, M., Muntel, J., Botella, E., Hessling, B., Kleijn, R.J., Le Chat, L., Lecointe, F., et al. (2012) Global network
Page IV
Scope and outline of this thesis
reorganization during dynamic adaptations of Bacillus subtilis metabolism. Science , 335, 1099–1103.
B.H. and J.M. performed the mass spectrometry analysis and analyzed the mass spectrometry data.
Article VI
Maass, S., Sievers, S., Zühlke, D., Kuzinski, J., Sappa, P.K., Muntel, J., Hessling, B., Bernhardt, J., Sietmann, R., Völker, U., Hecker, M., Becher, D., 2011. Efficient, global-scale quantification of absolute protein amounts by integration of targeted mass spectrometry and two-dimensional gel-based proteomics. Anal. Chem. 83, 2677–2684.
B.H., S.S., performed mass spectrometry analysis. B.H. and J.M. assisted S.M. in establishing the workflow.
Article VII
Hessling, B., Büttner, K., Hecker, M. and Becher, D. Global relative quantification with LC-MALDI – cross-validation with LTQ-Orbitrap proves reliability and reveals complementary ionization preferences. Submitted to Mol. Cell Proteomics .
B.H. performed the experiments, B.H. analyzed the data, B.H. wrote the manuscript
Confirmed, Greifswald xx.09.2012 ______
Prof. Dr. Michael Hecker
Page V
Introduction
1 Introduction
The term proteomics follows consequently from genomics and describes the study of a full set of proteins encoded by a genome. The starting point for the large-scale analysis of a cell's proteome was the invention of highly sensitive two-dimensional polyacrylamide gel electrophoresis (2D-PAGE) by O'Farrell in 1975 ((1)). This technique allowed a visualization of changed protein concentrations in an organism followed along when exposed to changed environmental conditions.
The early drawback of this technique was the limited capacity of protein identifications, so that visual changes in the protein pattern of a 2D-gel could be addressed to physiological adaptations of the organism.
1.1 Technical requirements for mass spectrometry-driven proteomics
The essential milestones for modern mass spectrometry-driven proteomics were one the one hand the technical advances made in the genome sequencing that today allows the sequencing of whole organisms in short periods of time and at low costs. The other break-through was the invention of soft ionization techniques that enabled the ionization of large biomolecules like proteins and peptides in an intact state, transfer them into gas phase and so render them accessible to mass spectrometry analysis. The two predominantly used soft ionization techniques in proteomic research are electrospray ionization (ESI), established and successfully applied by Fenn 1984 (2), and matrix assisted laser desorption/ionization (MALDI) introduced by Hillenkamp, Karas and Tanaka between 1985 and 1987 (3, 4). Their application in mass spectrometry and importance for life science was honored by awarding Fenn and Tanaka with the Noble price in chemistry 2002.
Even so the MALDI process is not understood in every detail crucial steps can be summarized in very brief and widely accepted theory that is illustrated in Figure 1 A. For MALDI analyses analyte and matrix are mixed in solution and spotted on a target plate. With solvent vaporization analyte and matrix molecules co-crystallize, so that analyte molecules are embedded in matrix molecules. The MALDI process itself consists of two steps. The first is the desorption process where energy from a laser beam is absorbed by matrix molecules and then transferred to the analyte, which drives the desorption process of these molecules. The second step is the ionization of the analyte through protonation or deprotonation. The charge in this process is again transferred from matrix molecules, resulting in
Page 1
Introduction
primarily singly charged analyte ions. MALDI process in total includes a variety of very complex events that are still debated and under research (5).
Also ESI is a multi-step process. Initially a liquid solution containing the analyte is passed through a capillary to a small diameter needle, on which a high electric tension is maintained. Due to the high electric field the emerging liquid generates a spray of very small and highly charged droplets. In a second step these droplets shrink drastically through solvent evaporation and finally gas phase ions are generated (Figure 1 B). Two different theories try to explain this last crucial step of bringing the charged ions into gas phase. The “Ion Evaporation Model” (IEM) proposed by Iribarne and Thomson in 1976 (6) can be summarized briefly as the assumption that ions directly eject from the supercharged droplets when they reach a critical size through the evaporation process. The “Charge Residue Model” (CRM) originally described by Dole in 1986 (7) is based on ongoing division of the highly charged droplets until one single ion remains in residue. Both theories have been proven to be valid under certain conditions (8, 9). The electrospray ionization delivers mostly multiple charged ions that can then be applied to a mass analyzer.
Figure 1: Schemata of A) matrix assisted laser desorption/ionization (MALDI) and B) electrospray ionization (ESI).
Both ionization methods can be equipped with a variety of different mass analyzers. For proteomic research five different types of mass analyzers are commonly used: Quadrupole (Q), Ion Trap (IT), Time Of Flight (TOF) mass analyzer, Fourier-transform ion cyclotron resonance (FTICR) mass analyzer and most recently invented by Makarov the Orbitrap (10). All these instrument types differ in their physical
Page 2
Introduction
principles and their performance characteristics like mass resolution, mass accuracy, sensitivity, and scan rate. Hybrid instruments combine two or more mass analyzers to not only complement the strengths of different mass analyzer types but also allow tandem mass spectrometry (MS/MS) measurements that are restricted for most analyzers in non-hybrid instruments. MS/MS measurement not only allows determination of an analyte's mass but also to fragment it and further characterize it by mass assignment of the produced fragments.
Whereas modern instruments using electrospray ionization for proteomic research are equipped with a broad variety of different mass analyzers, the predominantly matrix assisted laser desorption/ionization instrument is coupled with two time of flight mass analyzers (MALDI-TOF-TOF).
1.2 The two prevalent proteomic workflows utilizing mass spectrometry
The combination of whole genome sequence information and accessibility of proteins and peptides to mass spectrometry were the basement for modern microbial proteomics but were per se not sufficient to meet the inherent claim of proteomics to be genome-wide. The complexity of proteomes, even of microbial organisms, was a huge challenge for early mass spectrometry driven analysis. The 2D-PAGE technique is able to drastically reduce this complexity through separating proteins by their p I and mass and thus allows a following MS analysis. Protein spots on a 2D-gel contain only one or very few proteins that can afterwards be digested with a site specific protease like trypsin and identified by its so called peptide mass fingerprint (PMF) in a mass analyzer. This PMF identification is based on a comparison of the mass spectrum of the digested protein to an in silico digest of the whole theoretical proteome of the analyzed organism. Sufficient overlap of the mass spectrum and a theoretical mass spectrum of one protein of the in silico digest results in positive protein identification. Especially MALDI-TOF and MALDI- TOF-TOF instruments were ever applied for this analysis type, as they allowed fast and robust identification of many proteins in short time.
Despite reduction of protein complexity the second great benefit of the 2D-PAGE analysis is its generally quantitative nature. Altered protein amounts in two compared samples result in changed spot intensities after staining with a protein specific dye. This information in addition to protein identification by MS analysis allows following physiological adaptations of an organism to altered environmental conditions by observed protein amount changes on large-scale (Figure 2 A).
Page 3
Introduction
2D-PAGE analyses have great benefits like (I) high separation capacity, that is based on protein level, which additionally to the reduced complexity also allows to identify different protein isoforms and so to discover even previously unknown post-translational modifications, (II) an inherent quantitative nature through protein amount dependent spot intensities and (III) intuitive visualization of observed adaptations. But the technique is also limited as it is restricted to proteins with certain physicochemical properties. Especially very hydrophobic proteins like membrane proteins or very basic or acidic proteins are hard to access by 2D-PAGE analysis.
Therefore, methods based on liquid chromatography (LC) separation of digested protein samples combined with MS/MS analysis of eluting peptides were established and quickly gained relevance in proteome research. As ESI instruments can be coupled online to an LC-system they are predominantly used in these workflows. Even with an upstream pre-fractionation on protein level sample complexity is the biggest issue for this analysis type. One LC-run consists of the peptides derived from the digest of a subset of proteins, which extends the complexity drastically. Not only the huge amount of different peptides but also the large dynamic range, in which these peptides occur, raise challenges for sample fractionation and MS analysis. Technical advances made in MS instrument development have been tremendous over the last years, making especially ESI instruments faster and more sensitive. Although peptide complexity of most samples still exceed instruments capabilities, modern instrumentation and workflows allow identification of 50% to 70% of all predicted proteins of certain organisms as shown in recent studies (11–14).
As MS is per se not quantitative due to varying detector responses, differential ionization yields for different peptides, and other factors (15), special workflows have been established to also gain comprehensive data from samples analyzed with LC-MS/MS. Chemical and isotopic labeling techniques allow highly accurate determination of relative quantities. Especially isotopic labeling techniques like SILAC (stable isotope labeling by/with amino acids in cell culture) (16) or 15 N metabolic labeling (17) are widely used. The observed ratios between peptides derived from unlabeled and isotopically labeled proteins in these workflows are highly quantitative, because pairs of peptides do not chemically differ and they can be analyzed during the same experiment. The incorporation of the isotopic label already in cell culture allows mixing of unlabeled and labeled samples early in sample separation and so maintain samples in comparable states through the following preparation steps (Figure 2 B).
Page 4
Introduction
Figure 2: Schematic of two prevalent workflows used for microbial proteomics.
Besides relative quantification workflows also absolute protein quantification workflows have been established. Even though these workflows lack either the accuracy or the protein coverage of most relative quantification workflows, the additional value achieved by this type of data is highly important for answering demanding questions of proteomics and systems biology and will be further optimized in future.
Page 5
The history of mass spectrometry driven microbial proteomics
2 The history of mass spectrometry driven microbial proteomics (article I)
The history and importance of mass spectrometry in microbial proteomics is reviewed and exemplary shown for the Gram-positive bacterium B. subtilis in article I. The world-wide interest in B. subtilis as a simple model for understanding cell differentiation and adaptation on a molecular scale has generated extensive knowledge of the genetics and biochemistry of this organism. Additionally, the long-lasting use of the genus Bacillus in biotechnology has initiated further research of genetics and cell physiology.
An important basement for mass spectrometry driven proteomic analyses of B. subtilis was the published genome sequence in 1997 (18). The sub-proteome of soluble proteins located in the cytosole was the first deeply analyzed fraction of B. subtilis as these proteins were accessible for 2D-PAGE-based analysis. The bacterium's response to different types of stress and starvation stimuli have been characterized in various studies. 2D-PAGE in conjunction with radioactive metabolic pulse labeling (e.g. 35 S methionine) allows distinguishing between newly synthesized and already accumulated proteins in a cell. The use of dual channel image for image analysis and quantification and MALDI-TOF for identification lead to a deep analysis of the cytosolic proteins and various post-translational modifications in B. subtilis by this workflow as review in detail by Hecker et al. (19).
Even though the 2D-PAGE technique has been optimized for the analysis of the B. subtilis proteome by applying narrow range and alkaline pH gradients, highly sensitive fluorescent stains and the newly designed high-performance electrophoresis (HPE) system, the majority of the 2,800 predicted cytosolic proteins is still restricted from 2D-PAGE based analysis.
The introduction of gel-free approaches based on different pre-fractionation strategies and LC-MS/MS methods could drastically increase the proteome coverage. Whereas 2D-PAGE based studies could analyze up to 750 different proteins in B. subtilis (20, 21), Wolf and co-workers as well as Hahne and co- workers could increase the number of identified proteins to more than 1,700, of which more than 1,100 were predicted to be localized in the cytosole covering about 40% of the theoretical proteome (22, 23).
For quantification chemical labeling techniques like isobaric tags for relative and absolute quantitation (ITRAQ) as well as isotopic labeling like 15 N and SILAC have been applied in these ESI-LC-MS/MS based approaches.
Page 6
The history of mass spectrometry driven microbial proteomics
With the introduction of gel-free approaches the highly hydrophobic sub-proteome of membrane proteins were no longer excluded from proteome analyses. According to predictions based on the TMHMM (trans-membrane helices prediction by hidden Markov model) algorithm (24) about one quarter carry at least one trans-membrane domain. The first successful study of B. subtilis membrane proteins was achieved in 2004 by Bunai et al. (25) profiling ABC-transporter solute-binding proteins and identifying 256 predicted membrane proteins. In 2008 Hahne et al. (26) combined a membrane protein enrichment strategy and a membrane shaving procedure with proteinase K and were able to identify more than 500 of the predicted 1073 integral membrane proteins in one of the so far biggest proteomic study in B. subtilis targeting membrane proteins.
Besides intracellular located and membrane bound proteins a large group of B. subtilis proteins are cell surface associated or located in the extracellular space. Approximately 300 proteins contain a signal peptide that determines their translocation via different secretion pathways.
The group of proteins secreted in the extracellular space can be precipitated from the grown media and was analyzed in several studies under various growth conditions mainly using 2D-PAGE based approaches (27–33).
The cell surface associated proteins are accessible through a more recently developed approach of enrichment of surface-exposed proteins based on biotinylation and subsequent neutravidin affinity chromatography which can then be analyzed via LC-MS/MS (34).
In the so far biggest comprehensive analysis of B. subtilis all previously described enrichment strategies were applied to obtain a global view on cytosolic, integral membrane, membrane, surface and extracellular proteins (11). A combination of LC-MS/MS and 15 N metabolic labeling allowed the identification of 2,142 and quantification of 2,048 proteins during transition from a growing to non- growing state induced by glucose starvation (Figure 3).
Page 7
The history of mass spectrometry driven microbial proteomics
Figure 3: Summary of identified B. subtilis proteins achieved by Otto and co-workers (11). This summary considers the predicted localization of the proteins in the cell. (A) Identified proteins colored according to the predicted localization in the cell displayed in a theoretical 2D- gel. (B) Numbers of identified proteins in the different investigated protein fractions (EMF, enriched membrane protein fraction; MSF, membrane shaving fraction; BEF, biotinylation- enriched fraction of surface- associated proteins).
A similarly high coverage of the B. subtilis proteome was presented by Macek and coworkers using SILAC for quantitation to investigate B. subtilis differential proteome expression in response to gluconeogenetic growth on succinate and phosphate starvation (35). Besides the high number of 1,730 quantified proteins the analyses of global phosphorylation allowed one of the most detailed characterizations of post-translational modification dynamics in bacteria so far.
Page 8
LC-ESI-MS for global proteomics and post-translational modification analysis
3 LC-ESI-MS for global proteomics and post-translational modification analysis
2D-PAGE in combination with MALDI-TOF-TOF is a powerful and useful tool for the investigation of bacterial proteomes. However, the history of proteomics in B. subtilis has demonstrated that LC-ESI-MS approaches became more and more relevant as they can overcome the 2D-PAGE-based protein restrictions.
3.1 Global proteome studies (article II)
The capability of delivering highly accurate quantities for large numbers of proteins that is given with modern ESI-instruments and appropriate sample preparation was used in article II for a global proteome analysis of S. aureus in response to the antibiotic vancomycin. The Gram-positive bacterium S. aureus is an important human pathogen, especially due to the emergence of methicillin resistant strains (MRSA) that renders it a worldwide and severe health threat.
Vancomycin is a glycopeptide antibiotic that inhibits the cell wall synthesis of Gram-positive bacteria and is regarded as a drug of last resort for treatment of MRSA infections. But the occurrence of few vancomycin resistant strains (VRSA) and specially the increasing number of vancomycin intermediate susceptible strains (VISA) necessitate a better understanding of the cellular response of S. aureus to vancomycin presence.
Genome-wide transcriptome studies have analyzed vancomycin stress in S. aureus before, but proteomic studies in comparable scale have not been published so far. Article II describes such a global proteomic analysis of vancomycin stress in S. aureus . The use of state-of-the-art modern ESI-MS instruments in combination with in vivo 15 N metabolic labeling and extensive pre-fractionation allowed the quantification of nearly 1,400 proteins in response to vancomycin induced stress. Special enrichment strategies ensured high sub-proteome coverage not only for intracellular proteins but also for membrane bound and associated proteins as well as for extracellular and surface associated proteins (Figure 4).
Page 9
LC-ESI-MS for global proteomics and post-translational modification analysis
Figure 4: Fractionation workflow for the global analysis of vancomycin induced stress in S. aureus . Most crucial steps in the preparation of the six sub-fractions are displayed.
The visualization of global data in Voronoi treemaps that cluster proteins according to their function, so that functionally related elements are localized in close neighborhood to each other, and statistical filtering of the quantitative data were used to identify the major adaptive mechanism of S. aureus to the changed environmental conditions. We could observe plausible adaptations like the increase of proteins involved in cell wall synthesis, in the synthesis of amino acids that are essential for peptidoglycan synthesis or in protein quality control of exported proteins. Other observations like the decrease of autolysis related proteins or the decrease of iron uptake systems cannot be explained as a direct response to vancomycin presence, but confirm with results of transcriptomic studies analyzing cell wall stress in S. aureus .
One of the most striking results of this study was the uniform decreased amount of proteins with a virulence related function. This contradicts most transcriptomic studies analyzing similar conditions in S. aureus .
Page 10
LC-ESI-MS for global proteomics and post-translational modification analysis
Besides the clustering of proteins by functional categories in a Voronoi treemap we also clustered proteins by their affiliation to global regulons. This allowed the identification of the main regulators in vancomycin stressed S. aureus cells. The highest increase in amount showed the group of proteins regulated by the two-component system VraS/VraR. The strength and consistency of augmentation of these proteins clearly points at the activation of this regulatory system as the main and most specific response to the changed environmental conditions in our experiment. Kuroda et al. reported that this two-component system positively controls the expression of 46 genes in S. aureus N315 under vancomycin presence (36), Utaida et al. could show that it is the main regulator of a hypothesized cell wall stress stimulon (37).
Figure 5 indicates that another important regulator in S. aureus in our experiment was the alternative sigma factor SigB. The cluster of proteins under positive SigB control mainly increased, whereas negatively regulated proteins primarily decreased in amount after vancomycin addition.
The activation of the SigB regulon in S. aureus by vancomycin or at related cell wall stress conditions has not been reported so far, but an induced SigB expression has been recognized in some VISA strains after vancomycin addition (38). Increased SigB activity has also been observed in teicoplanin resistant strains (39); teicoplanin is another closely related glycopeptide antibiotic. Activation of the SigB regulon in B. subtilis after cell wall stress induction by glycopeptides like vancomycin has been observed in several studies (40, 41).
The use of extensive pre-fractionation and state-of-the-art LC-ESI-MS allowed for the first time a deep insight on the proteomic level into the cellular response of S. aureus coping with vancomycin induced stress. The visualization of global data in functionally clustered Voronoi treemaps in combination with a statistical filtering of the quantitative data pointed out some major adaptations of the bacterium to the changed environmental conditions and allowed the identification of the main regulatory systems implicated in the response to vancomycin in S. aureus .
Page 11
LC-ESI-MS for global proteomics and post-translational modification analysis
Figure 5: Regulon Voronoi treemap of vancomycin stressed S. aureus COL. Each cell represents one quantified protein which is associated with other functionally related proteins in parent convex-shaped categories. These are again summarized in higher-level categories. The treemap design is based on hierarchically structured regulatory data according to classification of “The Institute for Genome Research” (black borders: regulon/thin black borders within the regulons: operon/smallest cells: proteins). + / − depict regulons being induced ( + ) or repressed ( − ) depending on the regulator assigned to the area.
Page 12
LC-ESI-MS for global proteomics and post-translational modification analysis
3.2 Investigation of post-translational modification (article III and IV)
The proteome of an organism is not sufficiently defined by the total number of genes. Various known post-translational modifications (PTMs) change physicochemical attributes of proteins like their charge, hydrophobicity or conformation and subsequently create multiple gene products from a signal DNA sequence.
Investigation of PTMs in bacteria was for a long time exclusively based on 2D-PAGE based analysis as separation is based on proteins which allows the routinely detection of different protein isoforms if they differ in their p I and mass. Technical advances in the field of high-accuracy MS now allow not only the identification of the modified protein but also the exact localization of the modification site and so renders investigation of known PTMs possible for gel-free LC-MS/MS approaches.
One of the most important and best characterized PTM in bacteria is the reversible phosphorylation of proteins that allows the cell to adapt rapidly to changed environmental conditions by regulating protein activity.
Special enrichment strategies combined with high-accuracy and fast acquisition ESI-MS instruments have been used to extensively characterize the Ser, Thr and Tyr phosphoproteome of different bacterial species. Protein phosphorylations on either His or Asp, although commonly used in bacterial two- component regulatory systems, are difficult to detect by 2D-PAGE as well as LC-MS based approaches, because of its chemical instability at lower pH values (pH < 8). Recently, a bacterial protein arginine kinase, McsB, was identified and its activity characterized in vitro (42). Two McsB substrates, which were phosphorylated on arginine residues, were identified in vitro (42, 43) suggesting that McsB-mediated protein arginine phosphorylation could occur in bacteria, but neither the occurrence of the modification in vivo itself nor a physiological function for it has yet been described.
In Article III a global phosphoproteome studie was carried out to distinctively detect Arg phosphorylations in B. subtilis . As in vivo approaches in B. subtilis wild-type so far failed in proofing its existence, the analysis was carried out in an ywlE mutant. YwlE phosphatase has been identified as the cognate McsB phosphatase in vitro and was shown to antagonize the activity of McsB in vivo (43–47). The observation strongly suggests that YwlE may act as a general protein arginine phosphatase so that protein phosphorylation on arginine residues could be more stable and enriched in an ywlE deletion strain.
Page 13
LC-ESI-MS for global proteomics and post-translational modification analysis
118 unique arginine phosphorylation sites in 87 proteins have been identified in an ywlE mutant strain with a global, quantitative, label-free, gel-free, and site-specific approach using high accuracy ESI-MS in combination with biochemical enrichment of phosphopeptides from digested cell lysates using TiO 2 chromatography.
With this study we demonstrated that the recently identified modification of proteins by arginine phosphorylation (42) occurs and is functional in vivo . It has also been shown that detection of arginine phosphorylation is McsB/YwlE dependent, which strongly indicates that both proteins are protein arginine kinase and phosphatase rather than tyrosine kinase and phosphatase as previously suggested (44, 48).
A comprehensive transcriptome analysis of the ywlE deletion mutant with the wild type strain demonstrated the strong regulatory potential of this protein modification, which, in addition to the large number of detected arginine phosphorylation sites, indicates for a high importance of this PTM in physiological and cellular processes in B. subtilis .
The importance of PTMs in microbial cell physiology and the possibilities in their detection through state-of-the-art proteomics becomes also apparent in the analysis of S-bacillithiolation in B. subtilis (Article IV). Protein S-thiolation is a post-translational thiol-modification that controls redox-sensing transcription factors and protects active site cysteine residues against irreversible oxidation.
In this study we investigated the global response, regulatory mechanisms, and post-translational thiol- modifications that contribute to the resistance of the bacterium to the strong oxidant hypochlorid acid. Transcriptomic and redox proteomic approaches identified the OhrA peroxiredoxin as the major resistant determinant to NaOCl. The application of highly accurate LC-ESI-MS/MS enabled the distinct identification of S-bacillithiolations of the OhrR repressor, two methionine synthases MetE and YxjG, the inorganic pyrophosphatase PpaC, and the 3-D-phosphoglycerate dehydrogenase SerA and revealing it as major protection mechanisms against hypochlorite stress in B. subtilis .
4 Proteomics for systems biology
Systems biology describes the study of an organism as a complex system instead of focusing on individual parts of it (49). It includes comprehensive data at all molecular levels from DNA over RNA to proteins and metabolites. Genomic as well as transcriptomic data could already meet the requirements
Page 14
Proteomics for systems biology on data quality for quite a while as they deliver genome-wide and highly accurate data. The recent progress made in proteomics now enables similar data quality on protein level as well and so allowed its integration to systems biology approaches.
4.1 Dynamic adaptations of B. subtilis metabolism analyzed in a systems biology approach (article V)
In article V a systems biology approach was used to analyze the dynamic interplay between metabolic and regulatory networks systematically. We investigated dynamic shifts in availability of the preferred carbon sources, glucose and malate of B. subtilis. To elucidate the cellular adaptation mechanisms, we determined transcript, protein and absolute metabolite abundances as well as promoter activities providing dynamic data with up to 24 time points for two shift experiments that cover short-term metabolic to longer-term protein level adaptation (Figure 6).
To acquire the highest possible quality for the proteomic data we combined LC-ESI-Orbitrap and 2D- PAGE-MALDI-TOF approaches. The LC-ESI-Orbitrap analysis obtained highly accurate protein quantities on large-scale, using a 15 N metabolic labeling approach. Only four data points of both analyzed nutrient shifts were analyzed by this method, because of the long analysis time. To also gain highly time resolved data twelve data points were analyzed with a 2D-PAGE-MALDI-TOF-TOF approach. Although the amount of protein quantities per data point is lower and of lower accuracy, this approach allows a comparably fast analysis of large time series experiments. Measurement times can also be reduced strongly, since mass spectrometry based identifications of one data point can be transferred via gel image analysis to all other time points and quantifications are based on image analysis as well.
For both proteomic data types confidence-weighted averages were calculated and replicate time series data were afterwards smoothed and interpolated by Bayesian multi-curve regression analysis. Proteomic data as well as all other obtained data types were integrated into different modeling systems to understand how previously known regulatory mechanisms are combined to affect the investigated nutrition shift. The used methodology could discern the key regulatory events for both nutrient shifts and it could be shown that the overall control strategy of B. subtilis can be rationalized in terms of its evolutionary advantages; however, these advantages, and therefore the overall control design, depend on quantitative system characteristics.
Page 15
Proteomics for systems biology
Figure 6: Overview of our experimental design, computational analysis, and information flow. The dynamic shift experiments yielded measured data from which non-measurable quantities were inferred by statistical and model-based analysis. Arrows show the flow of information, including the use of inferred data as parameters for subsequent modeling steps.
4.2 A newly developed approach to obtain absolute protein quantities (article VI)
Besides highly accurate and time resolved data, systems biology is also in need for large-scale absolute protein quantities. This allows e.g. calculating copy numbers of proteins in a cell, which is an important input for modeling systems. In general there are two different approaches to obtain absolute quantities.
Page 16
Proteomics for systems biology
The first is based on targeted mass spectrometry utilizing synthetic peptides with incorporated stable isotopes as internal standards to quantify cellular peptides generated by proteolysis (50). Such pairs of native and heavy peptides are typically analyzed in ESI-triple-quadrupole-type mass spectrometers implementing multiple reaction monitoring (MRM) - a highly selective and sensitive acquisition mode (51). However, reaching a system-wide perspective by quantifying all proteins of interest constitutes a considerable workload and financial burden.
The second workflow to obtain absolute protein quantities is entirely mass spectrometry-based and uses LC-ESI-MS data to compare protein abundance indices which take into account the number of sequenced peptides per protein (52), by relating averages of ion signal intensities of the three most abundant peptides of proteins (53), or by measuring spectral counts which represent the number of matched MS/MS spectra for a protein (54). These workflows can provide estimations of protein concentrations on global-scale.
Article VI describes a new workflow integrating targeted mass spectrometric analyses of selected anchor proteins with quantification of all proteins accessible on calibrated two-dimensional gels. It allows the easy, large-scale determination of absolute protein concentrations using MRM LC-ESI-MS for the quantification of anchor proteins and MALDI-TOF-MS for the identification of proteins separated with 2D-PAGE.
The set of absolutely quantified proteins that was used for the calibration of all identified protein spots present on the analyzed 2D-gel needed to fulfill three selection criteria: (I) presence on 2D-gels under examined conditions, (II) arrangement in good quality spots on gels which can be reliably detected by custom software packages, and (III) the sum of anchor spots should cover the molecular mass and isoelectric point range examined. After calculating the absolute protein amount of anchor proteins based on targeted MRM analysis, their absolute amount on 2D-gels could be calculated as the same extract was used for 2D-PAGE and MRM analysis. In the next step we calibrated a 2D-gel based on the absolute amount of anchor proteins on the gel. Such an MS-calibrated 2D-gel constitutes the basis for an estimation of the absolute abundance of all proteins visible on this gel just by relating their spot intensities to the ones of anchor proteins.
This workflow is based on several assumptions: (I) the efficiency of protein digestion is supposed to be 100%, (II) there is no protein-specific bias regarding the rehydration to the immobilized pH-gradient- strip, isoelectric focusing, and the transfer to the second dimension, and (III) that protein staining on 2D-
Page 17
Proteomics for systems biology gels is independent of protein sequence and linear over a wide range of protein levels. These assumptions were basis for the development of an optimized workflow presented in Figure 7.
Figure 7: Workflow for absolute quantification of proteins by calibration of 2D-gels via targeted mass spectrometry: (a) all steps involved in sample preparation; (b) partial parallel analysis of the sample by targeted mass spectrometry and 2D-PAGE for absolute quantification.
To illustrate the power of this approach we have chosen to analyze the adaptation of B. subtilis and S. aureus to glucose starvation as proof-of-principle examples. Protein concentrations at three different time points along the bacterial growth curves - during exponential growth (t0), in early stationary phase (t1), and in late stationary phase (t2) - have been determined.
The innovative implementation of MS calibrated 2D-gels enabled reliable estimation of the abundance of 467 proteins of B. subtilis 482 proteins of S. aureus at three time points during glucose starvation. To determine accuracy of these protein concentrations we bootstrapped the MRM-based result, i.e., the real absolute quantity obtained for a specific anchor protein was compared with the value calculated
Page 18
Proteomics for systems biology from spot intensity comparisons to the other anchor proteins on the gel. The bootstrapping provided an average error rate of 1.77-fold for B. subtilis and 1.36-fold for S. aureus considering all time points and all anchor proteins.
Furthermore, we validated our quantification approach with an independent test set of proteins that were available through a QconCAT (quantification concatemer) analysis (55) resulting in good correlation between both methods with an average error of 1.97-fold. Thus, the newly developed approach reaches similar precision as other current methods for estimating absolute protein abundance on a proteome-wide scale (14).
Additionally to the calibration of one 2D-gel via absolute quantification of few anchor proteins it was possible to transfer absolute data of an MS-calibrated 2D-gel onto other gels. This allows estimating absolute protein abundance of samples, which were not analyzed by targeted MS directly, significantly expanding the application potential of this novel approach.
5 Exchangeability and complementarity of ESI-MS and MALDI-MS
The application of the two generally used ionization techniques in MS-based proteomic research is today largely separated into MALDI-MS instruments for the analysis of 2D-PAGE workflows and ESI-MS instruments for liquid chromatography based workflows.
Both instrument types offer advantages for the analyses of these two predominantly used workflows. MALDI-TOF-TOF instruments are especially suited for the fast and reliable analysis of a high number of lowly complex samples that typically follow from 2D-PAGE analysis. Even though coupling with LC instruments is possible, it mostly requires an additional fractionation step. The direct online coupling is for most instruments permitted, because the mass analyzer is a closed vacuum chamber and sample insertion is based on a lock chamber.
The enabled online coupling and high acquisition speed of modern ESI-MS instruments meet the requirements for large LC-MS analysis. Therefore, the great majority of these workflows are combined with this type of instrument. Especially in MS-based, relative quantification workflows LC-MALDI-MS stands back behind the widespread LC-ESI-MS workflows. Although it had been shown that shot-to-shot intensity variation, which is a general drawback for quantitative MALDI analysis, can be overcome with a
Page 19
Exchangeability and complementarity of ESI-MS and MALDI-MS suitable experimental setup (56), the lack of established software tools for the analysis of these data is evident and still hampers application of LC-MALDI-MS in large-scale experiments.
5.1 Global relative quantification with LC-MALDI-MS (article VII)
In article VII we described for the first time a global analysis of in vivo 15 N labeled samples with a LC-MS workflow using the 5800 MALDI-TOF-TOF analyzer (AB Sciex, Foster City, CA, USA). In our workflow proteins were pre-fractionated on a 1D-SDS-gel, tryptically digested, and the resulting peptide mixtures were separated by reversed-phase liquid chromatography. The LC eluate was then fractionated, which allowed offline coupling with a MALDI-TOF-TOF instrument. Data was analyzed with the Mascot Distiller software package (Matrix Science, London, UK).
Samples taken for this analysis have been the very same cytosolic samples used for the analysis of vancomycin induced stress in S. aureus (article II). As we here used same sample preparation and LC- system, we could cross-validate our results to a well established workflow based on ESI-LTQ-Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany).
Analysis of the six different cytosolic samples (three exponentially grown biological replicates, three vancomycin stressed biological replicates) resulted in identification of 848 proteins and 6,552 peptides in total. A comparison with identification numbers obtained by ESI-LTQ-Orbitrap analysis from the same samples revealed a considerably higher sensitivity for the LTQ-Orbitrap in our analysis setup. The Venn diagram in Figure 8 shows a very big overlap of identified proteins and the number of proteins exclusively identified with our LC-MALDI-TOF-TOF workflow was with only 39 very low. Identifications on peptide level showed a significantly lower relative overlap, indicating for a general complementary of both applied mass spectrometer types.
Figure 8: Venn diagram of a) identified proteins and b) identified peptides.
Page 20
Exchangeability and complementarity of ESI-MS and MALDI-MS
Approximately half of all peptides identified by MALDI-TOF-TOF analysis could also be quantified against the corresponding 15 N labeled peptide of the spiked-in standard using the Mascot Distiller software (between 49% and 51% quantification efficiency for the six different samples). In total, quantitative data for 554 proteins could be obtained, up to 415 in a single biological replicate.
The achieved quantification efficiency of about 50% was significantly lower than in our ESI-LTQ-Orbitrap analysis (about 80%). A main reason for this discrepancy might be the lower number of MS1 scans per time frame of the LC-run in our LC-MALDI analysis. Higher time-resolved data enable a better statistical analysis of a peptide's elution peak. The higher the number of data points in such an elution peak is, the less error-prone data processing in the quantification process will be. MS1 scans on a MALDI-TOF-TOF instrument take much more time than on the LTQ-Orbitrap, as one main MS1 spectrum is an average of multiple sub-spectra recorded on different locations of a spot. This averaging of different spectra is especially necessary in quantitative analyses to overcome the MALDI-specific problem of poor reproducibility of signal intensities (56). This inhibits a comparable time resolution of MS1 scans, as the analysis time would increase unreasonably.
The obtained quantities resulting from our new approach were of high quality. 14 N/ 15 N ratios for the same peptides of the different biological replicates analyzed with MALDI-TOF-TOF were in good correlation, as illustrated in Figure 9 a, where peptide ratios of two exponentially growing biological replicates are plotted against each other. Calculated trend lines for these plots of biological replicates showed slopes near to one and an average coefficient of determination of 0.71. As these replicates include a biological variance, the correlation is high on technical level. The cross-validation of MALDI- TOF-TOF generated peptide ratios with the LTQ-Orbitrap generated data from the same sample is shown in Figure 9 b. The very good correlation of LC-MALDI generated quantities with the ones obtained by the well-established quantification workflow using Census quantification software (57) and an LTQ-Orbitrap proves the general reliability of the new workflow.
Page 21
Exchangeability and complementarity of ESI-MS and MALDI-MS
Figure 9: Scatter plots of a) two replicates analyzed with MALDI-TOF-TOF and b) same sample analyzed with MALDI-TOF-TOF and with LTQ-Orbitrap. a) log2 ratios of two exponentially grown biological replicates against the internal standard, both analyzed with MALDI-TOF-TOF, are plotted against each other. b) log2 ratios one exponentially grown biological replicate against the internal standard, one analyzed with MALDI-TOF-TOF and the other with LTQ-Orbitrap, are plotted against each other. Linear trendline (black) with linear equation (y) and coefficient of determination (R 2) was calculated.
We were able to obtain reproducible and reliable data with a LC-MALDI-TOF-TOF based workflow using a combination of averaging enough laser shots over a large sample area and isotopically labeled peptides as internal standards, which were two proposed strategies to overcome the MALDI-specific problems in quantification (56).
5.2 Complementary of MALDI and ESI (article VII)
The analysis of the same samples separated with the same chromatographic systems and parameters but two different mass spectrometers in article VII allowed us to investigate whether peptides with certain biochemical features were preferentially detected with either of them.
Existing literature postulates such tendencies against certain peptide properties when using either MALDI or ESI for ionization. The large dataset of 13,000 identified peptides, of which more than 5,200 could have been detected with both mass spectrometers, enabled the a statistically meaningful comparison of an ESI and a MALDI instrument regarding their biases in the ionization of peptides.
To assign occurring differences between the two data sets to the different mass analyzers distinctively, we wanted to equal not only sample preparation, but also data processing for both data sets as far as possible. Therefore, we performed the same database search for the three exponentially grown
Page 22
Exchangeability and complementarity of ESI-MS and MALDI-MS replicates analyzed with MALDI-TOF-TOF as we did in the previous study for the LTQ-Orbitrap generated data.
To emphasize differences between the two detection techniques we focused on peptides that were preferentially identified by only one mass spectrometer. The comparison of relative abundances of the different amino acids showed significant differences between the two data sets generated with either MALDI-TOF-TOF or LTQ-Orbitrap (Figure 10).
Figure 10: Relative abundance of amino acids in peptides preferentially identified by MALDI-TOF-TOF (dark grey) and LTQ-Orbitrap (light grey bars).
Whereas the MALDI-based analysis showed biases against peptides containing larger proportions of arginine, aromatic amino acids and proline, the ESI instrument seemed to favor lysine, aliphatic amino acids and amino acids containing a carboxyl or hydroxyl group.
We could proof that these tendencies for specific amino acid classes affect the sensitivity of each ionization technique for peptides with certain biochemical properties and so result in a general complementary of ESI and MALDI in peptide analysis.
6 Conclusion This thesis demonstrates that LC-MS based workflows equipped with state-of-the-art electrospray ionization mass spectrometers are capable of analyzing microbial proteomes in quantitative and global manner and can overcome restrictions of 2D-PAGE-MALDI-MS workflows. Also the extensive analysis of
Page 23
Conclusion
post-translational modifications of proteins has been proven to not only be possible but often to exceed classical 2D-PAGE based workflows.
Despite that, 2D-PAGE-MALDI-MS based analyses still have major advantages, which can often complement LC-ESI-MS workflows. It has been shown that fast and reliable quantification of several hundred proteins in many samples by 2D-PAGE-MALDI-MS in conjunction with global and highly accurate quantification of few time points by LC-ESI-MS in the same time course experiment is of great benefit for systems biology approaches.
Combination of the same workflows in a newly developed approach allowed absolute protein quantification on large-scale in a rapid, cost-efficient manner, which is of high interest for a broad research community.
Finally, we were able to proof that global quantitative LC-MS analyses are feasible with MALDI-TOF-TOF instruments. This should not only encourage laboratories equipped with such an instrument type to perform large-scale quantitative proteomic experiments. The observed complementary preferences in peptide ionization for both ionization techniques should also motivate a reconsideration of the obligate separation of MALDI-MS for 2D-PAGE workflows and ESI-MS for LC based workflows.
Page 24
References
7 References
1. O’Farrell,P.H. (1975) High resolution two-dimensional electrophoresis of proteins. J. Biol. Chem , 250 , 4007–4021.
2. Yamashita,M. and Fenn,J.B. (1984) Electrospray ion source. Another variation on the free-jet theme. J. Phys. Chem. , 88 , 4451–4459.
3. Karas,M., Bachmann,D. and Hillenkamp,F. (1985) Influence of the wavelength in high-irradiance ultraviolet laser desorption mass spectrometry of organic molecules. Anal. Chem. , 57 , 2935– 2939.
4. Tanaka,K., Waki,H., Ido,Y., Akita,S., Yoshida,Y., Yoshida,T. and Matsuo,T. (1988) Protein and polymer analyses up to m/z 100 000 by laser ionization time-of-flight mass spectrometry. Rapid Communications in Mass Spectrometry , 2, 151–153.
5. Knochenmuss,R. and Zenobi,R. (2003) MALDI ionization: the role of in-plume processes. Chem. Rev. , 103 , 441–452.
6. Iribarne,J.V. and Thomson,B.A. (1976) On the evaporation of small ions from charged droplets. The Journal of chemical physics , 64 , 2287.
7. Dole,M., Mack,L.L., Hines,R.L., Mobley,R.C., Ferguson,L.D. and Alice,M.B. (1968) Molecular Beams of Macroions. The Journal of Chemical Physics , 49 , 2240–2249.
8. Znamenskiy,V., Marginean,I. and Vertes,A. (2003) Solvated Ion Evaporation from Charged Water Nanodroplets. J. Phys. Chem. A , 107 , 7406–7412.
9. Marginean,I., Znamenskiy,V. and Vertes,A. (2006) Charge Reduction in Electrosprays: Slender Nanojets as Intermediates. J. Phys. Chem. B , 110 , 6397–6404.
10. Makarov (2000) Electrostatic axially harmonic orbital trapping: a high-performance technique of mass analysis. Anal. Chem. , 72 , 1156–1162.
11. Otto,A., Bernhardt,J., Meyer,H., Schaffer,M., Herbst,F.-A., Siebourg,J., Mäder,U., Lalk,M., Hecker,M. and Becher,D. (2010) Systems-wide temporal proteomic profiling in glucose-starved Bacillus subtilis. Nat Commun , 1, 137.
12. Becher,D., Hempel,K., Sievers,S., Zühlke,D., Pané-Farré,J., Otto,A., Fuchs,S., Albrecht,D., Bernhardt,J., Engelmann,S., et al. (2009) A Proteomic View of an Important Human Pathogen – Towards the Quantification of the Entire Staphylococcus aureus Proteome. PLoS ONE , 4, e8176.
13. de Godoy,L.M.F., Olsen,J.V., Cox,J., Nielsen,M.L., Hubner,N.C., Fröhlich,F., Walther,T.C. and Mann,M. (2008) Comprehensive mass-spectrometry-based proteome quantification of haploid versus diploid yeast. Nature , 455 , 1251–1254.
Page 25
References
14. Malmström,J., Beck,M., Schmidt,A., Lange,V., Deutsch,E.W. and Aebersold,R. (2009) Proteome-wide cellular protein concentrations of the human pathogen Leptospira interrogans. Nature , 460 , 762–765.
15. Bunai,K., Nozaki,M., Kakeshita,H., Nemoto,T. and Yamane,K. (2005) Quantitation of de novo localized (15)N-labeled lipoproteins and membrane proteins having one and two transmembrane segments in a Bacillus subtilis secA temperature-sensitive mutant using 2D- PAGE and MALDI-TOF MS. J. Proteome Res. , 4, 826–836.
16. Ong,S.-E., Blagoev,B., Kratchmarova,I., Kristensen,D.B., Steen,H., Pandey,A. and Mann,M. (2002) Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach to expression proteomics. Mol. Cell Proteomics , 1, 376–386.
17. Oda,Y., Huang,K., Cross,F.R., Cowburn,D. and Chait,B.T. (1999) Accurate quantitation of protein expression and site-specific phosphorylation. Proc. Natl. Acad. Sci. U.S.A. , 96 , 6591–6596.
18. Kunst,F., Ogasawara,N., Moszer,I., Albertini,A.M., Alloni,G., Azevedo,V., Bertero,M.G., Bessières,P., Bolotin,A., Borchert,S., et al. (1997) The complete genome sequence of the gram-positive bacterium Bacillus subtilis. Nature , 390 , 249–256.
19. Hecker,M., Antelmann,H., Büttner,K. and Bernhardt,J. (2008) Gel-based proteomics of Gram-positive bacteria: A powerful tool to address physiological questions. Proteomics , 10.1002/pmic.200800278.
20. Büttner,K., Bernhardt,J., Scharf,C., Schmid,R., Mäder,U., Eymann,C., Antelmann,H., Völker,A., Völker,U. and Hecker,M. (2001) A comprehensive two-dimensional map of cytosolic proteins of Bacillus subtilis. Electrophoresis , 22 , 2908–35.
21. Eymann,C., Dreisbach,A., Albrecht,D., Bernhardt,J., Becher,D., Gentner,S., Tam,L.T., Büttner,K., Buurman,G., Scharf,C., et al. (2004) A comprehensive proteome map of growing Bacillus subtilis cells. Proteomics , 4, 2849–76.
22. Wolff,S., Otto,A., Albrecht,D., Zeng,J.S., Büttner,K., Glückmann,M., Hecker,M. and Becher,D. (2006) Gel-free and gel-based proteomics in Bacillus subtilis: a comparative study. Mol Cell Proteomics , 5, 1183–92.
23. Hahne,H., Mäder,U., Otto,A., Bonn,F., Steil,L., Bremer,E., Hecker,M. and Becher,D. (2010) A comprehensive proteomics and transcriptomics analysis of Bacillus subtilis salt stress adaptation. J. Bacteriol , 192 , 870–882.
24. Krogh,A., Larsson,B., von Heijne,G. and Sonnhammer,E.L. (2001) Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J. Mol. Biol. , 305 , 567– 580.
25. Bunai,K., Ariga,M., Inoue,T., Nozaki,M., Ogane,S., Kakeshita,H., Nemoto,T., Nakanishi,H. and Yamane,K. (2004) Profiling and comprehensive expression analysis of ABC transporter solute- binding proteins of Bacillus subtilis membrane based on a proteomic approach. Electrophoresis , 25 , 141–155.
Page 26
References
26. Hahne,H., Wolff,S., Hecker,M. and Becher,D. (2008) From complementarity to comprehensiveness-- targeting the membrane proteome of growing Bacillus subtilis by divergent approaches. Proteomics , 8, 4123–36.
27. Antelmann,H., Darmon,E., Noone,D., Veening,J.-W., Westers,H., Bron,S., Kuipers,O.P., Devine,K.M., Hecker,M. and van Dijl,J.M. (2003) The extracellular proteome of Bacillus subtilis under secretion stress conditions. Mol. Microbiol. , 49 , 143–156.
28. Antelmann,H., Scharf,C. and Hecker,M. (2000) Phosphate starvation-inducible proteins of Bacillus subtilis: proteomics and transcriptional analysis. J. Bacteriol. , 182 , 4478–4490.
29. Antelmann,H., Tjalsma,H., Voigt,B., Ohlmeier,S., Bron,S., van Dijl,J.M. and Hecker,M. (2001) A proteomic view on genome-based signal peptide predictions. Genome Res. , 11 , 1484–1502.
30. Antelmann,H., Yamamoto,H., Sekiguchi,J. and Hecker,M. (2002) Stabilization of cell wall proteins in Bacillus subtilis: a proteomic approach. Proteomics , 2, 591–602.
31. Hirose,I., Sano,K., Shioda,I., Kumano,M., Nakamura,K. and Yamane,K. (2000) Proteome analysis of Bacillus subtilis extracellular proteins: a two-dimensional protein electrophoretic study. Microbiology (Reading, Engl.) , 146 ( Pt 1) , 65–75.
32. Jongbloed,J.D.H., Antelmann,H., Hecker,M., Nijland,R., Bron,S., Airaksinen,U., Pries,F., Quax,W.J., van Dijl,J.M. and Braun,P.G. (2002) Selective contribution of the twin-arginine translocation pathway to protein secretion in Bacillus subtilis. J. Biol. Chem. , 277 , 44068–44078.
33. Jongbloed,J.D., Martin,U., Antelmann,H., Hecker,M., Tjalsma,H., Venema,G., Bron,S., van Dijl,J.M. and Müller,J. (2000) TatC is a specificity determinant for protein secretion via the twin-arginine translocation pathway. J. Biol. Chem. , 275 , 41350–41357.
34. Hempel,K., Pané-Farré,J., Otto,A., Sievers,S., Hecker,M. and Becher,D. (2010) Quantitative cell surface proteome profiling for SigB-dependent protein expression in the human pathogen Staphylococcus aureus via biotinylation approach. J. Proteome Res. , 9, 1579–1590.
35. Soufi,B., Kumar,C., Gnad,F., Mann,M., Mijakovic,I. and Macek,B. (2010) Stable isotope labeling by amino acids in cell culture (SILAC) applied to quantitative proteomics of Bacillus subtilis. J. Proteome Res. , 9, 3638–3646.
36. Kuroda,M., Kuroda,H., Oshima,T., Takeuchi,F., Mori,H. and Hiramatsu,K. (2003) Two-component system VraSR positively modulates the regulation of cell-wall biosynthesis pathway in Staphylococcus aureus. Mol. Microbiol , 49 , 807–821.
37. Utaida,S., Dunman,P.M., Macapagal,D., Murphy,E., Projan,S.J., Singh,V.K., Jayaswal,R.K. and Wilkinson,B.J. (2003) Genome-wide transcriptional profiling of the response of Staphylococcus aureus to cell-wall-active antibiotics reveals a cell-wall-stress stimulon. Microbiology , 149 , 2719– 2732.
38. Chen,H.-Y., Chen,C.-C., Fang,C.-S., Hsieh,Y.-T., Lin,M.-H. and Shu,J.-C. (2011) Vancomycin activates σ(B) in vancomycin-resistant Staphylococcus aureus resulting in the enhancement of cytotoxicity. PLoS ONE , 6, e24472.
Page 27
References
39. Bischoff,M. and Berger-Bächi,B. (2001) Teicoplanin stress-selected mutations increasing sigma(B) activity in Staphylococcus aureus. Antimicrob. Agents Chemother. , 45 , 1714–1720.
40. Mascher,T., Margulis,N.G., Wang,T., Ye,R.W. and Helmann,J.D. (2003) Cell wall stress responses in Bacillus subtilis : the regulatory network of the bacitracin stimulon. Molecular Microbiology , 50 , 1591–1604.
41. Cao,M., Wang,T., Ye,R. and Helmann,J.D. (2002) Antibiotics that inhibit cell wall biosynthesis induce expression of the Bacillus subtilis sigma(W) and sigma(M) regulons. Mol Microbiol , 45 , 1267–76.
42. Fuhrmann,J., Schmidt,A., Spiess,S., Lehner,A., Turgay,K., Mechtler,K., Charpentier,E. and Clausen,T. (2009) McsB is a protein arginine kinase that phosphorylates and inhibits the heat-shock regulator CtsR. Science , 324 , 1323–1327.
43. Elsholz,A.K.W., Hempel,K., Michalik,S., Gronau,K., Becher,D., Hecker,M. and Gerth,U. (2011) Activity control of the ClpC adaptor McsB in Bacillus subtilis. J. Bacteriol. , 193 , 3887–3893.
44. Kirstein,J., Zühlke,D., Gerth,U., Turgay,K. and Hecker,M. (2005) A tyrosine kinase and its activator control the activity of the CtsR heat shock repressor in B. subtilis. EMBO J. , 24 , 3435–3445.
45. Elsholz,A.K.W., Michalik,S., Zühlke,D., Hecker,M. and Gerth,U. (2010) CtsR, the Gram-positive master regulator of protein quality control, feels the heat. EMBO J. , 29 , 3621–3629.
46. Hahn,J., Kramer,N., Briley,K.,Jr and Dubnau,D. (2009) McsA and B mediate the delocalization of competence proteins from the cell poles of Bacillus subtilis. Mol. Microbiol. , 72 , 202–215.
47. Kirstein,J., Strahl,H., Molière,N., Hamoen,L.W. and Turgay,K. (2008) Localization of general and regulatory proteolysis in Bacillus subtilis cells. Mol. Microbiol. , 70 , 682–694.
48. Musumeci,L., Bongiorni,C., Tautz,L., Edwards,R.A., Osterman,A., Perego,M., Mustelin,T. and Bottini,N. (2005) Low-molecular-weight protein tyrosine phosphatases of Bacillus subtilis. J. Bacteriol. , 187 , 4945–4956.
49. Kitano,H. (2002) Systems Biology: A Brief Overview. Science , 295 , 1662–1664.
50. Gerber,S.A., Rush,J., Stemman,O., Kirschner,M.W. and Gygi,S.P. (2003) Absolute quantification of proteins and phosphoproteins from cell lysates by tandem MS. PNAS , 100 , 6940–6945.
51. Lange,V., Malmström,J.A., Didion,J., King,N.L., Johansson,B.P., Schäfer,J., Rameseder,J., Wong,C.-H., Deutsch,E.W., Brusniak,M.-Y., et al. (2008) Targeted Quantitative Analysis of Streptococcus pyogenes Virulence Factors by Multiple Reaction Monitoring. Mol Cell Proteomics , 7, 1489– 1500.
52. Ishihama,Y., Schmidt,T., Rappsilber,J., Mann,M., Hartl,F.U., Kerner,M.J. and Frishman,D. (2008) Protein abundance profiling of the Escherichia coli cytosol. BMC Genomics , 9, 102.
53. Silva,J.C., Gorenstein,M.V., Li,G.-Z., Vissers,J.P.C. and Geromanos,S.J. (2006) Absolute Quantification of Proteins by LCMSE A Virtue of Parallel ms Acquisition. Mol Cell Proteomics , 5, 144–156.
Page 28
References
54. Lu,P., Vogel,C., Wang,R., Yao,X. and Marcotte,E.M. (2007) Absolute protein expression profiling estimates the relative contributions of transcriptional and translational regulation. Nat. Biotechnol. , 25 , 117–124.
55. Beynon,R.J., Doherty,M.K., Pratt,J.M. and Gaskell,S.J. (2005) Multiplexed absolute quantification in proteomics using artificial QCAT proteins of concatenated signature peptides. Nat. Methods , 2, 587–589.
56. Szájli,E., Fehér,T. and Medzihradszky,K.F. (2008) Investigating the quantitative nature of MALDI-TOF MS. Mol. Cell Proteomics , 7, 2410–2418.
57. Park,S.K., Venable,J.D., Xu,T. and Yates,J.R. (2008) A quantitative analysis software tool for mass spectrometry-based proteomics. Nat. Methods , 5, 319–322.
Page 29
References
Page 30
Article I
Page 31
Page 32
Proteomics 2011, 11, 2971–2980 DOI 10.1002/pmic.201100090 2971
REVIEW From the genome sequence to the protein inventory of Bacillus subtilis
Do¨rte Becher, Knut Buttner,. Martin Moche, Bernd Heßling and Michael Hecker
Institute for Microbiology, Ernst Moritz Arndt University Greifswald, Greifswald, Germany
Owing to the low number of proteins necessary to render a bacterial cell viable, bacteria are Received: February 16, 2011 extremely attractive model systems to understand how the genome sequence is translated Revised: April 7, 2011 into actual life processes. One of the most intensively investigated model organisms is Accepted: April 20, 2011 Bacillus subtilis. It has attracted world-wide research interest, addressing cell differentiation and adaptation on a molecular scale as well as biotechnological production processes. Meanwhile, we are looking back on more than 25 years of B. subtilis proteomics. A wide range of methods have been developed during this period for the large-scale qualitative and quantitative proteome analysis. Currently, it is possible to identify and quantify more than 50% of the predicted proteome in different cellular subfractions. In this review, we summarize the development of B. subtilis proteomics during the past 25 years.
Keywords: Bacteria / Global protein analysis / Microbiology / Methods / Profiling / Proteome maps
1 Introduction The soil-dwelling, Gram-positive bacterium Bacillus subtilis has become a widely used model for functional The release of the first complete bacterial genome sequence genomics by joint efforts of many groups in the USA, Japan, in 1995 opened the era of functional genomics [1]. In the and Europe [3, 4]. The world-wide interest in B. subtilis as a same year, the term ‘‘proteome’’ was coined to describe the simple model for understanding cell differentiation and complete set of proteins synthesized under specific physio- adaptation on a molecular scale has produced extensive logical conditions [2]. Since then, the genome sequences of knowledge of the genetics and biochemistry of this organ- hundreds of organisms, including the human, have been ism. Additionally, the long-lasting use of the genus Bacillus determined. However, a genome sequence alone does not in biotechnology has instigated further research of genetics explain life; it provides only the ‘‘blueprint of life.’’ It is the and cell physiology [5, 6]. As a result of a joint research proteome which translates this ‘‘blueprint’’ into cellular program of approximately 30 European and Japanese functions, and therefore proteins can be regarded as the groups, the B. subtilis genome sequence was published in main players of life. The integration of multilevel cellular 1997, revealing more than 4100 genes [7]. In 2009, this expression profiles (transcriptomics, proteomics, and meta- sequence was updated [8]. With the whole genome sequence bolomics) into a systems biology perspective allows attaining available, mRNA profiling of gene expression and proteome a new quality in the understanding of life. Microbial model analyses on a global scale became possible. Although we are organisms play a particularly important role in this respect looking back on more than 25 years of proteome research in due to their low complexity, only a few hundred different B. subtilis [9, 10], still, to almost one-third of its genes, no proteins together with essential nonproteinogenous specific function has been assigned yet [3, 11]. Therefore, components render a bacterial cell viable. the elucidation of the function of the respective gene products remains a big challenge. Almost 10 years ago, the Correspondence: Dr. Do¨ rte Becher, Institute for Microbiology, first steps were taken by the construction of a mutant library Ernst Moritz Arndt University Greifswald, Friedrich-Ludwig- containing single mutations in each gene of unknown Jahn-Strasse 15, 17487 Greifswald, Germany function and a comprehensive phenotype screening E-mail: [email protected] program to obtain some indications of the specific function Fax: 149-3834-864202 of the unknown proteins [12]. Meanwhile, B. subtilis has also Abbreviation: HPE, high-performance electrophoresis been chosen as a model organism for systems biology used
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com 2972 D. Becher et al. Proteomics 2011, 11, 2971–2980 within the European research consortium BaSysBio (http:// Furthermore, even previously unknown post-translational www.basysbio.eu/). Moreover, B. subtilis is an ideal experi- modifications can be discovered and structurally character- mental system for studying the mechanisms of gene regu- ized in detail as exemplarily shown for succinylation [21]. lation, metabolism and differentiation in order to apply the Using traditional 2-DE and stains (e.g. Colloidal methodological approaches and findings to related micro- Coomassie or silver nitrate) about 500 proteins in the pI organisms such as the human pathogen Staphylococcus range of 4–7 are accessible. This number can be increased aureus. by applying narrow range pH gradients (pI 4–5, 4.5–5.5, Proteomics started in 1975 when O’Farrell published the 5–6, and 5.5–6.7) and gels covering the alkaline pH range revolutionary 2-DE separation technique of proteins that 7–12. Thus, up to 750 proteins of exponentially growing allows the separation of thousands of proteins [13]. Soon cells could be identified, covering more than 40% of the afterwards, Ruth Van Bogelen and Fred Neidhardt used this predicted cytosolic proteins [22, 23]. However, with tradi- method to address the crucial issues of the cell physiology of tional stains the low-abundant proteins remain almost Escherichia coli [14]. Twenty years later, the sequences of completely inaccessible. Highly sensitive staining is entire genomes provided the basis for large-scale protein required for successful identification of these low-abundant identification by sensitive MS techniques. Concurrently, the proteins in 2-DE. Fluorescent dyes like Lava Purple, development of mild ionization methods for peptides has Flamingo, or Krypton clearly outperform the conventional also promoted the high-throughput protein identification by stains like Colloidal Coomassie or silver nitrate in terms of MS [15–17]. The combination of both the development of sensitivity or MS compatibility and are the appropriate very sensitive and the highly reproducible 2-D protein choice for the analysis of low-abundant proteins. Addition- separation techniques and the development of MS of ally, a new high-performance electrophoresis (HPE) system peptides greatly enhanced the establishment of proteomics was recently introduced whose separation in the second as a new field of life sciences. At present, physiological dimension is performed in a horizontal SDS-PAGE with processes can be monitored on a large scale. Proteomics can nonfluorescent plastic-backed flatbed gels (Serva, Heidel- visualize cellular events never seen before in such detail berg, Germany). These HPE gels offer higher reproduci- with a high impact on understanding bacterial physiology bility than traditional vertical SDS-PAGE and considerably such as the responses of bacterial cells to stress and star- improved resolution since they can be run with a substan- vation [18, 19]. tially higher electrical field strength. In HPE gels, low- As a result of the development of gel-free proteomics abundant proteins which are frequently ‘‘buried’’ by high- techniques that are much more sensitive than gel-based abundant proteins in overcrowded regions of traditional gels proteomics approaches, the identification and even quanti- and small proteins are much better detectable (Fig. 2). tation of entire proteomes of simple organisms such as The combination of HPE and fluorescent stains allows bacteria become realistic goals. The identification of all the identification of more than 1550 spots in a single 2-D gel proteins expressed in a cell is an essential step toward a in the standard pI range of 4–7 (Moche et al., in prepara- comprehensive understanding how proteins differentially tion). Interestingly, the 1550 spots identified in the HPE translate the genome sequence into actual life processes gels in this pI region belong to o900 proteins, indicating under the given conditions. In this review, methods for the that many proteins occur in isoforms. The respective 2-D gel identification and the quantitation of the protein inventory and a list of all identified proteins are provided as of B. subtilis are summarized. Important milestones in the Supporting Information material (Supporting Information development of proteomics techniques and the simulta- Fig. S1 and Supporting Information Table S1). neous developments in B. subtilis proteomics are given in a Gel-based proteomics techniques have been extensively short overview (Fig. 1). used to characterize the regulation of gene expression and metabolism of B. subtilis in response to different types of stress or starvation stimuli [24–26]. Coupled with radioactive 2 Cytosolic proteins metabolic pulse labeling (e.g. with 35S-methionine), the gel- based approach allows the distinction of newly synthesized About 2800 cytosolic proteins are predicted from the proteins from proteins already accumulated in the cell genome sequence. For the analysis of cytosolic proteins, the before imposition of the stimulus. The dual channel 2-DE is still a powerful tool to address physiological issues. imaging technique is particularly suited to detect the Most metabolic pathways and the most obvious stress or proteins induced or repressed by the stimulus [27]. This starvation responses are visualized by gel-based proteomics technique allows a rapid assignment of proteins to groups of in a convenient, intelligible and intuitive way even for physiologically or regulatorily related proteins (stimulons, unclassified bacteria like the Riftia pachyptila endosymbionts regulons) collected in signature libraries by a comparison of (‘‘metabolism at a glance’’ [20]). One important advantage of protein amount (protein staining) versus protein synthesis 2-DE is that the separation is based on the proteins and not (phosphoimaging) [18, 28]. A comprehensive stress and on the peptides. Hence, in contrast to gel-free methods, starvation proteomics signature library has been created different protein isoforms can routinely be detected. which is a valuable toolbox to predict the physiological state
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com Proteomics 2011, 11, 2971–2980 2973
Figure 1. Short overview about milestones in the development of proteomics techniques and their application in B. subtilis. of cells (main nutrients, growth rate, stress/starvation more than 500 proteins were detected by only GeLC-MS/MS stimuli, etc.) grown in a bioreactor, in a biofilm or under techniques, underlining the higher sensitivity of gel-free infection-relevant conditions [29–33]. Protein modifications approaches. such as protein phosphorylation, protein formylation, The next step was the development of quantitation protein damage by oxidation and protection of reactive SH- methods for gel-free proteome analyses. At that time, new groups against oxidative damage were analyzed in the methods based on in vitro labeling with stable isotopes follow-up studies [34–40]. Furthermore, protein stability and mostly by chemical linkage of tag and biomolecule had been proteolysis can be monitored on a proteome-wide scale with established. One of the samples is usually labeled with a pulse chase labeling [33, 41]. The gel-based analysis of ‘‘light’’ tag. The other is provided with the heavy isotope- B. subtilis has been reviewed in detail earlier [19]. enriched variant of the tag [47–49]. In all these methods, As 2-D gels cover only about 25% of the proteome, quantitation is based on ion signal intensity observed at the different methods of protein prefractionation in combina- MS level, whereas the method developed by Ross and tion with LC-MS/MS were employed to increase the coworkers – named isobaric tagging for relative and absolute proteome coverage [42–44]. In 2006, Wolff et al. performed quantitation (iTRAQ) – supports a quantitation based on the the first gel-free analysis of B. subtilis using strong-ion- reporter ion signals measured at the MS/MS level [50]. The exchange (SCX) chromatography as a prefractionation design of the iTRAQ tag allows a multiplex quantitation of method in combination with LC-MS/MS (2-D-LC-MS/MS) up to eight samples – a big advantage in time course [45]. In this study, the number of identified cytosolic experiments. This iTRAQ method was used by Wolff et al. proteins could be increased to about 1000 proteins. Recently, for the quantitative proteome analysis of the cellular heat Hahne et al. identified 1710 proteins by gel-free proteome shock response in B. subtilis [45]. The results obtained by the analysis [46]. Of the 2816 cytosolic proteins predicted from gel-free approach were confirmed by a simultaneously the genome sequence about 860 were identified in growing performed 2-D gel quantitation. cells by use of 1-DE for prefractionation in combination with Two years later, Dreisbach et al. established the metabolic LC-MS/MS (GeLC-MS/MS). In total, 1156 cytosolic proteins labeling as a reproducible and reliable method for the were identified covering about 40% of the predicted analysis of enriched membrane proteins [51]. They proteome. The actual proteome coverage is even higher compared the quantitative analysis of proteins using Stable because only a part of the genome is expressed under Isotope Labeling with Amino acids in Cell culture (SILAC) defined conditions used in the studies. To assess the actual [52, 53] and 14N/15N metabolic labeling [54]. Both SILAC proteome coverage, a correlation to transcriptomic data and 14N/15N labeling provided very reliable quantitation of indicating which genes are transcribed under the experi- the membrane fraction. While SILAC offers a more mental conditions is useful. In addition to the highly straightforward analysis, a better coverage was achieved by expressed proteins also identified by a gel-based approach, 14N/15N labeling. But it has to be considered that SILAC
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com 2974 D. Becher et al. Proteomics 2011, 11, 2971–2980
study [55]. Additionally, osmotic upregulation of a large number of genes that do not belong to these regulons was observed. In total, osmotic upregulation of about 500 B. subtilis genes was detected. A global approach for absolute quantitation has been developed by Maass et al. to meet the requirements of systems biology for B. subtilis [56]. In this approach, the advantages of gel-based and gel-free quantitation are combined to a new fast and cost-effective strategy. In this strategy, the absolute intracellular concentrations of differ- ent proteins were initially determined for a subset of anchor proteins by isotopically labeled internal standard peptides. These anchor proteins were then used to calibrate 2-D gels, allowing the calculation of the absolute abundance of all detectable proteins (Fig. 3).
3 Membrane proteins
According to the TMHMM algorithm for the prediction of trans-membrane domains (TMDs) [57] about one quarter of the B. subtilis proteins carry one or more trans-membrane domains. The analysis of this protein subgroup remained a difficult task for a long time. One reason was the need for alternative analysis methods optimized for the specific physical properties of integral membrane proteins such as
Figure 2. Comparison of (A) traditional 2-DE system with (B) newly developed horizontal 2-DE system (HPE) (Moche et al., in preparation). Shown is the low-molecular-weight section of a traditional 2-DE (A) and of the new HPE (B). The displayed sections are not completely identical because of different separation characteristics of the different gel types in this region (features of HPE gels are superior) but the obviously higher resolution and better detection of small proteins in the new HPE system is clearly demonstrated. requires one or even better two auxotrophies, allowing simultaneous labeling with lysine and arginine. The 14N/15N labeling approach is amendable to bacteria even if auxotrophic mutants are difficult to be constructed, but the analysis is more labor-intensive owing to the lack of commercially available analysis software [51]. The 14N/15N labeling was applied by Hahne et al. for the analysis of the osmotic shock response on a global scale, Figure 3. Comparative illustration of traditional visual presenta- including more than 1700 cytosolic proteins and membrane tion of 2-D data and absolute quantitation by combination of proteins [46]. In this study, the first attempt was made to MRM, protein quantitation, and image analysis. Data for four target the dynamic changes of the proteome during the proteins involved in central carbon metabolism of B. subtilis are stress response by setting up a time course experiment with exemplarily illustrated. False color overlays of 2-D spot tiles are several time points before and during the stress response. utilized to visualize the trend in protein regulations. The sample corresponding to exponential growth was colored green and all This study revealed a well-coordinated induction of gene others red. Additionally, the absolute concentration in ng/mgof expression after an osmotic up-shift that involved large parts total protein and in molecules per cell (rounded) are given for the of the SigB, SigW, SigM, and SigX regulons and supports four proteins in growing cells (t0) as well as cells in early (t1)and the results obtained earlier by Ho¨per et al. in a 2-D gel-based late (t2) stationary phase induced by glucose exhaustion.
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com Proteomics 2011, 11, 2971–2980 2975 their high hydrophobicity and limited solubility in solutions global study of growing B. subtilis to an extent of about 880 compatible with 2-DE [58, 59]. proteins, 260 of the 880 identified proteins were integral The first successful study of B. subtilis membrane membrane proteins [22]. Four years later, the application of proteins was achieved in 2004 by Bunai et al. [60] profiling a method initially developed by Wolff et al. for S. aureus ABC-transporter solute-binding proteins. In the same year, enabled Hahne et al. to identify more than 500 of the 1073 Dreisbach and coworkers included membrane proteins in a predicted integral membrane proteins (i.e. almost 50%) by
Figure 4. (A) Gene expression of growing B. subtilis compared with average expression during the time course in cells entering the stationary phase. (B) Relative protein amount determined in cytosolic fraction of growing B. subtilis compared with the average protein amount during the investigated time course. To visualize the differences in expression, level/protein amount compared with the average level color coding was applied as follows: blue, decreased level (dec.); grey, same level as average (avg.); and orange, increased level (inc.). Adapted from [76].
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com 2976 D. Becher et al. Proteomics 2011, 11, 2971–2980 combining a membrane protein enrichment and a The metabolic labeling with 15N was used to characterize the membrane shaving procedure with proteinase K [61–63]. global cellular response of B. subtilis after osmotic shock in 2010 [46]. The most extensive description and quantitation of the 4 Cell surface associated and B. subtilis proteome has been provided recently by Otto et al. extracellular proteins [76]. This study represents the first comprehensive study on concomitant large-scale transcriptome and proteome As B. subtilis lacks an outer membrane, it is able to directly dynamics over a time course in B. subtilis coping with secrete large amounts of extracellular proteins into the physiological perturbations such as glucose starvation in environment. There are at least four distinct secretion this study. For proteomic profiling, a combination of in vivo pathways for approximately 300 proteins predicted to be 15N metabolic labeling and shotgun MS analysis was carried located extracellularly [64, 65]. The translocation route of a out for five different cellular fractions (cytosolic, integral protein and its final destination are determined by the membrane, membrane, surface and extracellular proteome presence of particular signal peptides. Most of the proteins fraction) [76]. transported across the cytoplasmic membrane contain For an intuitive display of such large ‘‘omics’’ data sets an N-terminal signal peptide that is cleaved by a signal comprising relative expression data and functional/regula- peptidase I (SPase I) after translocation which allows the tory classification, Voronoi treemaps were adapted [79]. protein to be released into the growth medium. For the After treemap construction and optimization, functionally analysis of secreted proteins, 2-D gel-based studies were related elements appear in close neighborhood to each carried out, considering different growth conditions such as other. Expression data were visualized in the treemaps using growth in LB medium or under phosphate starvation a color gradient. For the illustration of protein expression [66–72]. Proteins could be identified which were predicted to levels, colors from a range of blue (lower than average), grey be secreted because of an N-terminal signal peptide with a (equal to average), and orange (higher than average) were SPase I cleavage site and the lack of retention signals applied to Voronoi cells (Fig. 4). Each cell in the graph [64, 73]. On the other hand, several unpredicted secreted displays a single gene locus that is linked to other func- proteins were found which either possess signal peptides tionally related elements in parent convex-shaped categories. and retention signals (lipoproteins, cell wall-binding These are again summarized in higher-level categories. proteins and membrane proteins) or lack signal peptides Gene functional data are based on KEGG orthology (contamination of cytoplasmic proteins, phage-related (e.g. main level [metabolism]/first sublevel [carbohydrate proteins and flagella-related proteins). The identified extra- metabolism]/second sublevel [glycolysis]) [76]. cellular proteins are involved in uptake, transport or utili- This analysis led to the identification of 2145 proteins zation of carbohydrates, proteins, nucleotides, lipids, and and the quantitation of 2048 proteins. Detecting about 52% phosphate, in the cell-wall metabolism, in environmental of the predicted proteome of B. subtilis, it provides the most detoxification or they have flagella- and phage-related func- detailed proteomic picture of the B. subtilis starvation tions [68, 73]. response to date (Fig. 5). This coverage had been even A new approach for the analysis of surface-associated higher if the proteins synthesized in response to stress proteins comprises the biotinylation of surface-exposed stimuli would have been included. For instance, the proteins with subsequent enrichment via neutravidin affinity approximately 500 sporulation proteins are not considered chromatography. This method has been proven to be applic- in this study. able for the qualitative and quantitative studies of The Voronoi treemaps allow tracing back the stringent S. aureus [74, 75] and has been successfully applied for the control, the SigB-dependent general stress response, and the analysis of surface-associated proteins in B. subtilis as well [76]. carbon metabolism such as the TCA (tricarboxylic acid) cycle and gluconeogenesis. Most of the genes encoding proteins that are only required for main vegetative functions such as 5 Global quantitative proteome analysis growth and reproduction are no longer active in glucose- in growing and glucose-starved starved cells. The kinetics of those vegetative proteins in B. subtilis cells starved cells indicates a tightly controlled proteolysis of proteins no longer required in nongrowing cells. The The first global gel-based proteome analyses of B. subtilis hypothesis is that those ‘‘unemployed proteins’’ are no revealed that most proteins of the cytosolic metabolic path- longer integrated into functional complexes [80] and thereby ways are detectable [22, 23, 77]. Proteome studies were no longer protected against a proteolytic attack by the Clp therefore used to delineate active metabolic pathways and to machine [75, 81, 82]. The degradation of ribosomes in analyze their regulation [24, 33, 78]. They are still an glucose-starved cells is probably a main nutrient resource in appropriate tool to rapidly assess the cell metabolism. In the nongrowing cells [75]. first global gel-free analysis in 2006 [45], iTRAQ was applied Besides detailed information on the expression rates of for quantitation to analyze the cellular heat stress response. proteins, reliable information about the localization of
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com Proteomics 2011, 11, 2971–2980 2977
Figure 5. Summary of identified B. subtilis proteins achieved in [76]. This summary considers the predicted localization of the proteins in the cell. (A) Identified proteins colored according to the predicted localization in the cell displayed in a theoretical 2-D gel. (B) Numbers of identified proteins in the different investigated protein fractions (EMF, enri- ched membrane protein frac- tion; MSF, membrane shaving fraction; BEF, biotinylation- enriched fraction of surface- associated proteins). almost all detected proteins in the cell was gained. Several ization of the protein inventory dynamics during transition proteins were exclusively identified in one of the five from a growing to a nongrowing state induced by glucose proteomic fractions, indicating their localization in the starvation. cytosol or targeting to the cytosolic membrane, the cell wall A similarly high coverage of the B. subtilis proteome was or to the extracellular milieu. This information is of special presented by Macek and coworkers using SILAC for quan- interest for proteins of still unknown function as it provides titation to investigate B. subtilis differential proteome clues for the analysis of their role. expression in response to gluconeogenetic growth on succinate and phosphate starvation [83]. In total, 1730 proteins of the B. subtilis proteome could be accurately and 6 Concluding remarks and outlook reproducibly quantified. Notably, Macek et al. demonstrated that this approach is fully compatible with the measurement In summary, we have not only identified the majority of of global phosphorylation dynamics at the level of phos- proteins synthesized in growing B. subtilis cells but also phorylation sites, which has not been achieved in bacteria provide quantitative results for many proteins differentially so far. expressed in response to environmental stimuli such as In 2009, a new perspective was opened by Malmstro¨m osmotic stress or nutrient limitation, a frequent situation in et al. [84] who developed and applied a MS-based strategy to the natural habitat of B. subtilis. Furthermore, we present an determine the absolute quantity (i.e. average number of extensive list of all detected cytosolic, membrane-bound and protein copies per cell in a defined cell population) for more secreted proteins. This accessible protein inventory of living than 50% of the Leptospira interrogans proteome B. subtilis cells can be used for detailed physiological (Malmstro¨m et al., 2011, this issue). Methods like this pave analyses as exemplarily shown by the in-depth character- the way toward the mathematical modeling and simulation
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com 2978 D. Becher et al. Proteomics 2011, 11, 2971–2980 of biological processes in systems biology studies and [7] Kunst, F., Ogasawara, N., Moszer, I., Albertini, A. M. et al., corroborate the power of state-of-the-art MS-based proteo- The complete genome sequence of the Gram-positive mic approaches. bacterium Bacillus subtilis. Nature 1997, 390, 249–256. The progress in bacterial proteomics will have a great [8] Barbe, V., Cruveiller, S., Kunst, F., Lenoble, P. et al., From a impact on future studies on the cell physiology of B. subtilis. consortium sequence to a unified sequence: the Bacillus Not only protein expression profiling in response to envir- subtilis 168 reference genome a decade later. Microbiology onmental stimuli can now be accomplished with a very high 2009, 155, 1758–1775. proteomic coverage; the virulence protein inventory of [9] Streips, U. N., Polio, F. W., Heat shock proteins in Bacilli. bacterial pathogens can also be captured, analyzed, and J. Bacteriol. 1985, 162, 434–437. quantified. This will dramatically improve our knowledge [10] Richter, A., Hecker, M., Heat-shock proteins in Bacillus about the virulence and adaptation strategies of pathogens subtilis: a two-dimensional gel electrophoresis study. FEMS and about host–pathogen interactions [75]. Microbiol. Lett. 1986, 36, 69–71. Protein expression profiling, however, is only the first [11] Lammers, C. R., Flo´ rez, L. A., Schmeisky, A. G., Roppel, S. F. step toward the understanding of life processes. Using a et al., Connecting parts with processes: SubtiWiki and combination of gel-based and gel-free proteomics, the final SubtiPathways integrate gene and pathway annotation for destination and fate of each single protein can be followed Bacillus subtilis. Microbiology 2010, 156, 849–859. on a proteome-wide scale [19]. Detailed studies on the [12] Schumann, W., Ehrlich, D., Ogasawara, N. (Eds.), Functional dynamics of the membrane proteome, on the phosphopro- Analysis of Bacterial Genes: A Practical Manual, Wiley, teome and its role in signal transduction, on the damage, Weinheim 2001. the repair and the degradation of proteins, or on the protein [13] O’Farrell, P. H., High resolution two-dimensional electro- interaction network will provide a new quality in under- phoresis of proteins. J. Biol. Chem. 1975, 250, 4007–4021. standing cell physiology and the underlying dynamic [14] Neidhardt, F. C., Wirth, R., Smith, M. W., Van Bogelen, R., molecular processes [19]. Selective synthesis of plasmid-coded proteins by Escher- ichia coli during recovery from chloramphenicol treatment. The authors thank Holger Kock for kindly revising the J. Bacteriol. 1980, 143, 535–537. manuscript. This work was supported by the Grants from the [15] Hillenkamp, F., Karas, M., Mass spectrometry of peptides Bundesministerium fur. Bildung und Forschung (0313812C, and proteins by matrix-assisted ultraviolet laser desorption/ 0313978A, 0315784A, 0315592B, and 03ZIK011), and the EU ionization. Methods Enzymol. 1990, 193, 280–295. (BaSysBio LSHG-CT-2006-037469). [16] Hillenkamp, F., Karas, M., Beavis, R. C., Chait, B. T., Matrix- assisted laser desorption/ionization mass spectrometry of The authors have declared no conflict of interest. biopolymers. Anal. Chem. 1991, 63, 1193A–1203A. [17] Fenn, J. B., Mann, M., Meng, C. K., Wong, S. F. et al., Electrospray ionization for mass spectrometry of large biomolecules. Science 1989, 246, 64–71. 7 References [18] Hecker, M., Vo¨ lker, U., Towards a comprehensive under- standing of Bacillus subtilis cell physiology by physiologi- [1] Fleischmann, R. D., Adams, M. D., White, O., Clayton, R. A. cal proteomics. Proteomics 2004, 4, 3727–3750. et al., Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 1995, 269, 496–512. [19] Hecker, M., Antelmann, H., Buttner, K., Bernhardt, J., Gel-based proteomics of Gram-positive bacteria: a powerful [2] Wasinger, V. C., Cordwell, S. J., Cerpa-Poljak, A., Yan, J. X. tool to address physiological questions. Proteomics 2008, 8, et al., Progress with gene-product mapping of the Molli- 4958–4975. cutes: Mycoplasma genitalium. Electrophoresis 1995, 16, 1090–1094. [20] Markert, S., Arndt, C., Felbeck, H., Becher, D. et al., Physiological proteomics of the uncultured endosymbiont [3] Sonenshein, A. L., Hoch, J. A., Losick, R. (Eds.), Bacillus of Riftia pachyptila. Science 2007, 315, 247–250. subtilis and its Closest Relatives: from Genes to Cells, American Society for Microbiology Press, Washington, DC [21] Rosen, R., Becher, D., Buttner, K., Biran, D. et al., Probing 2002. the active site of homoserine trans-succinylase. FEBS Lett. [4] Hecker, M., A proteomic view of cell physiology of Bacillus 2004, 577, 386–392. subtilis – bringing the genome sequence to life. Adv. [22] Eymann, C., Dreisbach, A., Albrecht, D., Bernhardt, J. et al., Biochem. Eng. Biotechnol. 2003, 83, 57–92. A comprehensive proteome map of growing Bacillus [5] Zweers, J. C., Barak, I., Becher, D., Driessen, A. J. et al., subtilis cells. Proteomics 2004, 4, 2849–2876. Towards the development of Bacillus subtilis as a cell [23] Buttner,. K., Bernhardt, J., Scharf, C., Schmid, R. et al., A factory for membrane proteins and protein complexes. comprehensive two-dimensional map of cytosolic Microb. Cell. Fact. 2008, 7, 10. proteins of Bacillus subtilis. Electrophoresis 2001, 22, [6] Harwood, C. R., Bacillus subtilis and its relatives: molecular 2908–2935. biological and industrial workhorses. Trends Biotechnol. [24] Mader, U., Homuth, G., Scharf, C., Buttner, K. et al., Tran- 1992, 10, 247–256. scriptome and proteome analysis of Bacillus subtilis gene
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com Proteomics 2011, 11, 2971–2980 2979
expression modulated by amino acid availability. [40] Liebeke, M., Pother, D. C., van Duy, N., Albrecht, D. et al., J. Bacteriol. 2002, 184, 4288–4295. Depletion of thiol-containing proteins in response to [25] Hochgrafe, F., Wolf, C., Fuchs, S., Liebeke, M. et al., Nitric quinones in Bacillus subtilis. Mol. Microbiol. 2008, 69, oxide stress induces different responses but mediates 1513–1529. comparable protein thiol protection in Bacillus subtilis and [41] Gerth, U., Kock, H., Kusters,. I., Michalik, S. et al., Staphylococcus aureus. J. Bacteriol. 2008, 190, 4997–5008. Clp-dependent proteolysis down-regulates central meta- bolic pathways in glucose-starved Bacillus subtilis. J. [26] Tobisch, S., Zuhlke, D., Bernhardt, J., Stulke, J. et al., Role of Bacteriol. 2008, 190, 321–331. CcpA in regulation of the central pathways of carbon catabolism in Bacillus subtilis. J. Bacteriol. 1999, 181, [42] Wolters, D. A., Washburn, M. P., Yates J. R., 3rd, An 6996–7004. automated multidimensional protein identification tech- nology for shotgun proteomics. Anal. Chem. 2001, 73, [27] Bernhardt, J., Buttner, K., Scharf, C., Hecker, M., Dual 5683–5690. channel imaging of two-dimensional electropherograms in Bacillus subtilis. Electrophoresis 1999, 20, 2225–2240. [43] Link, A. J., Eng, J., Schieltz, D. M., Carmack, E. et al., Direct analysis of protein complexes using mass spectrometry. [28] Bernhardt, J., Weibezahn, J., Scharf, C., Hecker, M., Bacillus Nat. Biotechnol. 1999, 17, 676–682. subtilis during feast and famine: visualization of the overall regulation of protein synthesis during glucose starvation by [44] Washburn, M. P., Wolters, D., Yates J. R., 3rd, Large-scale proteome analysis. Genome Res. 2003, 13, 224–237. analysis of the yeast proteome by multidimensional protein identification technology. Nat. Biotechnol. 2001, 19, [29] Voigt, B., Schweder, T., Becher, D., Ehrenreich, A. et al., A 242–247. proteomic view of cell physiology of Bacillus licheniformis. Proteomics 2004, 4, 1465–1490. [45] Wolff, S., Otto, A., Albrecht, D., Zeng, J. S. et al., Gel-free and gel-based proteomics in Bacillus subtilis: a comparative [30] Voigt, B., Hoi le, T., Jurgen, B., Albrecht, D. et al., The study. Mol. Cell. Proteomics 2006, 5, 1183–1192. glucose and nitrogen starvation response of Bacillus liche- . niformis. Proteomics 2007, 7, 413–423. [46] Hahne, H., Mader, U., Otto, A., Bonn, F. et al., A compre- hensive proteomics and transcriptomics analysis of Bacillus [31] Hoi le, T., Voigt, B., Jurgen, B., Ehrenreich, A. et al., The subtilis salt stress adaptation. J. Bacteriol. 2010, 192, phosphate-starvation response of Bacillus licheniformis. 870–882. Proteomics 2006, 6, 3582–3601. [47] Gygi, S. P., Rist, B., Gerber, S. A., Turecek, F. et al., Quan- [32] Bandow, J. E., Hecker, M., Proteomic profiling of cellular titative analysis of complex protein mixtures using isotope- stresses in Bacillus subtilis reveals cellular networks and coded affinity tags. Nat. Biotechnol. 1999, 17, 994–999. assists in elucidating antibiotic mechanisms of action. Prog. [48] Chakraborty, A., Regnier, F. E., Global internal standard Drug Res. 2007, 64, 79, 81–101. technology for comparative proteomics. J. Chromatogr. A [33] Tam le, T., Antelmann, H., Eymann, C., Albrecht, D. et al., 2002, 949, 173–184. Proteome signatures for stress and starvation in Bacillus [49] Schmidt, A., Kellermann, J., Lottspeich, F., A novel strategy subtilis as revealed by a 2-D gel image color coding for quantitative proteomics using isotope-coded protein approach. Proteomics 2006, 6, 4565–4585. labels. Proteomics 2005, 5, 4–15. [34] Hochgrafe, F., Mostertz, J., Pother, D. C., Becher, D. et al., [50] Ross, P. L., Huang, Y. N., Marchese, J. N., Williamson, B. S-Cysteinylation is a general mechanism for thiol protection et al., Multiplexed protein quantitation in Saccharomyces of Bacillus subtilis proteins after oxidative stress. J. Biol. cerevisiae using amine-reactive isobaric tagging reagents. Chem. 2007, 282, 25981–25985. Mol. Cell. Proteomics 2004, 3, 1154–1169. [35] Hochgrafe, F., Mostertz, J., Albrecht, D., Hecker, M., [51] Dreisbach, A., Otto, A., Becher, D., Hammer, E. et al., Fluorescence thiol modification assay: oxidatively modified Monitoring of changes in the membrane proteome proteins in Bacillus subtilis. Mol. Microbiol. 2005, 58, during stationary phase adaptation of Bacillus subtilis 409–425. using in vivo labeling techniques. Proteomics 2008, 8, [36] Eymann, C., Becher, D., Bernhardt, J., Gronau, K. et al., 2062–2076. Dynamics of protein phosphorylation on Ser/Thr/Tyr in [52] Ong, S. E., Blagoev, B., Kratchmarova, I., Kristensen, D. B. Bacillus subtilis. Proteomics 2007, 7, 3509–3526. et al., Stable isotope labeling by amino acids in cell culture, [37] Bandow, J. E., Becher, D., Buttner, K., Hochgrafe, F. et al., SILAC, as a simple and accurate approach to expression The role of peptide deformylase in protein biosynthesis: a proteomics. Mol. Cell. Proteomics 2002, 1, 376–386. proteomic study. Proteomics 2003, 3, 299–306. [53] de Godoy, L. M., Olsen, J. V., de Souza, G. A., Li, G. et al., [38] Mostertz, J., Hochgrafe, F., Jurgen, B., Schweder, T. et al., Status of complete proteome analysis by mass spectro- The role of thioredoxin TrxA in Bacillus subtilis: a proteo- metry: SILAC labeled yeast as a model system. Genome mics and transcriptomics approach. Proteomics 2008, 8, Biol. 2006, 7, R50. 2676–2690. [54] Conrads, T. P., Alving, K., Veenstra, T. D., Belov, M. E. et al., [39] Pother, D. C., Liebeke, M., Hochgrafe, F., Antelmann, H. Quantitative analysis of bacterial and mammalian et al., Diamide triggers mainly S Thiolations in the cyto- proteomes using a combination of cysteine affinity plasmic proteomes of Bacillus subtilis and Staphylococcus tags and 15N-metabolic labeling. Anal. Chem. 2001, 73, aureus. J. Bacteriol. 2009, 191, 7520–7530. 2132–2139.
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com 2980 D. Becher et al. Proteomics 2011, 11, 2971–2980
[55] Hoper, D., Bernhardt, J., Hecker, M., Salt stress adaptation dimensional protein electrophoretic study. Microbiology of Bacillus subtilis: a physiological proteomics approach. 2000, 146, 65–75. Proteomics 2006, 6, 1550–1562. [71] Jongbloed, J. D., Antelmann, H., Hecker, M., Nijland, R. [56] Maass, S., Sievers, S., Zuhlke, D., Kuzinski, J. et al., Efficient, et al., Selective contribution of the twin-arginine transloca- global-scale quantification of absolute protein amounts by tion pathway to protein secretion in Bacillus subtilis. J. Biol. integration of targeted mass spectrometry and two-dimen- Chem. 2002, 277, 44068–44078. sional gel-based proteomics. Anal. Chem. 2011, 83, [72] Jongbloed, J. D., Martin, U., Antelmann, H., Hecker, M. 2677–2684. et al., TatC is a specificity determinant for protein secretion [57] Krogh, A., Larsson, B., von Heijne, G., Sonnhammer, E. L., via the twin-arginine translocation pathway. J. Biol. Chem. Predicting transmembrane protein topology with a hidden 2000, 275, 41350–41357. Markov model: application to complete genomes. J. Mol. [73] Tjalsma, H., Antelmann, H., Jongbloed, J. D., Braun, P. G. Biol. 2001, 305, 567–580. et al., Proteomics of protein secretion by Bacillus subtilis: [58] Santoni, V., Molloy, M., Rabilloud, T., Membrane proteins separating the ‘‘secrets’’ of the secretome. Microbiol. Mol. and proteomics: un amour impossible? Electrophoresis Biol. Rev. 2004, 68, 207–233. 2000, 21, 1054–1070. [74] Hempel, K., Pane´ -Farre´, J., Otto, A., Sievers, S. et al., [59] Rabilloud, T., Membrane proteins and proteomics: love is Quantitative cell surface proteome profiling for SigB- possible, but so difficult. Electrophoresis 2009, 30, dependent protein expression in the human pathogen S174–S180. Staphylococcus aureus via biotinylation approach. J. Proteome Res. 2010, 9, 1579–1590. [60] Bunai, K., Ariga, M., Inoue, T., Nozaki, M. et al., Profiling and comprehensive expression analysis of ABC transporter [75] Becher, D., Hempel, K., Sievers, S., Zuhlke, D. et al., A solute-binding proteins of Bacillus subtilis membrane based proteomic view of an important human pathogen – towards on a proteomic approach. Electrophoresis 2004, 25, 141–155. the quantification of the entire Staphylococcus aureus proteome. PLoS One 2009, 4, e8176. [61] Wu, C. C., MacCoss, M. J., Howell, K. E., Yates, J. R., 3rd, A method for the comprehensive proteomic analysis of [76] Otto, A., Bernhardt, J., Meyer, H., Schaffer, M. et al., membrane proteins. Nat. Biotechnol. 2003, 21, 532–538. Systems-wide temporal proteomic profiling in glucose- starved Bacillus subtilis. Nat. Commun. 2010, 1, 137. [62] Wolff, S., Hahne, H., Hecker, M., Becher, D., Complementary analysis of the vegetative membrane proteome of the [77] Ohlmeier, S., Scharf, C., Hecker, M., Alkaline proteins of human pathogen Staphylococcus aureus. Mol. Cell. Bacillus subtilis: first steps towards a two-dimensional alkaline master gel. Electrophoresis 2000, 21, 3701–3709. Proteomics 2008, 7, 1460–1468. [78] Tam le, T., Eymann, C., Antelmann, H., Albrecht, D. et al., [63] Hahne, H., Wolff, S., Hecker, M., Becher, D., From comple- Global gene expression profiling of Bacillus subtilis in mentarity to comprehensiveness--targeting the membrane response to ammonium and tryptophan starvation as proteome of growing Bacillus subtilis by divergent revealed by transcriptome and proteome analysis. J. Mol. approaches. Proteomics 2008, 8, 4123–4136. Microbiol. Biotechnol. 2007, 12, 121–130. [64] Tjalsma, H., Bolhuis, A., Jongbloed, J. D., Bron, S. et al., [79] Balzer, M., Deussen, O., Voronoi treemaps. In Proc IEEE Signal peptide-dependent protein transport in Bacillus Symposium on Information Visualization, 2005, InfoVis subtilis: a genome-based survey of the secretome. Micro- 05:49–56. biol. Mol. Biol. Rev. 2000, 64, 515–547. [80] Meyer, F. M., Gerwig, J., Hammer, E., Herzberg, C. et al., [65] van Dijl, J. M., Braun, P. G., Robinson, C., Quax, W. J. et al., Physical interactions between tricarboxylic acid cycle Functional genomic analysis of the Bacillus subtilis Tat enzymes in Bacillus subtilis: evidence for a metabolon. pathway for protein secretion. J. Biotechnol. 2002, 98, Metab Eng 2011, 13, 18–27. 243–254. [81] Gerth, U., Kruger, E., Derre, I., Msadek, T. et al., Stress [66] Antelmann, H., Darmon, E., Noone, D., Veening, J. W. et al., induction of the Bacillus subtilis clpP gene encoding a The extracellular proteome of Bacillus subtilis under homologue of the proteolytic component of the Clp secretion stress conditions. Mol. Microbiol. 2003, 49, protease and the involvement of ClpP and ClpX in stress 143–156. tolerance. Mol. Microbiol. 1998, 28, 787–802. [67] Antelmann, H., Scharf, C., Hecker, M., Phosphate starvation- [82] Michalik, S., Liebeke, M., Zuhlke, D., Lalk, M. et al., Proteo- inducible proteins of Bacillus subtilis: proteomics and lysis during long-term glucose starvation in Staphylococcus transcriptional analysis. J. Bacteriol. 2000, 182, 4478–4490. aureus COL. Proteomics 2009, 9, 4468–4477. [68] Antelmann, H., Tjalsma, H., Voigt, B., Ohlmeier, S. et al., A [83] Soufi, B., Kumar, C., Gnad, F., Mann, M. et al., Stable proteomic view on genome-based signal peptide predic- isotope labeling by amino acids in cell culture (SILAC) tions. Genome Res. 2001, 11, 1484–1502. applied to quantitative proteomics of Bacillus subtilis. [69] Antelmann, H., Yamamoto, H., Sekiguchi, J., Hecker, M., J. Proteome Res. 2010, 9, 3638–3646. Stabilization of cell wall proteins in Bacillus subtilis:a [84] Malmstrom, J., Beck, M., Schmidt, A., Lange, V. et al., proteomic approach. Proteomics 2002, 2, 591–602. Proteome-wide cellular protein concentrations of the [70] Hirose, I., Sano, K., Shioda, I., Kumano, M. et al., Proteome human pathogen Leptospira interrogans. Nature 2009, 460, analysis of Bacillus subtilis extracellular proteins: a two- 762–765.
& 2011 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim www.proteomics-journal.com
Article II
Page 43
Page 44
Title: Global proteome analysis of vancomycin stress in Staphylococcus aureus
Bernd Hessling1, Florian Bonn1, Florian‐Alexander Herbst1,2, Gerd‐Martin Rappen1, Jörg Bernhardt1,
Michael Hecker1 and Dörte Becher1
1Ernst‐Moritz‐Arndt‐University of Greifswald, Institute for Microbiology, F.‐L. Jahnstrasse 15, 17489
Greifswald, Germany
2UFZ ‐ Helmholtz Centre for Environmental Research, Department of Proteomics, Permoserstrasse 15,
04318 Leipzig, Gremany
Corresponding author: Dörte Becher
email: dbecher@uni‐greifswald.de
Tel.: (+49)3834‐864230
Fax: (+49)3834‐864202
Running title: Vancomycin stress in Staphylococcus aureus
1
List of abbreviations
GeLC‐MS 1D‐gel liquid chromatography mass spectrometry
LB Luria‐Bertani medium
MRSA methicillin resistant Staphylococcus aureus
S. aureus Staphylococcus aureus
TEAB triethylammonium bicarbonat
VISA vancomycin resistant Staphylococcus aureus
VRSA vancomycin resistant Staphylococcus aureus
VSSA vancomycin susceptibility Staphylococcus aureus
LTQ linear trap quadrupole
2
Summary
Vancomycin is one of the few remaining treatment options for multi resistant Staphylococcus aureus infections. Several transcriptomic and proteomics studies have investigated the bacterium’s cellular response to vancomycin, but quantitative proteomic studies have been limited in the number of proteins and restricted to certain sub‐cellular compartments so far.
Here, we combined the enrichment of different proteomic sub‐fractions with in vivo metabolic labeling and shotgun proteomics to analyze the vancomycin induced stress response. Quantitative data for about
1,400 proteins could be obtained, which allowed for the first time an in‐depth analysis of global proteome changes in response to vancomycin. Voronoi treemaps in conjunction with statistical data filtering identified major adaptations of the bacterium to the presence of vancomycin, e.g. the selective induction and repression of certain amino acid synthesis pathways and the decreased abundance of the majority of virulence related proteins. Whereas most findings correlate well with existing transcriptome studies, but the consistently decreased amount of virulence factors has not been observed so far.
The clustering of proteins according to their affiliation to different regulons revealed the two component system VraSR and the alternative sigma factor SigB as two important regulators active under vancomycin induced stress conditions in S. aureus.
3
Introduction
The gram‐positive bacterium Staphylococcus aureus is an important human pathogen which causes a wide range of infections from skin infection (1) to endocarditis (2) and septicemia (3). Especially the emergence of methicillin resistant strains (MRSA) rendered S. aureus a worldwide and severe health threat.
The glycopeptide vancomycin is the treatment of choice against MRSA strains. The antibiotic inhibits the cell wall synthesis in S. aureus by binding via hydrogen bonds to the C‐terminal D‐Ala‐D‐Ala residue of the peptidoglycan precursor, forming a stable complex and thus excluding the precursor from being used for cell wall synthesis (4, 5).
Vancomycin resistance emerged first in enterococci in 1988 (6). Resistance was caused by the vanA gene that altered the specific binding site in the peptidoglycan precursor from D‐Ala‐D‐Ala to D‐Ala‐D‐lactate which no longer binds vancomycin. In 2002 the first fully resistant S. aureus strains, which had incorporated the vanA gene from enterococci, appeared. Fortunately only few cases of vancomycin resistant S. aureus (VRSA) have been observed, indicating that resistance is not spreading rapidly (7).
In 1997 the first clinical isolates with reduced vancomycin susceptibility appeared (8, 9). The emergence of so‐called vancomycin intermediate susceptibility S. aureus (VISA) strains in community and hospital acquired infections is of great concern in the medical community as peptidoglycan antibiotics are one of the few remaining treatments of MRSA caused infections.
Unlike VRSA strains, VISA strains have no common genetic determinant of their reduced susceptibility.
Strains with reduced vancomycin susceptibility have shown changes in cell wall composition and synthesis (10), reduced autolytic activity (11, 10), regulatory events (12, 13) or a combination of these and some other adaptations. Increased tolerance of clinical isolates towards vancomycin seems to be based on a gradual adaptive process under vancomycin selective pressure (14). Transcriptome studies of 4 vancomycin susceptible S. aureus (VSSA) strains under vancomycin stress and comparative studies of isogenetic VSSA and VISA strains showed similar regulatory adaptations, making the investigation of vancomycin stress in susceptible S. aureus strains highly relevant for the medical community (15).
Observed physiological adaptations may also refer to mechanism that can increase tolerance against the antibiotic.
Because of this relevance, cell wall stress and particularly vancomycin induced stress have been analyzed in several transcriptomic and proteomic studies. Genome‐wide microarray studies have delivered global transcriptional profiles, whereas proteome studies have been limited to certain fractions. A global quantitative proteomic study of vancomycin stressed S. aureus, covering all sub‐cellular fractions of the bacterium, has not been carried out so far. Previous studies in S. aureus have shown that state‐of‐the‐art proteomics combined with extensive pre‐fractionation and in vivo labeling with stable isotopes allows the identification and quantification of nearly the entire predicted proteome (16).
Here, we used these techniques to analyze the influence of vancomycin on the proteome of the vancomycin susceptible S. aureus strain COL. Quantitative data for nearly 1,400 proteins could be recorded, covering the majority of cytosolic as well as membrane bound and associated, cell wall bound and associated and extracellular proteins. The large data set assists the understanding of the cellular response of S. aureus to vancomycin induced stress and its influence on virulence and other cellular processes. Visualization tools and statistical filtering detected major adaptive mechanisms e.g. the specific increase of proteins synthesizing amino acids which are essential for the peptidoglycan synthesis or the decrease of most proteins with a virulence related function. Besides identification of the two‐ component system VraSR as the main regulatory system also an activation of proteins under SigB, SarA and Fur control was detected.
5
Experimental procedures
Cultivation
S. aureus COL (17) was grown under agitation at 37°C in Luria‐Bertani (LB) medium (Gibco BRL,
Wiesbaden, Germany). At an OD540 of 0.5 vancomycin (Roth, Karlsruhe, Germany) was added at a concentration of 4.5 mg/l. This led to a decreased growth rate about 30 min after addition of the antibiotic. Cells were harvested 100 min after stress induction at an OD540 of about 1.4. Unstressed control samples were grown in LB as well and harvested during exponential growth at an OD540 of 1.4 as well (Figure 1). The cultivations were carried out in triplicates to obtain three biological replicates.
(Insert Figure 1)
Quantification strategy
14N/15N quantification was done by addition of an internal standard before mass spectrometry measurement according to MacCoss et al. (18). This standard was a combined pool of vancomycin stressed cells and exponentially growing cells, grown in an 15N enriched Bioexpress medium (Cambridge
Isotope Laboratories, Andover, MA), that was supplemented with 5 g/l glucose. The standard was mixed with equal amounts of vancomycin stressed and unstressed samples grown in LB as early during sample preparation as possible. Any protein loss during the cell lysis, digestion and measurement will be accounted for by equally affecting the respective 15N‐labeled protein.
Fractionation and proteome analysis
Cytosolic fraction
Harvested cultures were centrifuged for 15 min at 7,000 g at 4°C. Pelleted cells were then washed in TBS
(50mM Tris, 150 mM NaCL, pH 8.0) and resuspended in 50 mM triethylammonium bicarbonat buffer
(TEAB). 500 μL of glass beads with 0.1 mm diameter were added and cells were disrupted using the 6
Precellys 24 homogenizer (Peq Lab, Erlangen, Germany) for 2x30 s at 6,800 rpm. Cell debris and glass beads were separated from the proteins by centrifugation for 10 min at 4°C and 21,500 g. A second centrifugation step of 30 min at 4°C and 21,500 g removed insoluble and aggregated proteins to gain the cytosolic fraction of soluble proteins. The protein concentration was determined with the Roti‐
Nanoquant protein assay (Roth, Karlsruhe, Germany) according to the manufacturer's instructions, and the same protein amount of 15N labeled standard was added. The sample was analyzed by a 1D‐gel based liquid chromatography mass spectrometry (GeLC‐MS) workflow described by Dreisbach et al. (19).
The protein extract was separated by 1D SDS PAGE, the gel was cut into 12 pieces which were tryptically digested at 37°C over night. Resulting peptide mixtures were separated with reversed phase column chromatography (Waters BEH 1.7 μm, 100 μm inner diameter x 100 mm, Waters Corporation, Milford,
Mass., USA) operated on a nanoACQUITY‐UPLC (Waters Corporation, Milford, Mass., USA). Peptides were first concentrated and desalted on a trapping column (Waters nanoACQUITY UPLC column,
Symmetry C18, 5 μm, 180 μm inner diameter x 20 mm, Waters Corporation, Milford, Mass., USA) for 3 min at a flow rate of 10ml/min with 99% buffer A (0.1% acetic acid). Subsequently the peptides were eluted and separated with a non‐linear 80‐min gradient from 5–60% acetonitrile (ACN) in 0.1% acetic acid at a constant flow rate of 400 nl/min.
MS and MS/MS data were acquired by online coupling of the LC with an LTQ‐Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany).
Membrane enriched fraction
The membrane enrichment was carried out according to the protocol published by Eymann et al. (20), leaving out the extraction of proteins by n‐dodecyl‐β‐d‐maltoside treatment. The resulting protein solution was separated by a 1D SDS PAGE, the gel was cut in 10 pieces which were tryptically digested.
7
Peptide separation and MS and MS/MS analyses were performed in the same way as for the cytosolic fraction.
Membrane shaving fraction
Preparation of integral membrane proteins was performed as described by Wolff et al. (21). The peptide containing solution was loaded on a nanoACQUITY UPLC System (Waters Corporation), equipped with an analytical column (nanoACQUITY UPLC column, BEH130 C18, 1.7 μm, 100 μm x 100 mm; Waters
Corporation) operated at 60°C at 400nl/min. Peptides were loaded directly on the column, and after washing for 30 min with 99% buffer A (0.1% acetic acid) the peptides were eluted from the column in a
5‐h linear gradient from 5‐90% buffer B (90% ACN, 0.1% acetic acid). MS and MS/MS data were acquired with triplicate measurements of every sample on an LTQ‐Orbitrap (Thermo Fisher Scientific).
Biotinylation fraction
The biotinylation fractionation was accomplished as described previously by Hempel et al. (22). The resulting protein solution was fractionated with 1D SDS PAGE, gels were cut in 5 pieces, tryptically digested and peptides were separated using reversed‐phase chromatography as described for the cytosolic fraction. MS and MS/MS measurement was carried out with the LTQ‐FT‐ICR Hybrid Mass
Spectrometer (Thermo Fisher Scientific).
Cell wall shaving fraction
Cell wall shaving was carried out according to the trypsin shaving protocol described by Hempel et al.
(23). The peptide containing solution was consequently subjected to reversed phase chromatography and LTQ‐Orbitrap measurement as described for the cytosolic fraction.
Extracellular fraction
8
After pelleting the cells by centrifugation, proteins in the medium were precipitated using TCA as described previously by Antelmann et al. (24). Proteins were then separated by 1D SDS PAGE, and gels were cut in 10 pieces which were tryptically digested. Chromatography and MS and MS/MS measurements were performed in the same way as for the cytosolic fraction.
Mass spectrometry measurements
LTQ‐Orbitrap and LTQ‐FT‐ICR Hybrid Mass Spectrometer
For MS and MS/MS analysis a full survey scan in the Orbitrap (m/z 300 ‐ 2000) with a resolution of
30,000 was followed by MS/MS experiments of the five most abundant precursor ions acquired in the linear trap quadrupole (LTQ) via CID. The ion target values were 1x106 for a full MS in the Orbitrap and
1x104 for MS/MS in the LTQ. Precursors were dynamically excluded for 30 s, unassigned charge states as well as singly charged ions were rejected.
Same parameters were used for the LTQ‐FT‐ICR Hybrid Mass Spectrometer measurements, with the exception that the resolution of the survey scan in the FT‐ICR mass spectrometer was 50,000.
Data analysis
The *.dta files extracted from *.raw files using Bioworks Browser 3.3.1 SP1 (Thermo Fisher Scientific) with no charge state deconvolution and deisotoping performed on the data were searched with
SEQUEST version v28 (rev.12) (Thermo Fisher Scientific) against a S. aureus COL target‐decoy protein sequence database. This database was composed of all protein sequences of S. aureus COL extracted from the National Center for Biotechnology Information (NCBI) bacteria genomes
(http://www.ncbi.nlm.nih.gov/sites/entrez?Db=%20genome&Cmd=Retrieve&dopt=Protein%20%C3%BE
Table&list_uids=610). A set of the reversed sequences created by BioworksBrowser 3.2 EF2 as well as common contaminants such as keratin was appended resulting in 5,864 database entries in total. The
9 searches were performed in two iterations: First for the membrane shaving approach, the following search parameters were applied: enzyme type, none; peptide tolerance, 10 ppm.; tolerance for fragment ions, 1 amu; b‐ and y‐ion series; an oxidation of methionine (15.99 Da) and a carboxyamidomethylation
(57.02 Da) of cysteine were considered as variable modifications with a maximum of three modifications per peptide. For all other MS analyses the following search parameters were used: enzyme type, trypsin allowing two missed cleavage sites; peptide tolerance, 10 ppm.; tolerance for fragment ions, 1 amu.; b‐ and y‐ion series; variable modification, oxidation of methionine (15.99 Da) and carboxyamidomethylation of cysteine (57.02 Da); a maximum of three modifications per peptide was allowed. In the second iteration, the mass shift of all amino acids completely labeled with 15N‐nitrogen was taken into account in the search parameters. Resulting *.dta and *.out files were assembled and filtered using DTASelect (version 2.0.25) (parameters GeLC‐MS: ‐y 2 ‐c 2 ‐C 4 ‐‐here ‐‐decoy Reverse_ ‐p 2
‐t 2 ‐u ‐‐MC 2 ‐i 0.3 ‐‐fp 0.005; parameters membrane shaving: ‐‐nostats ‐1 1.9 ‐2 2.2 ‐3 3.3 ‐4 3.75 ‐i
0.299 ‐a false ‐p 1 ‐y 0 ‐d 0.1 ‐t 2 ‐u –here). The protein false‐positive rate was calculated for each analysis according to Peng et al. (25) and never exceeded 0.5% on protein and peptide level.
Relative quantification was carried out according to MacCoss et al. (18). The crude search results served as the base for the further analysis using the software Census (26) to obtain quantitative data of 14N peaks (sample) and 15N peaks (standard). The software extracts the respective mass traces of the two cognate isotopologues within a defined scan range centered on the fragment scan, leading to successful identification of the respective peptide. The ratio of the peak intensities is subsequently calculated for all overview scans contained in the chosen peak boundaries. Peptide ratios and combined protein ratios are finally exported (R2 values > 0.7 and only unique peptides; proteins failing to be relatively quantified were checked manually in the graphical user interface for on/off proteins). Proteins relatively quantified with at least two peptides were taken into account for the subsequent analysis. All quantification results of the same sample were median centered, ratios were log 2 transformed and averaged over the
10 biological replicates. The resulting protein ratios of the control samples were subtracted from the protein ratios of vancomycin stress samples to eliminate the internal standard.
11
Results and discussion
Identification/Quantification numbers and sub‐cellular localization
To cover as many proteins as possible of the whole proteome of S. aureus, several pre‐fractionation steps were combined with LC‐MS/MS identification. By fractionation and enrichment of different sub‐ cellular compartments the whole proteome consisting of extracellular (TCA precipitation), cell wall bound and associated (cell wall shaving and biotinylation), membrane bound and associated (membrane enrichment and membrane shaving) and cytoplasmic proteins (cytosolic fraction) was covered (Figure 2).
This led to the identification of 1,666 proteins, covering about 64% of the theoretical proteome of S. aureus COL (27). Quantification was done using a 14N/15N metabolic labeling approach. This mass spectrometry based method delivered quantities for 1,345 proteins in vancomycin stressed S. aureus cells. As vancomycin acts at the cell surface and the S. aureus transcriptional response have been shown to mainly affect membrane and cell wall associated proteins (28, 29), quantification of these proteins was of highest interest. Even for these fractions, which are hard to access by commonly used proteomic approaches, the proportion of identified and quantified proteins was remarkably high. Due to the special enrichment strategies quantitative data could be produced for nearly 50% of the proteins belonging to the membrane and about 70% of the proteins of the cell surface and extracellular sub‐proteome (Figure
3), according to their predicted sub‐cellular localization (30). This enables the detailed analysis not only of some of the major cellular responses of S. aureus to vancomycin stress but also of the influence of the antibiotic on virulence.
(Insert Figure 2)
(Insert Figure 3)
12
Voronoi treemap visualization
The high number of quantitated proteins not only offers the possibility to gain a deep insight into the physiological adaptation of S. aureus to vancomycin induced stress but is also a challenge in terms of visualization of relevant and consistent changes. To give a global, but also comprehensive overview of the cellular response of S. aureus COL to vancomycin induced stress we compiled Voronoi treemaps from the quantitative protein data. The treemap concept was originally established by Shneidermann (31).
Balzer and Deussen (32) refined it and implemented Voronoi treemaps which are based on irregular, convex elements for treemap construction. This concept has been adapted to visualize proteomic and transcriptomic data before and has proven to be a powerful visualization tool (33, 34). Figure 4 shows a
Voronoi treemap in which proteins are clustered according to their function based on TIGR classification, so that functionally related elements are localized in close neighborhood to each other. This visualization concept highlights major cellular responses due to vancomycin addition at a glance.
(insert Figure 4)
In‐depth analysis of proteome changes after vancomycin addition
The most striking changes in the treemap are the slightly but consistently decreased amounts of ribosomal proteins, the strongly decreased amount of proteins with a pathogenesis related function and the increased amount of capsular polysaccharide synthesis proteins. The physiological meaning of those observations will be discussed as followed.
Cell wall synthesis
Vancomycin interferes directly with the cell wall synthesis by binding to the D‐Ala‐D‐Ala residue of the
UDP‐MurNAc‐pentapeptide (5, 35), an essential precursor in the peptidoglycan synthesis. The existing
13 transcriptomic studies analyzing vancomycin and related peptidoglycan induced stress in S. aureus all showed an increased expression of many genes whose products are involved in peptidoglycan synthesis
(28, 15, 12). In Figure 4 the peptidoglycan biosynthesis cluster shows mainly increased protein amounts.
But this increase is not consistent across the whole cluster and is mostly not as distinct as recent transcriptome studies may have suggested.
Besides the global visualization of protein amount changes in Voronoi treemaps we also looked for statistically significant changes. We used the TMEV (36) software to carry out a Student's t‐test. The application of a p‐value threshold of 0.1 resulted in significantly increased or decreased amounts for more than 200 proteins. Five of these proteins are involved in peptidoglycan synthesis and showed an increased amount. In Table 1 log 2 ratios of vancomycin stress samples against control samples are listed for significantly changed proteins for selected functional categories (entire proteomic data is part of the supplemental material).
The significant increase of five proteins with peptidoglycan related function correlates well with the slight increase of main parts of the functional cluster in Figure 4.
Amino acid metabolism and transport
The different sub‐clusters in the parental amino acid biosynthesis cluster show differential abundance changes (Figure 4). Proteins involved in the synthesis of amino acids belonging to the pyruvate family decreased, whereas glutamate family synthesis proteins and some proteins belonging to the aspartate family mainly increased in amount after vancomycin addition.
Two increased proteins in the glutamate family cluster were the two components of glutamate synthase
GltB and GltD catalyzing the reaction from 2‐oxoglutarate to L‐glutamate under NADPH consumption.
14
The NADH dependent glutamate dehydrogenase GluD, catalyzing the same conversion and also the reverse reaction, showed a significantly decreased amount after vancomycin addition. Glutamine is essential for the peptidoglycan synthesis as it is part of the pentapeptide in the peptidoglycan precursors. This might explain the increase of GltB and GltD. A surprising aspect was the slight but significant decrease in the amount of the glutamine synthase FemC, because an increase of this protein would be assumed if glutamine synthesis needs to be increased. Existing literature reported only the induction of GltD in two transcriptomic studies (28, 29), both analyzing vancomycin stressed S. aureus cells.
All increased proteins in the aspartate family cluster take part in the synthesis of lysine. The proteins
LysC, Asd, DapA, DapB, DapD and LysA (SACOL1801) catalyze the stepwise conversion from aspartate to lysine. Besides the very consistent increase of all proteins involved in this particular conversion, augmentations of the proteins DapA, DapB and LysA were as well significant, according to our performed t‐test (Table 1). Gene induction of dapA and dapD under cell wall stress conditions have been documented before by two transcriptome studies (28, 29). As lysine is alongside with glutamine and alanine the third essential amino acid of the pentapeptide of the peptidoglycan precursor, an increase of its synthesis proteins can be explained under the assumption of an overall increase of peptidoglycan synthesis.
Decreased amounts or inconsistent changes of proteins involved in the synthesis of other amino acids like the pyruvate and aromatic amino acid family might be caused by a general decrease of protein synthesis due to the lower growth rate of the vancomycin stressed samples (Figure 1). That proves the specificity of the increase of proteins involved in the synthesis of amino acids that are essential for the peptidoglycan synthesis.
15
In addition to amino acid synthesis related proteins also proteins contributing to amino acid transport increased significantly after vancomycin addition. Three members of an amino acid ABC transporter system doubled in amount, two amino acid permeases increased more slightly but still significantly (Table
1). We believe that this transporter activation is another strategy for the cell to increase the pool of amino acids that are required for the peptidoglycan synthesis.
Virulence
The clusters in Figure 4 that represent proteins implicated in the pathogenesis and cell adhesion in S. aureus showed very homogeneous and strong decreases in amount after vancomycin addition. Only four proteins have a red or orange color, indicating an increased amount, few remained unchanged, and the great majority of proteins in these functional clusters decreased strongly in amount. Three of the increased proteins are involved in the synthesis of staphyloxanthin, the fourth is a putative hemolysin.
The t‐test revealed significant changes for 29 proteins with a virulence related function (Table 1), 27 of them decreased in amount. The two increasing proteins were the phosphodiesterase SACOL1305 and
SarA, which positively and negatively regulates the expression of several virulence related genes.
A down‐regulation of some virulence factors was observed by previous transcriptome studies that analyzed cell wall stress in S. aureus (28, 12, 29), but the great majority of virulence genes in these and topically related studies have been shown to be up‐regulated.
Only a study comparing the gene expression of isogenic strains recovered from a patient undergoing extensive vancomycin therapy revealed that strains isolated later, which developed an increased tolerance towards the antibiotic, showed a down‐regulation for most virulence related genes (13).
Capsular Polysaccharide
16
The Voronoi treemap in Figure 4 displays a uniform increase of proteins synthesizing the capsular polysaccharide in S. aureus. 15 out of the 19 proteins which effectuate the synthesis of this extracellular structure could be quantified in this study, all but one showed an increased amount. Two of these augmentations were significant according to the performed t‐test (Table 1).
Induction of these proteins has, as far as we know, not been reported by any proteomic or transcriptomic study analyzing cell wall stress in S. aureus. Conversely, an overexpression of these genes is regarded as one of the few uniform adaptations in VISA strains (37, 38, 13), leading to an increased tolerance towards vancomycin (7). However, the effect of an increased capsule production and its relevance on the reduced susceptibility is not understood.
Ribosomal proteins
The group of ribosomal proteins in Figure 4 decreased in amount after vancomycin addition. This decrease was only slight but very homogenous over nearly all 45 quantified ribosomal proteins. Also the large number of 9 significant protein amount reductions in this group is striking (Table 1). Besides the ribosomal proteins also other proteins with a translation related function (such as tRNA synthetases and translation factors), showed similar changes (Figure 4).
These changes were most likely not a directed response to the antibiotic addition but were caused by the accompanied decreased growth rate of the vancomycin stressed samples (Figure 1).
Autolysis
Significant changes could be documented for four proteins with an autolysis related function (Table 1).
SACOL2302 is a negative regulator of autolysis genes in S. aureus (39) and doubled its amount after vancomycin addition. The other three proteins (IsaA, SACOL0507 and SACOL2666) all have a peptidoglycan catabolic function and decreased strongly in amount. Induction of the negative autolysis
17 regulator as well as repression of SACOL0507, a peptidoglycan hydrolase, have been recognized in transcriptomic studies of S. aureus coping with comparable conditions (12, 28). The decrease of IsaA, another protein with a supposed autolytic function (40), matches well with the results of Kuroda et al.
(28). The fourth autolysis related protein with significantly decreased amount is an N‐acetylmuramoyl‐L‐ alanine amidase. As far as we know, no significant change of the protein or according gene has been reported in any S. aureus cell wall stress study yet.
The reduced growth rate and the resulting low cell turnover rate may contribute to a reduced autolysis.
Yet the level of decrease of the mentioned proteins and the fact that reduced autolytic activity is a common feature of VISA strains (41–43) may implicate a directed adaptation to the changed environmental conditions.
Iron uptake
Two members of different iron compound ABC transporters as well as IsdC, one of the iron‐regulated surface determinants which facilitates extracellular heme uptake and extraction of free iron from the heme molecule (44), were reduced significantly in amount (Table 1) after vancomycin addition.
SACOL1541, a FUR family protein, repressing iron uptake genes, and SACOL1952, a protein with an iron storage function, increased in amount.
Sobral and coworkers came to similar findings when analyzing the transcriptome of cell wall stress in S. aureus COL (29). Their study revealed down‐regulation for 46 genes with iron transport function.
The decreased need for iron and the increased iron storage may be related to the reduced cell turnover rate in both experiments, but it has also been shown that there is an increased expression of sir genes in
VISA strains as well (13).
18
(Insert Table 1)
Protein quality control/Proteolysis
The functional cluster of proteins with protein folding and stabilization function is consistently increased in amount as well as many proteins with a degrading function towards other proteins, peptides and glycopeptides (Figure 4).
The performed t‐test also revealed significant changes for a number of proteins with proteolytic and especially protein quality control function in the cytosol, membrane and cell surface fractions, most of them increasing in amount after vancomycin exposure.
The HtrA homologue SACOL1777, which degrades misfolded and mislocalized polypeptides (45), and
PrsA, involved in posttranslocational folding of several secreted proteins (46), both act on the cell surface of S. aureus and increased nearly twofold and fourfold in amount after stress induction.
The membrane bound serine protease SACOL1455 and FtSH, a protein with a proteolytic as well as protein quality control related function (47), both increased in amount.
Also three Clp ATPases showed significant changes after vancomycin exposure. ClpX, which is part of the
Clp proteasome, slightly decreased in amount. It has a chaperone function and specifically recognizes, unfolds and translocates substrates into the ClpP proteolytic chamber for degradation (48, 49). ClpP binding is proposed to be mediated by the ClpP recognition tripeptide (50). The other two Clp ATPases
ClpB and ClpL, which both increased in amount, lack this ClpP binding motif and thus are proposed not to be part of the Clp proteasome. Both proteins have been shown to be essential for induction of thermo tolerance as they possess the ability to mediate solubilization of protein aggregates (51).
19
Overexpression of the protein quality control genes SACOL1777, SACOL1897, SACOL1455 have been recognized by several cell wall stress studies in S. aureus (28, 12, 29) as well as in the analysis of VISA strains (13), whereas significant changes for Clp ATPases have, to our knowledge, not been reported in literature so far.
Regulatory processes
Besides the clustering of proteins by functional categories in a Voronoi treemap, there is as well the possibility to cluster proteins by their affiliation to global regulons. Constructing such a treemap with quantitative proteome or transcriptome data allows the comprehensive visualization of main regulators for certain physiological conditions.
Figure 5 shows a Voronoi treemap based on transcriptional regulatory units. Protein amount alterations along certain regulons possess remarkably high homogeneity.
Most striking is the increased amount of proteins regulated by the two component system VraS/VraR after vancomycin addition. The strength and consistency of augmentation of these proteins clearly points at the activation of this regulatory system as the main and most specific response to the changed environmental conditions in our experiment. Kuroda et al. reported 2003, that this two‐component system positively controls the expression of 46 genes in S. aureus N315 under vancomycin presence (28), and Utaida et al. could show that it is the main regulator of a hypothesized cell wall stress stimulon (12).
15 out of the 46 positively VraSR‐regulated proteins according to Kuroda et al. changed significantly their amount in our experiment (Table 2), all but one, the lipoteichoic acid synthase LtaS, increased in amount.
The 14 increased proteins include some of the strongest observed changes in our experiment, with some proteins showing up to four fold amplification after vancomycin addition. The two component system proteins VraS and VraR auto‐induce their expression and have been found in more than twofold amount in stressed samples. SACOL1944 is localized in the same operon as VraS and VraR and augmented 20 strongly as well. The protein function of SACOL1944 is not exactly known but it could be shown that it is essential for transcriptional activation through VraR and was suggested to play a role in sensing the yet unknown trigger of the regulatory system (52). The importance of the regulatory system in the adaptation to vancomycin induced stress also becomes apparent by the functions of positively controlled and significantly changed proteins. With proteins taking part in peptidoglycan biosynthesis (Pbp2,
MurAB), protein quality control and proteolysis (SACOL1455, SACOL1977, PrsA) and an autolysis gene repressor (SACOL2302), members of some of the most prominent observed adaptations that have been discussed already are controlled by Vras/VraR.
(Insert Table 2)
Figure 5 indicates that another important regulator in S. aureus in our experiment was the alternative sigma factor SigB. The cluster of proteins under positive SigB control mainly increased, whereas negatively regulated proteins primarily decreased in amount after vancomycin addition.
An assumed regulatory function of SigB in our vancomycin stressed samples can also be confirmed by the statistically significantly altered proteins after stress induction. All 16 of the negatively SigB‐controlled proteins decreased, whereas 22 out of 29 positively controlled proteins increased in amount (Table 3).
The other seven proteins that are positively controlled by SigB decreased in amount after antibiotic addition. Interestingly, one of these ”divergent” proteins was SigB itself. This observation seems to contradict a SigB regulation in our analysis. But Pané‐Farré and coworkers obtained that environmentally activated SigB transcription is not accompanied by strong accumulation of the sigma factor itself and proposed that it is independent from such an accumulation but could rely on mechanisms that influence transcription initiation or competitiveness for core RNA polymerase (53). The discrepancy observed by
Pané‐Farré and coworkers between increased transcription of the SigB operon and the accumulation of
21
SigB and the large number of possible indirect effects in SigB mediated regulation in S. aureus (54) make the unambiguous identification of its activation difficult. However, the observed quantity changes of SigB controlled proteins strongly suggest an important regulatory function in our study.
The activation of the SigB regulon in S. aureus by vancomycin or related cell wall stress conditions has not been reported so far, but an induced SigB expression has been recognized in some VISA strains after vancomycin addition (55). Increased SigB activity has also been observed in teicoplanin resistant strains
(56); teicoplanin is another closely related glycopeptide antibiotic. Activation of the SigB regulon in
Bacillus subtilis after cell wall stress induction by glycopeptides like vancomycin has been observed in several studies(57, 58).
(Insert Table 3)
Another regulon with mainly decreased protein amounts is the one controlled by the ferric uptake regulator Fur. The observations made for iron transport and storage proteins and the slight but significant increase of the Fur family repressor SACOL1541 correlate well with these findings.
The regulator protein SarA, which controls the expression of many genes with a virulent related function, was significantly increased (Table 1). Figure 5 shows an increase of positively and decrease of negatively
SarA controlled proteins. Even though the changes are not as consistent across the whole cluster as for the already mentioned regulons, they strongly indicate an activation of the regulator protein. The cluster of proteins regulated by the two‐component system SaeRS is largely decreased in amount. The two genes of this two‐component system have been shown to be up‐regulated in cell wall stressed S. aureus
N315 by Kuroda et al. (28), but in our experiment both members of the regulatory system remained unchanged after stress induction.
22
(Insert Figure 5)
Conclusion
The present study allowed for the first time a deep insight on the proteomic level into the cellular response of S. aureus coping with vancomycin induced stress. Unlike previous genome‐wide transcriptome studies it monitored changes not on mRNA but on protein level that directly reflect physiologically relevant adaptations. Due to extensive fractionation in sample preparation the analysis is not limited to certain sub‐cellular localization. The visualization of global data in functionally clustered
Voronoi treemaps in combination with a statistical filtering of the quantitative data pointed out some major adaptations of the bacterium to the changed environmental conditions. We could observe plausible adaptations like the increase of proteins involved in cell wall synthesis, in the synthesis of amino acids that are essential for peptidoglycan synthesis or in protein quality control of exported proteins. Observations like the decrease of autolysis related proteins or the decrease of iron uptake systems cannot be explained as a direct response to vancomycin presence.
Most of our results, like the influence of vancomycin on amino acid synthesis, iron uptake or autolysis, fit well with the observations of transcriptomic studies on cell wall stress in S. aureus. Other findings were unexpected and have previously not been recognized. For example, the uniformly decreased amount of virulence related proteins after vancomycin addition contradicts most transcriptional studies. In literature it has been presumed that treatment of a VISA caused infection would lead to an increased,
SigB mediated, pathogenicity (55). Our study showed that in the VSSA strain COL there is most likely an activation of the SigB regulon, but the abundance of most virulence related proteins is strongly decreased after vancomycin addition.
23
Besides activation of the regulon controlled by the alternative sigma factor SigB the two component system VraSR was the major regulatory system implicated in the response to vancomycin in our study.
VraSR controlled proteins showed the strongest and most significant increase in amount after stress induction. Additionally we could identify SarA and Fur as important regulators in S. aureus coping with vancomycin stress.
Acknowledgements
We thank Sebastian Grund and Daniela Wieseler for excellent technical assistance and Holger Kock for critically reading the manuscript.
This work was supported by grants of the DFG (SFB/Transregio 34) and EU (BaSysBio LSHG‐CT‐2006‐
037469).
24
References
1. Henning,C., Hillborgh,U., Lindvall,K., Marqvardsen,O., Sellers,J., Wåhlin,S. and Ransjö,U. (1979)
Comparison of Staphylococcus aureus carriage and skin infection rates in hospital and office
employees. J Hyg (Lond), 83, 437–444.
2. Federspiel,J.J., Stearns,S.C., Peppercorn,A.F., Chu,V.H. and Fowler,V.G.,Jr (2012) Increasing US rates of
endocarditis with Staphylococcus aureus: 1999‐2008. Arch. Intern. Med., 172, 363–365.
3. Villevieille,T., Vincenti‐Rouquette,I., Koeck,J.L., Petitjeans,F., Rousseau,J.M. and Brinquin,L. (2001)
[Fatal septicemia from Staphylococcus aureus resistance to glycopeptides]. Ann Fr Anesth
Reanim, 20, 498–499.
4. Chatterjee,A.N. and Perkins,H.R. (1966) Compounds formed between nucleotides related to the
biosynthesis of bacterial cell wall and vancomycin. Biochem. Biophys. Res. Commun., 24, 489–
494.
5. Perkins,H.R. (1969) Specificity of combination between mucopeptide precursors and vancomycin or
ristocetin. Biochem. J., 111, 195–205.
6. Leclercq,R., Derlot,E., Duval,J. and Courvalin,P. (1988) Plasmid‐mediated resistance to vancomycin and
teicoplanin in Enterococcus faecium. N. Engl. J. Med., 319, 157–161.
7. Howden,B.P., Davies,J.K., Johnson,P.D.R., Stinear,T.P. and Grayson,M.L. (2010) Reduced Vancomycin
Susceptibility in Staphylococcus aureus, Including Vancomycin‐Intermediate and Heterogeneous
Vancomycin‐Intermediate Strains: Resistance Mechanisms, Laboratory Detection, and Clinical
Implications. Clin. Microbiol. Rev., 23, 99–139.
25
8. Hiramatsu,K., Aritaka,N., Hanaki,H., Kawasaki,S., Hosoda,Y., Hori,S., Fukuchi,Y. and Kobayashi,I. (1997)
Dissemination in Japanese hospitals of strains of Staphylococcus aureus heterogeneously
resistant to vancomycin. Lancet, 350, 1670–1673.
9. Hiramatsu,K., Hanaki,H., Ino,T., Yabuta,K., Oguri,T. and Tenover,F.C. (1997) Methicillin‐resistant
Staphylococcus aureus clinical strain with reduced vancomycin susceptibility. J. Antimicrob.
Chemother., 40, 135–136.
10. Boyle‐Vavra,S., Labischinski,H., Ebert,C.C., Ehlert,K. and Daum,R.S. (2001) A spectrum of changes
occurs in peptidoglycan composition of glycopeptide‐intermediate clinical Staphylococcus aureus
isolates. Antimicrob. Agents Chemother., 45, 280–287.
11. Sieradzki,K. and Tomasz,A. (1997) Inhibition of cell wall turnover and autolysis by vancomycin in a
highly vancomycin‐resistant mutant of Staphylococcus aureus. J. Bacteriol., 179, 2557–2566.
12. Utaida,S., Dunman,P.M., Macapagal,D., Murphy,E., Projan,S.J., Singh,V.K., Jayaswal,R.K. and
Wilkinson,B.J. (2003) Genome‐wide transcriptional profiling of the response of Staphylococcus
aureus to cell‐wall‐active antibiotics reveals a cell‐wall‐stress stimulon. Microbiology, 149, 2719–
2732.
13. McAleese,F., Wu,S.W., Sieradzki,K., Dunman,P., Murphy,E., Projan,S. and Tomasz,A. (2006)
Overexpression of genes of the cell wall stimulon in clinical isolates of Staphylococcus aureus
exhibiting vancomycin‐intermediate‐ S. aureus‐type resistance to vancomycin. J. Bacteriol., 188,
1120–1133.
14. Mwangi,M.M., Wu,S.W., Zhou,Y., Sieradzki,K., de Lencastre,H., Richardson,P., Bruce,D., Rubin,E.,
Myers,E., Siggia,E.D., et al. (2007) Tracking the in vivo evolution of multidrug resistance in
26
Staphylococcus aureus by whole‐genome sequencing. Proc. Natl. Acad. Sci. U.S.A., 104, 9451–
9456.
15. McCallum,N., Spehar,G., Bischoff,M. and Berger‐Bächi,B. (2006) Strain dependence of the cell wall‐
damage induced stimulon in Staphylococcus aureus. Biochim. Biophys. Acta, 1760, 1475–1481.
16. Becher,D., Hempel,K., Sievers,S., Zühlke,D., Pané‐Farré,J., Otto,A., Fuchs,S., Albrecht,D., Bernhardt,J.,
Engelmann,S., et al. (2009) A Proteomic View of an Important Human Pathogen – Towards the
Quantification of the Entire Staphylococcus aureus Proteome. PLoS ONE, 4, e8176.
17. Shafer,W.M. and Iandolo,J.J. (1979) Genetics of staphylococcal enterotoxin B in methicillin‐resistant
isolates of Staphylococcus aureus. Infect. Immun., 25, 902–911.
18. MacCoss,M.J., Wu,C.C., Liu,H., Sadygov,R. and Yates,J.R. (2003) A correlation algorithm for the
automated quantitative analysis of shotgun proteomics data. Anal. Chem, 75, 6912–6921.
19. Dreisbach,A., Otto,A., Becher,D., Hammer,E., Teumer,A., Gouw,J.W., Hecker,M. and Völker,U. (2008)
Monitoring of changes in the membrane proteome during stationary phase adaptation of
Bacillus subtilis using in vivo labeling techniques. Proteomics, 8, 2062–2076.
20. Eymann,C., Dreisbach,A., Albrecht,D., Bernhardt,J., Becher,D., Gentner,S., Tam,L.T., Büttner,K.,
Buurman,G., Scharf,C., et al. (2004) A comprehensive proteome map of growing Bacillus subtilis
cells. Proteomics, 4, 2849–76.
21. Wolff,S., Hahne,H., Hecker,M. and Becher,D. (2008) Complementary analysis of the vegetative
membrane proteome of the human pathogen Staphylococcus aureus. Mol. Cell Proteomics, 7,
1460–1468.
27
22. Hempel,K., Pané‐Farré,J., Otto,A., Sievers,S., Hecker,M. and Becher,D. (2010) Quantitative cell
surface proteome profiling for SigB‐dependent protein expression in the human pathogen
Staphylococcus aureus via biotinylation approach. J. Proteome Res., 9, 1579–1590.
23. Hempel,K., Herbst,F.‐A., Moche,M., Hecker,M. and Becher,D. (2011) Quantitative Proteomic View on
Secreted, Cell Surface‐Associated, and Cytoplasmic Proteins of the Methicillin‐Resistant Human
Pathogen Staphylococcus aureus under Iron‐Limited Conditions. J. Proteome Res., 10, 1657–
1666.
24. Antelmann,H., Tjalsma,H., Voigt,B., Ohlmeier,S., Bron,S., van Dijl,J.M. and Hecker,M. (2001) A
proteomic view on genome‐based signal peptide predictions. Genome Res., 11, 1484–1502.
25. Peng,J., Elias,J.E., Thoreen,C.C., Licklider,L.J. and Gygi,S.P. (2003) Evaluation of multidimensional
chromatography coupled with tandem mass spectrometry (LC/LC‐MS/MS) for large‐scale protein
analysis: the yeast proteome. J. Proteome Res., 2, 43–50.
26. Park,S.K., Venable,J.D., Xu,T. and Yates,J.R. (2008) A quantitative analysis software tool for mass
spectrometry‐based proteomics. Nat. Methods, 5, 319–322.
27. Gill,S.R., Fouts,D.E., Archer,G.L., Mongodin,E.F., Deboy,R.T., Ravel,J., Paulsen,I.T., Kolonay,J.F.,
Brinkac,L., Beanan,M., et al. (2005) Insights on evolution of virulence and resistance from the
complete genome analysis of an early methicillin‐resistant Staphylococcus aureus strain and a
biofilm‐producing methicillin‐resistant Staphylococcus epidermidis strain. J. Bacteriol., 187,
2426–2438.
28. Kuroda,M., Kuroda,H., Oshima,T., Takeuchi,F., Mori,H. and Hiramatsu,K. (2003) Two‐component
system VraSR positively modulates the regulation of cell‐wall biosynthesis pathway in
Staphylococcus aureus. Mol. Microbiol, 49, 807–821.
28
29. Sobral,R.G., Jones,A.E., Des Etages,S.G., Dougherty,T.J., Peitzsch,R.M., Gaasterland,T., Ludovice,A.M.,
de Lencastre,H. and Tomasz,A. (2007) Extensive and genome‐wide changes in the transcription
profile of Staphylococcus aureus induced by modulating the transcription of the cell wall
synthesis gene murF. J. Bacteriol., 189, 2376–2391.
30. Zhou,M., Boekhorst,J., Francke,C. and Siezen,R.J. (2008) LocateP: genome‐scale subcellular‐location
predictor for bacterial proteins. BMC Bioinformatics, 9, 173.
31. Shneiderman,B. (1991) Tree visualization with Tree‐maps: A 2‐d space‐filling approach. ACM
Transactions on Graphics, 11, 92–99.
32. Balzer,M., Deussen,O. and Lewerentz,C. (2005) Voronoi treemaps for the visualization of software
metrics. In Proceedings of the 2005 ACM symposium on Software visualization.pp. 165–172.
33. Bernhardt,J., Funke,S., Hecker,M. and Siebourg,J. (2009) Visualizing gene expression data via voronoi
treemaps. In Proceedings of the 2009 Sixth International Symposium on Voronoi Diagrams.pp.
233–241.
34. Otto,A., Bernhardt,J., Meyer,H., Schaffer,M., Herbst,F.‐A., Siebourg,J., Mäder,U., Lalk,M., Hecker,M.
and Becher,D. (2010) Systems‐wide temporal proteomic profiling in glucose‐starved Bacillus
subtilis. Nat Commun, 1, 137.
35. Bordet,C. and Perkins,H.R. (1970) Iodinated vancomycin and mucopeptide biosynthesis by cell‐free
preparations from Micrococcus lysodeikticus. Biochem. J., 119, 877–883.
36. Saeed,A.I., Sharov,V., White,J., Li,J., Liang,W., Bhagabati,N., Braisted,J., Klapa,M., Currier,T.,
Thiagarajan,M., et al. (2003) TM4: a free, open‐source system for microarray data management
and analysis. BioTechniques, 34, 374–378.
29
37. Cui,L., Lian,J.‐Q., Neoh,H.‐M., Reyes,E. and Hiramatsu,K. (2005) DNA microarray‐based identification
of genes associated with glycopeptide resistance in Staphylococcus aureus. Antimicrob. Agents
Chemother., 49, 3404–3413.
38. Howden,B.P., Smith,D.J., Mansell,A., Johnson,P.D.R., Ward,P.B., Stinear,T.P. and Davies,J.K. (2008)
Different bacterial gene expression patterns and attenuated host immune responses are
associated with the evolution of low‐level vancomycin resistance during persistent methicillin‐
resistant Staphylococcus aureus bacteraemia. BMC Microbiol., 8, 39.
39. Brunskill,E.W. and Bayles,K.W. (1996) Identification and molecular characterization of a putative
regulatory locus that affects autolysis in Staphylococcus aureus. J. Bacteriol., 178, 611–618.
40. Sakata,N., Terakubo,S. and Mukai,T. (2005) Subcellular location of the soluble lytic transglycosylase
homologue in Staphylococcus aureus. Curr. Microbiol., 50, 47–51.
41. Boyle‐Vavra,S., Carey,R.B. and Daum,R.S. (2001) Development of vancomycin and lysostaphin
resistance in a methicillin‐resistant Staphylococcus aureus isolate. J. Antimicrob. Chemother., 48,
617–625.
42. Boyle‐Vavra,S., Challapalli,M. and Daum,R.S. (2003) Resistance to autolysis in vancomycin‐selected
Staphylococcus aureus isolates precedes vancomycin‐intermediate resistance. Antimicrob.
Agents Chemother., 47, 2036–2039.
43. Koehl,J.L., Muthaiyan,A., Jayaswal,R.K., Ehlert,K., Labischinski,H. and Wilkinson,B.J. (2004) Cell wall
composition and decreased autolytic activity and lysostaphin susceptibility of glycopeptide‐
intermediate Staphylococcus aureus. Antimicrob. Agents Chemother., 48, 3749–3757.
30
44. Skaar,E.P. and Schneewind,O. (2004) Iron‐regulated surface determinants (Isd) of Staphylococcus
aureus: stealing iron from heme. Microbes Infect., 6, 390–397.
45. Clausen,T., Kaiser,M., Huber,R. and Ehrmann,M. (2011) HTRA proteases: regulated proteolysis in
protein quality control. Nature Reviews Molecular Cell Biology, 12, 152–162.
46. Jacobs,M., Andersen,J.B., Kontinen,V. and Sarvas,M. (1993) Bacillus subtilis PrsA is required in vivo as
an extracytoplasmic chaperone for secretion of active enzymes synthesized either with or
without pro‐sequences. Molecular Microbiology, 8, 957–966.
47. Hengge,R. and Bukau,B. (2003) Proteolysis in prokaryotes: protein quality control and regulatory
principles. Molecular Microbiology, 49, 1451–1462.
48. Hoskins,J.R., Pak,M., Maurizi,M.R. and Wickner,S. (1998) The role of the ClpA chaperone in
proteolysis by ClpAP. PNAS, 95, 12135–12140.
49. Hoskins,J.R., Singh,S.K., Maurizi,M.R. and Wickner,S. (2000) Protein binding and unfolding by the
chaperone ClpA and degradation by the protease ClpAP. PNAS, 97, 8892–8897.
50. Kim,Y.I., Levchenko,I., Fraczkowska,K., Woodruff,R.V., Sauer,R.T. and Baker,T.A. (2001) Molecular
determinants of complex formation between Clp/Hsp100 ATPases and the ClpP peptidase. Nat.
Struct. Biol., 8, 230–233.
51. Frees,D., Chastanet,A., Qazi,S., Sørensen,K., Hill,P., Msadek,T. and Ingmer,H. (2004) Clp ATPases are
required for stress tolerance, intracellular replication and biofilm formation in Staphylococcus
aureus. Molecular Microbiology, 54, 1445–1462.
31
52. McCallum,N., Meier,P.S., Heusser,R. and Berger‐Bächi,B. (2011) Mutational analyses of open reading
frames within the vraSR operon and their roles in the cell wall stress response of Staphylococcus
aureus. Antimicrob. Agents Chemother., 55, 1391–1402.
53. Pané‐Farré,J., Jonas,B., Hardwick,S.W., Gronau,K., Lewis,R.J., Hecker,M. and Engelmann,S. (2009)
Role of RsbU in controlling SigB activity in Staphylococcus aureus following alkaline stress. J.
Bacteriol., 191, 2561–2573.
54. Pané‐Farré,J., Jonas,B., Förstner,K., Engelmann,S. and Hecker,M. (2006) The σB regulon in
Staphylococcus aureus and its regulation. International Journal of Medical Microbiology, 296,
237–258.
55. Chen,H.‐Y., Chen,C.‐C., Fang,C.‐S., Hsieh,Y.‐T., Lin,M.‐H. and Shu,J.‐C. (2011) Vancomycin activates
σ(B) in vancomycin‐resistant Staphylococcus aureus resulting in the enhancement of cytotoxicity.
PLoS ONE, 6, e24472.
56. Bischoff,M. and Berger‐Bächi,B. (2001) Teicoplanin stress‐selected mutations increasing sigma(B)
activity in Staphylococcus aureus. Antimicrob. Agents Chemother., 45, 1714–1720.
57. Mascher,T., Margulis,N.G., Wang,T., Ye,R.W. and Helmann,J.D. (2003) Cell wall stress responses in
Bacillus subtilis: the regulatory network of the bacitracin stimulon. Molecular Microbiology, 50,
1591–1604.
58. Cao,M., Wang,T., Ye,R. and Helmann,J.D. (2002) Antibiotics that inhibit cell wall biosynthesis induce
expression of the Bacillus subtilis sigma(W) and sigma(M) regulons. Mol Microbiol, 45, 1267–76.
32
59. Bischoff,M., Dunman,P., Kormanec,J., Macapagal,D., Murphy,E., Mounts,W., Berger‐Bächi,B. and
Projan,S. (2004) Microarray‐based analysis of the Staphylococcus aureus sigmaB regulon. J.
Bacteriol, 186, 4085–4099.
60. Ziebandt,A.K., Weber,H., Rudolph,J., Schmid,R., Höper,D., Engelmann,S. and Hecker,M. (2001)
Extracellular proteins of Staphylococcus aureus and the role of SarA and sigma B. Proteomics, 1,
480–493.
61. Ziebandt,A.‐K., Becher,D., Ohlsen,K., Hacker,J., Hecker,M. and Engelmann,S. (2004) The influence of
agr and sigmaB in growth phase dependent regulation of virulence factors in Staphylococcus
aureus. Proteomics, 4, 3034–3047.
33
Figure Legends
Figure 1: Growth curves of control and vancomycin treated S. aureus cultivations.
Stress samples (grey curve) were grown to an OD540 of 0.5, exposed to 4.5 mg/l vancomycin (indicated by time point zero) and harvested 100 min after addition of the antibiotic at an OD540 of 1.4. Unstressed control samples
(black curve) were harvested at the same OD540 of 1.4.
Figure 2: Fractionation workflow.
Most crucial steps in the preparation of the six sub‐fractions are displayed.
Figure 3: Number of quantified proteins and their predicted sub‐cellular localization.
Column charts of the quantified proteins showing the assignment to their predicted sub‐cellular localization(30) on absolute (a) and relative scale (b). Black blocks represent proteins that have been quantified in our experiment, grey blocks relate to proteins that have only been identified. The white fractions represent predicted proteins in our protein database which could not be identified in our study.
Figure 4: Functional Voronoi treemap of vancomycin stressed S. aureus COL.
Each cell represents one quantified protein which belongs to other functionally related proteins in parental, convex‐shaped categories. These are again summarized in higher‐level categories. Functional clustering is based on
TIGR gene classification. To visualize differences in protein amount after vancomycin addition color coding was applied as following: blue shows decreased amount, grey unchanged amount and orange increased protein amount.
34
Figure 5: Regulon Voronoi treemap of vancomycin stressed S. aureus COL.
Each cell represents one quantified protein which is associated with other functionally related proteins in parent convex‐shaped categories. These are again summarized in higher‐level categories. The treemap design is based on hierarchically structured regulatory data according to TIGR classification (black borders: regulon/thin black borders within the regulons: operon/smallest cells: proteins). + / − depict regulons being induced ( + ) or repressed ( − ) depending on the regulator assigned to the area.
35
Tables
Table 1: Subset of proteins with significantly changed amount after vancomycin addition, sorted by functional categories
Accession Protein functional category Function according to NCBI log2 number name ratio Amino acid biosynthesis and transport SACOL0630 ‐ AS permease amino acid permease 0.32 SACOL1728 ‐ AS permease amino acid permease 0.44 SACOL1431 DapB lysine synthesis dihydrodipicolinate reductase 0.40 SACOL1432 DapD lysine synthesis 2,3,4,5‐tetrahydropyridine‐2,6‐dicarboxylate N‐ 0.26 succinyltransferase SACOL1435 LysA lysine synthesis diaminopimelate decarboxylase 0.21 SACOL2450 ‐ AS ABC transporter amino acid ABC transporter, permease protein 0.84 SACOL2451 ‐ AS ABC transporter amino acid ABC transporter, amino acid‐binding 0.78 protein SACOL2453 ‐ AS ABC transporter amino acid ABC transporter, ATP‐binding protein 0.88 Peptidoglycan synthesis SACOL0743 UppP petidoglycan synthesis undecaprenyl pyrophosphate phosphatase 0.27 SACOL1410 FemA petidoglycan synthesis femA protein 0.23 SACOL1490 Pbp2 petidoglycan synthesis penicillin‐binding protein 2 0.45 SACOL2074 Ddl petidoglycan synthesis D‐alanyl‐alanine synthetase A 0.34 SACOL2116 MurAB petidoglycan synthesis UDP‐N‐acetylglucosamine 1‐ 0.56 carboxyvinyltransferase Virulence SACOL0209 ‐ virulence staphylocoagulase precursor ‐1.87 SACOL0271 EsxA virulence hypothetical protein ‐0.52 SACOL0272 EsaA virulence hypothetical protein ‐0.83 SACOL0442 ‐ virulence staphylococcal enterotoxin, putative ‐0.43 SACOL0468 ‐ virulence superantigen‐like protein ‐0.53 SACOL0478 ‐ virulence superantigen‐like protein ‐0.51 SACOL0608 SdrC virulence sdrC protein ‐0.52 SACOL0610 SdrE virulence sdrE protein ‐0.58 SACOL0856 ClfA virulence clumping factor A ‐0.64 SACOL0858 Empbp virulence secretory extracellular matrix and plasma ‐0.95 binding protein SACOL0887 Sei virulence staphylococcal enterotoxin type I ‐0.56 SACOL0907 Seb virulence staphylococcal enterotoxin B ‐0.49 SACOL1056 SspB1 virulence cysteine protease precursor SspB ‐0.60 SACOL1164 ‐ virulence fibrinogen binding‐related protein ‐1.44 SACOL1168 Efb virulence fibrinogen‐binding protein ‐1.51 SACOL1305 ‐ virulence phosphodiesterase 0.23 SACOL1880 LukD virulence leukotoxin LukD ‐0.41 SACOL1881 LukE virulence leukotoxin LukE ‐0.62 SACOL1970 SspB2 virulence cysteine protease precursor SspB ‐0.86 SACOL2003 Hlb virulence phospholipase C ‐0.51 SACOL2194 HysA virulence hyaluronate lyase ‐0.37
36
Accession Protein functional category Function according to NCBI log2 number name ratio Virulence SACOL2418 ‐ virulence IgG‐binding protein SBI ‐0.36 SACOL2421 HlgC virulence gamma hemolysin, component C ‐0.66 SACOL2509 FnbB virulence fibronectin binding protein B ‐2.25 SACOL2652 ClfB virulence clumping factor B ‐0.73 SACOL2660 IsaB virulence immunodominant antigen B ‐1.06 SACOL2694 Geh virulence lipase ‐1.13 SACOL0672 SarA virulence regulator staphylococcal accessory regulator A 0.42 SACOL2384 SarZ virulence regulator staphylococcal accessory protein Z ‐0.21 Capsular Polysaccharide Biosynthesis SACOL0137 Cap5B capsular polysaccharide capsular polysaccharide biosynthesis protein 0.67 biosynthesis Cap5B SACOL0148 Cap5M capsular polysaccharide capsular polysaccharide biosynthesis 0.47 biosynthesis galactosyltransferase Cap5M Ribosomal Proteins SACOL1274 RpsB ribosomal protein 30S ribosomal protein S2 ‐0.20 SACOL1769 RpsD ribosomal protein 30S ribosomal protein S4 ‐0.22 SACOL2206 RpsI ribosomal protein 30S ribosomal protein S9 ‐0.15 SACOL2220 RplO ribosomal protein 50S ribosomal protein L15 ‐0.19 SACOL2224 RplF ribosomal protein 50S ribosomal protein L6 ‐0.15 SACOL2229 RplN ribosomal protein 50S ribosomal protein L14 ‐0.19 SACOL2234 RplV ribosomal protein 50S ribosomal protein L22 ‐0.21 SACOL2236 RplB ribosomal protein 50S ribosomal protein L2 ‐0.23 SACOL2237 RplW ribosomal protein 50S ribosomal protein L23 ‐0.19 Autolysis SACOL2302 ‐ autolysis regulator transcriptional regulator, putative 1.06 SACOL0507 ‐ autolysis/peptidoglycan LysM domain‐containing protein ‐0.93 catabolism SACOL2584 IsaA autolysis/peptidoglycan immunodominant antigen A ‐0.31 catabolism SACOL2666 ‐ autolysis/peptidoglycan N‐acetylmuramoyl‐L‐alanine amidase ‐1.16 catabolism Iron transport and storage SACOL1952 ‐ iron strorage ferritins family protein 0.61 SACOL0099 SirA iron uptake iron compound ABC transporter, iron compound‐ ‐0.43 binding protein SirA SACOL1141 IsdC iron uptake NPQTN cell wall surface anchor protein ‐0.55 SACOL2277 ‐ iron uptake iron compound ABC transporter, iron compound‐ ‐0.36 binding protein SACOL1541 ‐ Iron uptake regulator FUR family transcriptional regulator 0.23 Protein Quality Control/Proteolysis SACOL1897 PrsA protein quality control protein export protein PrsA, putative 1.94 SACOL0555 FtsH protein quality ATP‐dependent zinc metalloprotease FtsH 0.30 control/proteolysis SACOL1777 ‐ protein quality serine protease HtrA, putative 0.88 control/proteolysis SACOL0979 ClpB protein quality ATP‐dependent Clp protease, ATP‐binding 0.29 control/proteolysis subunit ClpB SACOL2563 ClpL protein quality ATP‐dependent Clp protease, putative 0.50 control/proteolysis 37
Accession Protein functional category Function according to NCBI log2 number name ratio Protein Quality Control/Proteolysis SACOL1721 ClpX protein quality ATP‐dependent protease ATP‐binding subunit ‐0.12 control/proteolysis SACOL1455 ‐ proteolysis carboxyl‐terminal protease 0.75
Table 2: Proteins with significantly altered protein amount whose gene expression is controlled by VraSR (Kuroda et al. 2003 (28)) accession protein functional category Function according to NCBI log2 number name ratio SACOL0778 LtaS Lipoteichoic acid synthesis lipoteichoic acid synthase ‐0.33 SACOL1088 ‐ ‐ hypothetical protein 0.53 SACOL1455 ‐ Proteolysis carboxyl‐terminal protease 0.75 SACOL1457 Crr sugar transport PTS system, IIA component 1.10 SACOL1490 Pbp2 peptidoglycan biosynthesis penicillin‐binding protein 2 0.45 SACOL1777 ‐ protein quality serine protease HtrA, putative 0.88 control/proteolysis SACOL1897 PrsA protein quality control protein export protein PrsA, putative 0.39 SACOL1942 VraR cell wall stress regulator DNA‐binding response regulator VraR 1.68 SACOL1943 VraS cell wall stress regulator sensor histidine kinase VraS 1.18 SACOL1944 ‐ ‐ hypothetical protein 1.80 SACOL2116 MurAB peptidoglycan biosynthesis UDP‐N‐acetylglucosamine 1‐ 0.56 carboxyvinyltransferase SACOL2302 ‐ autolysis regulator LytR homologue which is autolysine regulator 1.06 SACOL2352 TcaA antibiotic resistance tcaA protein 0.80 SACOL2436 ‐ ‐ hypothetical protein 2.08 SACOL2518 ‐ ‐ hypothetical protein 1.18
38
Table 3: Proteins with significantly altered protein amount whose gene expression is controlled by SigB accession protein function according to NCBI regulation SigB log2 number name reported in* regulation ratio SACOL0099 SirA iron compound ABC transporter iron 2 negative ‐0.43 compound‐binding protein SirA SACOL0118 SodA1 superoxide dismutase 1; 2 negative ‐0.23 SACOL0860 Nuc thermonuclease precursor 1; 2; 3 negative ‐0.43 SACOL0907 Seb seb; enterotoxin B 3 negative ‐0.49 SACOL0962 ‐ glycerophosphoryl diester phosphodiesterase 1; 3 negative ‐0.47 SACOL0985 ‐ surface protein 1; 2 negative ‐0.52 SACOL1056 SspB1 cysteine protease precursor SspB 1; 3 negative ‐0.60 SACOL1141 IsdC NPQTN cell wall surface anchor protein 2 negative ‐0.55 SACOL1142 IsdD hypothetical protein SACOL1142 2 negative ‐0.47 SACOL1870 ‐ hypothetical protein SACOL1870 3 negative ‐0.57 SACOL1880 LukD leukotoxin LukD 3 negative ‐0.41 SACOL1970 SspB2 cysteine protease precursor SspB 1; 2 negative ‐0.86 SACOL1985 ‐ hypothetical protein SACOL1985 3 negative ‐0.17 SACOL2003 Hlb phospholipase C 3 negative ‐0.51 SACOL2421 HlgC gamma hemolysin, component C 1 negative ‐0.66 SACOL2694 Geh lipase 1; 3 negative ‐1.13 SACOL0137 Cap5B capsular polysaccharide biosynthesis protein 1; 2 positive 0.67 Cap5B SACOL0148 Cap5M capsular polysaccharide biosynthesis 1; 2 positive 0.47 galactosyltransferase Cap5M SACOL0209 ‐ staphylocoagulase precursor 1 positive ‐1.87 SACOL0317 ‐ lipase precursor, interruption‐N 3 positive ‐1.23 SACOL0610 SdrE sdrE 1 positive ‐0.58 SACOL0630 ‐ amino acid permease 1; 2 positive 0.32 SACOL0638 ‐ phosphomevalonate kinase 1; 2 positive 0.35 SACOL0672 SarA accessory regulator A 1; 2 positive 0.42 SACOL0740 ‐ hypothetical protein SACOL0740 1 positive 0.17 SACOL0856 ClfA clumping factor A 1; 2 positive ‐0.64 SACOL1368 KataA catalase 2 positive 0.22 SACOL1728 ‐ amino acid permease 1 positive 0.44 SACOL1802 ‐ hypothetical protein SACOL1802 1; 2 positive 0.50 SACOL1912 ‐ hypothetical protein SACOL1912 1; 2 positive 0.28 SACOL1933 ‐ ThiJ/PfpI family protein 1; 2 positive 0.35 SACOL1987 ‐ hypothetical protein SACOL1987 1 positive 0.36 SACOL2054 RpoF RNA polymerase sigma factor SigB 1; 2 positive ‐0.15 SACOL2114 ‐ aldehyde dehydrogenase 1; 2 positive 0.28 SACOL2149 MtlD mannitol‐1‐phosphate 5‐dehydrogenase 1 positive 0.27 SACOL2301 ‐ formate dehydrogenase subunit alpha 1; 2 positive 0.18 SACOL2461 ‐ hypothetical protein SACOL2461 1; 2 positive 0.25 SACOL2484 ‐ alkylhydroperoxidase 1; 2 positive 0.53 SACOL2532 ‐ acetyltransferase 1 positive 0.21 SACOL2553 ‐ pyruvate oxidase 1; 2 positive 0.68 SACOL2563 ‐ ATP‐dependent Clp protease 1; 2 positive 0.50 SACOL2584 IsaA immunodominant antigen A 1 positive ‐0.31 SACOL2596 ‐ hypothetical protein SACOL2596 1; 2 positive 0.44 SACOL2597 ‐ alpha/beta fold family hydrolase 1; 2 positive 0.33 39 accession protein function according to NCBI regulation SigB log2 number name reported in* regulation ratio SACOL2666 ‐ N‐acetylmuramoyl‐L‐alanine amidase 3 positive ‐1.16 * Regulation reproted in 1) Bischoff et al. (59) 2) Pané‐Farré et al. (54) 3) Ziebandt et al. 2001 and 2004 (60, 61)
40
Figures
Figure1:
41
Figure 2
42
Figure 3:
43
Figure 4:
44
Figure 5:
45
Article III
Page 91
Page 92
Global impact of protein arginine phosphorylation on the physiology of Bacillus subtilis
Alexander K. W. Elsholza,1, Kürsxad Turgayb,c, Stephan Michalika, Bernd Hesslinga, Katrin Gronaua, Dan Oertela,2, Ulrike Mäderd, Jörg Bernhardta, Dörte Bechera, Michael Heckera, and Ulf Gertha,3 aInstitute of Microbiology, and dInterfaculty Institute for Genetics and Functional Genomics, Ernst-Moritz-Arndt-University Greifswald, D-17487 Greifswald, Germany; bInstitute of Microbiology, Leibniz University Hannover, D-30167 Hannover, Germany; and cInstitute of Biology–Microbiology, Freie Universität Berlin, D-14195 Berlin, Germany
Edited by Tania A. Baker, Massachusetts Institute of Technology, Cambridge, MA, and approved March 19, 2012 (received for review October 23, 2011)
Reversible protein phosphorylation is an important and ubiquitous the case of the Hsp100/Clp protein ClpC, because we observe that protein modification in all living cells. Here we report that protein the ClpC activity is regulated through a McsB-dependent phos- phosphorylation on arginine residues plays a physiologically sig- phorylation on two arginine residues. In addition, we provide evi- nificant role. We detected 121 arginine phosphorylation sites in dence that protein arginine phosphorylation plays a significant role 87 proteins in the Gram-positive model organism Bacillus subtilis for many other regulatory processes within the bacterial cell. This is fi in vivo. Moreover, we provide evidence that protein arginine phos- a rst and important step in characterizing the cellular functions of fi fi phorylation has a functional role and is involved in the regulation of this recently identi ed protein modi cation. many critical cellular processes, such as protein degradation, motil- Results ity, competence, and stringent and stress responses. Our results suggest that in B. subtilis the combined activity of a protein arginine Protein Arginine Phosphorylation Exists in Vivo. We attempted to identify phosphorylation on protein arginine residues in vivo in kinase and phosphatase allows a rapid and reversible regulation of B. subtilis protein activity and that protein arginine phosphorylation can play a wild-type strain with various proteomic approaches, a physiologically important and regulatory role in bacteria. but failed to detect it. However, the YwlE phosphatase had been identified as the cognate McsB phosphatase in vitro and was shown to antagonize the activity of McsB in vivo (8, 9, 11–13). This McsB | YwlE | ClpC | HSP100/Clp | phosphagen kinase observation strongly suggests that YwlE may act as a general protein arginine phosphatase so that protein phosphorylation on roteins perform crucial cellular functions ranging from catal- arginine residues could be more stable and enriched in an ywlE Pysis and signal transduction to providing structural building deletion strain. Therefore, we analyzed a ywlE mutant for global blocks. These complex 3D structures can be modified, which can protein phosphorylation. We were able to detect distinct arginine affect and modulate all their various activities. These post- phosphorylations in a ywlE mutant strain with a global, quantita- translational modifications are ubiquitous and important for tive, label-free, gel-free, and site-specific approach using high- regulation of protein activity in all living cells, allowing the cell to accuracy MS in combination with biochemical enrichment of MICROBIOLOGY adapt rapidly to the ever-changing environment. Among the var- phosphopeptides from digested cell lysates using TiO2 chroma- ious known posttranslational protein modifications, reversible tography. The enriched phosphopeptides were analyzed using protein phosphorylation has emerged as one of the most impor- nanoscale LC coupled to high-resolution hybrid mass spec- tant signal-transduction processes in all three domains of life (1). trometers (LTQ-Orbitrap Velos; see Materials and Methods for Until recently, analysis of the bacterial phosphoproteome details). Protein arginine residues are phosphorylated at one of was restricted to the use of 2D-PAGE and classic biochemical the amine nitrogens of the guanidinium group, thereby forming approaches, which give only limited information on the specific a phosphoramidate N-P linkage (7). All identified proteins that phosphorylation sites. Recent technical developments in the field are phosphorylated on arginine residues unveiled in their frag- of high-accuracy MS allowed the detection of phosphorylation sites ment ion series the characteristic fragment, which reflects the in proteins with high sensitivity, and has been used to characterize addition of a phosphate moiety to an arginine residue (Fig. 1A). the Ser, Thr, and Tyr phosphoproteome for different bacterial In total, we identified 118 unique arginine phosphorylation species (2, 3). However, protein phosphorylations on either His or sites in 87 proteins with very high confidence using stringent Asp, although commonly used in bacterial two-component regu- quality criteria for the validation of the phosphorylation site latory systems, are difficult to detect by MS because of its chemical (Table 1 and Fig. S1; see Table S1 for the full genotypes of instability at lower pH values (pH < 8). Nevertheless, only protein strains and plasmids). We also detected already known protein phosphorylation on serine, threonine, tyrosine, histidine, cysteine, phosphorylations on either serine, threonine, or tyrosine residues and aspartate residues has so far been described for bacteria by different methods and their important roles in signal transduction and physiology firmly established (3–6). Author contributions: A.K.W.E., K.T., S.M., M.H., and U.G. designed research; A.K.W.E., S.M., Recently, a bacterial protein arginine kinase, McsB, was iden- B.H., K.G., D.O., and U.M. performed research; A.K.W.E., K.T., S.M., B.H., K.G., U.M., J.B., tified and its activity characterized in vitro (7), and two McsB D.B., M.H., and U.G. analyzed data; and A.K.W.E., K.T., M.H., and U.G. wrote the paper. substrates, which were phosphorylated on arginine residues, were The authors declare no conflict of interest. identified in vitro (7, 8). A functional McsB kinase is required for This article is a PNAS Direct Submission. the controlled degradation of the global heat-shock regulator Freely available online through the PNAS open access option. – CtsR (9 11), as well as for the de-localization of competence Data deposition: The microarray data reported in this paper have been deposited in the proteins (12). These observations suggest that McsB-mediated Gene Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. protein arginine phosphorylation could occur in bacteria, but GSE31249). neither the occurrence of the modification in vivo itself nor 1Present address: Department of Molecular and Cellular Biology, Harvard University, Cam- a physiological function for it has yet been described. bridge, MA 02138. Here we demonstrate that protein arginine phosphorylation does 2Present address: Forschungszentrum Jülich GmbH, Institute for Bio- and Geo-Sciences take place and is functional inside living cells. We identified 121 IBG-1: Biotechnology, 52425 Jülich, Germany. arginine phosphorylation sites in 87 proteins in the Gram-positive 3To whom correspondence should be addressed. E-mail: [email protected]. Bacillus subtilis model organism in vivo. Moreover, we demon- This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. strated a physiological role for protein arginine phosphorylation in 1073/pnas.1117483109/-/DCSupplemental. www.pnas.org/cgi/doi/10.1073/pnas.1117483109 PNAS | May 8, 2012 | vol. 109 | no. 19 | 7451–7456 Table 1. Proteins phosphorylated on arginine residues and their phospho-sites detected in a ywlE mutant † † Protein* Phospho-site Protein* Phospho-site
AccD/YttI pR205 MtnK pR80, pR82, pR365 AhpF pR457 NadB pR104 AlaR pR79 Nin pR96 Apt pR85 OdhA pR66, pR621 ArgG pR262 OxdC pR12 AroA pR45, pR301 PdxS pR8 AroF pR127 PfkA pR233 AtpH pR159 PurA pR97 BdhA/YdjL pR13 RadC pR59 ClpC pR5, pR254 RecA pR58 ClpP pR13 RplK pR94 ComFA pR20, pR186, pR446 RplN pR7 ComFC pR74 RplW pR10 ComGA pR28, pR71, pR330 RpmE2 pR70 ComK pR65, pR157, RpmGA pR29 pR161, pR165, Fig. 1. Protein arginine phosphorylation and dephosphorylation in B. sub- pR186, pR191 tilis result from McsB/YwlE catalysis. (A) Collision-induced dissociation-MS/MS CtsR pR15 RpoB pR312, pR313, pR539, spectrum of the GatC phosphopeptide SRISIEEVK. The characteristic ions of pR693, pR694, the phosphorylated Arg of 236.0674 daltons, in particular the b2 and y8 ions, pR827, pR1106 are highlighted in red, reflecting the addition of a phosphate moiety (79.966 DivIVA pR102 RpoC pR803, pR335 Da) to an arginine residue. The observed mass difference of 235.8 Da be- FadB/YsiB pR230 RpsG pR10 tween the y7 and y8 ions is the strongest evidence and is within the expected FliY pR251 RpsH pR72 mass deviation range of 0.5 Da. (B) In vitro phosphorylation of potential FrlB/YurP pR48 RpsI pR20 target proteins by McsB that requires McsA for full activity (9): 1 μMofin- GapA pR199 RpsL pR123 dicated proteins were incubated in the presence of [γ-32P]ATP for 20 min and subsequently analyzed by SDS–PAGE and autoradiography. In the presence GatC pR3 RpsM pR3, pR93 of McsA, McsB was activated and able to phosphorylate itself as well as McsA GgaA pR85 RpsN pR41 and potential substrates (9). The position of phosphorylated McsA and McsB GltA pR904, pR914 ScoC/Hpr pR3 is indicated on the left side. The question mark indicates a low abundant Gmk pR134, pR149 SrfAB pR1378 contaminant of the McsB purification from E. coli. The star displays the GroEL/GroL pR35, pR116, pR282 SsbA/Ssb pR76 position of the different substrates phosphorylated by McsB as indicated in GtaB pR76 SsbB/YwpH pR55 the column heading. The right gel depicts the same reaction as shown in the GudB pR56, pR83, ThiC pR556 more left gel after the addition of YwlE. pR421, pR423 HemB pR217 ThyB pR102 on proteins such as in PtsH (14) and RsbRB (15) (Fig. S2B), HupA pR61 Tig pR90 which demonstrates that the method used permits the detection Icd pR180 YceE pR10 of other protein phosphorylations. In accordance with standards IlvB pR297 YciC pR58 for the identification of phosphorylation sites, a spectrum for IlvC pR298 YdcI pR473 each of the phosphopeptides is listed in Fig. S2A. These obser- IscS pR385 YdjO pR7 vations are clearly demonstrating that protein arginine phos- KatA pR366 YfmG pR474 phorylation exists as a posttranslational modification in vivo in LacA pR387 YheA pR51 bacteria. Under the conditions tested, the arginine phosphory- LeuB pR4 YlbN pR158 lated proteins detected are distributed among distinct physio- LeuC pR81 YloV pR255 logical classes of proteins, including regulators, metabolic LutB/YvfW pR33 YtxH pR50 enzymes, and stress and ribosomal proteins (Fig. S1). LutC/YvbY pR97 YvrO pR161 McsA/YacH pR169 YvyD pR6 YwlE Is a Protein Arginine Phosphatase. Because we were only able Mdh pR156 YvyG pR60 to detect protein arginine phosphorylation in a ywlE mutant and MenB pR78, pR269 YwpJ pR258 not in the wild-type strain (Table 1), whereas proteins phos- MtnA pR80, pR298 phorylated on either serine, threonine, or tyrosine could be detected in both the wild-type and a ywlE mutant strain in equal *Name of the arginine-phosphorylated protein; names are given according amounts (Fig. S2B), we hypothesized that YwlE may exclusively to ref. 26; when a protein has a different naming in the Uni-Prot database, fi this name was also given after the forward slash. act as a protein arginine phosphatase. To con rm that the de- † tection of protein arginine phosphorylation indeed depends on Site of the arginine phosphorylation. the presence of YwlE, we also analyzed a ywlE mutant protein extract that was treated in vitro with purified YwlE protein be- fore MS analysis. In addition, we overexpressed ywlE in trans in As noted above, it was reported that YwlE is the cognate phos- a ywlE mutant in vivo. In both cases, no protein arginine phos- phatase for the McsB kinase in vitro and also antagonizes McsB – phorylation could be detected, which indicates that the presence activity in vivo (8, 9, 11 13). Consequently, we wondered whether of YwlE somehow influences the stability of the detected protein McsB is solely responsible for the arginine phosphorylations phospho-arginine residues. Furthermore, thirty arginine phos- detected in the ywlE mutant (Table 1 and Fig. S1). Protein arginine phorylated proteins were detected in a complex with YwlE phosphorylations were massively diminished in a ywlE/mcsB dou- in vivo (Table S2). This close interaction of these arginine ble-deletion strain, demonstrating that McsB indeed is responsible phosphorylated proteins with YwlE suggests that the stability of for most of the protein arginine phosphorylations, which are sub- these modifications is indeed influenced by the YwlE protein. strate for the YwlE phosphatase. This observation confirms that
7452 | www.pnas.org/cgi/doi/10.1073/pnas.1117483109 Elsholz et al. YwlE specifically antagonizes McsB activity in vivo. However, we A D also detected four proteins with arginine phosphorylations in this mutant (RecA, Nin, BdhA, and AroF), which implies that an ad- ditional protein arginine kinase may exist in B. subtilis. To verify these results, we performed in vitro phosphorylation experiments with purified proteins that had been shown to be phosphorylated on arginine residues in vivo. Our results confirm that these proteins (GapA, PfkA, ScoC, SsbB, ComK, Icd) are indeed phosphorylated by the McsB kinase in vitro (Fig. 1B), implying that McsB is also required for their phosphorylation in vivo. However, the rate of McsB auto-phosphorylation appears to be much higher than the phosphorylation of the different B E substrates under these experimental conditions (Fig. 1B). Con- sistent with our in vivo observations, these proteins, when phos- phorylated in the presence of McsB and its activator McsA, are de-phosphorylated by the YwlE phosphatase in vitro (Fig. 1C). We were also able to confirm by high-accuracy MS (Fig. S3) that the phosphorylation sites that had been detected in vivo were also detected in the purified proteins that had been incubated with McsB in vitro, but not when YwlE was also present in vitro. This finding strongly suggests that McsB is responsible for most of the YwlE-dependent protein arginine phosphorylation that we have identified in vivo (Table 1). However, RecA, which was C F identified as a protein that is phosphorylated on arginine in- dependently of McsB in vivo, was also not phosphorylated by McsB in vitro (Fig. 1B). This finding implies the possibility that an additional protein arginine kinase exists in B. subtilis. In summary, these results suggest that YwlE acts as a protein arginine phosphatase that specifically de-phosphorylates arginine residues both in vitro and in vivo.
Counteracting Activity of Protein Arginine Phosphatase/Kinase Affects the Expression of Different Regulatory Networks. To characterize the regulatory impact of protein arginine phosphorylation in the phosphatase (ywlE) and the phosphatase/kinase (ywlE/mcsB) Fig. 2. Phosphorylation of protein arginine residues influences global gene mutant strains, we monitored the influence of protein arginine expression. Voronoi treemaps combine the value of each gene expression phosphorylation on global gene expression using microarray with its regulation and clusters the genes in the treemap because of its fi MICROBIOLOGY analyses. For the comparison of the strains, we generated Vor- regulatory classi cation according to Subtiwiki (25) (e.g., all of the SigB- onoi Treemaps (16); to visualize the main differences between dependent genes in the SigB cluster; multiple regulated genes appear in each of its regulatory subunits). To visualize differences in transcription level the strains we show the expression rate for each regulon (Fig. 2 increased transcription was highlighted in brown, decreased in blue, and and Table S3). genes with unchanged transcription were indicated in gray (light brown/ The observed changes in gene expression depicted in Fig. 2 ywlE fi blue stands for slightly increased/decreased induction, whereas dark brown/ demonstrate that a deletion has a signi cant impact on gene blue represents strong induction/repression). The plus and minus symbols expression compared with a wild-type strain (Fig. 2 A and B), depict regulons being either induced (+) or repressed (−) depending on the which indicates that protein arginine phosphorylation has a strong regulator assigned to the area. Subpanels illustrate comparisons of expo- influence on gene expression. The observation that the tran- nentially growing B. subtilis ywlE mutant vs. exponentially growing wild- scription of nearly all genes of different regulons was specifically type (A), stationary B. subtilis ywlE mutant vs. stationary wild-type (B), ex- changed in their expression implies regulatory arginine phos- ponentially growing B. subtilis ywlE/mcsB mutant vs. exponentially growing phorylation of individual regulatory proteins, such as activators or wild-type (C), stationary B. subtilis ywlE/mcsB mutant vs. stationary wild-type repressors. For example, ComK-dependent gene expression was (D), exponentially growing B. subtilis ywlE mutant vs. exponentially growing strongly up-regulated in a ywlE mutant (Fig. 2 A and B)andwe B. subtilis ywlE/mcsB (E), and stationary B. subtilis ywlE mutant vs. stationary observed that ComK was also specifically phosphorylated on ar- B. subtilis ywlE/mcsB (F). ginine residues in an ywlE mutant (Table 1), suggesting regulation of ComK activity by these specific phosphorylation sites. fi for these regulatory events. We did not detect a strong effect of To con rm that protein arginine phosphorylations are re- mcsB sponsible for the altered gene expression in the ywlE mutant, we the deletion compared with the wild-type under our ex- fi also analyzed a ywlE/mcsB double-mutant, where the protein perimental conditions. This nding may be related to the ob- fi arginine kinase McsB (7) is mutated and the amount of cellular servation that the McsB kinase is only activated under speci c mcsB arginine phosphorylation is dramatically decreased. If the stress conditions (8, 11), and thus a deletion of should only changes in gene expression observed in the ywlE mutant indeed exhibit a phenotype under these specific physiological conditions. depend on a McsB-dependent protein arginine phosphorylation, The expression of different regulons, such as the ComK, YesS, these changes should be restored to the wild-type levels in the Fnr, MtnR, AbrB, SinR, SigB, CtsR, AcoR, BirA, and NadR re- ywlE/mcsB mutant because protein arginine phosphorylation, gulons, and the stringent response are influenced by a McsB/YwlE- which caused the changed expression, should be absent in this dependent phospho-switch (Fig. 2 and Fig. S4 A and B), but the mutant. In fact, the expression pattern of different regulons in specific functional site for these regulatory events need to be this mutant was restored to wild-type levels compared with the addressed in further studies. Nevertheless, the observed differences ywlE mutant (Fig. 2 C and D), which demonstrates that the al- in gene expression for this cluster of genes caused by mutation of tered expression of these regulons in the ywlE mutant depends the arginine protein kinase/phosphatase genes suggest a direct in- on the presence of the protein arginine kinase McsB. This ob- volvement of protein arginine phosphorylation in gene regulation. servation suggests that McsB-dependent protein arginine phos- We also detected expression patterns for several regulons in phorylations that are de-phosphorylated by YwlE are responsible the ywlE/mcsB double-mutant that are not changed compared
Elsholz et al. PNAS | May 8, 2012 | vol. 109 | no. 19 | 7453 with the ywlE mutant (Fig. 2 E and F) but are massively altered compared with the wild-type (Fig. 2 C and D). These results imply that the expression of this group of genes (i.e., the SigD, Xpf, PyrR, Rok, HrcA, SigE, FadR, PucR, and YvrHB regulons) (Fig. 2 and Fig. S4 C and D) are influenced through the YwlE phosphatase but not in a McsB-dependent manner. Instead, this observation implies that the expression of these genes is influ- enced by YwlE-dependent destabilization of protein arginine phosphorylations that are not catalyzed by McsB. We could also confirm the observed changes for the transcrip- tion of selected genes observed in our transcriptomics experiments on the proteomic level using [35S]methionine pulse-labeling and 2D-PAGE (Fig. S4 E and F). These data indicate that protein phosphorylation on arginine residues has very broad regulatory potential and influences the expression of many important de- velopmental programs, such as competence development, motil- ity, stress and stringent response, and biofilm formation.
Two Phosphorylatable Residues of ClpC Are Required for McsB- Dependent ClpC Activity. To verify the physiological and func- tional significance of protein modification by arginine phos- phorylation in wild-type cells, we analyzed the influence of protein arginine phosphorylation on protein activity for a specific model substrate. We decided to investigate the arginine phos- phorylations of the Hsp100/Clp protein ClpC (R5 and R254) (Table 1) of the AAA+ protein family (17). In contrast to other Hsp100/Clp proteins, ClpC activity depends on the assistance of specific adaptor proteins (18). These adaptor proteins include MecA, which targets ComK for ClpCP-dependent degradation (19), and McsB, which in its kinase-active form targets CtsR for ClpCP-mediated degradation (10, 20). We hypothesized that the ClpC arginine phosphorylation might play an important role for the McsB-mediated ClpC ac- tivity. Using ClpC protein purified from B. subtilis ywlE mutant cells, we were able to detect five arginine phospho-sites in total (Fig. S5A), which could not be identified in ClpC isolated from the wild-type or in the ywlE/mcsB mutant, demonstrating the YwlE/McsB dependence of these modifications. Two of these sites (R5 and R254) were already identified in our previous ap- proach (Table 1), but three of them (R96, R380, R443) were not previously observed (Table 1). To analyze the effect of these ClpC arginine phosphorylations, we mutationally substituted the corresponding ClpC arginine residues and measured the adap- tor-dependent induction of the ClpC ATPase activity in vitro (Fig. 3 A and B). As previously reported (10, 20), all ClpC proteins showed very low ATPase activity in the absence of an adaptor protein. When McsB and its activator McsA were added to the assay, the ATPase activity of wild-type ClpC was fully induced. Some of the ClpC point mutants (R96, R380, R443) are also induced in the presence of McsA/B, implying that phos- phorylation of these arginine residues is not essential for the McsB-mediated activation of ClpC. However, the ClpCR5 and ClpCR245 variants were not, or only marginally, activated by the McsB kinase compared with wild-type ClpC. These results demonstrated that phosphorylation of two arginine residues of ClpC (R5 and R254) by McsB is required for the McsB-mediated ClpC activity in vitro. In contrast, MecA, another ClpC adaptor (18), was still able to induce the ATPase activity of these ClpC-variants ClpCR5A and
mutant variants of ClpC (all 1 μM) as indicated in the figure. Samples were taken at the indicated time points, separated on SDS–PAGE and Coomassie- stained McsB protein is depicted. (D) Radioactive [35S]methionine pulse- Fig. 3. McsB mediated ClpC activity involves phosphorylation of specific chase labeling and immunoprecipitation of CtsR after heat stress in B. subtilis ClpC arginine residues. (A and B) The ATPase rate of ClpC and the different wild-type, clpCR5K or clpCR254K mutant cells. (E) Quantification of CtsR sta- ClpC variants was determined in the presence of (A) McsA/McsB or (B) MecA. bility after heat stress in ClpCWT, ClpCR5K, and ClpCR254K cells in vivo (see D). On the x axis are the different proteins depicted that are present in the (F) Interaction of ClpC, ClpCR5K, and ClpCR254K with McsB (all 1 μM) analyzed sample. (C) In vitro degradation of McsB by ClpC and the indicated ClpC by surface plasmon resonance. The binding response was measured in res- mutants. The assay was carried out in the presence of pyruvate kinase (20 μg/ onance units (RU). McsB was immobilized on a CM5 chip and the indicated mL), 2 mM PEP, and 2 mM ATP with McsA, McsB, ClpP, and the corresponding proteins were passed over the chip surface.
7454 | www.pnas.org/cgi/doi/10.1073/pnas.1117483109 Elsholz et al. ClpCR254A (Fig. 3B). This observation indicates that these ClpC strate that the general MecA-dependent activity of ClpC is not variants are generally functional in their activity and only lack the influenced by ClpC arginine phosphorylation but that the phos- ability to be activated by McsB. phorylation of R5 and R254 is required for a functional activation To analyze the effect of the different ClpC-variants in the wild- of ClpC by McsB in B. subtilis cells where the YwlE protein ar- type background, we used a genetic system to introduce substitution ginine phosphatase is present. mutations into the clpC locus on the chromosome (11). Through Taken together, these in vitro and in vivo results also suggest this process we could investigate the impact of ClpC arginine that a specific McsB-dependent protein arginine phosphorylation phosphorylation on CtsR proteolysis by radioactive pulse-chase event is of functional importance in the wild-type cells and displays R5 R254 labeling and immunoprecipitation in vivo (Fig. 3D)andonMcsB a clear phenotype for clpC and clpC . degradation by using purified components in vitro (Fig. 3C). The observed in vivo stabilization of CtsR in the respective Discussion R5 R254 clpC and clpC strains and the in vitro stabilization of McsB In this work, we demonstrated that the recently identified modi- with ClpCR5 and ClpCR254 but not with the wild-type ClpC fication of proteins by arginine phosphorylation (7) occurs and is variants demonstrate that the McsB-dependent proteolysis of functional in vivo. The deletion of the gene for the protein argi- CtsR by ClpCP requires phosphorylation of ClpC on the specific nine phosphatase, YwlE, allowed us to identify the importance of identified arginine residues (R5 and R254) in vivo and in vitro protein arginine phosphorylation in vivo. In this strain, we were (Fig. 3 C–E). We have also confirmed that McsB interacts with able to identify 121 phosphorylation sites for 87 proteins in the low the different ClpC variants comparable to the wild-type ClpC GC, Gram-positive model organism B. subtilis. protein in vivo and in vitro (Fig. 3F and Fig. S5 D and E). In addition, the McsB/YwlE-dependent detection of protein Additional experiments demonstrated that the MecA-de- arginine phosphorylation/de-phosphorylation in vivo unambigu- pendent ClpC activity is independent from phosphorylation of ously demonstrates that McsB and YwlE are a protein arginine these specific ClpC arginine residues. As mentioned above, the kinase and phosphatase, rather than a tyrosine kinase and phos- ATPase activity of ClpCR5A and ClpCR254A was activated like phatase, as was previously suggested (9, 21). However, we also wild-type ClpC by MecA in vitro (Fig. 3B) and consequently observed McsB-dependent phosphorylation of tyrosine residues, MecA was degraded by ClpCR5 and ClpCR254 at a rate comparable but only in the immediate vicinity to a phospho-arginine residue of to the wild-type ClpCP in vitro (Fig. 4A). In addition, the amount ClpC (Fig. S6), and not when the arginine residue was mutated to R5 R254 of ComK is not altered in the clpC and clpC strains com- lysine or alanine. This observation suggests that a transfer of the pared with the wild-type in vivo (Fig. 4B). Furthermore, the phospho-group from the high-energy phosphoramidate N-P ComK-dependent transcriptional activation of the comG pro- linkage (7, 22) to the -OH group of the tyrosine can occur in vitro. moter, which is regulated by the controlled degradation of ComK Consistent with this hypothesis, the chemically more stable through the MecA/ClpCP complex, is also not changed in the phospho-tyrosines were detected in extracts of hydrolyzed phos- ClpC point mutants (Fig. 4C), demonstrating that ComK is still phorylated McsB and McsA protein preparations by TLC (9). degraded by these ClpC-variants, which cannot be phosphorylated Moreover, our results establish that protein arginine phos- on these specific arginine residues in vivo. These results demon- phorylation is a physiological significant protein modification in wild-type B. subtilis cells. Although we could not detect any protein arginine phosphorylation in wild-type cells, our data indicate that
protein arginine phosphorylations do have a significant physio- MICROBIOLOGY logical function in vivo. One possible explanation for this failure could be that the McsB kinase, which is responsible for most of the YwlE-dependent detected protein arginine phosphorylation, is not active under the tested conditions in wild-type cells (8, 11). Nevertheless, it is known that McsB is slightly activated in a ywlE mutant under all tested conditions (8), which would lead to the phosphorylation of McsB substrates, even under conditions where they are normally not targeted for phosphorylation by McsB. An additional explanation for the failure to detect protein arginine phosphorylation might also be that the high-energy phospho-arginine bond, which forms a thermodynamically rela- tively unstable protein modification (7), together with the pres- ence of the YwlE arginine protein phosphatase activity results in a very labile protein modification, which may exist only in a rel- atively short time window. This finding would also imply that such rather short-lived pro- tein modification could allow rapid regulation of protein activity by protein arginine kinases. Fast modulation of protein activity by reversible protein arginine phosphorylation, mediated through the interplay between protein arginine kinases and the YwlE protein arginine phosphatase activity, presumably allows B. subtilis cells to adapt to their ever-changing environment in an appro- priate manner. Our data strongly suggest that phosphorylation of two ClpC arginine residues (R5 and R254) by McsB is required for the McsB-mediated activity of ClpC (Fig. 3). However, ClpC activity Fig. 4. Dual-activity control of ClpC activity by two different adaptors. (A)In does not generally depend on phosphorylation of these two ar- vitro degradation of MecA by ClpC and indicated mutant variants. Coo- massie-stained MecA protein is depicted. (B) Amount of ComK in B. subtilis ginine residues. In contrast, MecA, another adaptor of ClpC that wild-type and mutant strains. Western-Blot analysis of ComK showing that is required for ComK degradation, is still able to induce these ComK only accumulates in a clpC mutant, whereas ComK is wild-type like ClpC variants to a level comparable to the wild-type protein (Fig. degraded in the clpCR5K and clpCR254K mutant cells. (C) ComK activity was 4). Thus, the phosphorylation event is not needed for general monitored by a comG-lacZ fusion in a clpC mutant and in clpC + wild-type, ClpC activity. Instead, it presents a new layer of activity control clpCR5K,orclpCR254K mutant cells. of ClpC by the different adaptor proteins.
Elsholz et al. PNAS | May 8, 2012 | vol. 109 | no. 19 | 7455 Unfortunately, we were not able to determine quantitatively coli DH5α (Invitrogen) was used for cloning experiments. DNA manipulation the extent of protein arginine phosphorylation. Thus, we cannot and other molecular biological procedures were carried out according to exclude that only a subpopulation of the identified protein ar- standard protocols. The biochemical assays were performed as previously de- ginine kinase substrates are phosphorylated under the respective scribed (9, 10). For details of MS, protein digestion, phosphopeptide enrich- experimental conditions. However, the presented results suggest ment, and identification of phosphate-containing peptides, see SI Materials and that protein arginine phosphorylation is present in vivo and that Methods. To illustrate the correlation of gene expression with gene function as many of the proteins modified by arginine phosphorylation that well as regulon allocation, we used a variant of Voronoi treemaps (16, 23). For we detected in the ywlE deletion strain might also have specific surface plasmon resonance, McsB was covalently attached to a CM5 chip surface physiological functions in B. subtilis wild-type cells. The activity according to the manufacturer’s instructions (Biacore), yielding 3,000 resonance of protein arginine kinases, such as McsB, is important to control units of immobilized McsB. ClpC, ClpCR5K,andClpCR254K (1 μM) in HBS buffer [10 protein function, but the regulation of protein activity likely also mM Hepes pH 7.4, 150 mM NaCl, 3 mM EDTA, 0.005% (vol/vol) surfactant P20] depends on the YwlE phosphatase activity, because we could was passed over the chip surface with immobilized McsB and a reference surface detect the protein arginine phosphorylation only in a ywlE de- at a flow rate of 20 μL/min using a Biacore 3000 instrument. The subsequent letion strain. This finding indicates that the counteracting activity injection of HBS buffer allowed monitoring of the dissociation. RNA isolation of protein arginine kinase and phosphatase is integrated into was performed as described previously (11). Transcriptome analysis was carried important physiological and cellular processes of B. subtilis in out using an Agilent custom microarray, as described previously (24). The which protein arginine phosphorylation are involved. microarray data have been deposited in the National Center for Biotechnology In addition, the strong regulatory potential of this more-sensi- Information’s Gene Expression Omnibus (GEO) database and are accessible tive protein modification became apparent in our transcriptomic through GEO Series accession no. GSE31249. experiments, comparing the transcriptome of a ywlE deletion with See Table S1 for the full genotypes of strains and plasmids. the wild-type strain (Fig. 2). Our experiments demonstrate that protein arginine phos- ACKNOWLEDGMENTS. We thank A. Tschirner, Marc Schaffer, and A. Wiechert phorylation and de-phosphorylation can play a role in bacterial for technical assistance; M. Zander for help with experiments; L. Petruschka for physiology. Protein arginine phosphorylation may participate in assistance with the Biacore experiments; J. Schüler, A. Spillner, H. Mehlan (Uni- many critical cellular processes, and therefore our study provides versity of Greifswald), M. Moser, and M. Berth (Decodon) for help with Voronoi the basis for a detailed functional analysis of this posttransla- treemaps; J. Stülke for sending plasmids; S. Hübner, J. Landmann, and N. Pietack fi (University of Göttingen) for creating plasmids pGP438, pGP385, and pGP393; tional modi cation in bacteria. D. Dubnau for discussion; and R. Switzer for critically reading the manuscript. This work was supported by Deutsche Forschungsgemeinschaft Grants HE1887/ Materials and Methods 7-4, HE1887/8-1, and SFB/TR34 (to M.H.), Tu106/6-1 and the Heisenberg program B. subtilis was grown in liquid medium or on LB agar plates with tetracycline (17 (to K.T.), and the European Union Bacell Health Program LSHG-CT-2004-503468 μg/mL), spectinomycin (200 μg/mL), or chloramphenicol (5 μg/mL). Escherichia (to M.H.).
1. Pawson T, Scott JD (2005) Protein phosphorylation in signaling—50 years and 14. Deutscher J, Pevec B, Beyreuther K, Kiltz HH, Hengstenberg W (1986) Streptococcal counting. Trends Biochem Sci 30:286–290. phosphoenolpyruvate-sugar phosphotransferase system: Amino acid sequence and 2. Macek B, et al. (2008) Phosphoproteome analysis of E. coli reveals evolutionary site of ATP-dependent phosphorylation of HPr. Biochemistry 25:6543–6551. conservation of bacterial Ser/Thr/Tyr phosphorylation. Mol Cell Proteomics 7: 15. Chen C-C, Yudkin MD, Delumeau O (2004) Phosphorylation and RsbX-dependent dephosphorylation of RsbR in the RsbR-RsbS complex of Bacillus subtilis. J Bacteriol 299–307. 186:6830–6836. 3. Macek B, et al. (2007) The serine/threonine/tyrosine phosphoproteome of the model 16. Bernhardt J, Funke S, Hecker M (2009) Visualizing Gene Expression Data via Voronoi bacterium Bacillus subtilis. Mol Cell Proteomics 6:697–707. Treemaps. Voronoi Diagrams in Science and Engineering. ISVD ’09, Sixth International 4. Hoch JA (2000) Two-component and phosphorelay signal transduction. Curr Opin Symposium, June 23–26, 2009. Copenhagen, Denmark. pp. 233–241. – Microbiol 3:165 170. 17. Kirstein J, Molière N, Dougan DA, Turgay K (2009) Adapting the machine: Adaptor 5. Deutscher J, Saier MH, Jr. (2005) Ser/Thr/Tyr protein phosphorylation in bacteria–For proteins for Hsp100/Clp and AAA+ proteases. Nat Rev Microbiol 7:589–599. – long time neglected, now well established. J Mol Microbiol Biotechnol 9:125 131. 18. Schlothauer T, Mogk A, Dougan DA, Bukau B, Turgay K (2003) MecA, an adaptor 6. Meins M, et al. (1993) Cysteine phosphorylation of the glucose transporter of Es- protein necessary for ClpC chaperone activity. Proc Natl Acad Sci USA 100:2306–2311. cherichia coli. J Biol Chem 268:11604–11609. 19. Turgay K, Hahn J, Burghoorn J, Dubnau D (1998) Competence in Bacillus subtilis is 7. Fuhrmann J, et al. (2009) McsB is a protein arginine kinase that phosphorylates and controlled by regulated proteolysis of a transcription factor. EMBO J 17:6730–6738. inhibits the heat-shock regulator CtsR. Science 324:1323–1327. 20. Kirstein J, et al. (2006) Adaptor protein controlled oligomerization activates the AAA+ 8. Elsholz AKW, et al. (2011) Activity control of the ClpC adaptor McsB in Bacillus subtilis. protein ClpC. EMBO J 25:1481–1491. J Bacteriol 193:3887–3893. 21. Musumeci L, et al. (2005) Low-molecular-weight protein tyrosine phosphatases of 9. Kirstein J, Zühlke D, Gerth U, Turgay K, Hecker M (2005) A tyrosine kinase and its Bacillus subtilis. J Bacteriol 187:4945–4956. activator control the activity of the CtsR heat shock repressor in B. subtilis. EMBO J 24: 22. Teague WE, Jr., Dobson GP (1999) Thermodynamics of the arginine kinase reaction. J – 3435–3445. Biol Chem 274:22459 22463. 23. Balzer M, Deussen O (2005) Voronoi treemaps for the visualization of software 10. Kirstein J, Dougan DA, Gerth U, Hecker M, Turgay K (2007) The tyrosine kinase McsB is metrics. Proceedings of the AMC Symposium on Software Visualization. May 14–15, a regulated adaptor protein for ClpCP. EMBO J 26:2061–2070. 2005, St. Louis, MO. 11. Elsholz AKW, Michalik S, Zühlke D, Hecker M, Gerth U (2010) CtsR, the Gram-positive 24. Winter T, et al. (2011) Characterization of the global impact of low temperature gas master regulator of protein quality control, feels the heat. EMBO J 29:3621–3629. plasma on vegetative microorganisms. Proteomics 11:3518–3530. 12. Hahn J, Kramer N, Briley K, Jr., Dubnau D (2009) McsA and B mediate the de- 25. Lammers CR, et al. (2010) Connecting parts with processes: SubtiWiki and SubtiPath- localization of competence proteins from the cell poles of Bacillus subtilis. Mol Mi- ways integrate gene and pathway annotation for Bacillus subtilis. Microbiology 156: crobiol 72:202–215. 849–859. 13. Kirstein J, Strahl H, Molière N, Hamoen LW, Turgay K (2008) Localization of general 26. Barbe V, et al. (2009) From a consortium sequence to a unified sequence: The Bacillus and regulatory proteolysis in Bacillus subtilis cells. Mol Microbiol 70:682–694. subtilis 168 reference genome a decade later. Microbiology 155:1758–1775.
7456 | www.pnas.org/cgi/doi/10.1073/pnas.1117483109 Elsholz et al.
Article IV
Page 99
Page 100 Research
© 2011 by The American Society for Biochemistry and Molecular Biology, Inc. This paper is available on line at http://www.mcponline.org S-Bacillithiolation Protects Against Hypochlorite Stress in Bacillus subtilis as Revealed by Transcriptomics and Redox Proteomics*□S
Bui Khanh Chi‡, Katrin Gronau‡, Ulrike Ma¨ der§, Bernd Hessling‡, Do¨ rte Becher‡, and Haike Antelmann‡¶
Protein S-thiolation is a post-translational thiol-modifica- Molecular & Cellular Proteomics 10: 10.1074/mcp. tion that controls redox-sensing transcription factors and M111.009506, 1–21, 2011. protects active site cysteine residues against irreversible oxidation. In Bacillus subtilis the MarR-type repressor OhrR was shown to sense organic hydroperoxides via Reactive oxygen species (ROS)1 are an unavoidable con- formation of mixed disulfides with the redox buffer bacil- sequence of the aerobic lifestyle of many organisms (1, 2). lithiol (Cys-GlcN-Malate, BSH), termed as S-bacillithiola- ROS can be generated by incomplete reduction of molecular tion. Here we have studied changes in the transcriptome oxygen during the respiratory chain. Pathogenic bacteria en- and redox proteome caused by the strong oxidant hypo- counter ROS, such as hydrogen peroxide (H O ), superoxide chloric acid in B. subtilis. The expression profile of NaOCl 2 2 anion and hypochloric acid as defense of the innate immune stress is indicative of disulfide stress as shown by the induction of the thiol- and oxidative stress-specific Spx, response during host-pathogen interactions. Upon bacterial CtsR, and PerR regulons. Thiol redox proteomics identi- infection, myeloperoxidase is released from activated macro- fied only few cytoplasmic proteins with reversible thiol- phages to produce the strong oxidant hypochloric acid from Ϫ oxidations in response to NaOCl stress that include GapA H2O2 and Cl (3, 4). and MetE. Shotgun-liquid chromatography-tandem MS ROS can damage all cellular macromolecules, such as pro- analyses revealed that GapA, Spx, and PerR are oxidized teins, lipids, carbohydrates, and nucleotides (2, 5). Cells ac- to intramolecular disulfides by NaOCl stress. Further- tivate the expression of antioxidant mechanisms to detoxify more, we identified six S-bacillithiolated proteins in ROS and to repair the damage. The response of bacteria to NaOCl-treated cells, including the OhrR repressor, two H O and organic hydroperoxides (ROOH) involves heme- methionine synthases MetE and YxjG, the inorganic py- 2 2 cofactor containing catalases and thiol-dependent peroxi- rophosphatase PpaC, the 3-D-phosphoglycerate dehy- dases as detoxification mechanisms (5, 6). Peroxidases use drogenase SerA, and the putative bacilliredoxin YphP. S-bacillithiolation of the OhrR repressor leads to up- conserved redox-active cysteine residues that function in re- regulation of the OhrA peroxiredoxin that confers to- duction of peroxides. These oxidative stress defense mech- gether with BSH specific protection against NaOCl. S- anisms are often controlled by redox-sensitive transcription bacillithiolation of MetE, YxjG, PpaC and SerA causes factors that undergo post-translational thiol-modifications hypochlorite-induced methionine starvation as sup- upon challenge with ROS leading either to activation or inac- ported by the induction of the S-box regulon. The mech- tivation of the transcription factors (7). Post-translational thiol- anism of S-glutathionylation of MetE has been de- modifications implicated in redox-sensing mechanisms are scribed in Escherichia coli also leading to enzyme inactivation and methionine auxotrophy. In summary, our 1 The abbreviations used are: ROS, reactive oxygen species; BSH, studies discover an important role of the bacillithiol redox bacillithiol; Brx, bacilliredoxin; CHP, cumene hydroperoxide; Cys, buffer in protection against hypochloric acid by S-bacilli- cysteine; GlcNAc, N-acetyl glucoseamine; GSH, glutathione; GST, thiolation of the redox-sensing regulator OhrR and of glutathione S-transferase; IAM, iodoacetamide; IP, immunoprecipita- four enzymes of the methionine biosynthesis pathway. tion; LMW, low molecular weight; Mal, malate; MSH, mycothiol; Met, methionine; MG, methylglyoxal; MHQ, methylhydroquinone; N5-THF, 5-methyltetrahydrofolate; N5,N10-THF, 5,10-methylenetetrahydrofo- Ј Ј From the ‡Institute for Microbiology and §Interfaculty Institute for late; PAPS, 3 -phosphoadenosine-5 -phosphosulfate;PPi, inorganic Genetics and Functional Genomics, Ernst-Moritz-Arndt-University of pyrophosphate; ROOH, organic hydroperoxide; ROH, organic alco- Greifswald, D-17487 Greifswald, Germany hol; RES, reactive electrophilic species; RuMP, ribulose-5-mono- Received March 10, 2011, and in revised form, July 2, 2011 phosphate; THF, tetrahydrofolate; TrxAB, thioredoxin/thioredoxin re- Published, MCP Papers in Press, July 12, 2011, DOI 10.1074/ ductase; RNS, reactive nitrogen species; FOX, ferrous oxidation mcp.M111.009506 xylenol orange.
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–1 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis known as thiol-disulfide redox-switches and include in most tions that contribute to the resistance of B. subtilis to the cases inter- or intramolecular disulfides and mixed disulfides strong oxidant hypochloric acid. Hypochloric acid is the active with low molecular weight (LMW) thiols (S-thiolations). The component of household bleach and widely used as antimi- best studied examples for peroxide-sensing thiol-based tran- crobial disinfectant to clean surfaces. The bactericidal effect scription factors are Escherichia coli OxyR (8–11) and yeast of hypochloric acid has been proposed to involve generation Yap1 transcription factor (7, 12, 13) that are activated by of ROS, such as superoxide anion and hydroxyl radical for- intramolecular disulfide bond formation to induce the antiox- mation in E. coli (27). Recent redox proteomics studies in idant defense mechanisms. E. coli using the OxICAT approach have shown that bleach In Bacillus subtilis, the major detoxification enzymes for causes strong disulfide formation and protein aggregation in a peroxides are catalase (KatA) and alkylhydroperoxide reduc- different set of proteins than H2O2 (28). As defense mecha- tase (AhpCF) that are controlled of the peroxide-sensing Fur nism against NaOCl stress, E. coli uses the redox controlled family metalloregulatory PerR repressor (5). PerR harbors a chaperone Hsp33 that is activated by NaOCl by the formation structural Zn-binding site coordinated by four cysteine resi- of intramolecular disulfides in the Zn-redox switch centers dues and a regulatory Fe or Mn binding site consisting of His resulting in Zn release, oxidative unfolding and dimerization and Asp residues. Inactivation of PerR is caused by Fe- (29). Hsp33 protects cells against NaOCl-induced protein ag- catalyzed oxidation of His37 and His91 to 2-oxohistidine in gregation. The mode of action has been also studied using the regulatory site (5, 14–16). The response to ROOH involves transcriptome analyses in pathogenic E. coli O157:H7 out- the MarR-type repressor OhrR in B. subtilis that is conserved break strains and the food-borne pathogen B. cereus in many other bacteria (6, 7). OhrR-like proteins control a ATCC14579 (30, 31). Both transcriptome analyses suggest a thiol-dependent peroxiredoxin that converts ROOH to organic major oxidative stress response mechanism of NaOCl. Regu- alcohols. OhrR proteins can be devided into the one and lons involved in the biosynthesis of sulfur and sulfur-contain- two-Cys families of redox sensing repressors. The OhrR pro- ing amino acids were up-regulated by NaOCl in both genome- tein of Xanthomonas campestris belongs to the two-Cys fam- wide studies. However, the mode of action of hypochloric ily that is oxidized to a Cys22-Cys127‘ intermolecular disulfide acid has not yet been investigated in B. subtilis. between both subunits of the OhrR dimer (17). One-Cys OhrR We have used transcriptomic and redox proteomic ap- proteins harbor one conserved N-terminal Cys with the pro- proaches coupled with shotgun-LC-MS/MS analyses to ana- totype of B. subtilis OhrR or Staphylococcus aureus SarZ and lyze the mode of action and reversible thiol-modifications by MgrA (7, 18). Cumene hydroperoxide (CHP) leads to Cys15 NaOCl stress in B. subtilis. We discovered that the major oxidation to sulfenic acid that is rapidly oxidized to S-thiolated resistant determinant to NaOCl is the OhrA peroxiredoxin that OhrR containing cysteine or the redox buffer bacillithiol (BSH) conferred specific protection against NaOCl toxicity. More- (19, 20). Thus, B. subtilis OhrR is controlled by S-cysteinyla- over, we identified S-bacillithiolations of the OhrR repressor, tion and S-bacillithiolation as redox-switch mechanism lead- two methionine synthases MetE and YxjG, the inorganic py- ing to inactivation of the OhrR repressor function and dere- rophosphatase PpaC, and the 3-D-phosphoglycerate dehy- pression of ohrA transcription. drogenase SerA as major protection mechanisms against hy- In previous studies, we investigated the global response, pochlorite stress in B. subtilis. post-translational modifications and specific regulatory mechanisms that are induced by reactive electrophilic spe- EXPERIMENTAL PROCEDURES cies (RES) in B. subtilis, such as diamide, quinones, or alde- Bacterial Strains and Growth Conditions—The bacterial strains hydes. RES deplete the cellular redox buffer cysteine leading used were B. subtilis wild-type strains 168 (trpC2), JH642 (trpC2 to induction of the Spx-regulon that controls thiol-disulfide attSP), and CU1065 (trpC2 pheA1) and mutant strains ⌬spx r ⌬ r ⌬ oxidoreductases (TrxAB) to restore the redox homeostasis (trpC2,spx::neo ) (32), ohrR (trpC2,ohrR::cm ), ohrA (trpC2, ohrA::cmr), ⌬sigB (trpC2,sigB::cmr), ⌬perR (trpC2,perR::cmr) (33), (21–23). B. subtilis encodes specific redox-sensing regulators HB9121 (CU1065 trpC2,ohrR::kmr ohrR-FLAG (Spcr) ohrA-cat lacZ of the MarR/DUF24-family that sense RES, but not ROS (7). (Neor) (19), HB2048 (CU1065 SPc2⌬2::Tn917::(ohrA-cat- These include the paralogous repressors YodB and CatR lacZ)ohrR::kan,thrC::pXTohrRC15S)(34), HB11002 (CU1065 trpC2, that are inactivated via intermolecular disulfide formation by bshA::mlsr), and HB11053 (CU1065 trpC2, bshB1:: spcr bshB2::cmr) diamide and quinones resulting in derepression of the (35). B. subtilis strains were cultivated under vigorous agitation at 37 °C in Belitsky minimal medium (BMM) as described previously (36). azoreductase (AzoR1), nitroreductase (YodC), and thiol-de- The antibiotics were used at the following concentrations: 1 g/ml pendent dioxygenase (CatE) catalyzing the detoxification of erythromycin, 25 g/ml lincomycin, 5 g/ml chloramphenicol, 10 the electrophiles (24–26). Other proteins of the MarR/ g/ml kanamycin, and 100 g/ml spectinomycin. Sodium hypochlo- DUF24-family (HxlR) and of the MerR/NmlR-family (AdhR) rite (15% stock solution), diamide, and cumene hydroperoxide were sense specifically aldehydes, such as formaldehyde and purchased from Sigma Aldrich. For NaOCl stress exposure, cells were grown in BMM to an optical methylglyoxal (23). density at 500 nm (OD500) of 0.4 and treated with 50, 75, or 100 M In this study, we were interested in the global response, NaOCl diluted freshly in destilled water. The growth experiments in regulatory mechanisms, and post-translational thiol-modifica- the presence of methionine were performed by addition of 75 M
10.1074/mcp.M111.009506–2 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
methionine either after inoculation of the culture or 30 and 60 min after versibly oxidized proteins in the Coomassie-stained thiol-redox NaOCl stress exposure. proteome was performed manually as described previously (24). In Gene deletions for construction of the ohrA mutant were generated brief, gel pieces were washed 3–5 times with 1 ml of 20 mM (w/v) using long-flanking-homology polymerase chain reaction (LFH-PCR) ammonium bicarbonate, pH 8.0/50% (v/v) acetonitrile (ACN) for 30 as described previously (25). Primers ohrA-F1 (5Ј-TGCAGCTGATTG- min and once with 1 ml 75% ACN for 30 min. Gel pieces were dried AGGATACG-3Ј) and ohrA-F2 (5Ј-GTTATCCGCTCACAATTCGCGGT- and the proteins in-gel digested with 20 g/l trypsin dissolved in CTGATGAAATGACCT-3Ј) were used to amplify the up fragment and water (Promega, Madison, WI) at 37 °C for 16 h. Tryptic peptides primers ohrA-R1 (5Ј-CGTCGTGACTGGGAAAACGGTGTGACGCTG- were eluted with 0.5% (w/v) trifluoroacetic acid/50% (v/v) ACN and CAAGTAAA-5Ј) and ohrA-R2 (5Ј-CCCTTCAATCTCCGAATCAA-3Ј)to 0.5 l of this peptide solution was spotted on the MALDI-targets. amplify the down fragment, respectively. Fragments were amplified Then, 0.5 l of matrix solution (50% (v/v) ACN/0.5% (w/v) trifluo- and joined together with the chloramphenicol cassette using Pfusion roacetic acid) saturated with ␣-cyano-4-hydroxy cinnamic acid was DNA polymerase (Invitrogen, Carlsbad, CA) as described (33). Inte- mixed with the spotted tryptic peptides and dried on the target for gration and deletion of the ohrA gene were confirmed by PCR and by 15 min. Northern blot analysis using digoxigenin-labeled RNA probes of the The matrix-assisted laser desorption ionization/time of flight corresponding gene. (MALDI-TOF)-TOF measurement of spotted peptide solutions was Analysis of NaOCl Concentrations in the Cell Culture Supernatant carried out on a Proteome-Analyzer 4800 (Applied Biosystems, Foster Using the FOX Assay—The concentrations of the remaining NaOCl in City, CA). The spectra were recorded in reflector mode in a mass the culture supernatants were determined using the FOX assay (37). range from 900 to 3700 Da with a focus mass of 2000 Da. For one FOX reagent was prepared by mixing 100 ml FOX I (100 mM sorbitol, main spectrum 25 subspectra with 100 shots per subspectrum were 125 M xylenol orange) and 1 ml FOX II (25 mM ammonious ferrous- accumulated using a random search pattern. If the autolytical frag- ϩ ϩ (II)sulfate in 2.5 M H2SO4). It was not possible to measure any NaOCl ment of trypsin with the monoisotopic (M H) m/z at 2211.104 concentrations in BMM with tryptophane and glutamate. Thus, cells reached a signal to noise ratio (S/N) of at least 10, an internal cali-
were grown to an OD500 of 0.4 in BMM, centrifuged and resuspended bration was automatically performed using this peak for one-point- in BMM without tryptophane and glutamate before the addition of 75 calibration. The peptide search tolerance was 50 ppm but the actual M NaOCl. Samples of 500 l medium were taken at different time RMS value was between 10 and 20 ppm. After calibration the peak points after NaOCl addition, mixed with 500 l FOX reagent and lists were created by using the “peak to mascot” script of the GPS incubated at room temperature for 60 min. The absorbance was Explorer™ software version 3.6 with the following settings: mass measured at 560 nm. Calibration curves were generated using NaOCl range from 900 to 3700 Da; peak density of 50 peaks per range of 200 concentrations in the range from 0 to 100 M diluted in BMM without Da; minimal area of 100 and maximal 200 peaks per protein spot and tryptophane and glutamate. minimal S/N ratio of 6. The peak lists were searched against a Bacillus Thiol Redox Proteome Analysis—The thiol redox proteome analysis subtilis sequence database extracted from UniprotKB release 12.7 was performed as described previously (38) with the modifications as (UniProt Consortium, Nucleic acids research 2007, 35, D193–197) explained (21). Cells were harvested before (control conditions) and using the Mascot search engine version 2.1.04 (Matrix Science Ltd, 10, 20, and 30 min after exposure to 50 M NaOCl stress, resus- London, UK). pended in urea/CHAPS alkylation buffer (8 M urea; 1% CHAPS; 1 mM MALDI-TOF-TOF MS/MS analysis was performed for the three EDTA; 200 mM Tris-HCl pH 8,0; 100 mM iodoacetamide (IAM)), son- strongest peaks of the TOF-spectrum. For one main spectrum 20 icated, alkylated for 20 min in the dark, precipitated with 100% sub-spectra with 125 shots per subspectrum were accumulated us- acetone, washed several times with 80% acetone and dried. After ing a random search pattern. The internal calibration was automati- resolving in urea/CHAPS buffer without IAM, 200 g of the protein cally performed as one-point-calibration if the mono-isotopic arginine ϩ ϩ extract were reduced with 10 mM Tris-(2-carboxyethyl)-phosphine (MϩH) m/z at 175.119 or lysine (MϩH) m/z at 147.107 reached a
(TCEP) and labeled with BODIPY FL C1-IA [N-(4,4-difluoro-5,7-di- S/N of at least 5. The peak lists were created by using the “peak to methyl-4-bora-3a,4a-diaza-s-indacene-3-yl)-methyl)-iodoacetamide] mascot” script of the GPS Explorer™ software version 3.6 with the (Invitrogen, Eugene, OR). The fluorescence-labeled protein extract following settings: mass range from 60 Da to a mass that was 20 Da was separated using 2D PAGE as described (21). The two-dimen- lower than the precursor mass; peak density of 5 peaks per 200 Da; sional (2D) gels were scanned using a Typhoon 9400 variable mode minimal area of 100 and maximal 20 peaks per precursor and a imager (Amersham Biosciences, Freiburg, Germany) for BODIPY- minimal S/N ratio of 5. Peptide mixtures that yielded a mowse score fluorescence and then stained with Colloidal Coomassie for protein of at least 50 in the reflector mode and a sequence coverage of at amounts. Quantitative image analysis was performed with the least 30% that were confirmed by subsequent MS/MS analysis were DECODON Delta 2D software (http://www.decodon.com). regarded as positive identification. The complete Mascot search re- The first alkylation protocol (38) that applied the TCA-precipitation sults including the MS and MS/MS data of all protein identifications step to harvest cells to stop thiol-disulfide exchanges was changed are shown in supplemental Figs. S1A–P. previously (21) for two reasons: (1) The 2D gels were of bad quality Immunoprecipitation (IP) and Nonreducing SDS-PAGE Analysis of and exhibited protein streaking during the isoelectric focusing (IEF) the OhrR-FLAG Protein—The OhrR-FLAG protein expressing B. sub- and (2) GapA was oxidized artificially by this TCA precipitation step tilis strain HB9121 was grown in BMM and treated with 50 M NaOCl
under control conditions but not if this TCA step was omitted (38) and at an OD500 of 0.4. Cells were harvested before (control conditions) GapA is an important redox-controlled cytoplasmic marker protein and 15 min after NaOCl-treatment in TE-buffer (10 mM Tris-HCl, pH8; and strongly oxidized in response to quinones and diamide (21, 22). 1mM EDTA) with 100 mM IAM. Cells were sonicated, the protein Because GapA was neither oxidized using our alkylation protocol in extracts obtained after repeated centrifugation and alkylated in the the redox proteome of control cells nor in the LC-MS/MS approach, dark for 20 min. OhrR-FLAG protein was purified by IP using anti- this indicates no artificial thiol-disulfide exchange during our sample FLAG M2-affinity agarose (Invitrogen) according to the instructions of preparations. the manufacturer. The precipitated OhrR-FLAG protein was eluted by Identification of Reversibly Oxidized Proteins in the Thiol-Redox boiling in non-reducing SDS sample buffer (4% SDS; 62.5 mM Tris- Proteome Using Matrix-Assisted Laser Desorption Ionization/Time HCl pH 8.0, glycerol) and separated using 15% nonreducing SDS- Of Flight-TOF Tandem MS (MS/MS)—Tryptic digestion of the re- PAGE. The OhrR-FLAG protein band of the expected size was cut
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–3 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
from the SDS-gel, tryptically digested as described above and ana- the manufacturer’s instructions. Synthesis and purification of fluores- lyzed by LTQ-Orbitrap mass spectrometry. cently labeled cDNA were carried out as descibed (43) with minor LTQ-Orbitrap Velos Mass Spectrometry and Identification of Post- modifications. In detail, 10 g of total RNA were mixed with random translational Thiol-modifications—B. subtilis wild-type and ⌬bshA primers (Promega) and spike-ins (Two-Color RNA Spike-In Kit, Agilent mutant cells were harvested before (control conditions) and 15 min Technologies, Santa Clara, CA). The RNA/primer mixture was incu- after exposure to 50 M NaOCl stress. Cells were resuspended in bated at 70 °C for 10 min followed by 5 min incubation on ice. Then, urea/CHAPS alkylation buffer with 100 mM IAM as described above the following reagents were added: 10 l of 5x First Strand Buffer and sonicated to obtain the alkylated protein extracts. The alkylated (Invitrogen), 5 lof0.1M DTT (Invitrogen), 0.5 l of a dNTP mix (10 mM protein extracts were separated using 15% nonreducing SDS-PAGE dATP, dGTP, and dTTP, 2.5 mM dCTP), 1.25 l of Cy3-dCTP or (200 g each per lane) and the complete lanes were cut into 10 gel Cy5-dCTP (GE Healthcare) and 2 l of SuperScript II reverse tran- pieces and digested with trypsin as described above. Peptides eluted scriptase (Invitrogen). The reaction mixture was incubated at 42 °C for from tryptic digests of gel pieces were subjected to a reversed phase 60 min and then heated to 70 °C for 10 min. After 5 min on ice, the column chromatography (self packed C18 column, 100-mi.D.x200 RNA was degraded by incubation with 2 units of RNaseH (Invitrogen) mm) operated on a Easy-nLC II (Thermo Fisher Scientific, Waltham, at room temperature for 30 min. Labeled cDNA was then purified MA). Elution was performed by a binary gradient of buffer A (0.1% using the CyScribe GFX Purification Kit (GE Healthcare). The individ- (v/v) acetic acid) and B (99.9% (v/v) ACN, 0.1% (v/v) acetic acid) over ual samples were labeled with Cy5, whereas the reference pool was a period of 100 min with a flow rate of 300 nl/min. MS and MS/MS labeled with Cy3. 500 ng of Cy5-labeled cDNA and 500 ng of Cy3- data were acquired with the LTQ-Orbitrap-Velos mass spectrometer labeled cDNA were hybridized together to the microarray following (Thermo Fisher Scientific) equipped with a nanoelectrospray ion Agilent’s hybridization, washing and scanning protocol (Two-Color source. The Orbitrap Velos was operated in data-dependent MS/MS Microarray-based Gene Expression Analysis, version 5.5). Data were mode using the lock-mass option for real time recalibration. After a extracted and processed using the Feature Extraction software (ver- survey scan in the Orbitrap (r ϭ 30,000) MS/MS data were recorded sion 10.5, Agilent Technologies). For each gene on a microarray, the for the 20 most intensive precursor ions in the linear ion trap. Singly error-weighted average of the log ratio values of the individual probes charged ions were not taken into account for MS/MS analysis. was calculated using the Rosetta Resolver software (version 7.2.1, Post-translational modifications of proteins were identified by Rosetta Biosoftware). Genes showing induction or repression ratios searching all MS/MS spectra in “dta” format against an B. subtilis of at least threefold in three independent experiments were consid- target-decoy protein sequence database (8294 entries) using Sorcer- ered as significantly induced. The averages ratios and standard de- er™-SEQUEST® (Sequest version 2.7 rev. 11, Thermo Electron in- viations for all induced or repressed genes are calculated from three cluding Scaffold_3_00_02, Proteome Software Inc., Portland, OR). independent transcriptome experiments after 10 min of exposure to The target-decoy database includes the complete proteome set of B. NaOCl stress and listed in supplemental Table S1 and S2. All microar- subtilis 168 (4105 database entries) that was extracted from Uni- ray datasets are available in the GEO database under accession protKB release 12.7 (UniProt Consortium, Nucleic acids research numbers [GSE27637]. Hierarchical Clustering Analysis—Clustering of gene expression 2007, 35, D193–197) (39) and an appended set of 4147 reversed profiles was performed using Cluster 3.0 (44). The transcriptome data sequences and 42 sequences of common laboratory contaminants sets were derived from previous publications and this study and created by BioworksBrowser version 3.2 (Thermo Electron Corp.) included log2-fold expression changes 10 min after exposure of B. according to Elias et al. (40). The Sequest search was carried out subtilis to diamide (1 mM)(45), methylhydroquinone (MHQ) (0.33 mM) considering the following parameter: a parent ion mass tolerance 10 (46), catechol (2.4 mM) (47), formaldehyde (1 mM), methylglyoxal (MG) ppm, fragment ion mass tolerance of 1.00 Da. Up to two tryptic (2.8 and 5.6 mM)(23), and 50 M NaOCl. After hierarchical clustering, miscleavages were allowed. Methionine oxidation (ϩ15.994915 Da) the output was visualized using TreeView (48). For the clustering 630 and cysteine carbamidomethylation (ϩ57.021464 Da) were set as genes were selected that are induced by RES and NaOCl stress in B. variable modifications. Multiple Sequest searches were performed for subtilis (e.g. CtsR, CymR, Spx, PerR, ArsR, CsoR, CzrA, OhrR, YodB, either intramolecular disulfide bonds (Ϫ2.01565 Da), S-cysteinyla- YvaP (CatR), MhqR, LexA, SigmaD, AdhR, and HxlR regulons) tions (ϩ119.004099 Da for C H NO S) or S-bacillithiolations 3 7 2 (supplemental Table S4). (ϩ396.083866 Da for C H N O S) as variable post-translational 13 22 2 10 Northern Blot Experiments—Northern blot analyses were per- cysteine modifications, allowing a maximum of three modifications formed as described (49) using RNA isolated from B. subtilis wild-type per peptide in each Sequest search. cells before (control) and 10 min after treatment with 50 M NaOCl, 1 Proteins were identified by at least two peptides applying a strin- mM diamide and 100 M CHP, respectively. Hybridizations specific for gent SEQUEST filter. Sequest identifications required at least ⌬Cn ohrA, nfrA, cysK, katA, ohrA, azoR1, catE, and yitJ were performed scores of greater than 0.10 and XCorr scores of greater than 1.9, 2.2, with the digoxigenin-labeled RNA probes synthesized in vitro using T7 3.3, and 3.75 for singly, doubly, triply and quadruply charged pep- RNA polymerase from T7 promoter containing internal PCR products tides. The complete CID MS/MS spectra of the modified Cys-con- of the respective genes using the primer sets described previously taining peptides and the corresponding b and y fragment ion series (23, 25, 35) and the primer pairs yitJ-for, 5Ј CCGAACAGCAGTCTTC- are given in detail in supplemental Figs. S2 and S3. The Sequest CTTC 3Ј and yitJ-T7-rev, 5Ј CTAATACGACTCACTATAGGGAGAC- search results are submitted as “dta” and “out” files to the PRIDE CGTTTTACCGCTTTATCCA 3Ј for yitJ. database (http://www.ebi.ac.uk/pride/) (41) and deposited under the accession numbers 17516–17659. RESULTS Transcriptome Analysis—For microarray analysis, B. subtilis wild- type cells were grown in minimal medium to OD500 of 0.4 and har- Determination of the Growth-inhibitory Concentration of vested before and 10 min after exposure to 50 M NaOCl. Total RNA NaOCl in B. subtilis—At first we determined the concentration was isolated by the acid phenol method as described (42). For tran- that inhibited the growth of B. subtilis wild-type cells. Exposure scriptome analysis, 35 g RNA were DNase-treated using the RNase- Free DNase Set (Qiagen) and purified using the RNA Clean-Up and of cells to 50 M NaOCl stress caused a lag in growth for 60 min Concentration Micro Kit (Norgen). The quality of the RNA preparations and then growth was resumed with a similar growth rate as the was assessed by means of the Agilent 2100 Bioanalyzer according to untreated control (Fig. 1). This growth profile is very similar to
10.1074/mcp.M111.009506–4 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis that of diamide (26) indicating that cells are able to detoxify the tal Table S3 and visualized in the corresponding diagram in oxidant and to repair the protein damage within this time frame. Fig. 2. Transcriptional induction of selected thiol-stress spe- Growth of B. subtilis is completely inhibited by 100 M NaOCl. cific genes by NaOCl, diamide and CHP was verified by Transcriptome Analysis of the NaOCl Stress Response in B. additional Northern blot analyses (Fig. 3). In the following subtilis—To analyze the mode of action of NaOCl in B. subtilis, sections we have sorted the transcriptome datasets into thiol- we conducted genome-wide transcriptome analyses of cells and oxidative stress specific and general stress-induced treated for 10 min with 50 M NaOCl stress. Genes that were regulons. Ͼ3-fold up-regulated and down-regulated by NaOCl were Induction of the CtsR and Spx Regulons by NaOCl Stress is sorted according to the known stress regulons in sup- Indicative of Disulfide Stress—The microarray data showed plemental Tables S1 and S2. The transcriptome data revealed the up-regulation of the thiol- or electrophile-stress specific the significant induction of 430 genes and the repression of CtsR and Spx regulons in B. subtilis. Among the CtsR and Spx 400 genes by NaOCl stress in three biological transcriptome regulons, the clpE (28-fold), trxA and nfrA (40-fold) genes replicates. Representative genes of the most strongly up- displayed the highest induction ratios. The fold-changes of regulated genes and regulons are listed in supplemen- Spx-controlled genes are similar to diamide stress as verified by the Northern blots (Fig. 3). Because the Spx and CtsR regulons are generally induced by electrophiles, the global response to NaOCl is indicative of disulfide stress (23, 50). In contrast, the CymR regulon that functions in Cys biosynthesis (51) was strongly up-regulated by RES, but not by NaOCl stress. Among the CymR regulon genes, only cysK was 3-fold induced by NaOCl stress. This indicates that NaOCl probably does not deplete the pool of the redox buffer cysteine in B. subtilis. Induction of the S-box Regulon by NaOCl Stress Indicates Methionine Starvation—Interestingly, the S-box regulon genes that are regulated by an S-adenosyl methionine riboswitch mechanism were induced by NaOCl stress (supple- mental Table S1, Fig. 3). The S-box regulon is induced by methionine starvation when S-adenosyl methionine levels are FIG.1.Growth curves in response to sub-lethal concentrations of NaOCl stress in B. subtilis. B. subtilis wild type was grown in low. The S-box regulon gene products function in methionine biosynthesis (52–55). The microarray data showed the induc- minimal medium to an OD500 of 0.4 and exposed to 50, 75, and 100 M of NaOCl indicated by time point zero. tion of the methionine synthases encoding metE (5,3-fold),
FIG.2.Transcriptome changes of NaOCl-induced regulons (CtsR, Spx, PerR, OhrR, ArsR, CzrA, SigmaD, and SigmaB) and of the pstSCAB operon. Fold-changes are average induction ratios of genes induced in NaOCl-treated cells versus untreated cells calculated from three transcriptome replicates with standard deviations given as errors bars. Shown are only representative genes of each regulon that are more than fivefold induced by NaOCl in supplemental Tables S1 and S3.
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–5 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
structural Zn-binding site as confirmed by the liquid chroma- tography (LC)-MS/MS results (Table I). Induction of the OhrR-controlled OhrA Peroxiredoxin by NaOCl Stress—Interestingly, the ohrA gene was most strongly up-regulated (220-fold) in the transcriptome by NaOCl. The redox-sensing OhrR repressor controls the OhrA peroxire- doxin that confers resistance to CHP (6, 7). The Northern blots show that derepression of ohrA transcription occurs by CHP and NaOCl stress at similar levels (Fig. 3). These data suggest that OhrR is a specific determinant of the response to NaOCl and ROOH. Induction of Selective RES-specific MarR/DUF24 Regulons IG Northern blot analysis of selected thiol-stress specific F .3. by NaOCl Stress—We have shown that RES are sensed by genes of the Spx, CymR, S-box, PerR, OhrR, and MarR/DUF24 regulons in response to NaOCl, diamide, and CHP stress. Tran- members of the redox-sensing MarR/DUF24 family. The script analysis of nfrA, trxA, and katA indicates the induction of the YodB and CatR repressors are specific sensors for quinones Spx and PerR-regulons by NaOCl and diamide stress. The CymR- and diamide (7, 24–26, 50). The transcriptome and Northern controlled cysK gene is strongly induced by diamide. The S-box blot results showed the induction of the YodB-regulon genes regulon gene yitJ responds most strongly to NaOCl stress. The ohrA azoR1 (10-fold), yodB (3-fold), and yodC (5-fold) by NaOCl. gene is controlled by the OhrR repressor and strongly induced by NaOCl and CHP. The YodB and CatR-controlled azoR1 and catDE The CatR-controlled catDE operon was threefold induced by genes are strongly up-regulated by diamide. Cells were grown to an NaOCl. The inductions of the YodB and CatR regulons by
OD500 of 0.4 and harvested before (control conditions, co) and 10 min NaOCl are much lower as by diamide and quinones (Fig. 3) after exposure to 50 M NaOCl, 100 M CHP, or 1 mM diamide. The confirming the specific roles of the azoreductases and dioxy- RNA isolation and Northern blot hybridization was performed as genases in detoxification of diamide and quinones. In addi- described in the Methods section. The arrows point toward the sizes of the specific transcripts. tion, the genes encoding further DUF24-like redox sensors, yybR (35-fold), ykvN (19-fold), ydzF (6-fold), and ydeP (6-fold) yxjG (2,5-fold), and yxjH genes (2-fold), the metIC operon were induced by NaOCl stress indicating that these respond (3–4-fold) encoding cystathionine ␥-synthase and -lyase probably to disulfide stress via thiol-based redox switches. that are involved in cystathionine and homocysteine biosyn- The formaldehyde-sensing DUF24-type regulator HxlR con- thesis, the yoaD gene (3,6-fold) that encodes a 3-D-phospho- trols the hxlAB operon encoding the enzymes of the ribulose- glycerate dehydrogenase required for serine biosynthesis and 5-monophosphate (RuMP) pathway (58). The hxlAB operon the yitJ gene (7,4-fold) that encodes a bifunctional homocys- was 8-fold induced by NaOCl stress. teine S-methyltransferase in the N-terminal part and 5,10- Induction of Metal Ion Efflux Systems, Motility and Che- methylenetetrahydrofolate (N5,N10-THF) reductase in the motaxis by NaOCl Stress—Besides thiol- and oxidative stress C-terminal part (see Fig. 7). The NAD(P)H-dependent N5,N10- responses, the CzrA, ArsR, and CsoR regulons were strongly THF reductase activity is required to reduce N5,N10-THF to up-regulated by NaOCl stress that control the operons for 5-methyltetrahydrofolate (N5-THF) as methyl-group donor for metal ion efflux systems czcD-trkA (32-fold), arsR-yqcK- the methionine synthase MetE (56). The cysH operon that arsBC (57-fold), and yvgXYZ (3-fold) (59–61). Furthermore, encodes gene products for sulfur assimilation and cysteine the SigmaD regulon genes for flagella assembly, motility and biosynthesis and the metK gene encoding a S-adenosyl me- chemotaxis were strongly induced by NaOCl stress (23, 62). thionine synthetase were repressed in the transcriptome by Metal ion uptake systems and SigmaD regulon genes were NaOCl stress. The cysH operon belongs also to the CymR induced by NaOCl and RES. regulon and is rather induced by cysteine starvation than by Induction of the SigmaB-dependent General Stress Re- methionine starvation (57). The metK gene displayed also a sponse by NaOCl Stress—In contrast to electrophiles, NaOCl drop of expression at later time points upon methionine star- stress caused the 3–15-fold induction of 52 genes belonging vation in previous studies (52). Thus, metK and cysH are not to the SigmaB general stress response regulon. Among the induced by NaOCl stress in contrast to other S-box regulon SigmaB-controlled genes, 20 genes were relatively weakly genes listed above. induced (3-fold) and 32 genes in the range of 5–13-fold. Induction of the PerR-dependent Oxidative Stress Re- Representative genes of the SigmaB regulon include dps sponse by NaOCl Stress—NaOCl caused the induction of the (5-fold), gsiB (8-fold), gspA (13-fold), ydaST (6-fold), yfhK (7- oxidative stress specific PerR regulon genes katA (73-fold) fold), yfkM (6-fold), yflT (13-fold) ohrB (8-fold), ysnF (7-fold), and ahpCF (7–13-fold) (Figs. 2 and 3). The PerR regulon genes and ytxGHJ (6–13-fold). The induction of the SigmaB re- were similar strongly induced by NaOCl and diamide indicat- sponse might be caused as a result of starvation or even ing that strong oxidants that cause disulfide stress lead oxidative stress response caused by hypochloric acid which probably to oxidation of the conserved Cys residues in the requires further studies. It is interesting to note that NaOCl
10.1074/mcp.M111.009506–6 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis triggers the induction of thiol- and oxidative stress responses Thiol-redox Proteomics Identifies GapA, MetE, MtnA, LeuC as well as general stress responses. and PurQ as Reversibly Oxidized Proteins by NaOCl Stress— Other Genes Induced Strongly by NaOCl Stress—Finally, Next, we were interested in the changes in the thiol-redox many genes with unknown functions are up-regulated in the proteome by NaOCl stress to identify proteins with reversible NaOCl transcriptome (supplemental Table S1). The putative thiol modifications. In brief, reduced thiol groups were Naϩ/Hϩ antiporter nhaX (30-fold) and the major facilitator blocked with IAM and reversibly oxidized proteins were re- superfamily encoding ybcLM operon (42-fold) could be in- duced and labeled using the fluorescence dye BODIPY FL volved in sodium efflux. The high induction of the pstSCAB C1-IA (21, 38). The fluorescence-labeled proteins representing operon (30-fold) by NaOCl might be PhoPR-independent be- the disulfide proteome were separated using two-dimensional cause other PhoPR-regulon genes were not induced, such as gels and the fluorescence image (red) overlaid with the Coo- phoA, phoB, and phoD encoding alkaline phosphatases and massie-stained protein amount image (green) (Fig. 5A,5B). phosphodiesterases. Finally, the yhdJ gene encoding the pu- The oxidation ratios were quantified as BODIPY-fluorescence tative GCN5-N-acetyltransferase was 67-fold induced by levels versus protein amount levels of the reversibly oxidized NaOCl. proteins (Fig. 5C). In agreement with previous studies, pro- Repression of the Stringent Controlled RelA-regulon by teins that form disulfides under control conditions include NaOCl Stress—There are 400 genes in the transcriptome Adk, AhpC, AccB, Eno, Tpx, PdhD, GuaB, CysH, CysJ (YvgR), datasets that displayed more than 3-fold decreased expres- LeuC, YceC, and YumC (Fig. 5A)(21, 22). The identities of sion ratios in response to NaOCl stress (supplemen- these oxidized proteins were verified using MALDI-TOF-TOF tal Table S2). Among the repressed genes are the stringent MS/MS (supplemental Figs. S1A–P). Surprisingly, we did not detect a strong increase of revers- controlled RelA regulon genes involved in translation, ATP ible thiol-oxidations in the redox proteome in many cytoplas- generation, cell wall biosynthesis, and turnover. The PurR and mic proteins after treatment of cells with 50 M NaOCl stress. RyrR regulons controlling purine and pyrimidine biosynthesis Instead, NaOCl rather caused specific oxidation of few selec- were repressed by NaOCl as result of the slower growth rate. tive proteins, including GapA, MetE, MtnA, LeuC, and PurQ The osmostress-responsive uptake systems for choline and (Fig. 5B,5C). The glyceraldehyde-3-phosphate dehydrogen- glycine betaine encoded by the opuA, opuB, and opuC oper- ase (GapA) was identified as most strongly oxidized protein ons were strongly repressed by NaOCl stress (63). with a 10-fold increased fluorescence/protein ratio (Fig. 5B, Hierarchical Clustering Analyses for RES and NaOCl Indi- 5C). The redox proteome analysis revealed also strongly in- cates a Disulfide Stress Signature of NaOCl—Next, we per- creased oxidation ratios for the cobalamin-independent me- formed a hierarchical clustering analysis using selected tran- thionine synthase MetE. Furthermore, the methylthioribose-1- scriptome datasets for diamide (45), MHQ (46–47), catechol phosphate isomerase MtnA (YkrS) is oxidized by NaOCl (47), formaldehyde, methylglyoxal (23), and NaOCl. In total, stress that is involved in methionine biosynthesis via the sal- 630 genes were selected for the clustering analysis that vage pathway (see Fig. 7). showed inductions of at least threefold in any of these tran- Identification of Proteins with Intramolecular Disulfides Un- scriptome datasets (supplemental Table S4). The complete der Control and NaOCl Stress Conditions Using Shotgun-LC- cluster can be arranged in 14 major groups of genes which MS/MS Analyses—We were interested to identify proteins share a similar expression profile by electrophiles and NaOCl with reversible thiol-oxidations, including intramolecular (Fig. 4). The cluster analysis confirmed the common induction disulfides, S-cysteinylations, and S-bacillithiolations under of the CtsR, Spx (node 9), ArsR, CzrA, and SigmaD regulons NaOCl stress conditions. We are aware that we exclude the (nodes 11, 12) by RES and NaOCl as indicator for disulfide identifications of intramolecular disulfides of proteins with stress. The strong induction of the CymR regulon by RES but Cys residues that are separated by tryptic digestion result- not by NaOCl is visualized in cluster 13. The quinone and ing in cross-linked intermolecular disulfides after enzymatic diamide-specific MarR/DUF24 regulons controlled by YodB, digestion. Alkylated protein extracts of control and NaOCl- CatR, and MhqR clustered with the PerR regulon in nodes 6, treated cells were separated using nonreducing SDS-PAGE, 8, and 10. The aldehyde-specific HxlR, AdhR, LexA regulons tryptically in-gel digested and analyzed using LTQ-Orbitrap- and the formate uptake system and formate dehydrogenase Velos mass spectrometry as described in the Methods encoding yrhFG and yrhDE operons (23) clustered in nodes 11 section. and 14. The NaOCl-specific inductions of the OhrR and Eight proteins with intramolecular disulfides were identified SigmaB regulons, the ydzF and yfkN genes encoding DUF24- in wild-type proteome samples at control and NaOCl stress family regulators and the pstSCAB operon are shown in clus- conditions, including Adk, CysJ (YvgR), GltA, PdhD, TopA, ters 1, 3, 4. In conclusion, the transcriptome comparisons YutI, YdjI, and Zwf (Table I and supplemental Figs. S2A–2K). showed that NaOCl elicits a Spx, CtsR-, and PerR-controlled Adk, PdhD, and CysJ were detected also in the redox pro- disulfide and oxidative stress response and a selective OhrR- teome as reversibly oxidized proteins under control condi- and SigmaB response in B. subtilis. tions (Fig. 5A). The dihydrolipoamide dehydrogenase E3 sub-
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–7 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
FIG.4.Hierarchical clustering analysis of RES and NaOCl induced gene expression profiles. Log2-fold changes of gene expression ratios were clustered for 1 mM diamide (Dia), 2.4 mM catechol (Cat), 0.5 mM methylhydroquinone (MHQ), 1 mM formaldehyde, 2.8 mM and 5.6 mM methylglyoxal (MG-2 and MG-5), and 50 M NaOCl stress using the Treeview software. The cluster analysis resulted in 14 different nodes that are enriched for RES and NaOCl-induced regulons (CtsR, Spx, ArsR, CzrA, CsoR, SigmaD), the RES-specific CymR regulon, quinone- responsive regulons (PerR, YodB, CatR (YvaP), MhqR), aldehyde-responsive regulons (HxlR, AdhR, LexA) and NaOCl-specific regulons (SigmaB, OhrR). For the cluster analysis 630 genes were selected as listed in supplemental Table S4 that are induced by RES and NaOCl stress. Red indicates induction and green repression under the stress specific conditions. unit of the pyruvate dehydrogenase complex (PdhD) forms an redox proteome (Fig. 5A) nor during tryptic digestion in control active site Cys47-Cys52 intramolecular disulfide used for oxi- extracts indicating that no oxidation occurred during our sam- dation of the dihydrolipoamide to lipoamide (64). The adenylate ple preparation protocol. The intramolecular C152-C156 disul- kinase Adk and the topoisomerase 1 (TopA) form intramolecular fide bond in GapA‘s catalytic center was detected as triply disulfides in Zn binding sites (Table I). Further proteins with charged peptide YDAANHDVISNASC152(-2)TTNC156LAPFAK intramolecular disulfides include those with Fe-S clusters or at an m/z ϭ 842,0466 (supplemental Fig. S2B). Fe-binding sites, such as the glutamate synthase large subunit The intramolecular disulfides for Spx and PerR were iden- GltA (GOGAT), the sulfite reductase flavoprotein alpha subunit tified only in cell extracts of NaOCl-treated bshA mutant cells. CysJ (YvgR) and the Fe-S cluster assembly protein YutI. The redox-sensing regulator Spx is activated by a thiol-
Three redox-sensitive proteins were identified that are oxi- disulfide redox switch in the N-terminal C10xxC13 motif in dized to intramolecular disulfides only under NaOCl stress response to diamide stress (65, 66). This C10-C13 intramo- conditions, including GapA, PerR, and Spx (Table I). GapA is lecular disulfide was identified as doubly charged peptide ϩ ϭ most strongly oxidized in response to NaOCl stress in the M1( 16)VTLYTSPSC10(-2)TSC13Ratanm/z 781,8378 in redox proteome (Fig. 5B). GapA was neither oxidized in the NaOCl-treated cells (supplemental Fig. S2F). The peroxide-
10.1074/mcp.M111.009506–8 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–9 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis sensing PerR repressor is inactivated in response to hydrogen phosphoglycerate dehydrogenase SerA (supplemental Fig. peroxide by iron-catalyzed His-oxidation (14). The intramolec- S3E, Table II) and the thiol-disulfide oxidoreductase YphP ular disulfide was detected in the structural Zn-binding site of (supplemental Fig. S3F, Table II). The MetE protein was iden-
PerR as doubly charged peptide LEIYGVC136QEC139(-2)SK at tified as specific target for reversible thiol-oxidation by NaOCl an m/z ϭ 685,3092 (supplemental Fig. S2E). This suggests stress in the redox proteome (Fig. 5B,5C). Moreover, MetE, that PerR senses disulfide stress probably via oxidation of the PpaC and SerA have been shown to be modified by S- Cys residues in the Zn binding site. cysteinylations in response to diamide stress (67). Identification of S-bacillithiolations in OhrR, MetE, YxjG, For MetE we detected the triply charged peptide VPS- PpaC, SerA, and YphP in the Proteome of NaOCl-treated ϩ ϭ TEEMYNIIVDALAVC719( BSH)PTDR at an m/z 944.7604 Cells Using Shotgun-LC-MS/MS Analysis—The OhrR repres- ϩ and the doubly charged peptide FWVNPDC730( BSH)- sor is inactivated by an S-thiolation mechanism including GLK at an m/z ϭ 787,8297 containing the additional mass of cysteine and BSH in response to CHP (7, 19). Our data BSH (ϩ396 Da) at Cys719 and Cys730 (Fig. 5F,5G; showed that NaOCl stress also inactivates the OhrR repressor supplemental Fig. S3B, Table II). In addition, the triply charged because ohrA transcription was derepressed. Thus, we ana- Cys730 peptide with an S-cysteinylation modification was lyzed whether OhrR forms mixed disulfides with cysteine or identified as peptide FWVNPDC (ϩCys)GLK at an m/z ϭ BSH in response to NaOCl. Because OhrR is a low abundant 730 433,1952 containing the additional mass of Cys (ϩ119 Da) regulatory protein, a FLAG-epitope-tagged OhrR protein was (supplemental Fig. S3B; Table II). Interestingly, the CID enriched using immunoprecipitation (IP) from NaOCl-treated MS/MS spectra of the bacillithiolated MetE peptides revealed cells as in previous studies (19). Purified OhrR-FLAG protein a characteristic loss of malate from the precursor ions and was tryptically digested and analyzed by LTQ-Orbitrap MS/MS from abundant y-fragment ions as indicated by the loss of 134 analysis as described in the Methods section. In protein sam- Da. The precursor ions minus 134 Da are most abundant in ples containing OhrR-FLAG protein from NaOCl-treated cells, the MS/MS spectra of the bacillithiolated peptides identified the Cys15-peptide was identified as triply charged peptide at an for MetE, PpaC and SerA. Thus, the bacillithiolated and cys- m/z ϭ 684.98 containing the additional mass of BSH (ϩ396 Da) teinylated Cys730 peptides of MetE showed different MS/MS (supplemental Fig. S3A, Table II). This indicates that OhrR is inactivated by an S-bacillithiolation mechanism in response to fragment ion spectra (supplemental Fig. S3B). Moreover, the NaOCl stress. To verify that only Cys15 of OhrR is involved in abundance of the precursor ions minus malate leads also to redox-sensing of NaOCl stress, we analyzed ohrA transcription decreased intensities of the b and y ions in the MS/MS spec- in an ohrRC15S mutant (34). The Northern blot results in tra of the bacillithiolated PpaC and SerA peptides supplemental Fig. S3A indicate that ohrA transcription is com- (supplemental Fig. S3D, 3E). pletely abolished in the ohrRC15S mutant. This confirms that For YxjG, we detected the triply charged peptide YVSLD- ϩ ϩ ϭ regulation of ohrA transcription requires only inactivation of the QLC341( IAM)LSPQC346( BSH)GFASTEEGNK at an m/z redox-sensing Cys15 of OhrR. 981,4183 with the additional mass of 396 Da at Cys346 Next, we searched our LC-MS/MS results obtained from (supplemental Fig. S3C, Table II). YxjG is homolog to the whole proteome tryptic digests of NaOCl-treated cells for C-terminal part of MetE and the essential Zn-binding Cys730 S-thiolation modifications. We identified five cytoplasmic pro- residue of MetE aligns with Cys346 in YxjG. This suggests teins that were modified by S-bacillithiolation, including two that also Cys346 of YxjG could be essential for methionine methionine synthase paralogs MetE and YxjG (Fig. 5F,5G, synthase activity. Thus, homolog active site Cys residues are supplemental Fig. S3B, 3C, Table II), the inorganic pyrophos- S-bacillithiolated in the paralogous methionine synthases phatase PpaC (supplemental Fig. S3D, Table II), the 3-D- MetE and YxjG.
FIG.5.ABC. The thiol-redox proteome (red) in comparison to the protein amount image (green) at control conditions (A) and after exposure to 50 M NaOCl (B) in the wild type. Reduced protein thiols in cell extracts were alkylated with IAM followed by reduction of oxidized protein thiols with TCEP and labeling with BODIPY FL C1-IA. Proteins with reversible thiol-modifications in the control and NaOCl redox proteome are labeled in white and newly oxidized proteins in NaOCl-treated cells are labeled in red. The oxidized proteins were identified by MALDI-TOF-TOF MS/MS as shown in detail in supplemental Fig. S1A–P. C, The fluorescence/protein amount ratios are quantified as oxidation ratios at control conditions (control) and 10, 20, and 30 min after exposure to 50 M NaOCl stress as shown in the diagram. DEFG. Close-ups of the main NaOCl-sensitive proteins MetE and GapA in the thiol-redox proteome of the wild type (D) and bshA mutant (E) and CID MS/MS spectra of the S-bacillithiolated Cys719 and Cys730 peptides of MetE (F, G). Figs. D and E show sections of reversibly oxidized proteins after NaOCl stress in the thiol-redox proteome (red) in comparison to the protein amount image (green) at control conditions (co) and 10, 20, and 30 min after exposure to 50 M NaOCl. Figs. F and G show the CID MS/MS spectra of the S-bacillithiolated Cys719-and Cys730-peptides of MetE identified in the wild-type proteome using LTQ-Orbitrap-Velos mass spectrometry as described in the Methods section. The MS/MS spectra show the characteristic abundant precursor ions that have lost malate indicated by parent-134 ⌬ (HOOC-CH2-CHOH-COOH). The Xcorr and Cn scores and peptide masses are listed in Table II and the corresponding b and y fragment ion series for the modified peptides are given in detail in supplemental Fig. S3B.
10.1074/mcp.M111.009506–10 Molecular & Cellular Proteomics 10.11 oeua ellrPoemc 10.11 Proteomics Cellular & Molecular
TABLE I Proteins with intramolecular disulfides identified at control and NaOCl stress conditions. Cytoplasmic proteins were harvested in IAM-urea/thiourea-buffer, separated using SDS-PAGE, tryptically digested and analyzed in a LTQ Orbitrap-Velos mass spectrometer as described in the Methods section. Peptides with intramolecular disulfides that showed a mass difference of Cys-2 Da were identified for Adk, GltA, CysJ, PdhD, TopA, YdjI, YutI, and Zwf in control and NaOCl samples. The table includes the m/z of the precursor ions, charges, Xcorr and ⌬Cn scores of the disulfide linked peptides. The complete CID MS/MS spectrum of all peptides and the b and y fragment ion series are given in supplemental Fig. S2A–K as indicated in column SI-Fig.
m/z precursor ⌬Cn SI-Fig. Protein Function Redox-active Cys Disulfide-Peptide ion XCorr-score score Charge
b S2A Adk Adenylate kinase Zn-binding: Cys130,133, IC130 SVC133 (-2)GTTYHLVFNPPK 938,9590 4,6515 0,6794 2 150,153 S2B GapAa,b Glyceraldehyde 3-phosphate Cys152 redox-active catalytic YDAANHDVISNASC (-2) 842,0466 5,8694 0,7600 3 152 in Resistance Hypochlorite Confers S-Bacillithiolation dehydrogenase site TTNC156 LAPFAK S2C GltA Glutamate synthase large subunit 3Fe-4S: C113, 1119, 1124 AC1119 (-2) 674,6567 5,2172 0,7056 3 HLDTC1124 PVGVATQNPELR b S2D PdhD dihydrolipoamide dehydrogenase Cys47 and Cys52 redox-active ATLGGVC47 (-2)LNVGC52 IPSK 765,3950 3,6256 0,5999 2 E3 subunit of pyruvate disulfide as catalytic centre dehydrogenase a S2E PerR Fur-family repressor of the Zn-binding site:Cys96,99, LEIYGVC136 QEC139 (-2)SK 685,3092 2,6948 0,7523 2 peroxide regulon 136,139 a S2F Spx Regulator of the disulfide stress Cys10, 13 redox-active sites M1(ϩ16)VTLYTSPSC10 (-2)TSC13 R 781,8378 2,9770 0,7398 2 response S2G TopA DNA topoisomerase 1 3 Zn-fingers: Cys579,582, 599,605 Cys619,622,641,647 C619 PSC622 (-2)GEGNIVER 631,2697 2,3857 0,7171 2 Cys660,663,680,683 S2H YdjI Unknown Unknown TGEANFC315 PNC318 (-2)GQK 683,7802 2,8661 0,6388 2 S2I YutI Putative nitrogen fixation protein, Cys79 and Cys82 Fe-S-cluster LLGAC79 GSC82 (-2)PSSTITLK 774,8928 4,3033 0,6283 2 Fe-S cluster assembly or repair binding b S2J YvgR (CysJ) Sulfite reductase ͓NADPH͔ Cys427 and Cys431 probably GVC427 (-2)SILC431 AER 524,7494 2,1040 0,5121 2 flavoprotein alpha-component Fe-binding
10.1074/mcp.M111.009506–11 S2K Zwf Glucose-6-phosphate Unknown LDYC407 (-2)SNC410 NDELNTPEAYEK 1109,9501 4,8708 0,8275 2 dehydrogenase a The intramolecular disulfide of GapA was detected in NaOCl-treated wild-type and bsh mutant cells and the disulfides for PerR and Spx were detected in NaOCl-treated bsh mutant cells. b Adk, GapA, PdhD, and YvgR (CysJ) were also identified in the redox proteome as reversible oxidized proteins (Fig. 5A,5B). .subtilis B. 10.1074/mcp.M111.009506–12 in Resistance Hypochlorite Confers S-Bacillithiolation
TABLE II Proteins with S-Bacillithiolations and S-Cysteinylations identified in NaOCl-treated cells. Cytoplasmic proteins of the wild type were harvested in IAM-buffer, separated using SDS-PAGE, tryptically digested and analyzed in a LTQ Orbitrap-Velos™ mass spectrometer as described in the Methods section. Peptides with S-bacillithiolations were identied for OhrR, MetE, YxjG, PpaC, SerA, and YphP indicated by the additional mass of 396 Da. Peptides with S-cysteinylations were identified for MetE and PpaC with a mass difference of 119 Da. The table includes the m/z of the precursor ions, charges, Xcorr, and ⌬Cn scores of the S-thiolated peptides. The complete CID MS/MS spectrum of all peptides and the b and y fragment ion series are given in supplemental Fig. S3A–F as indicated in column SI-Fig. m/z ⌬Cn SI-Fig. Protein Function Redox-active Cys S-thiolated Peptide XCorr-score Charge precursor ion score
S3A OhrR Redox-sensing MarR-type Cys15 redox-sensing LENQLC15 (ϩBSH)FLLYASSR 684,9800 2,9444 0,5816 3 repressor for organic hydroperoxide resistance a S3B MetE Cobalamin-independent Cys 647, 730 essential, Zn-binding VPSTEEMYNIIVDALAVC719 (ϩBSH)PTDR 944,7604 4,9737 0,7545 3 methionine synthase active site in B. sub. Cys719 non-essential, FWVNPDC730 (ϩBSH)GLK 787,8297 2,8014 0,6097 2 S-cysteinylated by diamide in
B. sub. subtilis B. Cys645 non-essential, FWVNPDC730 (ϩCys)GLK 433,1952 3,2330 0,5435 3 S-glutathionylated in E. coli S3C YxjG Cobalamin-independent Cys251, 346 homolog to Zn- YVSLDQLC341 (ϩIAM)LSPQC346 - 981,4173 4,9686 0,7521 3 methionine synthase binding Cys647, 730 in MetE of (ϩBSH)GFASTEEGNK B. sub. oeua ellrPoemc 10.11 Proteomics Cellular & Molecular S3D PpaC inorganic pyrophosphatase Cys158 SPTC158 (ϩBSH)TDQDVAAAK 851,8456 3,4552 0,8838 2 S-cysteinylated by diamide in SPTC158 (ϩCys)TDQDVAAAK 713,3041 3,2977 0,8356 2 B. sub. S3E SerA 3-D-phosphoglycerate Cys410 S-cysteinylated by ISSSESGYDNC410 (ϩBSH)ISVK 992,9086 3,6448 0,8507 2 dehydrogenase diamide in B. sub. S3F YphP Thiol-disulfide isomerase Cys53 redox-active AEGTTLVVVNSVC53 (ϩBSH)GC55 - 815,3752 4,077 0,6036 3 (ϩIAM)AAGLAR a MetE was also identified as reversibly oxidized in the redox proteome after NaOCl stress. The identification of the essential Zn-binding ligands Cys647 and 730 of B. subtilis MetE is based on similarity to the E. coli MetE enzyme and derived from the UniprotKB database entry P80877. S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
The inorganic pyrophosphatase PpaC was S-bacillithio- that the OhrA peroxiredoxin and the BSH redox buffer play lated at Cys158 and the same site was also S-cysteinylated most essential roles in NaOCl detoxification. by diamide (67). For PpaC the Cys158-containing doubly The Growth Defect in Response to NaOCl Stress can be ϩ charged peptide SPTC158( BSH)TDQDVAAAK was detected Attributed to Methionine Limitation—The methionine synthase at an m/z ϭ 851,8456 Da with the additional mass of 396 Da MetE catalyzes the final step in the methionine biosynthesis, (supplemental Fig. S3D, Table II). We also detected the doubly the chemically difficult methyl group transfer of methyltetra- charged S-cysteinylated Cys158-containing peptide of PpaC hydrofolate (N5-THF) to homocysteine (70) (Fig. 7). Diamide ϩ ϭ SPTC158( Cys)TDQDVAAAK at an m/z 713,3041 indicated stress leads to methionine auxotrophy that is caused by MetE by the additional mass of 119 Da (supplemental Fig. S3D, inactivation via S-cysteinylation in B. subtilis and S-glutathio- Table II). Thus, for MetE and PpaC S-bacillithiolations and nylation in E. coli (38, 67, 70, 71). S-bacillithiolation of the S-cysteinylations were observed. active site Cys residues of MetE and YxjG most likely leads to The phosphoglycerate dehydrogenase SerA was S-bacilli- enzyme inactivation and methionine auxotrophy by NaOCl thiolated at Cys410 in response to NaOCl stress and the same stress. This methionine starvation phenotype was supported site was S-cysteinylated by diamide stress (67). The S-bacil- by the induction of the S-box regulon genes in the transcrip- ϩ lithiolated Cys410-peptide ISSSESGYDNC410( BSH)ISVK tome (Fig. 7, supplemental Table S1). Using Northern blot was detected as doubly charged peptide at an m/z ϭ analysis the transcriptional induction of the S-box regulon 992,9086 with the additional mass of 396 Da (supple- gene yitJ was verified during the time of starvation after ex- mental Fig. S3E, Table II). posure to 50 M NaOCl (Fig. 7). Interestingly, the thiol-disulfide isomerase YphP (68) was Furthermore, growth experiments were performed in the identified as S-bacillithiolated at the redox-active Cys53 presence of extracellular methionine to confirm the methio- residue. The triply charged peptide AEGTTLVVVNSVC53- nine starvation phenotype. Cells were treated with of 75 l ϩ ϩ ϭ ( BSH)GC55( IAM)AAGLAR was observed at an m/z NaOCl and 30 or 60 min later 75 M extracellular methionine 815,3752 with the additional mass of 396 Da at Cys53 and the was added. The growth was resumed immediately after me- carbamidomethylation at Cys55 (supplemental Fig. S3F, thionine addition (Fig. 8A). To control that the added methio- Table II). nine does not simply remove the remaining oxidants from the The OhrA Peroxiredoxin and the BSH Redox Buffer are supernatant, the NaOCl concentrations in the culture super- Essential for Protection Against NaOCl Stress—Next, we an- natants were monitored using the FOX-assay. The results alyzed if the growth of ohrA and ohrR mutant strains is af- showed that 66% of NaOCl is consumed and detoxified by fected by NaOCl treatment. The growth of an ohrA mutant the cells after 30 min and 93% after 60 min supporting that the was strongly impaired by 50 M NaOCl compared with that of added methionine abolished the methionine starvation phe- the wild type whereas the ohrR mutant was more resistant to notype (Fig. 8B). In another experiment, cells were inoculated 75 M NaOCl (Fig. 6). Thus, the ohrA encoded peroxiredoxin in minimal medium containing 75 M methionine, grown to an is a specific determinant of NaOCl resistance in B. subtilis and OD500 of 0.4 and treated with 75 M NaOCl. Cells were able to perhaps involved in detoxification of this strong oxidant. grow without any lag phase in the methionine-supplemented Next, we monitored the growth of bshA and bshB1B2 mu- medium after treatment with 75 mM NaOCl (Fig. 8C). This tants with defects in the BSH biosynthesis enzymes (35). The indicates that NaOCl stress causes methionine auxotrophy growth of NaOCl-treated bshA and bshB1B2 mutants was that can be abolished by methionine addition. strongly impaired compared with the wild type (Fig. 6). While B. subtilis contains about 1 mol/g BSH and 0.58 mol/g the growth of the wild type was resumed 60 min after treat- Cys as redox buffers (35). To analyze whether Cys can replace ment with 50 M NaOCl, the bsh mutants required 180 min to BSH in post-translational modification and MetE inactivation, resume growth. This NaOCl-sensitive phenotype of bsh mu- bshA mutant cells were exposed to 60 M NaOCl and 30 and tants points to a major role of BSH in NaOCl detoxification 60 min later 60 M methionine was added. Methionine was similar as glutathione (GSH) in E. coli (69). able to restore also the growth of bshA mutant cells indicating Finally, we investigated whether the PerR, Spx and SigmaB that the growth defect of bshA mutants could be attributed to regulons confer protection against NaOCl. However, the perR methionine auxotrophy (supplemental Fig. S4A). The tran- mutant that overproduces the catalase KatA and the alkylhy- scriptional induction of the S-box regulon gene yitJ in the droperoxide reductase AhpCF was not resistant to NaOCl bshA mutant by 50 M NaOCl stress supports the methionine stress (data not shown). The spx mutant displayed increased starvation phenotype (Fig. 7). In addition, the redox proteome sensitivities toward NaOCl stress, but the growth was also of NaOCl-treated bshA mutant cells revealed similar MetE impaired under non-stress conditions due to the pleiotropic oxidation ratios as observed for the wild type (Fig. 5E). Using phenotype of the spx mutant (Fig. 6). The growth of the sigB the shotgun LC-MS/MS approach we identified the S-cystei- mutant was only slightly impaired by NaOCl indicating per- nylated Cys730-peptide of MetE in the bshA mutant proteome haps a role of the SigmaB-controlled OhrA-paralog OhrB in (supplemental Fig. S5). These results indicate that Cys can NaOCl detoxification. These results lead to the conclusion replace BSH for post-translational modifications and inactiva-
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–13 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
FIG.6. The OhrA peroxiredoxin and BSH protect cells against NaOCl toxicity. Growth phenotype of B. subtilis wild type (wt) in comparison to the ⌬ohrR (A), ⌬ohrA (B), ⌬bshA (C), ⌬bshB1,B2 (D), ⌬sigB (E), and ⌬spx (F) mutant strains that were treated with 50 or 75 M
NaOCl at an OD500 of 0.4. tion of MetE. However, the bshA mutant was strongly impaired also in host tissue damage and diseases. The redox proteome in NaOCl detoxification and consumption compared with the in response to NaOCl has been recently studied in E. coli wild type because about 50% of NaOCl was left in the medium using the OxICAT approach and several redox-sensitive pro- after 60 min in the bshA mutant (supplemental Fig. teins could be identified that are sensitive to NaOCl-directed S4B). This indicates that BSH is more efficient in NaOCl detox- reversible oxidation (28, 29). In B. subtilis, only few NaOCl- ification than Cys. sensitive proteins were identified in the 2D gel-based thiol- redox proteome. One reason for the observed differences in DISCUSSION the redox proteome analyses between E. coli and B. subtilis Hypochloric acid is widely used as disinfectant and pro- could be that only abundant cytoplasmic proteins with revers- duced in activated macrophages by the enzyme myeloperoxi- ible thiol-modifications are visualized by the 2D gel-based dase as first defense line upon bacterial infections. Hypochlo- approach (e.g. MetE and GapA) and that the gel-free OxICAT ric acid is beneficial by killing invading bacteria but can result approach is much more sensitive for identification of oxidized
10.1074/mcp.M111.009506–14 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
FIG.7.Pathways for sulfur assimilation, cysteine and methionine biosynthesis, and induction of S-box regulon genes in response to NaOCl stress. The S-box regulon gene products that are induced in the transcriptome as a result of methionine starvation provoked by the S-bacillithiolation of MetE and YxjG are bold-faced in gray. The transcriptome induction ratios are shown below the gene products. The gene products of the sulfate assimilation pathway (Sat, CysC, CysH, CysJI, CysE) are not induced upon NaOCl stress (except for CysK). The inset shows the transcriptional induction of yitJ in the wild type and ⌬bshA mutant using Northern blot analysis before (co) and at different times (10, 30, 60, 90, and 120 min) after exposure to 50 M NaOCl stress. Abbreviations: APS, adenosine-5Ј-phosphosulfate; PAPS, 3Ј-phosphoad- enosine-5Ј-phosphosulfate; OAS, O-acetylserine; N5,N10-THF, 5,10-methylenetetrahydrofolate; N5 -THF, 5-methyltetrahydrofolate; THF, tetrahydrofolate. This figure was adapted from Tomsic et al., 2008 (52). proteins of lower abundance. However, previous 2Dgel-based GapA protein. This confirms previous results in E. coli using redox proteomics analyses in B. subtilis have shown that only the OxICAT approach suggesting that GapA is oxidized to an toxic peroxide concentrations cause increased reversible intramolacular disulfide by NaOCl stress (28). GapDH activity thiol-oxidations and that only few proteins are oxidized upon was also inactivated by NaOCl in endothelial cells accompa- paraquat stress (38). Moreover, no reversible thiol-modifica- nied by thiol depeletion (74). GapA has been shown as redox- tions could be monitored using the 2D gel-based approach in controlled cytoplasmic enzyme in prokaryotes and yeast and response to nitric oxide (NO) in B. subtilis (72). In contrast is susceptible also for inactivation by S-sulfenylation, S-sul- again, several targets for reversible thiol-modifications were fonylation, S-glutathionylation, S-nitrosylation, and intermo- identified by NO stress in E. coli by the 2D gel-based ap- lecular aggregation in response to ROS and RNS (13). GapA proach (73). Thus, there might be differences in the extents of is also strongly oxidized by diamide and quinones in B. subtilis reversible thiol-modifiations in response to ROS and RNS (21–22). The reversible inactivation of GapA‘s glycolytic activ- between E. coli and B. subtilis. To address this point, more ity in B. subtilis by the oxidative mode of quinones has been sensitive gel-free quantifications of reversibly oxidized pro- shown in previous studies (21). Inactivation of GapA leads teins using the OxICAT approach are required in the future to probably to re-routing of glucose-6-phosphate to the pentose quantify the changes in the redox proteome of B. subtilis more phosphate pathway. The function of the pentose phosphate comprehensively. Nevertheless, we show here that NaOCl pathway is to increase NADPH levels and provide reducing leads to S-bacillithiolation of selective proteins in B. subtilis equivalents for the thioredoxin/thioredoxin reductase system that play essential roles in protection against hypochloric acid to restore the redox balance of the cell (13). toxicity. S-bacillithiolation Protects Metabolic Enzymes Against Irre- GapA is Oxidized to an Intramolecular Disulfide Upon versible Overoxidation—We further demonstrate that six cy- NaOCl Stress—One of the strongest targets that forms intra- toplasmic proteins undergo reversible S-bacillithiolations by molecular disulfides upon NaOCl stress was identified as the NaOCl treatment. Protein S-thiolation is an important post-
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–15 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
FIG.8.NaOCl stress causes methionine auxotrophy that is abolished by methionine addition. Growth phenotypes of the B. subtilis wild type treated with 75 M NaOCl at an OD500 of 0.4. A, Methionine was added 30 or 60 min after exposure to 75 M NaOCl stress and the growth was resumed. B, The concentrations of consumed NaOCl of wild-type cells were monitored in the culture supernatants after exposure to 75 M NaOCl using the FOX assay. The NaOCl concentrations are given as mean values of three independent experiments with error bars. C, Methionine (75 M) was added after inoculation to the culture at an OD500 of 0,07 and 75 M NaOCl was added when cells had reached an OD500 of 0.4.
translational thiol-modification that occurs in response to inactivated by S-bacillithiolation after CHP challenge (19). oxidative stress in eukaryotic and prokaryotic cells and Here we found that S-bacillithiolation in response to NaOCl controls metabolic processes and redox-sensing transcrip- controls OhrR activity resulting in expression of the OhrA tion factors to counteract the cellular damage. In eu- peroxiredoxin as protection mechanism (Fig. 9). This ex- karyotes, S-glutathionylation is implicated in disease mech- pands to role of the thiol-dependent OhrA-like peroxiredox- anisms, including cell-signaling pathways associated with ins in detoxification of hypochloric acid. OhrA-like peroxire- viral infections and with tumor necrosis factor ␣-induced doxins catalyze the reduction of ROOH to their apoptosis (75). In E. coli the activity of the oxidative stress corresponding alcohols (79). The structure of the Pseu- responsive regulator OxyR is controlled by reversible S- domonas Ohr peroxiredoxin has been resolved, which is a glutathionylation in vitro (11). Moreover, the activities of homodimer consisting of a tightly folded ␣/ fold with two several metabolic enzymes, such as glyceraldehyde-3- active site cysteines located at the monomer interface on phosphate dehydrogenase, methionine synthase, and the opposite sites of the molecule (80, 81). The mechanism of PAPS reductase are inhibited by S-glutathionylation in hydroperoxide reduction is similar to the structurally unre- E. coli (13, 70, 76). Protein S-thiolation is thought to protect lated eukaryotic 2-Cys peroxiredoxins that exhibit very high active site Cys residues of essential enzymes against irre- specificity and reaction rates with peroxides (81). The cata- versible overoxidation to sulfonic acids. Hypochloric acid lytic mechanism of peroxiredoxins involves the attack of the ϭ ϫ shows very fast reaction rates with Cys residues (K2 3 hyproperoxide substrate by the active site Cys thiolate (the 7 Ϫ1 Ϫ1 10 M s ) that are seven orders of magnitude higher than peroxidatic Cys) that is oxidized to a sulfenic acid intermediate measured for peroxides and rapidly lead to irreversible ox- with the release of the alcohol. The peroxiredoxin sulfenic acid idation products, such as sulfinic and sulfonic acids (3, 4, undergoes intersubunit or intramolecular disulfide bond forma- 77, 78). Thus, S-bacillithiolation in B. subtilis serves to pro- tion with the resolving Cys located in the same or another tect active site Cys residues of key metabolic enzymes subunit of the dimer. The enzyme is regenerated by thiol-disul- against irreversible overoxidation by the strong oxidant hy- fide exchange with GSH or protein electron donors (82). Our pochloric acid. results show that OhrA confers specific protection against The OhrA Peroxiredoxin and the BSH Redox Buffer Confer NaOCl. This leads to the question by which mechanism OhrA is Protection Against NaOCl—In B. subtilis the OhrR repressor is able to detoxify hypochloric acid?
10.1074/mcp.M111.009506–16 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
FIG.9.Proposed defensive mechanisms against hypochloric acid stress in B. subtilis. Exposure of B. subtilis to NaOCl induces an oxidative, disulfide and general stress response (OhrR, Spx, CtsR, PerR, SigmaB). NaOCl leads to S-bacillithiolation of OhrR, MetE, YxjG, PpaC, SerA, and YphP and to intramolecular disulfide formation in GapA. The thiol-disulfide isomerase YphP could function as bacilliredoxin in reduction of S-bacillithiolated proteins. (1) The redox buffer BSH could be directly involved in NaOCl detoxification leading to BSSB formation. (2) S-bacillithiolation of OhrR causes induction of the OhrA peroxiredoxin that is involved in specific NaOCl detoxification. (3)
S-bacillithiolation of the inorganic pyrophosphatase PpaC could lead to decreased ATP sulfurylase activity as the removal of PPi is prevented. (4) S-bacillithiolation of the phosphoglycerate dehydrogenase SerA causes decreased levels of serine that is required for cysteine and methionine biosynthesis. (5) S-bacillithiolation of the active site Cys residues of MetE and YxjG leads to methionine auxotrophy to stop translation during the time of NaOCl detoxification (6) The glyceraldehyde-3-phosphate dehydrogenase GapA is inhibited causing decreased glycolysis. The stars indicate the redox-sensing Cys residues that are S-bacillithiolated in OhrR (Cys15), MetE (Cys730), YxjG (Cys346), or oxidized to an intramolecular disulfide in GapA (Cys152). Abbreviations: APS, adenosine-5Ј-phosphosulfate; N5-THF, 5-methyltetrahydrofo- late; THF, tetrahydrofolate; 3-PG, 3-D-phosphoglycerate, GA-3-P, glyceraldehyde-3-phosphate; 1,3-BPG, 1,3-Bisphosphoglycerate.
The reaction of hypochloric acid with Cys proceeds via Cys726 of E. coli MetE and with Cys620 and Cys704 of Ther- chlorination of the thiol group to form the unstable sulfenyl- motoga maritima MetE (84, 85). MetE is S-bacillithiolated at chloride intermediate that reacts with another thiol group to the non-essential Cys719 and the Zn-binding ligand Cys730 form a disulfide (3, 4). It could be possible that the OhrA and the methionine synthase paralog YxjG is S-bacillithiolated peroxiredoxin is attacked at the peroxidatic Cys by chlorina- at Cys346 that aligns with Cys730 in MetE. In addition, tion, resulting in OhrA sulfenylchloride formation and further Cys730 of MetE is also S-cysteinylated by NaOCl stress in the oxidation to the OhrA inter/intrasubunit disulfide between per- wild type as well as in the bshA mutant. The E. coli MetE oxidatic and resolving Cys residues (Fig. 9). The OhrA disul- protein is S-glutathionylated at the nonessential Cys645 in fide could be regenerated by the BSH redox buffer. response to diamide that is not conserved in B. subtilis MetE In addition, BSH likely plays a direct role in detoxification of (71). In E. coli MetE Cys645 is positioned at the entrance of  ␣ hypochloric acid as has been shown in E. coli for GSH (69). the active Zn site within a cleft between two 8 8 barrels and The reaction of GSH with hypochloric acid is spontaneous it‘s glutathionylation leads to conformational changes of the and does not involve conjugating enzymes such as glutathi- homocysteine binding active site (70, 85). In B. subtilis MetE one S-transferases (GSTs) (83). GSH is oxidized to GSSG by the non-essential Cys719 is about 20 Å apart from the active hypochloric acid nonenzymatically very efficiently as 1 Mol site Cys730. This indicates that thiol-disulfide exchange be- GSH reacted with 4 Mol hypochloric acid in vitro. Hypochloric tween Cys719 and Cys730 residues is not possible. The acid likely also oxidizes BSH directly to BSSB in B. subtilis. question arises how BSH get‘s access to the active site Zn Four Enzymes of the Methionine Synthesis Pathway (MetE, center in B. subtilis MetE ? Recent structural characterizations YxjG, PpaC, SerA) are S-Bacillithiolated by NaOCl Stress of the T. maritima MetE Zn center in the substrate-free and Leading to Methionine Auxotrophy—Besides OhrR, we iden- homocysteine-bound states have revealed an elastic nature tified four enzymes with S-bacillithiolations in response to of the catalytic Zn center upon substrate binding (84). Homo- NaOCl stress as MetE, YxjG, PpaC, and SerA. The methionine cysteine binding leads to an unexpected inversion of Zn ge- synthase MetE has two Zn-coordinating active-site Cys resi- ometry with displacement of the endogenous Zn ligand dues in positions 647 and 730 that align with Cys643 and Glu642 of MetE and movement of the Zn relative to the protein
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–17 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis scaffold. It is proposed that this Zn geometry inversion en- isomerase YphP was identified as novel S-bacillithiolated pro- hances the nucleophilic activation of the homocysteine thio- tein. YphP is a member of the DUF1094 family with a con- late that is required for methyl transfer (84). The dynamic served CXC motif and a thioredoxin-like structure (68). It has nature and flexibility of the Zn center could also provide been shown that YphP functions as thiol-disulfide isomerase access for oxidation of the Zn ligand Cys730 by BSH. in vitro, but the specific substrate is unknown. Phylogenetical There are also structural differences between GSH and BSH studies suggest that YphP could be a novel bacilliredoxin that that could explain why BSH has access to the active site Zn co-occurs with the enzymes of the BSH biosynthesis pathway center. GSH consists of the tripeptide L-␥-glutamylcysteinylgly- across different genomes of low GC Gram-positive bacteria cine with the Cys bound N-terminally and C-terminally by glu- (35). Because YphP is S-bacillithiolated at the active site tamate and glycine residues that could preclude the accessabil- Cys53, it is likely that YphP might function in de-bacillithiola- ity of the Cys thiol to the active site in MetE of E. coli. The tion of MetE, YxjG, PpaC and SerA upon return to nonstress structure of BSH was determined as N-cysteinyl-␣-D-gluco- conditions in vivo. Thus, our proteome-wide studies have saminyl L-malate (20) with an N-terminally located Cys that discovered physiological substrates of the methionine biosyn- enables access of the thiol to the active site Zn center of MetE. thesis pathway as targets for S-bacillithiolation by NaOCl and Our transcriptional data and growth assays support the a candidate bacilliredoxin that might catalyze the reduction of methionine starvation phenotype of NaOCl-treated cells that these protein mixed disulfides. is caused by MetE and YxjG inactivation via S-bacillithiolation. Why Causes NaOCl S-Bacillithiolation and Diamide S-Cys- The inactivation of MetE by oxidative stress has important teinylation in the Proteome?—Previous proteome analyses consequences for the cell. The initiation of translation requires revealed that diamide causes S-cysteinylation in six cytoplas- formyl-Met as start methionine and depletion of methionine is mic proteins but no S-bacillithiolations (67). Three of these discussed as checkpoint to stop translation (70). The inacti- S-cysteinylated proteins (MetE, PpaC and SerA) were identi- vation of MetE by S-bacillithiolation could cause inhibition of fied as S-bacillithiolated by NaOCl. This raises the question translation to prevent further protein damage allowing the cell why S-bacillithiolation was not observed by diamide stress? to detoxify the oxidant and to restore the thiol redox homeo- The answer could be the different thiol chemistries of diamide stasis. Another possibility could be that MetE inactivation and NaOCl. Diamide is an electrophilic azocompound that serves to increase cysteine levels. This has been discussed in does not oxidize redox-sensitive Cys residues to reactive response to diamide stress that leads to MetE inactivation via sulfenic acids or sulfenylchloride. This is supported by the fact S-cysteinylation (67). The CymR-controlled cystathionine that diamide stress does not lead to inactivation of the OhrR beta-synthase MccA and cystathionine lyase MccB are in- repressor (Fig. 3), which would require oxidation of the Cys15 volved in the methionine-to cysteine conversion (Mcc) path- thiolate which then undergoes S-bacillithiolation. Instead, di- way (51, 86). The mccAB operon is strongly up-regulated by amide forms directly disulfide bonds between two Cys-thiols diamide stress supporting that homocysteine is converted to via an Michael addition reaction of the electrophilic azogroup cysteine via this Mcc pathway. However, the mccAB operon with two nucleophilic Cys thiol groups (89). Thus, S-bacilli- was not induced by NaOCl stress, indicating that S-bacilli- thiolation requires probably strong oxidants such as NaOCl thiolation of MetE leads to methionine starvation, but not to that oxidize peroxidatic Cys residues first to reactive sulfenic conversion of homocysteine to cysteine. acids or sulfenylchloride. Besides MetE and YxjG, the pyrophosphatase PpaC and Another question is why S-cysteinylation and S-bacillithio- the 3-D-phosphoglycerate dehydrogenase SerA were identi- lation was observed after NaOCl stress ? The observed S- fied as S-bacillithiolated that also function in the methionine cysteinylations could originate from S-bacillithiolations that biosynthesis pathway. PpaC is an essential and conserved could be explained by the BSH biosynthesis pathway. The enzyme that catalyzes the hydrolysis of inorganic pyrophos- pathways for BSH biosynthesis and degradation have been phate (PPi). Pyrophosphate is generated in a number of ATP- identified in B. subtilis and the structures of the L-malic acid driven cellular processes, including also the ATP sulfurylation glycosyltransferase BaBshA and the GlcNAc-Mal deacetylase as first step of the sulfate assimilation pathway (87, 88). Effi- BaBshB have been resolved in B. anthracis (35, 90). There are cient removal of PPi is required to drive the forward direction two GlcNAc-Mal deacetylases in B. subtilis, BshB1 and in these reactions. Thus, inhibition of PpaC activity by S- BshB2 with similar activities. It is postulated that one of these thiolation could further contribute to methionine starvation. deacetylases could function as bacillithiol-S-conjugate ami- SerA is the first enzyme in the L-serine biosynthetic pathway dase (likewise to the Mca mycothiol-S-conjugate amidase in catalyzing the conversion of 3-D-phosphoglycerate to 3-phos- Mycobacteria) to cleave conjugates formed by reaction of phonooxypyruvate with the generation of NADH. Since serine electrophiles with BSH (35, 91, 92). In turn, BshB1 and/or is used as precursor for cysteine biosynthesis, it‘s inactivation BshB2 could also cleave GlcN-Mal from S-bacillithiolated could pronouce the methionine starvation phenotype. proteins leaving S-cysteinylated proteins. Thus, S-cysteiny- The S-Bacillithiolated Disulfide Isomerase YphP Could lated proteins observed after NaOCl stress could originate Function as Putative Bacilliredoxin—The putative disulfide from S-bacillithiolations that are cleaved by BshB in vivo (91).
10.1074/mcp.M111.009506–18 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
The CID MS/MS spectra of the S-bacillithiolated peptides also 3. Davies, M. J. (2011) Myeloperoxidase-derived oxidation: mechanisms of point to a fragmentation of BSH itself because characteristical biological damage and its prevention. J. Clin. Biochem. Nutr. 48, 8–19 4. Hawkins, C. L., Pattison, D. I., and Davies, M. J. (2003) Hypochlorite- fragment and precursor ions appeared that have lost malate. induced oxidation of amino acids, peptides and proteins. Amino Acids Thus, it is possible that further loss of GlcN in the collision cell 25, 259–274 leads to “artificial” S-cysteinylated peptides as a result of 5. Faulkner, M. J., and Helmann, J. D. (2011) Peroxide stress elicits adaptive changes in bacterial metal ion homeostasis. Antioxid Redox Signal. doi fragmentation. However, the redox buffer cysteine can be 10.1089/ars.2010.3682 also directly used for S-cysteinylations as shown by identifi- 6. Mongkolsuk, S., and Helmann, J. D. (2002) Regulation of inducible peroxide cation of S-cysteinylated MetE in the ⌬bshA mutant. Thus, stress responses. Mol. Microbiol. 45, 9–15 7. Antelmann, H., and Helmann, J. D. (2011) Thiol-based Redox Switches and cysteine can replace BSH for S-thiolation of MetE and OhrR Gene Regulation. Antioxid Redox Signal. 14, 1049–1063 (19, 35). Future proteomic approaches will be directed to 8. Zheng, M., Aslund, F., and Storz, G. (1998) Activation of the OxyR tran- quantify the levels of S-cysteinylations and S-bacillithiolations scription factor by reversible disulfide bond formation. Science 279, 1718–1721 in the proteome. 9. Lee, C., Lee, S. M., Mukhopadhyay, P., Kim, S. J., Lee, S. C., Ahn, W. S., In conclusion, we show here that S-bacillithiolation is a Yu, M. H., Storz, G., and Ryu, S. E. (2004) Redox regulation of OxyR regulatory redox-switch that occurs in response to hypochlo- requires specific disulfide bond formation involving a rapid kinetic reac- tion path. Nat. Struct. Mol. Biol. 11, 1179–1185 ric acid, that is produced in activated macrophages as part of 10. Choi, H., Kim, S., Mukhopadhyay, P., Cho, S., Woo, J., Storz, G., and Ryu, the host immune defense. Thus, Gram-positive pathogens, S. E. (2001) Structural basis of the redox switch in the OxyR transcription such as S. aureus could use BSH as weapon against the host factor. Cell 105, 103–113 11. Kim, S. O., Merchant, K., Nudelman, R., Beyer, W. F., Jr., Keng, T., De- immune system. OhrA-like peroxiredoxins are controlled by Angelo, J., Hausladen, A., and Stamler, J. S. (2002) OxyR: a molecular the one-Cys OhrR-family repressors SarZ and MgrA in S. code for redox-related signaling. Cell 109, 383–396 aureus, that are inactivated by S-thiolation mechanisms in 12. D’Autre´ aux, B., and Toledano, M. B. (2007) ROS as signalling molecules: vitro (18, 93). Furthermore, reversible thiol-modifications are mechanisms that generate specificity in ROS homeostasis. Nat. Rev. Mol. Cell Biol. 8, 813–824 also widespread in the redox proteome of S. aureus by di- 13. Brandes, N., Schmitt, S., and Jakob, U. (2009) Thiol-based redox switches amide stress including a MetE homolog (22, 94). Thus, our in eukaryotic proteins. Antioxid Redox Signal 11, 997–1014 studies provide important leads to future studies on Gram- 14. Lee, J. W., and Helmann, J. D. (2006) The PerR transcription factor senses H2O2 by metal-catalysed histidine oxidation. Nature 440, 363–367 positive pathogens to study the impact of S-bacillithiolations 15. Traore´ , D. A., El Ghazouani, A., Jacquamet, L., Borel, F., Ferrer, J. L., in redox-regulation of OhrR and MetE homologs as defense Lascoux, D., Ravanat, J. L., Jaquinod, M., Blondin, G., Caux-Thang, C., mechanisms against the host immune system. Duarte, V., and Latour, J. M. (2009) Structural and functional character- ization of 2-oxo-histidine in oxidized PerR protein. Nat. Chem. Biol. 5, Acknowledgments—We thank the Decodon company for support 53–59 16. Jacquamet, L., Traore´ , D. A., Ferrer, J. L., Proux, O., Testemale, D., Haze- with the Decodon Delta 2D software, Dirk Albrecht for the MALDI-TOF mann, J. L., Nazarenko, E., El Ghazouani, A., Caux-Thang, C., Duarte, V., analysis, Dana Clausen and Anja Wiechert for excellent technical and Latour, J. M. (2009) Structural characterization of the active form of assistance. We further thank two anonymous reviewers for valuable PerR: insights into the metal-induced activation of PerR and Fur proteins comments and suggestions to improve this manuscript. We are es- for DNA binding. Mol. Microbiol. 73, 20–31 pecially thankful for the advise of one reviewer that the precursor and 17. Panmanee, W., Vattanaviboon, P., Poole, L. B., and Mongkolsuk, S. (2006) y ions in the MS/MS spectra of the bacillithiolated peptides have lost Novel organic hydroperoxide-sensing and responding mechanisms for malate. We further thank the PRIDE team for support in processing OhrR, a major bacterial sensor and regulator of organic hydroperoxide and submission of the LC-MS/MS results to the PRIDE database. We stress. J. Bacteriol. 188, 1389–1395 are very grateful to Ahmed Gaballa and John D. Helmann for discus- 18. Chen, P. R., Brugarolas, P., and He, C. (2011) Redox Signaling in Human sion of the results prior to publication and for providing the B. subtilis Pathogens. Antioxid Redox Signal. 14, 1107–1118 19. Lee, J. W., Soonsanga, S., and Helmann, J. D. (2007) A complex thiolate ohrR-FLAG strain (HB9121), the ohrRC15S mutant strain (HB2048) switch regulates the Bacillus subtilis organic peroxide sensor OhrR. and the bshA und bshB1B2 mutants (HB11002 and HB11053). We Proc. Natl. Acad. Sci. U.S.A. 104, 8743–8748 ⌬ ⌬ ⌬ further thank Kazuo Kobayashi for the ohrR, sigB and perR 20. Newton, G. L., Rawat, M., La Clair, J. J., Jothivasan, V. K., Budiarto, T., mutants and Peter Zuber for the ⌬spx mutant. We are grateful to Hamilton, C. J., Claiborne, A., Helmann, J. D., and Fahey, R. C. (2009) Gottfried Palm for stimulating discussions about our model of hypo- Bacillithiol is an antioxidant thiol produced in Bacilli. Nat. Chem. Biol. 5, chloric acid chemistry and the structure of the MetE active site. 625–627 21. Liebeke, M., Po¨ ther, D. C., van Duy, N., Albrecht, D., Becher, D., Hochgra¨ fe, * This work was supported by a grant from the Deutsche Forsch- F., Lalk, M., Hecker, M., and Antelmann, H. (2008) Depletion of thiol- ungsgemeinschaft (AN 746/2-1) to H. A. containing proteins in response to quinones in Bacillus subtilis. Mol. □S This article contains supplemental Tables S1 to S4 and Microbiol. 69, 1513–1529 Figs. S1 to S5. 22. Po¨ ther, D. C., Liebeke, M., Hochgra¨ fe, F., Antelmann, H., Becher, D., Lalk, ¶ To whom correspondence should be addressed: Institute for M., Lindequist, U., Borovok, I., Cohen, G., Aharonowitz, Y., and Hecker, Microbiology, Ernst-Moritz-Arndt-University of Greifswald, D-17487 M. (2009) Diamide triggers mainly S Thiolations in the cytoplasmic pro- Greifswald, Germany. Tel.: ϩ49-3834-864237; Fax: ϩ49-3834- teomes of Bacillus subtilis and Staphylococcus aureus. J. Bacteriol. 191, 7520–7530 864202; E-mail: [email protected]. 23. Nguyen, T. T., Eiamphungporn, W., Ma¨ der, U., Liebeke, M., Lalk, M., Hecker, M., Helmann, J. D., and Antelmann, H. (2009) Genome-wide REFERENCES responses to carbonyl electrophiles in Bacillus subtilis: control of the 1. Imlay, J. A. (2003) Pathways of oxidative damage. Annu. Rev. Microbiol. 57, thiol-dependent formaldehyde dehydrogenase AdhA and cysteine pro- 395–418 teinase YraA by the MerR-family regulator YraB (AdhR). Mol. Microbiol. 2. Imlay, J. A. (2008) Cellular defenses against superoxide and hydrogen 71, 876–894 peroxide. Annu. Rev. Biochem. 77, 755–776 24. Chi, B. K., Albrecht, D., Gronau, K., Becher, D., Hecker, M., and Antelmann,
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–19 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
H. (2010) The redox-sensing regulator YodB senses quinones and di- 44. de Hoon, M. J., Imoto, S., Nolan, J., and Miyano, S. (2004) Open source amide via a thiol-disulfide switch in Bacillus subtilis. Proteomics 10, clustering software. Bioinformatics 20, 1453–1454 3155–3164 45. Leichert, L. I., Scharf, C., and Hecker, M. (2003) Global characterization of 25. Chi, B. K., Kobayashi, K., Albrecht, D., Hecker, M., and Antelmann, H. disulfide stress in Bacillus subtilis. J. Bacteriol. 185, 1967–1975 (2010) The paralogous MarR/DUF24-family repressors YodB and CatR 46. Nguyen, V. D., Wolf, C., Ma¨ der, U., Lalk, M., Langer, P., Lindequist, U., control expression of the catechol dioxygenase CatE in Bacillus subtilis. Hecker, M., and Antelmann, H. (2007) Transcriptome and proteome J. Bacteriol. 192, 4571–4581 analyses in response to 2-methylhydroquinone and 6-brom-2-vinyl- 26. Leelakriangsak, M., Huyen, N. T., To¨ we, S., van Duy, N., Becher, D., chroman-4-on reveal different degradation systems involved in the ca- Hecker, M., Antelmann, H., and Zuber, P. (2008) Regulation of quinone tabolism of aromatic compounds in Bacillus subtilis. Proteomics 7, detoxification by the thiol stress sensing DUF24/MarR-like repressor, 1391–1408 YodB in Bacillus subtilis. Mol. Microbiol. 67, 1108–1124 47. Tam le, T., Eymann, C., Albrecht, D., Sietmann, R., Schauer, F., Hecker, M., 27. Touati, D., Jacques, M., Tardat, B., Bouchard, L., and Despied, S. (1995) and Antelmann, H. (2006) Differential gene expression in response to Lethal oxidative damage and mutagenesis are generated by iron in delta phenol and catechol reveals different metabolic activities for the degra- fur mutants of Escherichia coli: protective role of superoxide dismutase. dation of aromatic compounds in Bacillus subtilis. Environ. Microbiol. 8, J. Bacteriol. 177, 2305–2314 1408–1427 28. Leichert, L. I., Gehrke, F., Gudiseva, H. V., Blackwell, T., Ilbert, M., Walker, 48. Eisen, M. B., Spellman, P. T., Brown, P. O., and Botstein, D. (1998) Cluster A. K., Strahler, J. R., Andrews, P. C., and Jakob, U. (2008) Quantifying analysis and display of genome-wide expression patterns. Proc. Natl. changes in the thiol redox proteome upon oxidative stress in vivo. Proc. Acad. Sci. U.S.A. 95, 14863–14868 Natl. Acad. Sci. U.S.A. 105, 8197–8202 49. Wetzstein, M., Vo¨ lker, U., Dedio, J., Lo¨ bau, S., Zuber, U., Schiesswohl, M., 29. Winter, J., Ilbert, M., Graf, P. C., Ozcelik, D., and Jakob, U. (2008) Bleach Herget, C., Hecker, M., and Schumann, W. (1992) Cloning, sequencing, activates a redox-regulated chaperone by oxidative protein unfolding. and molecular analysis of the dnaK locus from Bacillus subtilis. J. Bac- Cell 135, 691–701 teriol. 174, 3300–3310 30. Wang, S., Deng, K., Zaremba, S., Deng, X., Lin, C., Wang, Q., Tortorello, 50. Antelmann, H., Hecker, M., and Zuber, P. (2008) Proteomic signatures M. L., and Zhang, W. (2009) Transcriptomic response of Escherichia coli uncover thiol-specific electrophile resistance mechanisms in Bacillus O157:H7 to oxidative stress. Appl. Environ. Microbiol. 75, 6110–6123 subtilis. Expert Rev. Proteomics 5, 77–90 31. Ceragioli, M., Mols, M., Moezelaar, R., Ghelardi, E., Senesi, S., and Abee, 51. Even, S., Burguie` re, P., Auger, S., Soutourina, O., Danchin, A., and Martin- T. (2010) Comparative transcriptomic and phenotypic analysis of the Verstraete, I. (2006) Global control of cysteine metabolism by CymR in responses of Bacillus cereus to various disinfectant treatments. Appl. Bacillus subtilis. J. Bacteriol. 188, 2184–2197 Environ. Microbiol. 76, 3352–3360 52. Tomsic, J., McDaniel, B. A., Grundy, F. J., and Henkin, T. M. (2008) Natural 32. Nakano, M. M., Hajarizadeh, F., Zhu, Y., and Zuber, P. (2001) Loss-of- variability in S-adenosylmethionine (SAM)-dependent riboswitches: S- function mutations in yjbD result in ClpX- and ClpP-independent com- box elements in bacillus subtilis exhibit differential sensitivity to SAM In petence development of Bacillus subtilis. Mol. Microbiol. 42, 383–394 vivo and in vitro. J. Bacteriol. 190, 823–833 33. Hayashi, K., Kensuke, T., Kobayashi, K., Ogasawara, N., and Ogura, M. 53. Henkin, T. M. (2008) Riboswitch RNAs: using RNA to sense cellular metab- (2006) Bacillus subtilis RghR (YvaN) represses rapG and rapH, which olism. Genes Dev. 22, 3383–3390 encode inhibitors of expression of the srfA operon. Mol. Microbiol. 59, 54. Henkin, T. M. (2009) RNA-dependent RNA switches in bacteria. Methods 1714–1729 Mol. Biol. 540, 207–214 34. Fuangthong, M., and Helmann, J. D. (2002) The OhrR repressor senses 55. Grundy, F. J., and Henkin, T. M. (2003) The T box and S box transcription organic hydroperoxides by reversible formation of a cysteine-sulfenic termination control systems. Front. Biosci. 8, d20–31 acid derivative. Proc. Natl. Acad. Sci. U.S.A. 99, 6690–6695 56. Szegedi, S. S., Castro, C. C., Koutmos, M., and Garrow, T. A. (2008) 35. Gaballa, A., Newton, G. L., Antelmann, H., Parsonage, D., Upton, H., Betaine-homocysteine S-methyltransferase-2 is an S-methylmethionine- Rawat, M., Claiborne, A., Fahey, R. C., and Helmann, J. D. (2010) homocysteine methyltransferase. J. Biol. Chem. 283, 8939–8945 Biosynthesis and functions of bacillithiol, a major low-molecular-weight 57. Mansilla, M. C., Albanesi, D., and de Mendoza, D. (2000) Transcriptional thiol in Bacilli. Proc. Natl. Acad. Sci. U.S.A. 107, 6482–6486 control of the sulfur-regulated cysH operon, containing genes involved in 36. Stu¨ lke, J., Hanschke, R., and Hecker, M. (1993) Temporal activation of L-cysteine biosynthesis in Bacillus subtilis. J. Bacteriol. 182, 5885–5892 beta-glucanase synthesis in Bacillus subtilis is mediated by the GTP 58. Yurimoto, H., Hirai, R., Matsuno, N., Yasueda, H., Kato, N., and Sakai, Y. pool. J. Gen. Microbiol. 139, 2041–2045 (2005) HxlR, a member of the DUF24 protein family, is a DNA-binding 37. Nourooz-Zadeh, J., Tajaddini-Sarmadi, J., and Wolff, S. P. (1994) Mea- protein that acts as a positive regulator of the formaldehyde-inducible surement of plasma hydroperoxide concentrations by the ferrous oxida- hxlAB operon in Bacillus subtilis. Mol. Microbiol. 57, 511–519 tion-xylenol orange assay in conjunction with triphenylphosphine. Anal. 59. Moore, C. M., Gaballa, A., Hui, M., Ye, R. W., and Helmann, J. D. (2005) Biochem. 220, 403–409 Genetic and physiological responses of Bacillus subtilis to metal ion 38. Hochgra¨ fe, F., Mostertz, J., Albrecht, D., and Hecker, M. (2005) Fluores- stress. Mol. Microbiol. 57, 27–40 cence thiol modification assay: oxidatively modified proteins in Bacillus 60. Moore, C. M., and Helmann, J. D. (2005) Metal ion homeostasis in Bacillus subtilis. Mol. Microbiol. 58, 409–425 subtilis. Curr. Opin. Microbiol. 8, 188–195 39. Barbe, V., Cruveiller, S., Kunst, F., Lenoble, P., Meurice, G., Sekowska, A., 61. Smaldone, G. T., and Helmann, J. D. (2007) CsoR regulates the copper Vallenet, D., Wang, T., Moszer, I., Me´ digue, C., and Danchin, A. (2009) efflux operon copZA in Bacillus subtilis. Microbiology 153, 4123–4128 From a consortium sequence to a unified sequence: the Bacillus subtilis 62. Szurmant, H., and Ordal, G. W. (2004) Diversity in chemotaxis mechanisms 168 reference genome a decade later. Microbiology 155, 1758–1775 among the bacteria and archaea. Microbiol. Mol. Biol. Rev. 68, 301–319 40. Elias, J. E., and Gygi, S. P. (2007) Target-decoy search strategy for in- 63. Kempf, B., and Bremer, E. (1998) Uptake and synthesis of compatible creased confidence in large-scale protein identifications by mass spec- solutes as microbial stress responses to high-osmolality environments. trometry. Nat. Methods 4, 207–214 Arch. Microbiol. 170, 319–330 41. Vizcaino, J. A., Coˆte´ , R., Reisinger, F., Barsnes, H., Foster, J. M., 64. Massey, V., Gibson, Q. H., and Veeger, C. (1960) Intermediates in the Rameseder, J., Hermjakob, H., and Martens, L. (2010) The Proteomics catalytic action of lipoyl dehydrogenase (diaphorase). Biochem. J. 77, Identifications database: 2010 update. Nucleic Acids Res. 38, D736–742 341–351 42. Majumdar, D., Avissar, Y. J., and Wyche, J. H. (1991) Simultaneous and 65. Zuber, P. (2004) Spx-RNA polymerase interaction and global transcriptional rapid isolation of bacterial and eukaryotic DNA and RNA: a new ap- control during oxidative stress. J. Bacteriol. 186, 1911–1918 proach for isolating DNA. BioTechniques 11, 94–101 66. Zuber, P. (2009) Management of oxidative stress in Bacillus. Annu. Rev. 43. Charbonnier, Y., Gettler, B., Franc¸ ois, P., Bento, M., Renzoni, A., Vaudaux, Microbiol. 63, 575–597 P., Schlegel, W., and Schrenzel, J. (2005) A generic approach for the 67. Hochgra¨ fe, F., Mostertz, J., Po¨ ther, D. C., Becher, D., Helmann, J. D., and design of whole-genome oligoarrays, validated for genomotyping, dele- Hecker, M. (2007) S-cysteinylation is a general mechanism for thiol tion mapping and gene expression analysis on Staphylococcus aureus. protection of Bacillus subtilis proteins after oxidative stress. J. Biol. BMC Genomics 6, 95 Chem. 282, 25981–25985
10.1074/mcp.M111.009506–20 Molecular & Cellular Proteomics 10.11 S-Bacillithiolation Confers Hypochlorite Resistance in B. subtilis
68. Derewenda, U., Boczek, T., Gorres, K. L., Yu, M., Hung, L. W., Cooper, D., 82. Barford, D. (2004) The role of cysteine residues as redox-sensitive regula- Joachimiak, A., Raines, R. T., and Derewenda, Z. S. (2009) Structure and tory switches. Curr. Opin. Struct. Biol. 14, 679–686 function of Bacillus subtilis YphP, a prokaryotic disulfide isomerase with 83. Chesney, J. A., Eaton, J. W., and Mahoney, J. R., Jr. (1996) Bacterial a CXC catalytic motif. Biochemistry 48, 8664–8671 glutathione: a sacrificial defense against chlorine compounds. J. Bacte- 69. Masip, L., Veeravalli, K., and Georgiou, G. (2006) The many faces of riol. 178, 2131–2135 glutathione in bacteria. Antioxid Redox. Signal. 8, 753–762 84. Koutmos, M., Pejchal, R., Bomer, T. M., Matthews, R. G., Smith, J. L., and 70. Hondorp, E. R., and Matthews, R. G. (2004) Oxidative stress inactivates Ludwig, M. L. (2008) Metal active site elasticity linked to activation of cobalamin-independent methionine synthase (MetE) in Escherichia coli. homocysteine in methionine synthases. Proc. Natl. Acad. Sci. U.S.A. PLoS Biol. 2, e336 105, 3286–3291 71. Hondorp, E. R., and Matthews, R. G. (2009) Oxidation of cysteine 645 of 85. Pejchal, R., and Ludwig, M. L. (2005) Cobalamin-independent methionine cobalamin-independent methionine synthase causes a methionine limi- synthase (MetE): a face-to-face double barrel that evolved by gene tation in Escherichia coli. J. Bacteriol. 191, 3407–3410 duplication. PLoS Biol. 3, e31 72. Hochgra¨ fe, F., Wolf, C., Fuchs, S., Liebeke, M., Lalk, M., Engelmann, S., 86. Hullo, M. F., Auger, S., Soutourina, O., Barzu, O., Yvon, M., Danchin, A., and Hecker, M. (2008) Nitric oxide stress induces different responses but and Martin-Verstraete, I. (2007) Conversion of methionine to cysteine in mediates comparable protein thiol protection in Bacillus subtilis and Bacillus subtilis and its regulation. J. Bacteriol. 189, 187–197 Staphylococcus aureus. J. Bacteriol. 190, 4997–5008 87. Cooperman, B. S., Baykov, A. A., and Lahti, R. (1992) Evolutionary conser- 73. Brandes, N., Rinck, A., Leichert, L. I., and Jakob, U. (2007) Nitrosative vation of the active site of soluble inorganic pyrophosphatase. Trends stress treatment of E. coli targets distinct set of thiol-containing proteins. Biochem. Sci. 17, 262–266 Mol. Microbiol. 66, 901–914 88. Lee, H. S., Kim, Y. J., Lee, J. H., and Kang, S. G. (2009) Identification and 74. Pullar, J. M., Winterbourn, C. C., and Vissers, M. C. (1999) Loss of GSH and characterization of inorganic pyrophosphatase and PAP phosphatase thiol enzymes in endothelial cells exposed to sublethal concentrations of from Thermococcus onnurineus NA1. J. Bacteriol. 191, 3415–3419 hypochlorous acid. Am. J. Physiol. 277, H1505–1512 89. Kosower, N. S., and Kosower, E. M. (1995) Diamide: an oxidant probe for 75. Dalle-Donne, I., Rossi, R., Colombo, G., Giustarini, D., and Milzani, A. (2009) thiols. Methods Enzymol. 251, 123–133 Protein S-glutathionylation: a regulatory device from bacteria to humans. 90. Parsonage, D., Newton, G. L., Holder, R. C., Wallace, B. D., Paige, C., Trends Biochem. Sci 34, 85–96 Hamilton, C. J., Dos Santos, P. C., Redinbo, M. R., Reid, S. D., and 76. Lillig, C. H., Potamitou, A., Schwenn, J. D., Vlamis-Gardikas, A., and Claiborne, A. (2010) Characterization of the N-acetyl-alpha-D-gluco- Holmgren, A. (2003) Redox regulation of 3Ј-phosphoadenylylsulfate re- saminyl l-malate synthase and deacetylase functions for bacillithiol bio- ductase from Escherichia coli by glutathione and glutaredoxins. J. Biol. synthesis in Bacillus anthracis. Biochemistry 49, 8398–8414 Chem. 278, 22325–22330 91. Helmann, J. D. (2011) Bacillithiol, a New Player in Bacterial Redox Home- 77. Winterbourn, C. C., and Hampton, M. B. (2008) Thiol chemistry and spec- ostasis. Antioxid. Redox. Signal 15, 123–133 ificity in redox signaling. Free Radic. Biol. Med. 45, 549–561 92. Newton, G. L., Buchmeier, N., and Fahey, R. C. (2008) Biosynthesis and 78. Pattison, D. I., and Davies, M. J. (2001) Absolute rate constants for the functions of mycothiol, the unique protective thiol of Actinobacteria. reaction of hypochlorous acid with protein side chains and peptide Microbiol. Mol. Biol. Rev. 72, 471–494 bonds. Chem. Res. Toxicol. 14, 1453–1464 93. Poor, C. B., Chen, P. R., Duguid, E., Rice, P. A., and He, C. (2009) Crystal 79. Dubbs, J. M., and Mongkolsuk, S. (2007) Peroxiredoxins in bacterial anti- structures of the reduced, sulfenic acid, and mixed disulfide forms of oxidant defense. Subcell. Biochem. 44, 143–193 SarZ, a redox active global regulator in Staphylococcus aureus. J. Biol. 80. Lesniak, J., Barton, W. A., and Nikolov, D. B. (2002) Structural and func- Chem. 284, 23517–23524 tional characterization of the Pseudomonas hydroperoxide resistance 94. Wolf, C., Hochgra¨ fe, F., Kusch, H., Albrecht, D., Hecker, M., and Engel- protein Ohr. EMBO J. 21, 6649–6659 mann, S. (2008) Proteomic analysis of antioxidant strategies of Staphy- 81. Poole, L. B., and Nelson, K. J. (2008) Discovering mechanisms of signaling- lococcus aureus: diverse responses to different oxidants. Proteomics 8, mediated cysteine oxidation. Curr. Opin. Chem. Biol. 12, 18–24 3139–3153
Molecular & Cellular Proteomics 10.11 10.1074/mcp.M111.009506–21
Article V
Page 123
Page 124
REPORTS loosening needed for growth. Increased wall re- 7. D. Reinhardt et al., Nature 426, 255 (2003). 25. R. Vandiver, A. Goriely, Philos. Transact. A Math. Phys. modeling is expected to change elastic proper- 8. D. L. Rayle, R. E. Cleland, Plant Physiol. 99, 1271 (1992). Eng. Sci. 367, 3607 (2009). 26. V. Mirabet, P. Das, A. Boudaoud, O. Hamant, Annu. Rev. ties, which explains why fast-growing areas in the 9. D. J. Cosgrove, Nat. Rev. Mol. Cell Biol. 6, 850 (2005). 10. T. Vernoux, J. Kronenberger, O. Grandjean, P. Laufs, Plant Biol. 62, 365 (2011). peripheral region can stay in their linear range J. Traas, Development 127, 5157 (2000). 27. U. Kutschera, K. J. Niklas, J. Plant Physiol. 164, 1395 of elasticity for larger deformations (20–22); 11. D. Reinhardt, T. Mandel, C. Kuhlemeier, Plant Cell 12, (2007). while at the same time, they are stiffer than slow- 507 (2000). 28. J. A. Lockhart, J. Theor. Biol. 8, 264 (1965). growing regions for small strains. Our data 12. A. Fleming, S. McQueen-Mason, T. Mandel, C. Kuhlemeier, 29. T. Vernoux et al., Mol. Syst. Biol. 7, 508 (2011). Science 276, 1415 (1997). 30. R. Sablowski, Curr. Opin. Plant Biol. 10, 639 (2007). suggest that the functional distinction between 13. S. Pien, J. Wyrzykowska, S. McQueen-Mason, C. Smart, 31. R. K. Yadav, T. Girke, S. Pasala, M. Xie, G. V. Reddy, slow- and fast-growing regions in the shoot apex A. Fleming, Proc. Natl. Acad. Sci. U.S.A. 98, 11812 (2001). Proc. Natl. Acad. Sci. U.S.A. 106, 4941 (2009). is not only genetically defined (30, 31)butisen- 14. A. Peaucelle et al., Curr. Biol. 18, 1943 (2008). hanced by mechanical feedbacks. Such a mech- 15. A. Peaucelle et al., Curr. Biol. 21, 1720 (2011). Acknowledgments: We thank J. Dumais, A. Fleming, S. Robinson, D. Kwiatkowska, and P. Wojtaszek for reading the anism would stabilize and protect the critically 16. M. G. Heisler et al., PLoS Biol. 8, e1000516 (2010). 17. O. Hamant et al., Science 322, 1650 (2008). manuscript. This work was supported by the Swiss National Science Foundation, SystemsX.ch, and the European Union important stem cell niche from the numerous 18. P. Milani et al., Plant J. 67, 1116 (2011). New and Emerging Science and Technology grant MechPlant. sources of noise inherent in the chemistry of 19. Materials and methods are available as supporting N.N. received European Molecular Biology Organization biological systems. material on Science Online Long-Term and Roche Research Foundation fellowships. 20. C. Storm, J. J. Pastore, F. C. MacKintosh, T. C. Lubensky, References and Notes P. A. Janmey, Nature 435, 191 (2005). 1. T. Steeves, I. Sussex, Eds., Patterns in Plant Development 21. S. E. Whitney, M. G. Gothard, J. T. Mitchell, M. J. Gidley, Supporting Online Material (Cambrige Univ. Press, New York, 1988). Plant Physiol. 121, 657 (1999). www.sciencemag.org/cgi/content/full/335/6072/1096/DC1 2. D. Kwiatkowska, J. Dumais, J. Exp. Bot. 54, 1585 (2003). 22. R. R. Vincent, D. N. Pinder, Y. Hemar, M. A. Williams, Materials and Methods 3. O. Grandjean et al., Plant Cell 16, 74 (2004). Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 76, 031909 Figs. S1 to S8 4.G.V.Reddy,M.G.Heisler,D.W.Ehrhardt,E.M.Meyerowitz, (2007). References (32–41) Development 131, 4225 (2004). 23. J. Dumais, C. R. Steele, J. Plant Growth Regul. 19, 7 (2000). 5. S. Vanneste, J. Friml, Cell 136, 1005 (2009). 24. J. M. Selker, G. L. Steucek, P. B. Green, Dev. Biol. 153, 24 August 2011; accepted 30 January 2012 6. M. G. Heisler et al., Curr. Biol. 15, 1899 (2005). 29 (1992). 10.1126/science.1213100
to environmental change (9, 11),andanunantic- on July 10, 2012 Global Network Reorganization ipated complexity of unicellular organisms (7). Data interpretation, however, has generally been During Dynamic Adaptations restricted to multivariate statistical analysis meth- ods that indicate general but not mechanistic Bacillus subtilis relationships between different molecular entities. of Metabolism Focusing on single data types with sophisticated computational analysis has been informative (3) Joerg Martin Buescher,1* Wolfram Liebermeister,2* Matthieu Jules,3* Markus Uhr,4* Jan Muntel,5* but increases the risk of missing the functionally Eric Botella,7 Bernd Hessling,5 Roelco Jacobus Kleijn,1 Ludovic Le Chat,3 François Lecointe,3 relevant multilevel control mechanisms (2), lim- www.sciencemag.org Ulrike Mäder,5 Pierre Nicolas,6 Sjouke Piersma,8 Frank Rügheimer,15 Dörte Becher,5 iting the description of the underlying molecular Philippe Bessieres,6 Elena Bidnenko,3 Emma L. Denham,8 Etienne Dervyn,3 Kevin M. Devine,7 mechanisms and, hence, the depth of biological Geoff Doherty,9 Samuel Drulhe,15 Liza Felicori,10 Mark J. Fogg,11 Anne Goelzer,6 Annette Hansen,7 insight. Colin R. Harwood,12 Michael Hecker,5 Sebastian Hubner,7 Claus Hultschig,13 Hanne Jarmer,14 Edda Klipp,2 Aurélie Leduc,6 Peter Lewis,9 Frank Molina,10 Philippe Noirot,3 Sabine Peres,10 3 12 14 13 5 Nathalie Pigeonneau, Susanne Pohl, Simon Rasmussen, Bernd Rinn, Marc Schaffer, 1Institute of Molecular Systems Biology, ETH Zurich, 8093 Julian Schnidder,1 Benno Schwikowski,15 JanMaartenVanDijl,8 Patrick Veiga,6 Sean Walsh,13 Zurich, Switzerland. 2Theoretical Biophysics, Humboldt Uni- 11 4 3 1 versity Berlin, 10115 Berlin, Germany. 3INRA, UMR1319 Micalis, AnthonyJ.Wilkinson, Jörg Stelling, † Stéphane Aymerich, † Uwe Sauer † Downloaded from Jouy-en-Josas F78350, France. 4Department of Biosystems Sci- ence and Engineering, ETH Zurich, 4058 Basel, Switzerland. Adaptation of cells to environmental changes requires dynamic interactions between metabolic 5Institute for Microbiology, Ernst-Moritz-Arndt-University Greifs- and regulatory networks, but studies typically address only one or a few layers of regulation. wald, 17487 Greifswald, Germany. 6INRA, Mathématique For nutritional shifts between two preferred carbon sources of Bacillus subtilis, we combined Informatique et Génome UR1077, 78350 Jouy-en-Josas, France. 7 statistical and model-based data analyses of dynamic transcript, protein, and metabolite Smurfit Institute of Genetics, Trinity College Dublin, Dublin 2, Ireland. 8Department of Medical Microbiology, University of abundances and promoter activities. Adaptation to malate was rapid and primarily controlled Groningen and University Medical Center Groningen, 9700 RB posttranscriptionally compared with the slow, mainly transcriptionally controlled adaptation to Groningen, Netherlands. 9School of Environmental and Life glucose that entailed nearly half of the known transcription regulation network. Interactions Sciences,UniversityofNewcastle,Callaghan,NSW2308, Australia. 10Sysdiag CNRS Bio-Rad UMR 3145, Cap Delta/Parc across multiple levels of regulation were involved in adaptive changes that could also be achieved 11 by controlling single genes. Our analysis suggests that global trade-offs and evolutionary Euromédecine, 34184 Montpellier Cedex 4, France. York Struc- tural Biology Laboratory, Department of Chemistry, University of constraints provide incentives to favor complex control programs. York, York YO10 5YW, UK. 12Institute, of Cell and Molecular Biosciences, Centre for Bacterial Cell Biology, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4AX, UK. major challenge in biology is to under- acquisition of such data is technically demand- 13 stand the organization and interactions ing, few studies have reported transcript, protein, Center for Information Sciences and Databases, Department of Biosystems Science and Engineering, ETH Zurich, 4058 Aof the various functional and regulatory and metabolite abundances, and most studies Basel, Switzerland. 14Center for Biological Sequence Anal- networks in cells. The underlying complexity have been restricted to steady-state conditions ysis, Department of Systems Biology, Technical University of arises from the intertwined nonlinear and dy- (4–8). Consequently, only subsets of components Denmark, 2800 Kgs. Lyngby, Denmark. 15Institut Pasteur, Sys- namic interactions among a large number of have been monitored dynamically for very short- tems Biology Lab, Department of Genomes and Genetics, and CNRS URA 2171, F-75015 Paris, France. cellular components. To better understand these (9, 10) or long-term responses (11, 12)toenvi- *These authors contributed equally to this work. interacting molecular networks, the acquisition ronmental perturbations. These studies typically †To whom correspondence should be addressed. E-mail: of appropriate, preferably time-resolved quanti- revealed coordinated abundance changes (11, 12), [email protected] (J.S.); stephane.aymerich@ tative data is a prerequisite (1–3). Because the major transcriptional reconfigurations in response grignon.inra.fr (S.A.); [email protected] (U.S.)
www.sciencemag.org SCIENCE VOL 335 2 MARCH 2012 1099 REPORTS
To elucidate the dynamic interplay between two shift experiments that cover short-term meta- transcript-array platforms did not require further metabolic and regulatory networks systematically, bolic to longer-term protein-level adaptation (15). consolidation (figs. S3 and S4), but we gen- we investigated dynamic shifts in availability of To enable data integration, we developed erated consensus proteomics data by calculating the preferred carbon sources, glucose and malate, problem-driven, yet generic solutions in three areas confidence-weighted averages of protein abun- of the bacterium Bacillus subtilis (13). Because (table S1). To generate consistent data, we min- dances obtained from two-dimensional gels and these nutrients repress the use of other substrates imized biological variability (Fig. 1) by withdrawing liquid chromatography–mass spectrometry (SOM but are themselves used together, they represent a samples from the same bioreactor culture in trip- 1). Replicate time series of transcript and protein tractable model for analyzing dynamic decision- licate experiments (fig. S1) and, in a few cases, data were then smoothed and interpolated by making by cells when faced with compounds of from standardized small-scale cultivations (fig. S2). Bayesian multicurve regression analysis (SOM 1). similar nutritional value. We induced dynamic We used naming and formatting conventions with Because metabolite data showed changes on var- shifts by adding glucose or malate to cultures of unique identifiers for all considered constituents to ious time scales, we developed a special algorithm B. subtilis growing exponentially on the other permit the efficient exchange of data and knowl- that aligns time courses on the basis of Kalman substrate. To elucidate the cellular adaptation mech- edge. Furthermore, to maximize reliability and cov- smoothing (Fig. 2A and SOM 1) (16). anisms, we determined transcript, protein, and erage for subsequent data analysis and modeling, We exploited the profusion of data (table S1) absolute metabolite abundances, as well as pro- we combined overlapping dynamic data acquired to extend the genome annotation of B. subtilis moter activities (Fig. 1 and table S1) (14), provid- from different analytical platforms and on differ- by statistical data analysis and integration with ing dynamic data with up to 24 time points for ent time scales (Fig. 1). Data from the different prior knowledge (table S2). Analysis of mRNA
Dynamic Measured Modeling & Inferred experiments data data analyses data on July 10, 2012 www.sciencemag.org Downloaded from
Fig. 1. Overview of our experimental design, computational analysis, and data as parameters for subsequent modeling steps. ChIP, chromatin immuno- information flow. The dynamic shift experiments yielded measured data from precipitation; LC-MS, liquid chromatography–mass spectrometry; GFP, green which nonmeasurable quantities were inferred by statistical and model-based fluorescent protein; HPLC, high-performance liquidchromatography;OD, analysis. Arrows show the flow of information, including the use of inferred optical density.
1100 2 MARCH 2012 VOL 335 SCIENCE www.sciencemag.org REPORTS abundances identified 4393 transcriptionally active (table S1). For the metabolically important tran- ponent analysis (20) quantified the strengths of genomic regions, 109 of which were not previ- scription factors CcpA, CcpC, CcpN, and CggR all 2900 transcription factor interactions with target ously described, including 21 putative protein- (17–19), we identified DNA binding sites by chro- genes, and 1488 of the interactions were impli- coding sequences and 23 antisense RNAs (table matin immunoprecipitation–on-chip analysis cated in at least one shift (689 for malate addition; S3). Additionally, 2422 genomic regions were in malate plus glucose (table S1). We found 1244 for glucose addition) (SOM 2). Furthermore, differentially transcribed after the nutrient shifts 184 CcpA binding sites, most of which were lo- we identified 110 posttranscriptional regulation (table S1). Clustering of mRNA profiles and func- cated near to promoters of genes differentially events in protein synthesis (77 and 23 were spe- tional classification of differentially transcribed expressed immediately after glucose addition (fig. cific to malate and glucose addition, respectively) genes enabled detailed functional annotation S5). By filtering this group for high-scoring, pre- through a dynamic model that correlates time pro- of 46 genes (table S4) and probable function viously undetected CcpA sites, we generated an files of promoter activity, mRNA abundance, and assignment of 853 previously not annotated genes improved regulatory network topology (SOM 2) protein abundance for 300 genes for which all for subsequent analysis of regulatory events. three data types were available (tables S5 to S7). Metabolomics data revealed instantaneous Overall, our integrated analysis suggests that ap- malate uptake into cells grown on glucose but parently similar adaptation processes are mediated substantially delayed uptake of glucose into cells by fundamentally different control mechanisms, grown on malate (Fig. 2, B and C). Sensitivity namely a predominantly posttranscriptional regula- analysis of steady-state fluxes in the stoichiometric tion after malate addition and a greater reliance on network model in combination with proteomic transcriptional regulation after glucose addition. data predicted transcriptional regulation to be Next, we inferred the dynamic activity profiles more important for the dynamics induced by of 154 transcription factors by network-component glucose than by malate (Fig. 2D). Network com- analysis (20) from the transcript abundances of on July 10, 2012 ρ www.sciencemag.org Downloaded from
Fig. 2. (A) Processing of time-series data from parallel experiments. The example shown here is fructose 1,6-bisphosphate during the malate–to– glucose-plus-malate shift. Measurements from three independent cultures (black data points) yield an averaged and smooth time course (blue line). (B and C) Initial dynamic uptake rates of malate (green) and glucose (red) after addition to wild-type B. subtilis growing on the respective other substrate were obtained by fitting of splines. Black lines and shadings represent the specific growth rate and 95% confidence intervals, respectively (SOM 3). (D) The global importance of transcriptional regulation after glucose addition is confirmed by an increasing correlation (red) between flux sensitivity (measuring the impact of a change in environmental conditions Fig. 3. (A) Activities of the main transcriptional regulators controlling central metabolism during the shifts on a metabolic flux) and protein abundance over time. of glucose (GM) or malate (MG) to glucose plus malate. Blue curves represent log2 expression profiles of This increase does not occur after malate addition target genes; purple curves denote the inferred transcription factor activities. Straight lines in the middle (blue). White areas denote statistically significant sections indicate the relative contribution (proportional to color intensity) of different transcription factors correlation values with respect to an approximate to target gene expression changes (blue, activation; red, repression). (B and C) Averaged time profiles of 95% confidence interval; gray shaded areas denote transcripts (light blue), proteins (dark blue), metabolic fluxes (green), and metabolites (red) during the the absence of statistically significant correlation. dynamic shift experiments of glucose (B) or malate (C) to malate plus glucose.
www.sciencemag.org SCIENCE VOL 335 2 MARCH 2012 1101 REPORTS
their 1754 known target genes (Fig. 3A and SOM the deviations between the inferred transcription NadR, KipR, CcpN, FruR) are regulated by yet 2). One hundred twenty-seven transcription fac- factor activities and the measured transcription unknown effectors. Overall, both nutrient shifts tors changed their activity profile significantly in factor mRNA abundance provided evidence for induced substantial, global network reconfigu- at least one shift (61 for malate addition and 91 posttranslational regulation (SOM 2). Network ration at all levels of regulation. for glucose addition). A rapid change of activity component analysis predicted that the activity of To identify the relevant primary changes in (<5 min), most prominent for the transporter 51 transcription factors was modulated post- central metabolism, where essentially all genes regulators MalR (malate) and GlcT (glucose), and transcriptionally, and 39 of these (for example, were differentially expressed during one or both shifts, we correlated time courses of metabolic fluxes with those of the abundances of the corre- sponding enzymes. We estimated dynamic ex- tracellular rates by interpolation (SOM 3) and network-wide pseudodynamic metabolic fluxes with a stoichiometric network model and the intracellular metabolite concentrations (Fig. 3, B and C, and SOM 4). Positive correlations between fluxes and enzyme abundances indicated genetic rather than metabolic regulation of the reaction rates (21). Upon the addition of malate, the abun- dances of only the nicotinamide adenine dinucle- otide phosphate–dependent malic enzyme (YtsJ) and two acetate production enzymes (Pta and AckA) correlated with their respective reaction rates (R > 0.8) and thus putatively control flux (Fig. 4A). After the addition of glucose, enzymes for 11 central reactions putatively controlled flux, and in six cases increasing enzyme abundance on July 10, 2012 potentially overcame the bottleneck in glucose up- take. The corresponding genes are organized in two operons: ptsGHI for glucose transport and phosphorylation and cggR-gapA-pgk-tpiA-pgm- eno for lower glycolysis. This focused analysis of multiple “omics” data sets suggested the pts and cggR operons as primary control targets for adap- tation, where the half-maximal responses of the pts and cggR operon-encoded proteins within 30 www.sciencemag.org to 60 min of glucose addition could account for the ~60-min delay in glucose uptake (Fig. 2, B and C). We evaluated the relative contributions of both operons to the delay with a cggR dele- tion mutant and a mutant with constitutive ptsG expression (Fig. 4B and SOM 5). When grown on malate, both mutants used glucose immedi-
ately without the delay of the wild-type strain. Downloaded from As the cggR or ptsG expression alone is suffi- cient to facilitate immediate glucose uptake, it is not obvious why B. subtilis does not express these genes constitutively. Fig. 4. (A) Regulatory mechanisms relevant for control of metabolic fluxes during the two shifts. Key B. subtilis needs to optimize its use of two controlling enzymes are indicated by a positive correlation between reaction rate and enzyme abundance qualitatively distinct substrates: Glucose results after addition of malate (left) or glucose (right). Reactions were considered only if both enzyme abundance in high growth yields at low metabolic rates, and flux changed by more than one standard deviation and if the correlation coefficient was positive. The whereas malate results in low growth yields at size of the purple circles is proportional to the correlation coefficient (see inset). Transcription factors high metabolic rates (13). Similar trade-offs exist (middle) were identified by network component analysis to be responsible for changes in enzyme expression, between respiratory and fermentative strategies as indicated by color coding (gray denotes unknown factors). (B) Dynamics of glucose uptake rate upon in adenosine triphosphate generation, both of addition of glucose to a culture growing exponentially on malate in mutants with constitutive ptsG expres- which can confer condition-specific evolution- sion (green), cggR deletion (blue), and the respective parent strains (solid red, vertical shading and dashed ary advantages (22). To investigate optimal con- red, horizontal shading). Shaded areas denote 95% pointwise confidence intervals. gDCW,gramsofdrycell weight. (C) Predicted growth effects of different control strategies using a simplified dynamic model. In silico trol strategies for substrate usage in Bacillus,we competition experiments were carried out between two strains that differ in how they control malate and developed a simplified dynamic model that de- glucose uptake; both systems can be expressed constitutively or only upon substrate availability (SOM 6). scribes the substrate and biomass dynamics of Cocultures in a chemostat setting were pulsed every 3 hours with substrate (similar results were obtained with the shift experiments quantitatively (SOM 6). pulse frequencies of 1 and 9 hours), where the probability of selecting malate versus glucose as substrate was The maintenance cost for enzyme expression (r) varied. The cost of metabolic system maintenance (r) normalized by the estimated cost for B. subtilis (r0;the is a key parameter in the model; the estimated solid line indicates the estimated value of r; the dashed line to the right is the upper bound of the 95% value of r0 ≈ 0.06 implies that full expression confidence interval) constituted a free simulation parameter. Control strategies with higher growth rates were of a metabolic (substrate) system would reduce considered advantageous, and the best control strategies are identified by color coding. the maximal specific growth rate by ≈5%. In silico
1102 2 MARCH 2012 VOL 335 SCIENCE www.sciencemag.org REPORTS competition of strains that either constitutively ex- its evolutionary advantages; however, these ad- 16. S. Haykin, Kalman Filtering and Neural Networks press the metabolic systems or induce their ex- vantages, and therefore the overall control design, (Wiley-Interscience, New York, 2001). — 17. A. L. Sonenshein, Nat. Rev. Microbiol. 5, 917 (2007). pression only upon substrate availability allowed depend on quantitative system characteristics 18. T. Doan, S. Aymerich, Mol. Microbiol. 47, 1709 (2003). us to quantify the dynamic control effects. In sim- regulation is not beneficial per se. The dynamic 19. A. Goelzer et al., BMC Syst. Biol. 2, 20 (2008). ulations of a dynamically changing environment, data presented here may be used for further com- 20. J. C. Liao et al., Proc. Natl. Acad. Sci. U.S.A. 100, 15522 we varied the substrate probabilities and the main- putational analyses such as multivariate statistics (2003). r 21. S. Rossell et al., Proc. Natl. Acad. Sci. U.S.A. 103, 2166 tenance cost because of the uncertainties in the and large-scale structural or kinetic network mod- (2006). estimated r0 (Fig. 4C). Surprisingly, there almost els (24). We hope that our publicly available 22. T. Pfeiffer, S. Schuster, S. Bonhoeffer, Science 292, 504 always existed a unique, evolutionarily dominating tools for mathematical analyses and modeling will (2001). control strategy. The winning strategy depended facilitate future large-scale and dynamic systems 23. H. P. Bais, T. L. Weir, L. G. Perry, S. Gilroy, J. M. Vivanco, on quantitative parameters. Hence, active regula- biology studies in B. subtilis and other species. Annu. Rev. Plant Biol. 57, 233 (2006). 24. M. Heinemann, U. Sauer, Curr. Opin. Microbiol. 13, 337 tion or constitutive subsystem expression are not (2010). advantageous per se. Even with the uncertainties in References and Notes 1. U. Sauer, M. Heinemann, N. Zamboni, Science 316, 550 Acknowledgments: This work was funded through the r0, the experimentally observed strategy of consti- (2007). tutive malate usage capacity (Fig. 2B) but in- European Commission 7th Framework project BaSysBio 2. M. Ralser et al., Nat. Biotechnol. 27, 604 (2009). (LSHG-CT-2006-037469), coordinated by P. Noirot, the duced (and delayed) glucose-specific metabolism 3. G. Chechik et al., Nat. Biotechnol. 26, 1251 (2008). Department of Education, Science and Training (CG110055), confers an evolutionary advantage if B. subtilis 4. N. Ishii et al., Science 316, 593 (2007). and the National Health and Medical Research Council predominantly encounters malate (Fig. 4C). This 5. E. Yus et al., Science 326, 1263 (2009). (455646). Data, algorithms, and mathematical models have 6. M. Güell et al., Science 326, 1268 (2009). is biologically plausible given the organism’shab- been deposited in a publicly accessible relational database 7. J. F. Moxley et al., Proc. Natl. Acad. Sci. U.S.A. 106, (www.basysbio.eu/nutrientshift/ and table S1). itat in the vicinity of plant root systems that often 6477 (2009). secrete carboxylic acids such as malate (13, 23). 8. A. B. Canelas et al., Nat. Comm. 1, 145 (2010). Supporting Online Material Our systems approach helps reveal how pre- 9. M. T. A. P. Kresnowati et al., Mol. Syst. Biol. 2, 49 (2006). www.sciencemag.org/cgi/content/full/335/6072/1099/DC1 10. M. J. Brauer et al., Proc. Natl. Acad. Sci. U.S.A. 103, Materials and Methods viously known regulatory mechanisms are com- 19302 (2006). SOM Text (SOM 1 to 6) bined to effect nutritional transitions. Despite 11. S. Jozefczuk et al., Mol. Syst. Biol. 6, 364 (2010). Figs. S1 to S20 more than half of the B. subtilis gene comple- 12. P. H. Bradley, M. J. Brauer, J. D. Rabinowitz, Tables S1 to S8 ment being involved in the adaptive response O. G. Troyanskaya, PLOS Comput. Biol. 5, e1000270 (2009). References (25–28) 13. R. J. Kleijn et al., J. Biol. Chem. 285, 1587 (2010). Database S1 on July 10, 2012 to glucose, our methodology could discern the 14. E. Botella et al., Microbiology 156, 1600 (2010). key regulatory events. The overall control strat- 15. Materials and methods are available as supporting 12 April 2011; accepted 11 January 2012 egy of B. subtilis can be rationalized in terms of material on Science Online. 10.1126/science.1206871
quantitative exploration of transcriptome changes Condition-Dependent Transcriptome in Bacillus subtilis, whose natural habitat, the soil, is subject to severe environmental fluctua-
Reveals High-Level Regulatory tions (6). B. subtilis is also a laboratory model for www.sciencemag.org Gram-positive bacteria and is grown in industrial- Bacillus subtilis scale fermentors for the production of enzymes Architecture in and vitamins. We selected 104 conditions covering
1 2,3 4 4 1 Pierre Nicolas, * Ulrike Mäder, * Etienne Dervyn, * Tatiana Rochat, Aurélie Leduc, 1 4 4 4 1 INRA, UR1077, Mathématique Informatique et Génome, Nathalie Pigeonneau, Elena Bidnenko, Elodie Marchadier, Mark Hoebeke, F-78350 Jouy-en-Josas, France. 2Institute for Microbiology, 4 2 5 5 4 Stéphane Aymerich, Dörte Becher, Paola Bisicchia, Eric Botella, Olivier Delumeau, Ernst-Moritz-Arndt University Greifswald, D-17489 Greifs- 6 7 8 1 1 wald, Germany. 3Interfaculty Institute for Genetics and Functional Geoff Doherty, Emma L. Denham, Mark J. Fogg, Vincent Fromion, Anne Goelzer, Downloaded from 5 9 10 3 11 Genomics, Ernst-Moritz-Arndt-University Greifswald, D-17489 Annette Hansen, Elisabeth Härtig, Colin R. Harwood, Georg Homuth, Hanne Jarmer, 4 4 12 4 4 6 Greifswald, Germany. INRA, UMR1319 Micalis, F-78350 Matthieu Jules, Edda Klipp, Ludovic Le Chat, François Lecointe, Peter Lewis, Jouy-en-Josas, France. 5Smurfit Institute of Genetics, Trinity College 12 9 7 3 Wolfram Liebermeister, Anika March, Ruben A. T. Mars, Priyanka Nannapaneni, Dublin, Dublin 2, Ireland. 6School of Environmental and Life David Noone,5 Susanne Pohl,10 Bernd Rinn,13 Frank Rügheimer,14 Praveen K. Sappa,3 Sciences, University of Newcastle, Callaghan, NSW 2308, 7 Franck Samson,1 Marc Schaffer,2 Benno Schwikowski,14 Leif Steil,3 Jörg Stülke,15 Australia. Department of Medical Microbiology, University 16 5 8 7 of Groningen and University Medical Center Groningen, 9700 Thomas Wiegert, Kevin M. Devine, Anthony J. Wilkinson, Jan Maarten van Dijl, 8 2 3 1 4 RB Groningen, Netherlands. York Structural Biology Labora- Michael Hecker, Uwe Völker, Philippe Bessières, Philippe Noirot † tory, Department of Chemistry, University of York, York YO10 5YW, UK. 9Institute of Microbiology, Technical University of 10 Bacteria adapt to environmental stimuli by adjusting their transcriptomes in a complex manner, Braunschweig, D-38106 Braunschweig, Germany. Centre for the full potential of which has yet to be established for any individual bacterial species. Here, we report Bacterial Cell Biology, Institute, of Cell and Molecular Bio- sciences, Newcastle University of Newcastle, Newcastle upon the transcriptomes of Bacillus subtilis exposed to a wide range of environmental and nutritional Tyne, NE2 4AX, UK. 11Center for Biological Sequence Anal- conditions that the organism might encounter in nature. We comprehensivelymappedtranscription ysis, Department of Systems Biology, Technical University of Denmark, 2800 Kgs. Lyngby, Denmark. 12Theoretical Biophysics, units (TUs) and grouped 2935 promoters into regulons controlled by various RNA polymerase sigma 13 factors, accounting for ~66% of the observed variance in transcriptional activity. This global Humboldt University Berlin, 10115 Berlin, Germany. Center for Information Sciences and Databases, Department of Bio- classification of promoters and detailed description of TUs revealed that a large proportion of the systems Science and Engineering, ETH Zurich, 4058 Basel, detected antisense RNAs arose from potentially spurious transcription initiation by alternative sigma Switzerland. 14Systems Biology Lab and CNRS URA 2171, factors and from imperfect control of transcription termination. Institut Pasteur, 75724 Paris cedex 15, France. 15Department of General Microbiology, Georg-August-University Göttingen, D-37077 Göttingen, Germany. 16FN Biotechnologie, Hochschule acterial transcriptomes are surprisingly vironmental conditions have been investigated, Zittau/Görlitz, D-02763 Zittau, Germany. – complex (1 5) and include diverse and the full extent of transcriptome complexity remains *These authors contributed equally to this work. Babundant small RNAs and antisense RNAs to be established for a single bacterial species. †To whom correspondence should be addressed. E-mail: (asRNAs). Because only a small number of en- This prompted us to undertake a systematic and [email protected]
www.sciencemag.org SCIENCE VOL 335 2 MARCH 2012 1103
Article VI
Page 131
Page 132
ARTICLE
pubs.acs.org/ac
Efficient, Global-Scale Quantification of Absolute Protein Amounts by Integration of Targeted Mass Spectrometry and Two-Dimensional Gel-Based Proteomics † † € † †,‡ § † Sandra Maass, †Susanne Sievers,† Daniela Zuhlke,† Judith Kuzinski,§ Praveen K.† Sappa, Jan Muntel,† Bernd Hessling, J€org Bernhardt, Rabea Sietmann, Uwe V€olker, Michael Hecker, and D€orte Becher*, † Institute for Microbiology, Ernst-Moritz-Arndt-University Greifswald, Greifswald, Germany ‡ Leibniz Institute for Farm Animal Biology, Dummerstorf, Germany § Interfaculty Institute for Genetics and Functional Genomics, Ernst-Moritz-Arndt-University Greifswald, Greifswald, Germany bS Supporting Information
ABSTRACT: Knowledge on absolute protein concentrations is mandatory for the simulation of biological processes in the context of systems biology. A novel approach for the absolute quantification of proteins at a global scale has been developed and its applicability demonstrated using glucose starvation of the Gram-positive model bacterium Bacillus subtilis and the pathogen Staphylococcus aureus as proof-of-principle examples. Absolute intracellular protein concentrations were initially determined for a preselected set of anchor proteins by employ- ing a targeted mass spectrometric method and isotopically labeled internal standard peptides. Known concentrations of these anchor proteins were then used to calibrate two-dimensional (2-D) gels allowing the calculation of absolute abundance of all detectable proteins on the 2-D gels. Using this approach, concentrations of the majority of metabolic enzymes were determined, and thus a quantification of the players of metabolism was achieved. This new strategy is fast, cost-effective, applicable to any cell type, and thus of value for a broad community of laboratories with experience in 2-D gel-based proteomics and interest in quantitative approaches. Particularly, this approach could also be utilized to quantify existing data sets with the aid of a few standard anchor proteins.
odeling the complexity of a cell by not only measuring the choice for quantification, e.g., by comparing protein abun- Msingle matrix components such as RNAs, proteins, meta- dance indices which take into account the number of sequenced bolites, or fluxes, but rather by integrating these data and also peptides per protein,7 by relating averages of ion signal intensities capturing the interactions between these different layers is in the of the three most abundant peptides of proteins,8 or by measur- 1 focus of systems biology approaches. Such modeling efforts ing spectral counts which represent the number of matched MS/ absolutely depend on quantitative data, also for proteins because MS spectra for a protein.9 Such approaches can provide estima- they represent the central players in the complex cellular meta- tions of protein concentrations on a global scale, which has 2 bolic and adaptational network. recently been convincingly demonstrated by quantitation of Targeted mass spectrometric analyses are presently the meth- 3,4 about half of the predicted proteome of the human pathogen od of choice to determine absolute protein concentrations. Leptospira interrogans.10 Yet another and MS-independent work- fi Gerber et al. pioneered this eld utilizing synthetic peptides with flow was utilized to quantify the proteome of Saccharomyces incorporated stable isotopes as internal standards to quantify 11 5 cerevisiae. All open reading frames of this yeast were tagged with cellular peptides generated by proteolysis. Such pairs of native a high-affinity epitope and their expression followed by quanti- and heavy peptides are typically analyzed in triple-quadrupole- tative Western blotting. Even if this approach was extremely type mass spectrometers implementing multiple reaction mon- successful it imposed a tremendous workload and required the itoring (MRM)—a highly selective and sensitive acquisition 6 availability of a strain library containing tagged versions of all mode. However, reaching a systemswide perspective by quanti- yeast-specific proteins, which considerably limits easy application fying all proteins of interest constitutes a considerable workload to other species. and financial burden, which can currently only be handled by a few laboratories which have the resources of addressing such Received: December 3, 2010 analysis of absolute protein abundance at a proteome-wide scale. Accepted: February 21, 2011 Currently, entirely mass spectrometry-based strategies would be Published: March 11, 2011
r 2011 American Chemical Society 2677 dx.doi.org/10.1021/ac1031836 | Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE
The study presented here introduces a new workflow, which is cylinder and two hemispheres for the rod-shaped Bacillus cells. easy to use, does neither require genetic manipulation of cells nor Cell size determination of S. aureus cells using scanning electron exhaustive technical resources, and is thus easily applicable by a microscopy was carried out as follows: After a fixation step (2 h in broad proteomics community. Such easy, large-scale determina- 2.5% glutaraldehyde [v/v], 5 mM HEPES [pH 7.5], and 50 mM - tion of absolute protein concentrations was achieved by integrat- NaN3), samples were treated with amino acid sucrose solution ing targeted mass spectrometric analyses of selected anchor (2% arginine [w/v], 2% glycine [w/v], 2% glutamate [w/v], 2% proteins with quantification of all proteins accessible on cali- sucrose [w/v]) for 1.5 h, guanidine-tannin solution (2% brated two-dimensional (2-D) gels. guanidine [w/v], 2% tannin [w/v]) for 1.5 h, 1% osmium tetroxide (w/v) for 2 h, and 2% uranyl acetate (w/v) for 2 h ’ MATERIALS AND METHODS with washing steps in between. The samples were dehydrated in a graded series of aqueous ethanol solutions (10-100% [v/v]) Comparison of Protein Staining Methods. The five protein and then critical point-dried via amylacetate and CO2. Finally, dyes Coomassie Brilliant Blue (Sigma), Coomassie Blue Silver samples were mounted on aluminum stubs, sputtered with gold, (Sigma), Sypro Ruby (Invitrogen), Flamingo (Bio-Rad), and and examined in a DSM 940A electron microscope (Carl Zeiss Krypton (Thermo Scientific) were tested via two different AG, Germany). Cell length and width was determined with methods for their staining characteristics in respect to sensitivity, ImageJ software.13 The volume of S. aureus cells was calculated dynamic range, and protein sequence specificity. using the equation for prolate spheroids. fi ff - In a rst approach di erent concentrations (2.5 2500 ng At least 100 B. subtilis and S. aureus cells were dimensioned for -3 fi β mm nal) of bovine serum albumin (BSA), -lactoglobulin every sample. Therefore, a standard deviation could be calculated β ( -LG), and chicken serum albumin (CSA) were each spiked for each analyzed population. into 1 mL of gel solution and polymerized in a MiniProtean II cell Further Experimental Details. Further experimental details (Bio-Rad) equipped with 1 mm spacers. In the second approach on cell culture, protein preparation, determination of protein - μ β varying concentrations (0.0005 0.25 g) of -galactosidase, concentration, heavy peptides and protein digestion, targeted β BSA, CSA, carbonic anhydrase, -LG, and lysozyme were mixed MS analysis and global absolute protein quantification, and and separated via 1D-PAGE in a MiniProtean 3 Dodeca cell (Bio- validation are provided as Supporting Information. Rad), also equipped with 1 mm spacers. Gels and gel pieces were stained with the different dyes according to the manufacturer’s instructions. After image acquisition gray values were determined ’ RESULTS using ImageQuant 5.2 (GE Healthcare). For the first time reported, we combined targeted mass Two-Dimensional Polyacrylamide Gel Electrophoresis. spectrometry for the accurate determination of absolute protein Two-dimensional polyacrylamide gel electrophoresis (2-D fi 12 abundance of speci c proteins with the resolving power of 2-D PAGE) was performed as previously described using IPG strips PAGE to obtain a global view of a large number of protein - in the range of 4 7 (GE Healthcare). For protein extracts of concentrations. To illustrate the power of this approach we have μ Bacillus subtilis 24 cm IPG strips loaded each with 100 g chosen adaptation of B. subtilis, the model bacterium of Gram- were used. positive bacteria, and the human pathogen S. aureus, to glucose fi After 2-D PAGE gels were xed with 40% (v/v) ethanol and starvation as proof-of-principle examples. We determined pro- - 10% (v/v) acetic acid for 1 2 h and subsequently stained with tein concentrations at three different time points along the Flamingo (three to four technical replicates). For 2-D PAGE of bacterial growth curves—during exponential growth (t0), in μ Staphylococcus aureus samples, 70 g of protein was loaded onto early stationary phase (t1), and in late stationary phase (t2). fi 18 cm IPG strips, and 2-D gels were xed (see above) and stained Quantitative Calibration of 2-D Gels via Targeted MS. First fi with Krypton ( ve technical replicates). Stained gels were of all we selected several proteins to determine their absolute scanned (Typhoon 9400, GE Healthcare), and their images were abundance by spiking protein samples with isotopically labeled analyzed employing Delta2D 4.0 software (Decodon GmbH, peptides having the same amino acid sequence as the native Germany). A normalized spot volume in percent of intensity of peptides of the proteins chosen. We digested samples in solution all spots detected on the gel was assigned to each protein. with trypsin and analyzed the resulting peptide mixtures via Normalized volumes of multiple spots of the same protein were targeted MRM acquisition. The preselected proteins, which we added up. Standard deviations for such proteins were calculated called anchor proteins, had to fulfill the following selection as follows: criteria: (I) presence on 2-D gels of all conditions examined, qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi (II) arrangement in good quality spots on gels which can be 2 2 2 SD ¼ ðSD1Þ þðSD2Þ þ ::: þðSDnÞ reliably detected by custom software packages, and (III) the sum of anchor spots should cover the molecular mass and isoelectric point range examined. The peptides of anchor proteins selected Efficiency of Cell Disruption and Determination of Cell for targeted MS analysis fulfilled length and sequence criteria Size. To estimate cell lysis efficiency the number of intact cells described in detail elsewhere.14 Such targeted MS analysis was counted directly after sampling and cell disruption using a yielded the fractions of the anchor protein of the total protein light microscope (Axiolab re, Carl Zeiss AG, Germany). Bacterial content of the sample. Hence, the absolute amount of anchor cell size was determined at the different stages of growth. For this proteins on 2-D gels could be calculated, provided that the purpose B. subtilis cells were photographed with a size scale using identical protein extract was used for 2-D PAGE and MRM and a light microscope (Axiolab re) coupled to a digital camera the protein content was determined with a protein-composition- (ProgRes C5, Jenoptik AG, Germany). Length and width of the independent assay. After comparative assessment of the tradi- cells were determined on the photographs with the help of tional Coomassie-based Bradford assay and a ninhydrin-based ImageJ software.13 Volumes of cells were calculated assuming a assay, we have chosen the latter for determination of protein
2678 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE
that the two highly sensitive fluorescent dyes Krypton and Flamingo met the requirements of sequence unspecificity, sensi- tivity, and wide dynamic range in all tests (Figure 2 and Supporting Information S-1), and therefore they were used in this study. Anchor Proteins of B. subtilis and S. aureus. Seven proteins of B. subtilis were selected to be anchors—CitZ, Icd, Mbl, PyrR, SdhA, Upp, and YkrZ (Supporting Information Figure S-2). For S. aureus AtpA, GroEL, RplJ, SACOL2053, and Upp met the requirements to be used as anchors proteins (Supporting In- formation Figure S-2). The absolute quantification of peptides of anchor proteins via targeted MS (three technical replicates for B. subtilis, four technical replicates for S. aureus) revealed a high reproducibility. Only peptides of the B. subtilis CitZ and Icd in sample t0 displayed an error higher than 15% standard deviation (SD) leading to the exclusion of these two proteins from the list of anchors for this time point. For the calibration of 2-D gels we excluded proteins showing a SD higher than 35% among the technical replicates of 2-D gels.For this reason SACOL2053 hadtobeomittedasananchorprotein.Asaresultfour anchor proteins remained for the quantification of S. aureus proteins. Sequences of quantified peptides and their optimized MRM acquisition parameters are provided for B. subtilis and S. aureus in the Supporting Information (Tables S-1a and S-1b). To determine how accurate our strategy indeed captured absolute abundance of all proteins on a gel we bootstrapped the MRM-based result, i.e., the real absolute quantity obtained for a specific anchor protein was compared with the value calculated from spot intensity comparisons to the other anchor Figure 1. Workflow for absolute quantification of proteins by calibra- proteins on the gel (Supporting Information Figure S-3a). The tion of 2-D gels via targeted mass spectrometry: (a) all steps involved in bootstrapping provided an average error rate of 1.77-fold for B. sample preparation; (b) partial parallel analysis of the sample by targeted subtilis and 1.36-fold for S. aureus considering all time points and mass spectrometry and 2-D PAGE for absolute quantification. all anchor proteins. The maximal deviations between real abso- lute amount and a calculated value was a factor of 2.54-fold (Mbl at t0) and 2.06-fold (RplJ at t0), respectively. Furthermore, we content because it did not show any protein specificity and higher validated our quantification approach with an independent test sensitivity (data not shown). set of proteins that were available through a QconCAT analysis. In the next step we calibrated a 2-D gel based on the absolute The heavy-labeled QconCAT protein covered eight glycolytic amount of anchor proteins on the gel. Such an MS-calibrated 2-D enzymes and the pyruvate dehydrogenase complex of B. subtilis gel constitutes the basis for an estimation of the absolute (Supporting Information Table S-2). SRM analysis using the abundance of all proteins visible on this gel just by relating their QconCAT method (Supporting Information text and Figure spot intensities to the ones of anchor proteins (Figure 1). S-3b) was performed with the protein extracts of exponential The application of this workflow is based on several assump- ffi (t0) and early stationary phase (t1). In total 12 proteins were tions: (I) E ciency of protein digestion prior to MRM analyses is quantified and the absolute concentrations of the SRM assay supposed to be 100%. Although this value cannot really be compared with the values derived from the anchor-protein-aided achieved digestion protocol was optimized in terms of the calibration of the 2-D gel. Enolase (Eno) has to be omitted from digestion enzyme used, the enzyme/protein ratio, the duration the analysis due to improper chromatographic peaks and miss ff of digestion, and the composition of the digestion bu er reaching cleavages within the QconCAT protein. For glyceraldehyde- ffi always a digestion e ciency higher than 95% (data not shown). 3-phosphate dehydrogenase (GapA) the calculated error was fi (II) An additional postulate is that there is no protein-speci c significantly higher than the average error, which can probably be bias regarding the rehydration to the IPG-strip, isoelectric attributed to the fact that GapA can be detected in numerous focusing, and the transfer to the second dimension, which is spots overlapping with other proteins in the 2-D gel, thus especially important for the anchor proteins. On the basis of this compromising its quantification on 2-D gels in general. For the assumption amounts of anchor proteins were calculated as remaining 10 out of the 12 proteins the data fitted very well relative proportions of total protein extracts separated, thus yielding an average error of 1.97-fold. Thus, the approach rendering calculations independent of fractional losses of protein presented here reaches a similar precision as other current extracts. (III) Our approach also assumed a protein staining on methods for estimating absolute protein abundance on a pro- the gel which was independent of protein sequence and linear teome-wide scale.10 over a wide range of protein levels. Many commercially available Determination of Cellular Protein Concentrations. We dyes such as Coomassie do not meet these criteria. Experiments analyzed every sample that was harvested from the bacterial for the selection of appropriate staining procedures have shown culture with respect to the number of cells present in a given
2679 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE
- Figure 2. Dynamic range and protein sequence specificity of protein staining. Known in-gel concentrations (ng mm 3) of three different proteins [bovine serum albumin (BSA), β-lactoglobulin (β-LG), and chicken serum albumin (CSA)] are plotted against the respective signal intensities obtained - by staining (counts mm 3) with Coomassie silver blue (SB), Coomassie brilliant blue (BB), Sypro ruby, Flamingo, and Krypton (a). In a second approach (b) varying concentrations of six proteins [β-galactosidase, BSA, CSA, carbonic anhydrase (CA), β-LG, and lysozyme] were separated via 1D- PAGE and stained with the same dyes. volume of medium. Here incomplete cell lysis can be a possible after harvest but also after lysis of cell extracts. Actual efficiencies source of error. Therefore, we optimized cell lysis prior to this for both bacteria and each time point were always better than study aiming at a disruption efficiency of 100%. Additionally, we 99% for B. subtilis and 95% for S. aureus (Supporting Information determined disruption efficiency by not only counting intact cells Table S-3). Knowing the number of cells per volume culture as
2680 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE well as the disruption efficiency of the sample we could derive protein copy numbers per cell. Absolute protein amount per microgram of crude protein extract as well as copy numbers per cell for all proteins detected on 2-D gels can be found in the Supporting Information (Table S-4). Additionally, the intracel- lular concentration of the proteins is provided. This value could be calculated after a precise determination of the average size of B. subtilis and S. aureus cells at the different physiological con- ditions employing light and electron microscopy, respectively (Supporting Information Tables S-3 and S-5). The information on cellular protein concentration is just as important as knowing the concentration of metabolites, e.g., for metabolic flux analysis representing a major branch in the systems biology approach. The numbers obtained in this gel-based absolute quantification prove the wide dynamic range of our method. During exponen- tial growth the elongation factor Ef-Tu (gene product of tufA) represents the most abundant protein in B. subtilis with about 100 000 molecules per cell. A recent publication on absolute protein abundance in Escherichia coli, a bacterium of similar volume and proteome size as B. subtilis, determined the concen- Figure 3. Comparative illustration of traditional visual presentation of tration of elongation factor EF-Tu with about 90 000 cellular 2D data and absolute quantification by combination of MRM, protein 7 copies. In S. aureus, the alkaline shock protein 23 (Asp23) and quantification, and image analysis. Data for four proteins involved in superoxide dismutase SodA2 proved to be most abundant in central carbon metabolism [triose phosphate isomerase (TpiA), phos- growing cells with copy numbers of about 25 000 per cell. The phoglycerate kinase (Pgk), succinate dehydrogenase flavoprotein sub- elongation factor EF-Tu is present with 13 000 molecules per cell unit (SdhA), 2-oxoglutarate dehydrogenase E1 subunit (OdhA/SucA)] and is therefore one of the five most abundant proteins in S. of B. subtilis (left) and S. aureus (right) are exemplarily illustrated to aureus. demonstrate the power of the analysis. Appearance of the corresponding Protein estimation was also quite accurate for proteins of lower protein spots on 2-D gels is illustrated in spot tiles showing enlarged sections of image overlays. False color overlays are utilized to visualize abundance. Unfortunately, for B. subtilis and S. aureus absolute the trend in protein regulations. The sample corresponding to expo- concentrations have so far only been reported for a few proteins nential growth was colored green, and all others are red. Thus, shades of in literature, and thus, we cite here only ClpC as an example. The yellow indicate almost equal amounts in the samples compared, whereas number of ClpC molecules in growing cells was previously shades of green and red indicate overrepresentation during exponential determined for B. subtilis with 1500 molecules per cell,15 and phase or stationary phase, respectively. Additionally, the absolute - the study presented here yielded 700 molecules per cell, which is concentration in ng μg 1 of total protein and in molecules per cell well within the error tolerance of this method. (rounded) are given for the four proteins in growing cells (t0) as well as However, employing the workflow presented, we were also cells in early (t1) and late (t2) stationary phase. able to quantify proteins present at much lower concentrations. We have assessed DNA ligase (LigA) of S. aureus to exist in less data on glucose starvation in B. subtilis and S. aureus.16,17 As reported, than 20 copies during exponential growth. In B. subtilis DNA a down-regulation of glycolysis with simultaneous up-regulation of polymerase X (PolX) was calculated to be present with about 100 the tricarboxylic acid cycle and induction of gluconeogenesis are the molecules in every cell. Thus, with the method reported here a most pronounced regulatory effects in cells starving for glucose. A range of cellular protein abundances of about 3 orders of selection of a few regulated proteins illustrates the agreement of our magnitude can be covered, a range similar to that previously strategy with the traditional visual inspection of 2-D gels for each of obtained for E. coli.9 The minimal molecule numbers per cell the three time points analyzed (Figure 3). calculated were even lower (down to two molecules per cell for Some of the quantified proteins are also part of known protein the HexB protein of S. aureus). However, due to standard complexes for which stoichiometric information is available. To deviations higher than 35% these values are not as precise as additionally prove the accuracy of the method, the protein ratios the ones for proteins present at higher concentrations. Details on of components of complexes determined in this study were the assignment of calculated protein copy numbers per cell and compared to previously published values (Table 1). In agree- - mass fraction (ng μg 1 protein extract) of B. subtilis and S. aureus ment with previously reported data we observed the same ratios to different abundance classes are provided in the Supporting for components of the 2-oxoglutarate dehydrogenase complex Information (Figure S-4). The spread of cellular molecule and the pyruvate dehydrogenase complex.18,19 The same is true numbers is comparable to previously published mass-spectro- for the ratio of 1:4 observed for the proteins RplJ and RplL of the metry-based data on E. coli.9 large ribosomal subunit corresponding well to the literature The proteins determined to have the highest cellular concen- information obtained in E. coli.20,21 This also extends to the tration were usually involved in energy metabolism or protein RpsB/RpsF ratio in S. aureus. Interestingly, the RpsBB/RpsF biosynthesis (Supporting Information Figure S-5), whereas low- ratio in B. subtilis (2.4 to 1) was different to the one observed in E. coli abundant proteins often function in DNA replication and repair, which might indicate some special feature in B. subtilis. However, so which is in good accordance with previous findings.7 far this is speculative and awaits further experimental support. Further evidence for the reliability of the data presented is Transferability of Absolute Data between 2-D Gels. Having provided by the analyses of the dynamics of protein pattern in shown the feasibility to quantitatively calibrate entire 2-D gels on growing and glucose-starved cells compared to formerly published the basis of a comparatively small number of mass spectrometric
2681 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE
a Table 1. Stoichiometric Information for Selected Proteins 482 proteins of S. aureus (504 proteins identified) at three time points during glucose starvation. This number is in the same B. subtilis S. aureus literaturea,b organism range as quantification of 449 E. coli proteins by the mass- 9 OdhA/OdhB 1.0:2.4 1:2 E. coli (ref 18) spectrometry-based APEX method but lower than the quanti- fi 7 SucA/SucB 1.0:2.7 1:2 E. coli (ref 18) cation by Ishihama et al. who employed a normalized spectral fi PdhA/PdhB/PdhC 0.6:1.0:1.0 1.0:1.3:1.0 1:1:1 Gram-positives counting approach (emPAI) for quanti cation of 1103 E. coli (ref 19) proteins. Although 2-D PAGE of this study only spanned a pH of 4-7, RplJ/RpIL 1.0:4.0 1.0:4.0 1:4 E. coli (refs 20 it included virtually all enzymes involved in the main metabolic RpsB/RpsF 2.4:1.0 1.2:1.0 1:1 and 21) fl a pathways providing valuable information for metabolic ux Stoichiometric composition of known protein complexes was deter- simulations. Whereas proteins featuring isoelectric points out- mined using the absolute quantification approach presented in this study - “ ” side the standard range of 4 7 could be included by using and then compared to previously published data (column literature ). - - Literature values were extracted from the references indicated and distinct IPG strips (e.g., pH 3 10 or pH 6 11), one has to point correspond to the organisms that have been listed. out that some protein classes, such as hydrophobic integral membrane proteins, will not be accessible via the technique measurements, it would be another advantage of the method if it presented here because these proteins cannot be separated by was possible to transfer absolute data from one gel to another just 2-D PAGE due to their inherent biophysical properties. Never- theless, usage of 2-D PAGE constitutes a major advantage by relating intensities of corresponding protein spots. The fi benefit of such an approach would be the potential to be able compared to exclusively MS-based quanti cation strategies and to easily quantify entire time course experiments, for instance, is still a powerful tool for physiological analyses on a global 22 fi but only performing targeted MS experiments for a few or even scale. Post-translational modi cations resulting in a shift of the for only one of the time points. To answer this question again a proteins pI or molecular mass can be visualized on a 2-D gel, bootstrapping procedure was applied. For this purpose we easily providing information on the absolute abundance of all assumed that an MS-calibrated 2-D gel was available only for individual forms of proteins which appear in multiple spots the exponential growth phase. 2-D gels of early (t1) and late (protein isoforms). The use of an expanded set of anchor stationary phase (t2) were overlaid to this gel, and all protein spots proteins would also allow transfer of this method to zoom gels offering a high resolution of narrow pH ranges which even were quantified in relation to the sample from the exponential ff growth phase (t0). Hence, one indirectly obtains absolute protein enhances di erentiation of protein isoforms and can also increase the number of illustratable and identifiable proteins in general. amounts for samples t1 and t2. Because in this study MS calibration fi fl of 2-D gels was actually performed for t1 and t2 as well (intragel The accuracy of protein quanti cation by the work ow quantification), the real protein abundance for the two time points introduced in this study is critically dependent on the ability to wereknownandcouldbecomparedtotheresultsgainedviarelative stain proteins on 2-D gels in a sequence-independent and quantification to sample t0 (intergel quantification). The correlation concentration-dependent manner within a wide dynamic range. With the use of precisely known concentration ranges of six of intra- and intergel quantification is displayed in Figure 4 for both fl organisms. Notably, good correlation was obtained for both quanti- marker proteins, the uorescent dyes Flamingo and Krypton were proven to fulfill the requirements mentioned above, thus fication strategies with only slight scattering of a few data points. In fi most cases these outliers represent proteins that occur in multiple providing the basis for quanti cation of a wide array of diverse spots and overlap with other protein spots on the 2-D gel. Although proteins on 2-D gels with the aid of a few anchor proteins. there is almost no scattering of data points, one can observe a peculiar Furthermore, a potential bias of results might also be caused by shift affecting all points of the data series in the sample representing saturation of protein staining on 2-D gels. However, the use of fluorescent dyes and optimized scan parameters for 2-D gels can the late stationary phase of B. subtilis (Figure 4c). This systematic fi deviation was possibly caused by an error when the heavy peptide circumvent this bias. Quanti cation of highly abundant proteins is also an issue in MS-based approaches since saturation effects mixture was spiked into the sample before MRM acquisition. An 7 actual concentration of heavy peptides lower than assumed would have also been observed previously. Thus, the introduced 2-D gel-based strategy as well as recently reported MS-based ap- result in an underestimation of anchor proteins on the 2-D gel and fi consequently of all proteins on this gel. Except for this systematic proaches make use of simpli cation of actually highly complex error, which is also an issue of purely MS-based quantification peptide and protein characteristics and can therefore only be an estimation of abundance for the majority of proteins. The strategies and which could be minimized by analyzing more fi replicates, the comparison of intra- and intergel quantification proves precision of this and other current protein quanti cation meth- the option of indirect absolute quantification of protein concentra- ods (an about 2-fold error among 3 orders of magnitude) is certainly sufficient for general proteomics studies but might still tions by transferring calibration data between different 2-D gels. On fl the basis of a highly reproducible, high-resolution 2-D PAGE we have need further improvement in metabolism or ux-oriented pro- thus introduced a fast, convenient, and low-cost method for the large- teomics where precise determination of protein stoichiometries matters. In future 2-D PAGE as well as MS-based approaches will scale estimation of absolute protein abundance of complex samples fi and even of entire sample series. bene t from increasing resolution which will further improve the results of large-scale absolute quantifications. However, the fl ff ’ work ow presented in this study enables a rapid, cost-e ective, DISCUSSION and convenient determination of protein concentrations on a Our assay introduces a novel approach for the global absolute large scale with an accuracy which can compete with the most quantification of proteins. The innovative implementation of current MS-based approaches.10 Even if the fraction of proteome MS-calibrated 2-D gels enabled reliable estimation of the abun- analyzed here is lower compared to recent MS-based studies,10 dance of 467 proteins of B. subtilis (494 proteins identified) and the use of 2-D gels for absolute quantification offers a nearly
2682 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE
Figure 4. Analysis of correlation between intragel and intergel quantification for proteins of stationary phase. Scatter plot display of absolute amounts of proteins on a 2-D gel determined by direct MS-based gel calibration (intragel, X-axis) and protein amounts resulting from gel-to-gel comparison (intergel, Y-axis). Results are shown for protein samples of the early stationary phase t1 (a and b) and late stationary phase t2 (c and d). Left diagrams give the correlation for B. subtilis (a and c), right diagrams for S. aureus (b and d). For intergel calculations protein amounts were inferred by always relating spot intensities to those of gels from directly quantified exponential growth phase extracts (t0). An expected perfect correlation of intragel and intergel quantification is illustrated by the red line. Data of proteins exhibiting a standard deviation among 2-D gels of more than 35% are indicated in gray. complete coverage of metabolic pathways and provides an easy ’ ASSOCIATED CONTENT visualization of changes in protein patterns. In the last years highly sophisticated software packages became bS Supporting Information. Additional information as available supporting an overlay of many different 2-D gel images and noted in text. This material is available free of charge via the hence also the relative quantificationofproteinspotsonthesegels. Internet at http://pubs.acs.org. Employing such tools it was possible to transfer absolute data of an MS-calibrated 2-D gel onto other gels. Therefore, one can even ’ AUTHOR INFORMATION estimate absolute protein abundance of samples which were not Corresponding Author analyzed by targeted MS directly, which significantly expands the *Phone: þ49-3834-864230. Fax: þ49-3834-864202. E-mail: application potential of this novel approach. [email protected].
’ CONCLUSION ’ ACKNOWLEDGMENT After the introduction of quantitative Western blotting,11 flow We are grateful to V. Liebscher for help with mathematical and cytometry,23 and purely MS-based strategies7,10,24 for a global statistical data analysis, A. Hartmann for introduction to the QTRAP determination of absolute protein abundance, we herewith 4000, and Marc Burian for designing and validating the QconCAT present a different approach which is ready to use for a large protein. B. Knoke and colleagues are acknowledged for support and number of proteomics laboratories. Compared to other methods performing batch fermentations. Furthermore, we thank S. Baronick it does not require the availability of large mutant libraries or the for help in evaluation of cell disruption techniques. We are indebted synthesis of hundreds of synthetic peptides and achieves an to E. Klotz and S. Grund for excellent technical assistance. We also accuracy similar to MS measurements. Thus, the novel approach thank Decodon GmbH for providing Delta2D software. B. Voigt and combines the accuracy of targeted MS with the resolving power D. Albrecht are acknowledged for support in protein digestion and and intuitive visualization of protein patterns by 2-D PAGE identification. We are grateful to P. Levis, G. Doherty, and P. Noirot thereby providing access to cellular protein concentrations for a for stimulating discussions. This work was supported by Grants from large set of proteins. Provided that 2-D PAGE can be performed the Deutsche Forschungsgemeinschaft (SFB/TRR34), the Bundes- in a reproducible manner, we are convinced that this simple and ministerium f€ur Bildung und Forschung (0313978A, 0315784A, low-cost strategy will greatly benefit a broad systems biology 0315592B and 03ZIK011), and the EU (BaSysBio LSHG-CT-2006- community. 037469). The first three authors contributed equally to this work.
2683 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684 Analytical Chemistry ARTICLE
’ REFERENCES (1) Aebersold, R. Mol. Syst. Biol. 2005, 1, 2005.0005. (2) Souchelnytskyi, S. Proteomics Clin. Appl. 2005, 5, 4123–4137. (3) Malmstr€om, J.; Lee, H.; Aebersold, R. Curr. Opin. Biotechnol. 2007, 18, 378–384. (4) Pan, S.; Aebersold, R.; Chen, R.; Rush, J.; Goodlett, D. R.; McIntosh, M. W.; Zhang, J.; Brentnall, T. A. J. Proteome Res. 2009, 8, 787–797. (5) Gerber, S. A.; Rush, J.; Stemman, O.; Kirschner, M. W.; Gygi, S. P. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 6940–6945. (6) Lange, V.; Malmstr€om, J. A.; Didion, J.; King, N. L.; Johansson, B. P.; Sch€afer, J.; Rameseder, J.; Wong, C. H.; Deutsch, E. W.; Brusniak, M. Y.; B€uhlmann, P.; Bj€orck, L.; Domon, B.; Aebersold, R. Mol. Cell. Proteomics 2008, 7, 1489–1500. (7) Ishihama, Y.; Schmidt, T.; Rappsilber, J.; Mann, M.; Hartl, F. U.; Kerner, M. J.; Frishman, D. BMC Genomics 2008, 9, 102–119. (8) Silva, J. C.; Gorenstein, M. V.; Li, G.; Vissers, J. P. C.; Geromanos, S. J. Mol. Cell. Proteomics 2006, 5, 144–156. (9) Lu, P.; Vogel, C.; Wang, R.; Yao, X.; Marcotte, E. M. Nat. Biotechnol. 2007, 25, 117–124. (10) Malmstr€om, J.; Beck, M.; Schmidt, A.; Lange, V.; Deutsch, E. W.; Aebersold, R. Nature 2009, 460, 762–765. (11) Ghaemmaghami, S.; Huh, W. K.; Bower, K.; Howson, R. W.; Belle, A.; Dephoure, N.; O’Shea, E. K.; Weissmann, J. S. Nature 2003, 425, 737–741. (12) B€uttner, K.; Bernhardt, J.; Scharf, C.; Schmid, R.; M€ader, U.; Eymann, C.; Antelmann, H.; V€olker, A.; V€olker, U.; Hecker, M. Electrophoresis 2001, 22, 2908–2935. (13) Abramoff, M. D.; Magelhaes, P. J.; Ram, S. J. Biophotonics Int. 2004, 11,36–42. (14) Mayya, V.; Rezual, K.; Wu, L.; Fong, M. B.; Han, D. K. Mol. Cell. Proteomics 2006, 5, 1146–1157. (15) Gerth, U.; Kirstein, J.; Mostertz, J.; Waldminghaus, T.; Miethke, M.; Kock, H.; Hecker, M. J. Biotechnol. 2004, 186, 179–191. (16) Bernhardt, J.; Weibezahn, J.; Scharf, C.; Hecker, M. Genome Res. 2003, 13, 224–237. (17) Kohler, C.; Wolff, S.; Albrecht, D.; Fuchs, S.; Becher, D.; B€uttner, K.; Engelmann, S.; Hecker, M. Int. J. Med. Microbiol. 2005, 295, 547–565. (18) Pettit, F. H.; Hamilton, L.; Munk, P.; Namihira, G.; Eley, M. H.; Willms, C. R.; Reed, L. J. J. Biol. Chem. 1973, 248, 2582–2590. (19) Neveling, U.; Bringer-Meyer, S.; Sahm, H. Biochim. Biophys. Acta 1998, 1385, 367–372. (20) Hardy, S. J. S. Mol. Gen. Genet. 1975, 140, 253–274. (21) Tal, M.; Weissman, I.; Silberstein, A. J. Biochem. Biophys. Methods 1990, 21, 247–266. (22) Hecker, M.; Antelmann, H.; B€uttner, K.; Bernhardt, J. Proteo- mics 2008, 8, 4958–4975. (23) Newman, J. R. S.; Ghaemmaghami, S.; Ihmels, J.; Breslow, D. K.; Noble, M.; DeRisi, J. L.; Weissmann, J. S. Nature 2006, 441, 840–846. (24) Masuda, T.; Saito, N.; Tomita, M.; Ishihama, Y. Mol. Cell. Proteomics 2009, 8, 2770–2777.
2684 dx.doi.org/10.1021/ac1031836 |Anal. Chem. 2011, 83, 2677–2684
Article VII
Page 141
Page 142
Title: Global relative quantification with LC-MALDI – cross-validation with LTQ-
Orbitrap proves reliability and reveals complementary ionization preferences
Bernd Hessling 1, Knut Büttner 1, Michael Hecker 1 and Dörte Becher 1
1Ernst-Moritz-Arndt-University of Greifswald, Institute for Microbiology, F.-L. Jahnstrasse 15, 17489
Greifswald, Germany
Corresponding author: Dörte Becher
email: [email protected]
Tel.: (+49)3834-864230
Fax: (+49)3834-864202
Running title: Global relative quantification with LC-MALDI
1
List of abbreviations
GeLC -MS 1D -gel liquid chromatography mass spectrometry
LB Luria -Bertani medium
S. aureus Staphyloco ccus aureus
SILAC Stable isotope labeling by amino acids in cell culture
TEAB triethylammonium bicarbonat
2
Summary
Quantitative LC-MALDI is still an underrepresented method, especially in large-scale experiments. The additional fractionation step that is needed for most MALDI-TOF-TOF instruments, the comparatively long analysis time and a lack of established software tools for the analysis of the data render LC-MALDI a niche application for large quantitative analysis besides the widespread LC-ESI workflows.
Here, we used LC-MALDI in a proteome-wide relative quantification analysis of Staphylococcus aureus .
Samples were analyzed in parallel with an LTQ-Orbitrap, which allowed cross-validation with a well- established workflow. With nearly 850 proteins identified in the cytosolic fraction and quantitative data for more than 550 proteins obtained with the MASCOT Distiller software, we could prove that LC-MALDI is able to process highly complex samples. The good correlation of quantities determined by this method and by the LTQ-Orbitrap workflow confirmed the high reliability of our LC-MALDI approach for global quantification analysis.
A comparative analysis of identified proteins and peptides obtained with either MS instrument revealed biases against certain biochemical properties for both MALDI-TOF-TOF and LTQ-Orbitrap based approaches. These biases result in a general complementarity of the two approaches.
3
Introduction
1D-gel based liquid chromatography mass spectrometry (GeLC-MS) is a well-established technique in life science. In combination with in vivo labeling approaches like SILAC (Stable isotope labeling by amino acids in cell culture) (1) or 15 N labeling (2) it allows the relative quantification of a large numbers of proteins in a complex sample. Mass spectrometry (MS) measurements in such workflows are predominantly performed with electro spray ionization (ESI) based mass spectrometers. Online coupling of a liquid chromatography (LC)-system with fast MS spectra acquisition and high mass accuracy ESI instruments in conjunction with fractionation on both protein and peptide level enable the analysis of very complex samples in a relatively short period of time (3). Identification and quantification of data can be done in an automatic or semi-automatic manner with a variety of well-established software packages (4).
In proteomic research matrix-assisted laser desorption/ionization time-of-flight tandem mass spectrometers (MALDI-TOF-TOF) are mainly used for the analysis of non-complex protein samples, which do not require fractionation on peptide level by liquid chromatography. Especially the analysis of single protein spots resulting from 2D-PAGE separation of complex samples is primarily carried out with this MS technique, as it allows the fast and reliable analysis of a high number of lowly complex samples
(5).
In most LC-MALDI workflows the LC system and the MALDI instrument are coupled offline with a fractionation step in between. Online measurements of the LC eluate is not appropriate for most MALDI systems as the mass analyzer is a closed vacuum chamber and sample insertion is based on an lock chamber, which precludes the direct injection of an LC run. Also the measurement speed of most MALDI instruments is too low to analyze samples of mid and high complexity online, when large numbers of peptides elute in a short time frame. The comparatively low throughput in terms of single spectrum
4 acquisition and the restricted online coupling for most instruments is generally circumvented by offline coupling of an LC system and a MALDI-TOF-TOF instrument through a fractionation system (6). The decoupling from the chromatographic process makes MS measurements independent of the instrument's scan cycle time. The only restriction is the sample consumption in the ionization process.
Besides counteracting too long cycle times of the mass spectrometer the offline coupling also enables multiple measurements of the same LC run and therefore allows the selective analysis of single precursor ions after a first analysis of the data (7).
LC-MALDI for qualitative analysis stands back behind the widespread LC-ESI approaches. This is mainly due to the additional fractionation step causing longer analysis times. The discrepancy in usage is even bigger for relative quantification workflows. LC-MALDI is rarely employed for the analysis of in vivo labeled samples, especially not in large-scale experiments. Even though it had been shown that shot-to- shot intensity variation, which is a general drawback for quantitative MALDI analysis, can be overcome with a suitable experimental setup (8), the lack of established software tools for the analysis of these data is evident and still hampers application of LC-MALDI in large-scale experiments.
Here, we describe for the first time a global analysis of in vivo 15 N labeled samples with a GeLC-MS/MS workflow carried out with a MALDI-TOF-TOF instrument. In this workflow proteins were prefractionated on a 1D-SDS gel, tryptically digested, and the resulting peptide mixtures were separated by reversed phase liquid chromatography. The LC eluate was then fractionated, which allowed offline coupling with a MALDI-TOF-TOF instrument. Data was analyzed with the Mascot Distiller software package. The same samples were also measured with an LTQ-Orbitrap as described in Hessling et al.1. This second data set from the very same samples allowed cross-validation with a well-established workflow.
1 Hessling,B., Herbst,F.-A., Bonn,F., Rappen,G.-M., Bernhardt,J., Hecker,M. and Becher,D. Global proteome analysis of vancomycin stress in Staphylococcus Aureus. Submitted to Mol. Cell Proteomics
5
We proved that LC-MALDI is an appropriate option for the quantitative analysis of in vivo labeled samples on a proteome-wide scale. We identified nearly 850 proteins and quantified more than 550 proteins within a reasonable analysis time. Resulting protein ratios are in good correlation with existing
LTQ-Orbitrap data and should encourage groups equipped with a MALDI-TOF-TOF instrument to perform large-scale quantitative proteomic experiments.
The measurement of the same samples with two mass analyzers, one using electro spray ionization and the other matrix assisted laser desorption/ionization, also allowed investigating possible biases of one or the other mass analyzer against peptides with certain physiochemical characteristics. Existing comparative studies in this field have been few so far, applied divergent sample preparations like different LC-systems for peptide fractionation (9) and were relatively small in terms of identification numbers (10, 11). The large amount of data and the avoidance of any technical variations in the present study allowed the so far most comprehensive comparison of ESI and MALDI generated data and revealed physicochemically based biases in the detection of peptides, which confirms the generally complementary nature of the two ionization techniques.
6
Materials and methods
Sample preparation and chromatography
The same samples analyzed in this study by a LC-MALDI workflow had already been analyzed by LTQ-
Orbitrap (Thermo Fisher Scientific, Bremen, Germany) in Hessling et al.1 There, S. aureus COL (12) was grown under agitation at 37°C in Luria-Bertani (LB) medium (Gibco BRL, Wiesbaden, Germany). At an
OD 540 of 0.5, vancomycin (Roth, Karlsruhe, Germany) was added to a concentration of 4.5 mg/l. This led to a decreased growth rate about 30 min after antibiotic addition. Cells were harvested 100 min after stress induction at an OD 540 of about 1.4. Unstressed control samples were grown in LB and harvested during exponential growth at an OD 540 of 1.4 as well. The cultivations were carried out in triplicates to obtain three biological replicates.
Also the 15 N-labeled standard used for quantification was the same as in Hessling et al. 1 and was added to every sample before the mass spectrometry measurement (13). This standard was a combined pool of vancomycin stressed cells and exponentially growing cells, grown in an 15 N enriched Bioexpress medium (Cambridge Isotope Laboratories, Andover, MA, USA), that was supplemented with 5 g/l glucose. The standard was mixed with equal amounts of vancomycin stressed, and ,respectively, unstressed samples grown in LB as early during sample preparation as possible. Any protein loss during the cell lysis, digestion and measurement will be accounted for by equally affecting the respective 15 N- labeled protein.
Harvested cultures were centrifuged for 15 min at 7,000 g at 4°C. Pelleted cells were then washed in TBS
(50mM Tris, 150 mM NaCL, pH 8.0) and resuspended in 50 mM triethylammonium bicarbonat buffer
(TEAB). 500 μL of glass beads with 0.1 mm diameter were added and cells were disrupted using the
Precellys 24 homogenizer (Peq Lab, Erlangen, Germany) for 30 s at 6,800 rpm. Cell debris and glass beads were separated from the proteins by centrifugation for 10 min at 4°C and 21,500 g. A second
7 centrifugation step of 30 min at 4°C and 21,500 g should have removed insoluble and aggregated proteins to gain the cytosolic fraction of soluble proteins. The protein concentration was determined with the Roti-Nanoquant protein assay (Roth, Karlsruhe, Germany) according to the manufacturer's instructions, and the same protein amount of 15 N labeled standard was added. The sample was analyzed by the GeLC-MS workflow described by Dreisbach et al. (14). The protein extract was separated by 1D
SDS PAGE; the gel was cut into 12 pieces, which were tryptically digested at 37°C over night. Resulting peptide mixtures were separated with reversed phase column chromatography (Waters BEH 1.7 μm,
100 μm inner diameter x 100 mm, Waters Corporation, Milford, Mass., USA) operated on a nanoACQUITY-UPLC (Waters Corporation, Milford, Mass., USA). Peptides were firstly concentrated and desalted on a trapping column (Waters nanoACQUITY UPLC column, Symmetry C18, 5 μm, 180 μm inner diameter x 20 mm, Waters Corporation, Milford, Mass., USA) for 3 min at a flow rate of 1ml/min with
99% buffer A (0.1% acetic acid). Subsequently the peptides were eluted and separated with a non-linear
80-min gradient from 5 to 60% acetonitrile (ACN) in 0.1% acetic acid at a constant flow rate of 400 nl/min.
The cell wall shaving samples, that were only qualitatively analyzed to investigate possible detection preferences dependent on the peptide's hydrophobicity, were also the same as used by Hessling et al. 1.
The membrane shaving protocol was carried out as described by Wolff et al. (15). The peptide containing solution was loaded on a nanoACQUITY UPLC System (Waters Corporation), equipped with an analytical column (nanoACQUITY UPLC column, BEH130 C18, 1.7 μm, 100 μm x 100 mm; Waters
Corporation) operated at 60°C at 400nl/min. Peptides were loaded directly on the column, and after washing for 30 min with 99% buffer A (0.1% acetic acid) the peptides were eluted from the column in a
5-h linear gradient from 5-90% buffer B (90% ACN, 0.1% acetic acid).
MS analysis
8
MALDI-TOF-TOF measurements
For MALDI-TOF-TOF measurements the LC was online coupled with the Probot Microfraction Collector
(Dionex GmbH, Idstein, Germany). The LC eluate was mixed with matrix containing solution and spotted on a MALDI target plate, with a spot collection time of 15 s. Spotted targets were subsequently subjected to the 5800-MALDI-TOF-TOF Analyzer and measured using TOF/TOF™ Series Explorer™
Software V4.1.0 (AB Sciex, Foster City, CA, USA). MS Spectra were recorded in a mass range from 700 to
4,000 Da with a focus mass of 1,700 Da. For one main spectrum 15 sub-spectra with 200 shots per sub- spectrum were accumulated using a random search pattern and continuous stage movement.
Up to 20 precursor ions per spot were selected for MS/MS measurement using a Job Wide
Interpretation Method, with a fraction-to-fraction precursor mass tolerance of 200 ppm. For one main
MS/MS spectrum up to 25 sub-spectra with 250 shots per sub-spectrum were accumulated. The
DynamicExit TM Algorithm was enabled using highest threshold settings. These settings resulted in about
25,000 to 30,000 MS and MS/MS spectra per biological replicate with a MS analysis time of 20 to 30 hours.
ESI-LTQ-Orbitrap measurements
For LTQ-Orbitrap measurements the LC system was coupled online with the mass analyzer. Analyses were performed as described in Hessling et al. 1.
Data analysis
MALDI data analysis using Mascot
MzML files were extracted from the instrument’s internal Oracle database using the MS Data Converter
Beta version 1.2 (AB Sciex, Foster City, CA, USA) and then processed applying the default MALDI-TOF-
TOF processing options and searched with the Mascot search engine V2.2 (Matrix science, London, UK)
9 using Mascot Distiller version 2.4.2. All files belonging to one sample were processed together in an independently processed multi-file project and searched against a S. aureus COL target-decoy protein sequence database. This database was composed of all protein sequences of S. aureus COL extracted from the National Center for Biotechnology Information (NCBI) bacteria genomes
(http://www.ncbi.nlm.nih.gov/sites/entrez?Db=%20genome&Cmd=Retrieve&dopt=Protein%20%C3%BE
Table&list_uids=610). A set of the reversed sequences created by Bioworks Browser 3.2 EF2 as well as common contaminants such as keratin was appended resulting in 5,864 database entries in total. Search parameters were as follows: enzyme type, trypsin, allowing two missed cleavage sites; peptide tolerance, 150 ppm; tolerance for fragment ions, 0.5 amu; variable modifications, oxidation of methionine (15.99 Da) and carboxyamidomethylation of cysteine (57.02 Da); 15 N quantification was enabled and only singly charged peptides were taken into account. For identification, at least two peptides per protein had to exceed the Mascot identity or homology threshold, using a p-value threshold of 0.01. The protein false-positive rate was calculated for each analysis according to Peng et al. (16) and never exceeded 0.5% on protein or peptide level.
Quantification was performed with the Quantification Toolbox of the Mascot Distiller. Peptides needed to pass the following quality thresholds: correlation threshold 0.9, fraction threshold 0.8, standard error threshold 0.19. As files of gel bands belonging to the same sample were processed independently, more than one quantification value per peptide could be returned by Mascot Distiller. Using Excel (Microsoft
Corporation, Redmond, WA) an intensity weighted average of each peptide was calculated. The median of the quantification values of all peptides belonging to the same protein determines the protein quantification value. Protein quantification results were median centered and ratios were log 2 - transformed.
MALDI data analysis using Sequest
10
Protein identification using Sequest and DTASelect, that was only used for the direct comparison of
MALDI and ESI data, was done as described for LTQ-Orbitrap data in Hessling et al. 1 using the same mzML files as for the Mascot analysis.
MzML files were searched with SEQUEST version v28 (rev.12) (Thermo Fisher Scientific) against a
S. aureus COL target-decoy protein sequence database. This database was composed of all protein sequences of S. aureus COL extracted from the National Center for Biotechnology Information (NCBI) bacteria genomes
(http://www.ncbi.nlm.nih.gov/sites/entrez?Db=%20genome&Cmd=Retrieve&dopt=Protein%20%C3%BE
Table&list_uids=610). A set of the reversed sequences created by BioworksBrowser 3.2 EF2 as well as common contaminants such as keratin was appended. The searches were performed in two iterations:
First for the membrane shaving approach, the following search parameters were applied: enzyme type, none; peptide tolerance, 10 ppm; tolerance for fragment ions, 1 amu; b- and y-ion series; an oxidation of methionine (15.99 Da) and a carboxyamidomethylation (57.02 Da) of cysteine were considered as variable modifications with a maximum of three modifications per peptide.
For MS analysis of the cytosolic samples the following search parameters were used: enzyme type, trypsin, allowing two missed cleavage sites; peptide tolerance, 10 ppm; tolerance for fragment ions, 1 amu; b- and y-ion series; variable modification, oxidation of methionine (15.99 Da) and carboxyamidomethylation of cysteine (57.02 Da); a maximum of three modifications per peptide was allowed. In the second iteration, the mass shift of all amino acids completely labelled with 15 N-nitrogen was taken into account in the search parameters. For cytosolic samples the resulting *.dta and *.out files were assembled and filtered using DTASelect (version 2.0.25) (parameters -y 2 -c 2 -C 4 --here -- decoy Reverse_ -p 2 -t 2 -u --MC 2 -i 0.3 --fp 0.005). The SEQUEST search results for the membrane shaving samples were probabilistically validated with Scaffold V3.4.8 (Proteome Software, Portland,
11
Oregon, USA) applying 95% protein and peptide identification probability filters and a minimum of 2 identified peptides per protein. The protein false-positive rate was calculated for each analysis according to Peng et al. (16) and never exceeded 0.5% on protein or peptide level.
LTQ-Orbitrap data analysis
Data generated by LTQ-Orbitrap analysis were searched and quantitated as described in Hessling et al. 1.
12
Results and discussion
Identification
Analysis of the six different cytosolic samples (three exponentially grown biological replicates, three vancomycin stressed biological replicates) resulted in identification of 848 proteins in total. More than
50 % of these proteins could be found in at least five of these six samples (Figure 1). A maximum of 697 protein could be identified in one sample. The false positive rate was checked for each replicate according to Peng et al. (16) and never exceeded 1% on protein or peptide level.
The large number of proteins found in the majority of all analyzed samples proves the consistency and robustness of the method.
(InsertFigure 1)
Analysis of the same samples with an LTQ Orbitrap resulted in identification of 1165 proteins with comparable filter criteria and false discovery rates (Hessling et al. 1), 60% of them were detected in at least 5 out of six samples.
The Venn diagram in Figure 2 shows that the overlap of identifications for the two instruments is very high on protein level. The number of proteins exclusively identified with MALDI-TOF-TOF mass spectrometry (39) is low.
The number of identified peptides was 6,552 in the MALDI-TOF-TOF approach and 11,607 in the LTQ-
Orbitrap analysis. Even though the absolute numbers point at a considerably higher sensitivity for the
LTQ-Orbitrap in our analysis setup, the Venn diagram on peptide level shows a lower overlap than on protein level ( Figure 2). This indicates that the two mass spectrometry techniques are complementary.
13
In our case, an additional measurement with MALDI-TOF-TOF does not increase the number of identified proteins significantly, but would enhance the sequence coverage of the identified proteins.
(Insert Figure 2)
The sample preparation and fractionation parameters were chosen not only to yield good sensitivity, but also to allow sample analysis in an appropriate time frame and the direct comparison to the LTQ-
Orbitrap analysis.
Shortening sample collection times of the LC run fractionation would be advantageous for peptide detection but would be accompanied by an extended analysis time. With given parameters the analysis time for one sample was between 20 to 30 hours, and thus similar to the LTQ-Orbitrap analysis.
Existing studies that compared MALDI and ESI instruments with respect to their sensitivity in proteomic analysis came up with different results, but the detection levels of the two compared instruments did mostly not vary as strongly as in the present study (17, 3, 9, 18, 11). We used state-of-the-art mass spectrometers for both ionization techniques, and we suggest that the high effort recently put in the development of LC-MS optimized ESI-MS/MS instruments resulted in higher sensitivity of the ESI instrument in our study.
However, the complementary nature of the two ionization techniques that was recognized in many studies (10, 9) is still apparent. A strong indication for this is the observed discrepancy between very few exclusively identified proteins and a comparatively high number of exclusively identified peptides in the
MALDI-MS based data. This means that different peptides originating from the same protein and therefore present in the same amounts were specifically detected with either MS instrument. This observation may be linked to a preferential detection of peptides based on their biochemical properties.
14
Comparison of MALDI and ESI identifications
The analysis of the same samples separated with the same chromatographic systems and parameters but two different mass spectrometry instruments allowed us to investigate whether peptides with certain biochemical features were preferentially detected with either of them. Existing literature postulates such tendencies against certain peptide properties when using either MALDI or ESI for ionization. There are common hypotheses for both ionization methods, e.g. higher sensitivity of MALDI instruments towards more basic peptides (19). But these findings are based on the analysis of peptides with different biochemical properties on the same instrument. Different systematic studies carried out on either instrument type revealed that MALDI as well as ESI preferentially ionize peptides with a large proportion of hydrophobic amino acids (20, 21). Comparative proteomic studies of MALDI and ESI instruments that directly compare both ionization techniques have been sparse so far and were mostly too small in terms of identification numbers to allow a statistically relevant analysis of identified peptides (10, 9, 11).
This large dataset of nearly 13,000 identified peptides, of which more than 5,200 could be detected with both mass analyzers, now allowed a statistically meaningful comparison of an ESI and MALDI instrument regarding their biases in the ionization of peptides. To assign occurring differences between the two data sets to the different mass analyzers distinctively, we wanted to equal not only sample preparation, but also data processing for both data sets as far as possible. We performed the same database search for the three exponentially grown replicates analyzed with MALDI-TOF-TOF as we did in the previous study for the LTQ-Orbitrap generated data, which means that data were searched with SEQUEST algorithm and afterwards filtered with DTASelect towards a target false discovery rate of 0.5%. These results were compared with the corresponding LTQ-Orbitrap data.
15
To emphasize differences between the two detection techniques we focused on peptides that were preferentially identified by only one mass analyzer. A preferentially identified peptide was specified as a peptide that was detected in all three replicates with one mass analyzer and at most in one replicate with the other mass analyzer, or it was detected in two out of the three replicates with one mass analyzer but in no replicate with the other.
A comparison of the relative abundance of the different amino acids showed significant differences between the two data sets generated with either MALDI-TOF-TOF or LTQ-Orbitrap.
(Insert Figure 3)
A comparatively high proportion of arginine in the MALDI-TOF-TOF data, respectively lysine in the LTQ-
Orbitrap data, was observed. 65% of all MALDI identified peptides ended with a C-terminal arginine, in contrast to only 24% for peptides preferentially identified with the LTQ-Orbitrap. The preferred ionization of arginine containing tryptic peptides by MALDI was already observed by Krause and co- workers (22) as well as in different ionization comparison studies (9, 23).
Also the high proportion of the aromatic amino acids tyrosine, tryptophan and phenylalanine in MALDI identified peptides was already described (24, 9) and might be attributed to beneficial photo excitation of these moieties during ionization (19). The larger proportion of the secondary amino acid proline in the MALDI identified peptides is noticeable. Baumgart and coworkers have reported that MALDI generally prefers peptides with a large proportion of proline, arginine, phenylalanine, and leucine (20).
Apart from leucine, these observed biases are now additionally confirmed in a direct comparison to an
ESI instrument (Figure 3).
16
Besides lysine there were some more amino acids with a significantly higher proportion in the peptides preferentially detected with the LTQ-Orbitrap. These amino acids often share certain biochemical properties. The aliphatic amino acids alanine, valine, leucine and isoleucine all showed a moderately higher proportion in the LTQ-Orbitrap identified peptides. This is to a larger extent also true for methionine, which is not strictly aliphatic because of its sulphur containing side chain but shares chemical attributes of this group. Proline was the only exception in the class of aliphatic amino acids, as it occurred far more often in MALDI identified peptides. As already mentioned both ionization techniques on their own have been reported to favor hydrophobic over hydrophilic peptides (20, 21).
The direct comparison previously revealed a higher preference of ESI towards aliphatic peptides (9), which correlates well with our observations.
Also the amino acids with a carboxyl group containing side chain, aspartic acid and glutamic acid, and two amino acids with a hydroxyl group containing side chain, serine and threonine, could be found in higher proportion in LTQ-Orbitrap identified peptides than in MALDI identified peptides. The third proteinogenic, hydroxyl containing amino acid, tyrosine occurred as already mentioned more often in
MALDI data. The strong bias of MALDI for aromatic residues superimposes the relatively weak effect of the hydroxyl group in this amino acid. A preference of ESI for hydroxyl group containing peptides was already described (9), but not for carboxyl containing peptides.
The differential preferences of the two mass analyzers for certain amino acid classes correlate with differences in biochemical properties of the peptides preferentially identified with one of the two mass spectrometers.
The comparison of the isoelectric points (p I), calculated with ExPASy's Compute p I/Mw program, showed a higher proportion of peptides with lower p I values for the LTQ-Orbitrap detected peptides and
17 with higher p I values for the MALDI-TOF based peptides (Figure 4 a). The average p I of all exclusively identified peptides (6.01) was higher in the MALDI-TOF than in the LTQ-Orbitrap obtained data (5.28).
The bias of MALDI for peptides that contain a large proportion of basic amino acids becomes even more evident when comparing average p Is of all amino acids of the identified peptides. In Figure 4 b the
MALDI-TOF data based distribution curve of this average amino acid pI is clearly shifted to more basic values.
(Insert Figure 4)
We also calculated the GRAVY of the identified peptides according to Kyte and Doolittle (25). The distribution of peptides according to their GRAVY is shown in Figure 5 a. The obtained overlap of the two distribution curves seems to contradict the supposed bias of ESI-MS for peptides containing more hydrophobic amino acids. To further investigate a possible tendency for hydrophobic peptides by either ionization technique we additionally analyzed the membrane shaving fraction, purified and analyzed by
LTQ-Orbitrap in Hessling et al. 1, now with MALDI-TOF-TOF. This membrane shaving fraction especially contains the hydrophobic membrane spanning domains of peptides (15). We compiled distribution curves of these more hydrophobic peptides according to their GRAVY as well (Figure 5 b). The distribution curve of the preferentially LTQ-Orbitrap identified peptides was clearly shifted to higher
GRAVY values, which supports the reported bias of ESI for hydrophobic peptides (9) and our observed preference of ESI-MS for peptides containing aliphatic amino acids.
We believe that the high proportion of aliphatic amino acids in the membrane shaving fraction and the preferred detection of peptides containing these amino acids by ESI-MS resulted in the shifted distribution. The obtained differences between MALDI and ESI in the detection of peptides containing
18 aliphatic amino acids were small, and the lower proportion of these amino acids in the cytosolic samples may not have been sufficient to cause the same differences in these samples.
(Insert Figure 5)
An additional discriminating peptide attribute for detection by MALDI and ESI instruments was reported to be the peptide mass. Lasaosa and coworkers found a higher percentage of peptides with masses below 1,400 Da in their LC-MALDI approach and a larger proportion for peptides with masses higher than 1,400 Da in their LC-ESI analysis (11). A discriminating effect of peptide mass could not be observed in our analysis. Distribution curves of peptides detected with either instrument type showed variation, but a significant, mass-related bias could not be detected, which correlates with observations of Stapels and Barofsky (9)and Seymoure and coworkers (26).
(Insert Figure 6)
Most observed biases like the tendency of MALDI to prefer peptides containing arginine and aromatic amino acids or on the other hand of ESI to favor peptides containing hydrophobic amino acids support findings of the existing ESI-MALDI comparisons. The former studies compared technically different instruments but always a MALDI versus an ESI. Therefore common preferences between these studies can be attributed to the two applied ionization techniques. The reported biases are most likely a major reason for the complementary nature of MALDI-MS and ESI-MS that we observed.
Quantification
19
About half of all peptides identified by MALDI-TOF-TOF analysis could also be quantified against the corresponding 15 N labeled peptide of the spiked-in standard using the Mascot Distiller software
(between 49% and 51% quantification efficiency for the six different samples). In total, quantitative data for 554 proteins could be obtained, nearly 50% of them in at least five out of six samples. Up to 415 proteins could be quantified in a single biological replicate.
Quantification efficiency for peptides identified with the LTQ-Orbitrap instrument and processed with the Census software was around 80% for the different samples. This led to quantitative data for 1,100 proteins, up to 972 proteins could be quantified in a single biological replicate. The significantly higher quantification efficiency for the LTQ-Orbitrap data could be caused by the different data analysis software packages, but a quantification of LTQ-Orbitrap generated data with MALDI Distiller delivered similar quantification efficiencies as Census (80% to 90%). We suggest that the main reason for the lower quantification efficiency might be the lower number of MS1 scans per time frame of the LC-run.
As already described above, one fraction collected with the LC Probot represents 15 sec of the LC run. In the LTQ-Orbitrap analysis cycle times between two consecutive MS1 scans were less than two seconds.
These highly time-resolved data enable a better statistical analysis of a peptide's elution peak. The higher the number of data points in such an elution peak is, the less error-prone data processing in the quantification process will be. A comparable time resolution for LC-MALDI generated data is not achievable, as the analysis time would increase unreasonably. MS1 scans on a MALDI-TOF-TOF instrument take much more time than on the LTQ-Orbitrap, as one main MS1 spectrum is an average of multiple sub-spectra recorded on different locations of a spot. This is especially necessary in quantitative analyses to overcome the MALDI-specific problem of poor reproducibility (8) of signal intensities.
14 N/ 15 N ratios for the same peptides of the different biological replicates analyzed with MALDI-TOF-TOF were in good correlation, as illustrated in Figure 7 a where peptide ratios of two exponentially growing
20 biological replicates are plotted against each other. Calculated trend lines for these plots of biological replicates showed slopes near to one and an average coefficient of determination of 0.71. As these replicates include a biological variance, the correlation is high on technical level.
The cross-validation of MALDI-TOF-TOF generated peptide ratios with the LTQ-Orbitrap generated data from the same sample is shown in Figure 7 b. The very good correlation of our data that were processed with Mascot Distiller with data generated by a well-established quantification workflow (27) that used the Census software (28) proves the general reliability of the new workflow.
There have been reports about the non-quantitative nature of MALDI-TOF-TOF mass spectrometry that referred to its reproducibility problems in term of spot-to-spot variability (29–32). This is mainly caused by inhomogeneous samples due to so-called hot spot formation in the crystallization process (33). Also the ion suppression effect in the MALDI process hampers quantitative analysis with this ionization technique (34). Our results show that with averaging enough laser shots over a large sample area as well as the use of isotopically labeled peptides as internal standards, which were proposed strategies to overcome the MALDI-specific problems in quantification (8), we were indeed able to obtain reproducible and reliable data (Figure 7).
(Insert Figure 7)
Conclusion
We could prove that global quantitative LC-MS analysis is feasible to be combined with MALDI-TOF-TOF instruments. Modern instrumentation and software solutions can be used to produce high quality data in reasonable time. Although the comparison with a state-of-the-art ESI instrument showed discrepancies in terms of identification numbers and quantification efficiency, it also revealed a complementary nature of the ionization techniques. This complementarity results from tendencies of
21 either technique towards certain biochemical properties of peptides and might be taken into consideration when choosing an instrument for the analysis of samples with particular characteristics.
Acknowledgement
We thank Holger Kock for critically reading the manuscript. This work was supported by grants of the
DFG (SFB/Transregio 34) and EU (BaSysBio LSHG-CT-2006-037469).
22
References
1. Ong,S.-E., Blagoev,B., Kratchmarova,I., Kristensen,D.B., Steen,H., Pandey,A. and Mann,M. (2002)
Stable isotope labeling by amino acids in cell culture, SILAC, as a simple and accurate approach
to expression proteomics. Mol. Cell Proteomics , 1, 376–386.
2. Oda,Y., Huang,K., Cross,F.R., Cowburn,D. and Chait,B.T. (1999) Accurate quantitation of protein
expression and site-specific phosphorylation. Proc. Natl. Acad. Sci. U.S.A. , 96 , 6591–6596.
3. Kuzyk,M.A., Ohlund,L.B., Elliott,M.H., Smith,D., Qian,H., Delaney,A., Hunter,C.L. and Borchers,C.H.
(2009) A comparison of MS/MS-based, stable-isotope-labeled, quantitation performance on ESI-
quadrupole TOF and MALDI-TOF/TOF mass spectrometers. Proteomics , 9, 3328–3340.
4. Lau,K.W., Jones,A.R., Swainston,N., Siepen,J.A. and Hubbard,S.J. (2007) Capture and analysis of
quantitative proteomic data. PROTEOMICS , 7, 2787–2799.
5. Otto,A., Bernhardt,J., Hecker,M. and Becher,D. (2012) Global relative and absolute quantitation in
microbial proteomics. Curr. Opin. Microbiol. , 15 , 364–372.
6. Mueller,D.R., Voshol,H., Waldt,A., Wiedmann,B. and Van Oostrum,J. (2007) LC-MALDI MS and
MS/MS--an efficient tool in proteome analysis. Subcell. Biochem. , 43 , 355–380.
7. Chen,H., Rejtar,T., Andreev,V., Moskovets,E. and Karger,B.L. (2005) Enhanced characterization of
complex proteomic samples using LC-MALDI MS/MS: exclusion of redundant peptides from
MS/MS analysis in replicate runs. Anal. Chem. , 77 , 7816–7825.
23
8. Szájli,E., Fehér,T. and Medzihradszky,K.F. (2008) Investigating the Quantitative Nature of MALDI-TOF
MS. Molecular & Cellular Proteomics , 7, 2410 –2418.
9. Stapels,M.D. and Barofsky,D.F. (2004) Complementary Use of MALDI and ESI for the HPLC-MS/MS
Analysis of DNA-Binding Proteins. Analytical Chemistry , 76 , 5423–5430.
10. Bodnar,W.M., Blackburn,R.K., Krise,J.M. and Moseley,M.A. (2003) Exploiting the complementary
nature of LC/MALDI/MS/MS and LC/ESI/MS/MS for increased proteome coverage. J Am Soc
Mass Spectrom , 14 , 971–9.
11. Lasaosa,M., Delmotte,N., Huber,C.G., Melchior,K., Heinzle,E. and Tholey,A. (2009) A 2D reversed-
phase x ion-pair reversed-phase HPLC-MALDI TOF/TOF-MS approach for shotgun proteome
analysis. Anal Bioanal Chem , 393 , 1245–1256.
12. Shafer,W.M. and Iandolo,J.J. (1979) Genetics of staphylococcal enterotoxin B in methicillin-resistant
isolates of Staphylococcus aureus . Infect. Immun. , 25 , 902–911.
13. MacCoss,M.J., Wu,C.C., Liu,H., Sadygov,R. and Yates,J.R. (2003) A correlation algorithm for the
automated quantitative analysis of shotgun proteomics data. Anal. Chem , 75 , 6912–6921.
14. Dreisbach,A., Otto,A., Becher,D., Hammer,E., Teumer,A., Gouw,J.W., Hecker,M. and Völker,U. (2008)
Monitoring of changes in the membrane proteome during stationary phase adaptation of
Bacillus subtilis using in vivo labeling techniques. Proteomics , 8, 2062–2076.
15. Wolff,S., Hahne,H., Hecker,M. and Becher,D. (2008) Complementary analysis of the vegetative
membrane proteome of the human pathogen Staphylococcus aureus . Mol. Cell Proteomics , 7,
1460–1468.
24
16. Peng,J., Elias,J.E., Thoreen,C.C., Licklider,L.J. and Gygi,S.P. (2003) Evaluation of multidimensional
chromatography coupled with tandem mass spectrometry (LC/LC-MS/MS) for large-scale
protein analysis: the yeast proteome. J. Proteome Res. , 2, 43–50.
17. Hattan,S.J., Marchese,J., Khainovski,N., Martin,S. and Juhasz,P. (2005) Comparative Study of [Three]
LC−MALDI Workflows for the Analysis of Complex Proteomic Samples. J. Proteome Res. , 4, 1931–
1941.
18. Stapels,M.D., Cho,J.-C., Giovannoni,S.J. and Barofsky,D.F. (2004) Proteomic analysis of novel marine
bacteria using MALDI and ESI mass spectrometry. J Biomol Tech , 15 , 191–198.
19. Burkitt,W.I., Giannakopulos,A.E., Sideridou,F., Bashir,S. and Derrick,P.J. (2003) Discrimination Effects
in MALDI-MS of Mixtures of Peptides—Analysis of the Proteome. Aust. J. Chem. , 56 , 369–377.
20. Baumgart,S., Lindner,Y., Kühne,R., Oberemm,A., Wenschuh,H. and Krause,E. (2004) The
contributions of specific amino acid side chains to signal intensities of peptides in matrix-
assisted laser desorption/ionization mass spectrometry. Rapid Communications in Mass
Spectrometry , 18 , 863–868.
21. Cech,N.B. and Enke,C.G. (2000) Relating electrospray ionization response to nonpolar character of
small peptides. Anal. Chem. , 72 , 2717–2723.
22. Krause,E., Wenschuh,H. and Jungblut,P.R. (1999) The dominance of arginine-containing peptides in
MALDI-derived tryptic mass fingerprints of proteins. Anal. Chem. , 71 , 4160–4165.
23. Medzihradszky,K.F., Leffler,H., Baldwin,M.A. and Burlingame,A.L. (2001) Protein identification by in-
gel digestion, high-performance liquid chromatography, and mass spectrometry: peptide
analysis by complementary ionization techniques. J. Am. Soc. Mass Spectrom. , 12 , 215–221.
25
24. Valero,M.-L., Giralt,E. and Andreu,D. (1999) An investigation of residue-specific contributions to
peptide desorption in MALDI-TOF mass spectrometry. Letters in Peptide Science , 6, 109–115.
25. Kyte,J. and Doolittle,R.F. (1982) A simple method for displaying the hydropathic character of a
protein. J Mol Biol , 157 , 105–32.
26. Seymour,S., Booy,A., Gundry,R., Van Eyk,J. and Hunter,C. (2010) Assessing the Complementarities of
MALDI and ESI for Protein Identification in Complex Mixtures. AB SCIEX Technical Note.
27. Otto,A., Bernhardt,J., Meyer,H., Schaffer,M., Herbst,F.-A., Siebourg,J., Mäder,U., Lalk,M., Hecker,M.
and Becher,D. (2010) Systems-wide temporal proteomic profiling in glucose-starved Bacillus
subtilis . Nat Commun , 1, 137.
28. Park,S.K., Venable,J.D., Xu,T. and Yates,J.R. (2008) A quantitative analysis software tool for mass
spectrometry-based proteomics. Nat. Methods , 5, 319–322.
29. Tholey,A., Zabet-Moghaddam,M. and Heinzle,E. (2006) Quantification of peptides for the monitoring
of protease-catalyzed reactions by matrix-assisted laser desorption/ionization mass
spectrometry using ionic liquid matrixes. Anal. Chem. , 78 , 291–297.
30. Bucknall,M., Fung,K.Y.C. and Duncan,M.W. (2002) Practical quantitative biomedical applications of
MALDI-TOF mass spectrometry. J. Am. Soc. Mass Spectrom. , 13 , 1015–1027.
31. Dekker,L.J., Dalebout,J.C., Siccama,I., Jenster,G., Sillevis Smitt,P.A. and Luider,T.M. (2005) A new
method to analyze matrix-assisted laser desorption/ionization time-of-flight peptide profiling
mass spectra. Rapid Commun. Mass Spectrom. , 19 , 865–870.
32. Tholey,A. and Heinzle,E. (2006) Ionic (liquid) matrices for matrix-assisted laser desorption/ionization
mass spectrometry-applications and perspectives. Anal Bioanal Chem , 386 , 24–37.
26
33. Garden,R.W. and Sweedler,J.V. (2000) Heterogeneity within MALDI samples as revealed by mass
spectrometric imaging. Anal. Chem. , 72 , 30–36.
34. Knochenmuss,R., Dubois,F., Dale,M.J. and Zenobi,R. (1996) The Matrix Suppression Effect and
Ionization Mechanisms in Matrix-assisted Laser Desorption/Ionization. Rapid Communications in
Mass Spectrometry , 10 , 871–877.
27
Figure legends:
Figure 1: Frequency of protein identifications
The proportion of proteins that could be identified in one to six out of the six analyzed samples (three exponentially grown replicates and three vancomycin stressed replicates) is displayed. More than 50% of all 848 proteins were detected in at least 5 out of 6 samples (the 15 N standard that was spiked in all samples theoretically contained all proteins present in the two different physiological conditions).
Figure 2: Venn diagram of a) identified proteins and b) identified peptides.
The numbers over dark grey areas represent identifications in the GeLC-MALDI workflow. The numbers over light grey areas represent identification by GeLC-LTQ-Orbitrap from the same samples. Numbers in the intersection areas were identified with either method.
Figure 3: Relative abundance of amino acids in peptides preferentially identified by MALDI-TOF-TOF (dark grey) and LTQ-Orbitrap (light grey bars).
Figure 4: Basicity of identified peptides.
Relative occurence of peptides in a certain pI (pH 7) range is shown for peptides preferentially identified with
MALDI-TOF-TOF (dark grey) and LTQ-Orbitrap (light grey). In Diagram a the frequency of peptides according to the peptides p I; in diagram b according to the average p I of the amino acids of the peptides.
Figure 5: Hydrophobicity of identified peptides.
28
Relative occurrence of peptides in a certain GRAVY range is shown for peptides preferentially identified with
MALDI-TOF-TOF (dark grey) and LTQ-Orbitrap (light grey). Diagram a) is based on peptides identified in the cytosolic samples. Diagram b) is based on peptides identified in the membrane shaving fraction.
Figure 6: Molecular weight of identified peptides.
Relative occurrence of peptides in a certain molecular weight range is shown for peptides preferentially identified with MALDI-TOF-TOF (dark grey) and LTQ-Orbitrap (light grey).
Figure 7: Scatter plots of a) two replicates analyzed with MALDI-TOF-TOF and b) same sample analyzed with
MALDI-TOF-TOF and with LTQ-Orbitrap a) log2 ratios of two exponentially grown biological replicates against the internal standard, both analyzed with
MALDI-TOF-TOF, are plotted against each other. b) log2 ratios one exponentially grown biological replicate against the internal standard, one analyzed with MALDI-TOF-TOF and the other with LTQ-Orbitrap, are plotted against each other.
Linear trendline (black) with linear equation (y) and coefficient of determination (R 2) was calculated.
29
Figures:
Figure 1:
30
Figure 2:
Figure 3:
31
Figure 4:
Figure 5:
32
Figure 6:
Figure 7:
33
Affirmation
Hiermit erkläre ich, dass diese Arbeit bisher von mir weder an der Mathematisch- Naturwissenschaftlichen Fakultät der Ernst-Moritz-Arndt-Universität Greifswald noch einer anderen wissenschaftlichen Einrichtung zum Zwecke der Promotion eingereicht wurde.
Ferner erkläre ich, dass ich diese Arbeit selbständig verfasst und keine anderen als die darin angegebenen Hilfsmittel und Hilfen benutzt und keine Textabschnitte eines Dritten ohne Kennzeichnung übernommen habe.
------
Page 177
Curicculum vitae name: Bernd Heßling address: Steinstraße 37/38 17489 Greifswald date of birth: 24.08.1980 place of birth: Bocholt
PhD thesis:
Since Oct. 2006 Research scientist / PhD student at the Institute for Microbiology - Department of Microbial Physiology - Ernst-Moritz-Arndt-University Greifswald Topic of the PhD thesis: „Complementary application of ESI and MALDI based mass spectrometry in microbial proteomics“
Diploma in Biochemistry
Oct. 2001 – Oct. 2006 Diploma student for Biochemistry at the Ernst-Moritz-Arndt-University Greifswald, Diploma thesis at the Institute for Microbiology - Department of Microbial Physiology - Ernst-Moritz-Arndt-University Greifswald Topic of the diploma thesis „Untersuchungen an Low Abundance Proteinen von Bacillus subtilis “
Civilian service
Jul. 2000 - Jun. 2001 Civilian service at Caritasverband Dekanat Wesel e.V.
School
Aug. 1991 – Jun. 2000 St-Josef Gymnasium, Bocholt
Page 178
Acknowledgement
Mein größter Dank gilt Prof. Hecker für die Bereitstellung der interessanten Themen und die Möglichkeit, diese in den exzellent ausgestatteten Laboren frei bearbeiten zu dürfen. Des Weiteren danke ich Ihnen für die wissenschaftliche Beratung und die nicht selbstverständliche lückenlose Finanzierung während der gesamten Promotion.
Ein großes Dankeschön geht an Dörte und Knut, für die exzellente Betreuung und die vielen, am Ende doch noch fruchtbaren, wissenschaftlichen Diskussionen. Ihr habt mir die Möglichkeit gegeben mit vielen wissenschaftlichen Themen in Kontakt zu kommen, ohne dabei das Ziel aus den Augen zu verlieren. Außerdem danke ich Euch für die nette Arbeitsatmosphäre.
Auch Dirk danke ich für die Zusammenarbeit an den „guten alten“ und neueren MALDIs und die „sehr geistreichen“ Diskussionen abseits des wissenschaftlichen Alltags.
Der gesamten MS-Gruppe danke ich für die gute Atmosphäre in den Laboren, aber auch bei den glorreichen Paddeltouren oder Grillabenden. Speziell Jan sende ich besten Dank in die US und A. Ich hoffe wir können bald mal wieder ein Feierabendbier zusammen trinken.
Außerdem möchte ich den vielen technischen Assistenten der AG Hecker danken. Ich bin sicher, ich werde Euch noch viel mehr zu schätzen wissen, wenn ich nicht mehr in den Genuss einer perfekt funktionierenden Spülküche oder auf magische Weise auftauchenden selbstgepackten LC-Säulen komme. Danke Sebastian, Daniela, Katrin, Anke und noch vielen mehr.
Zu guter Letzt danke ich meiner Familie. Zum einen meinen Eltern, die mich immer unterstützten und auf die ich mich auch weiterhin jederzeit verlassen kann. Ich danke Euch auch dafür, dass ihr wenigsten versucht habt, nicht ständig zu fragen wann ich endlich mit der Promotion fertig bin. Zum anderen danke ich meiner nun eigenen kleinen Familie. Marlen, Du hast mir so oft wie möglich den Rücken frei gehalten, immer an mich geglaubt und meine sehr wenigen und äußerst kurzen Phasen schlechter Launen ausgehalten. Dann danke ich natürlich Dir Mathilda! Du bist in der Lage, jede schlechte Stimmung in Sekunden in Luft aufzulösen und bist daher immer die perfekte Ablenkung.
Page 179