Towards a phosphoproteome map ofCorynebacterium glutamicum

Anne K. Bendt1

Andreas Burkovski1

Steffen Schaffer2

Michael Bott2

Mike Farwick3

Thomas Hermann3

1Institut für Biochemie,Universität zu Köln,Köln, Germany

2Institut für Biotechnologie 1,Forschungszentrum Jülich,Jülich, Germany

3Degussa AG,Halle-Künsebeck, Germany

Towards a phosphoproteome map ofCorynebacterium glutamicum

In this study, the phosphoproteome of Corynebacterium glutamicum, an industriallyimportant soil bacterium of the Corynebacterium/Mycobacterium/Nocardia (CMN)group of Gram-positive bacteria, was investigated by two different detection methods:first, by in vivo radio-labeling using [33P]-phosphoric acid with subsequent autoradio-graphy and second, by immunostaining with phosphoamino acid-specific monoclonalantibodies. After two-dimensional gel electrophoresis (2-DE), around 60 [33P]-labeledprotein spots were visualized and around 90 antibody-decorated protein spotsdetected; 31 of the protein spots were detected with both methods. By peptide massfingerprinting, 41 different proteins were identified, namely 5-enolpyruvylshikimate3-phosphate synthase, aconitase, acyl-CoA carboxylase, acyl-CoA synthetase, ATP(synthase �- and �-chain), carbamoyl-phosphate synthase, citrate synthase, cysteinesynthase, DnaK, the elongation factors G, P, Ts and Tu, enolase, fructose bisphosphatealdolase, fumarase, Gap dehydrogenase, glutamine synthetase I, glycine hydroxy-methyltransferase, GroEL2, GTPase, heat-inducible transcriptional repressor DnaJ2, in-organic pyrophosphatase, isocitrate dehydrogenase, ketol-acid reductoisomerase, lac-tate dehydrogenase, leucine-tRNA ligase, lipoamide dehydrogenase, methionine syn-thase, O-acetylhomoserine sulfhydrylase, pyruvate carboxylase, pyruvate kinase,pyruvate oxidase, ribosomal protein S1, RNA polymerase (�-subunit), succinyl-CoA:CoAtransferase, transketolase and UDP-N-acetylmuramoyl-L-alanine ligase, besides a hy-pothetical 35k protein and a hypothetical glucose kinase. Both detection techniqueswere used to create a phosphoproteome map. Additionally, the influence of nitrogendeprivation on the phosphoproteome of C. glutamicum was investigated.

Keywords: Corynebacterium / Phosphorylation / Phosphoproteome / Two-dimensional gel electro-phoresis PRO 0494

1 Introduction

All organisms have to adapt their metabolism to changingenvironmental conditions, such as varying nutrient sup-ply, temperature, osmolality, etc. For a single cell theseadaptations occur on two different levels, regulation ofgene expression and protein activity. Both mechanismsoften rely on the modification of proteins, e.g., by phos-phorylation of distinct amino acid residues. Althoughphosphorylation cascades are typical for eukaryotes,also prokaryotic proteins are phosphorylated (e.g., see[1, 2]). Especially the phosphorylation cascade of bacteri-al two-component signal transduction systems is well-

investigated, while a global analysis of phosphorylatedproteins in bacteria to our knowledge has not been car-ried out yet.

We are interested in the physiology of Corynebacteriumglutamicum, a Gram-positive soil bacterium belonging tothe mycolic acid-containing actinomycetes. Due to theremarkable ability of this bacterium to excrete highamounts of glutamic acid under conditions of biotin lim-itation [3, 4], it is applied in fermentation processes onindustrial scale. By use of different C. glutamicum strainsnot only large amounts of L-glutamate (1 000 000 tons peryear) but also of L-lysine (450 000 tons per year) are pro-duced, in addition to smaller amounts of the industriallyless important amino acids L-alanine, L-isoleucine, and L-proline [5]. In contrast to closely related pathogenic spe-cies like Corynebacterium diphtheriae, Mycobacteriumleprae or Mycobacterium tuberculosis, C. glutamicum isgenerally recognized as a nonhazardous organism, whichis safe to handle. Furthermore, based on its extremelywell investigated central metabolism and well establishedmolecular biology tools, C. glutamicum is suitable as a

Correspondence: Dr. Andreas Burkovski, Institut für Biochemie,Universität zu Köln, Zülpicher Str. 47, D-50674 Köln, GermanyE-mail: [email protected]: +49-221-470-5091

Abbreviations: IgM, immunoglobulin M; pSer, phosphoserine;pThr, phosphothreonine; pTyr, phosphotyrosine

Proteomics 2003, 3, 1637–1646 1637DOI 10.1002/pmic.200300494

2003 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

1638 A. K. Bendt et al. Proteomics 2003, 3, 1637–1646

model organism for high G�C Gram-positive bacteria ingeneral and mycolic acid-containing actinomycetes inparticular.

Recently, we started an approach to elucidate the C. glu-tamicum proteome based on the differential fractionationof total cell protein in a cytoplasmic fraction and a mem-brane fraction prior to 2-DE [6, 7]. Additionally, sub-mapsof cell wall-associated and of secreted proteins wereestablished [8]. These proteome analyses were boostedby the C. glutamicum genome project initiated byDegussa AG [9] allowing the identification of proteins notonly based on microsequencing but also on mass spec-trometric techniques like ESI-MS or MALDI-TOF-MS [8,10]. Since we are interested in regulatory processes inC. glutamicum (e.g. [11–15]), we extended the proteomeanalysis towards the investigation of protein modifica-tions, namely phosphorylation, the most abundant cova-lent modification of proteins [16].

2 Materials and methods

2.1 Bacterial growth

C. glutamicum wild-type strain ATCC 13032 was culti-vated aerobically on a rotary shaker (110 rpm) at 30�C inMOPS-buffered minimal medium [17]. For nitrogen star-vation experiments, cells were precultivated for 8–10 h inBHI medium (Difco, Detroit, MI, USA), transferred to mini-mal medium for overnight growth and then used to inocu-late fresh minimal medium to an OD600 of approx. 0.7.When the exponential growth phase (OD600 4) wasreached, the culture was harvested (6 min, 30�C,3220�g) and resuspended in prewarmed minimal medi-um without nitrogen source. After 7 min (immunostaining)or 45 min (radioactive labeling) of nitrogen limitation, cellswere harvested in nearly iced killing buffer (20 mM Trisbase, 5 mM MgCl2, 20 mM NaN3, pH (HCl) 7.5), and usedfor protein sample preparation.

2.2 In vivo radio-labeling

C. glutamicum cells were labeled with [33P]-phosphoricacid (specific activity �111 TBq/mmol; Hartmann,Braunschweig, Germany) similar to a method describedfor the metabolic labelling of proteins using [35S]-methio-nine [12]. For phospho-labeling, however, the phosphateconcentration of the minimal medium was reduced to0.1 mM K2HPO4 and 0.2 mM KH2PO4. Exponentiallygrowing cells were harvested and resuspended in 15 mLprewarmed phosphate-reduced minimal medium withand without nitrogen source, respectively. After adding8 MBq [33P]-phosphoric acid, cells were incubated for

45 min under vigorous shaking, harvested and usedimmediately for protein sample preparation. Aliquotedprotein samples could be stored at �80�C until use, butbest detection results were obtained when they wereused immediately for the two-dimensional gel electropho-resis (2-DE). In order to determine if the lower phosphateconcentrations in the medium for [33P]-labelling results inphosphate limiting conditions for the cell we have ana-lyzed growth behavior and protein pattern (as judged byrunning 2-D gels with subsequent colloidal Coomassie-staining) of C. glutamicum cells grown in minimal mediumwith 5 mM phosphate (as used for standard cultivation)and 0.3 mM phosphate (as used for radiolabelling andimmunoblotting experiments) and could not observe anydifferences (data not shown). Therefore, the metabolicstates of the cells in the two media should be identicaland the data obtained be directly comparable.

2.3 Protein sample preparation

Total C. glutamicum cells were disrupted using glassbeads and a Q-BIOgene FastPrep FP120 instrument (Q-BIOgene, Heidelberg, Germany) by lysing the cells 4times for 30 s and 6.5 m/s in the presence of proteinaseinhibitor Complete (Roche, Basel, Switzerland) as recom-mended by the supplier, and 100 �M phosphatase inhibi-tor sodium orthovanadate (Sigma-Aldrich, Taufkirchen,Germany). To degrade radioactively labeled DNA, 5 �g/mL DNase I (Roche) was added. Proteins were separatedin the cytoplasmic and membrane-associated proteinfraction [6, 7] by ultracentrifugation. In this study, only thecytoplasmic proteins were further analyzed. Protein con-centrations were determined using the Roti-Nanoquantassay (Roth, Karlsruhe, Germany). Incorporated radio-activity was quantitated by scintillation counting using aBeckman scintillation counter LS 6500 (Beckman, Fullter-ton, CA, USA).

2.4 Two-dimensional gel electrophoresis

For isoelectric focusing (IEF) 18 cm precast IPG stripspI 4–7 and an IPGphor IEF unit (Amersham PharmaciaBiotech, Uppsala, Sweden) were used as described [7].300 �g of protein for Western blot analysis or 300 000 cpm(100–300 �g protein) for radiodetection were focused for68 000 Vh in a sample buffer containing 6 M urea, 2 M

thiourea, 4% CHAPS, 0.5% Pharmalyte (3–10) and0.4% DTT. The run for the second dimension wascarried out using precast 12–14% polyacrylamide lineargradient gels (ExcelGel Gradient XL 12–14; AmershamPharmacia Biotech) in the Multiphor II apparatus asdescribed [7]. After electrophoresis, radioactive 2-Dgels were stained with colloidal Coomassie Brilliant


Proteomics 2003, 3, 1637–1646 C. glutamicum phosphoproteome 1639

Blue, vacuum-dried on Whatman paper, exposed for3 days to imaging plates (FujiFilm, Tokyo, Japan) andscanned with a BAS 1800 (Raytest, Straubenhardt, Ger-many) at a resolution of 50 �m and a color depth of16 bits. For identification purposes the resulting autora-diographs were aligned with the corresponding imagesof the Coomassie-stained gels using the Delta2D soft-ware, Version 2.1 (Decodon, Greifswald, Germany). Non-radioactive 2-D gels were used for subsequent Westernblot analysis.

2.5 Western blot analysis

Western blot analysis was carried out using a semi-drysystem (Amersham Pharmacia Biotech) with a transferbuffer optimized for proteins with pI’s between 4 and 7.The anode buffer contained 192 mM glycine, 25 mM Tris-base and 10% methanol, pH (HCl) 8.5, the cathode bufferadditional 0.1% SDS. Before blotting, gels and PVDFmembranes were equilibrated in anode buffer for 15 min.The blotting procedure was carried out for 1.5 h with acurrent density of 0.8 mA/cm2. After electrotransfer, theprotein binding capacity of the membranes was saturatedfor 2 h in blocking solution containing 1% BSA (fraction V;Roche), 1% PVP-10 (Sigma-Aldrich), 1% PEG 3500(Sigma-Aldrich), 0.2% Tween 20 (ICN, Aurora, OH, USA)in 2�PBS (274 mM NaCl, 5.4 mM KCl, 20 mM Na2HPO4,3.6 mM KH2PO4, pH 7.2). The blocked PVDF membranes(Immobilon-P; Millipore, Bedford, MA, USA) were probedusing antibody sampler kits (Biomol, Hamburg, Germany)containing monoclonal antibodies purified from differentmouse clones against phosphoserine (pSer), phospho-threonine (pThr) and phosphotyrosine (pTyr), respectively.Routinely, the anti-pSer kit was used, containing immuno-globulin M (IgM) antibodies of the five clones 1C8, 4A3,4A9, 4H4 and 16B4. Antibodies were diluted in blockingsolution to a final concentration of 0.5 �g/mL each, mixed,and incubated overnight at 4�C with the membranes. Forfurther use, the diluted antibodies were stored at 4�C upto 3 months in the presence of 10 mM NaN3. Bound anti-bodies were visualized on Kodak scientific imaging films(Rochester, NY, USA) with chemiluminescence, using ahorseradish peroxidase (HRP)-coupled goat anti-mouseIgM antibody (Stressgen, Victoria, BC, Canada) in a dilu-tion of 1:10 000, and the West Pico supersignal chemilu-minescent substrate (Pierce, Rockford, IL, USA). Since noMALDI-TOF-MS spectra could be obtained from proteinsblotted onto PVDF membranes, protein spot identitieswere assigned by matching the chemiluminescenceimages with Coomassie-stained gels run in parallel, usingthe Delta2D software, Version 2.1 (Decodon, Greifswald,Germany).

2.6 In-gel digestion

For peptide mass fingerprinting using MALDI-TOF-MS,protein spots were either cut out of fresh Coomassie-stained 2-D gels or out of gels dried on Whatman paper.Excised protein spots were washed twice for 5 min andonce for 30 min with 50 mM NH4HCO3 and destained byincubating twice for 30 min in 50 mM NH4HCO3 in 50%acetonitrile. To shrink the sample, 100 �L of acetonitrilewere added and the gel pieces were incubated for 5 min.Low-abundance protein spots were further subjected toreduction (10 �L of 10 mM DTT for 30 min at 55�C) andalkylation (10 �L of 55 mM iodoacetamide for 30 min atroom temperature in the dark) prior to digestion. Aftervacuum-drying in a concentrator 5301 (Eppendorf, Ham-burg, Germany), 2.5–5 �L freshly prepared trypsin solu-tion was added until no further uptake of solution tookplace (10 �g/mL trypsin, sequence grade; Promega,Madison, WI, USA) in 25 mM NH4HCO3). Then the gelpiece was covered with 25 mM NH4HCO3 and incubatedovernight at 37�C. The supernatant was recovered andthe peptides were extracted from the gel pieces using25 mM NH4HCO3, 50% acetonitrile by incubating for 1 hwith frequent ultrasonication. Peptides were concen-trated by lyophilization of the pooled supernatants in aconcentrator 5301 to a final volume of 10 �L.

2.7 MALDI-TOF-MS and database searches

For MALDI-TOF-MS analysis, peptides were desaltedand concentrated using reversed-phase C18 pipettetips (ZipTip; Millipore). The peptides were eluted fromthe column with 1 �L of matrix (5 mg/mL recrystallized�-cyano-4-hydroxy-trans-cinnamic acid solution in 50%acetonitrile, 0.1% TFA) and directly spotted onto asample plate. MALDI mass spectra were obtained asdescribed by Schaffer and co-workers [10], using aVoyager-DE STR Biospectrometry Workstation (AppliedBiosystems) equipped with a delayed extraction deviceand calibrated with the Sequazyme Peptide Mass Stan-dard Kit (Applied Biosystems). Monoisotopic peptidemasses were determined and the mass lists used for in-house database searches of the C. glutamicum genomeperformed with the GPMAW software, Version 4.0(Lighthouse Data, Odense, Denmark) and MS-Fit(htpp://www.Prospector.ucsf.edu). Standard search pa-rameters were chosen as follows: tryptic digest with themaximum number of one missed cleavage, fixed car-bamidomethylation of cysteines, considered oxidizationof methionines and considered phosphorylation ofserine, threonine, and tyrosine. The peptide mass toler-ance was set to 100 ppm.



Figure 1. Cytoplasmic proteins of C. glutamicum. 100 �g proteins were separated by 2-DE in therange of pI 4–7 and Mr 5–100 kDa. Identified phosphorylated proteins are numbered (see Table 1 fordetails). Molecular weight markers are indicated at the left and pI values at the top of the panel.

3 Results and discussion

3.1 In vivo labeling with [33P]-phosphoric acid

To create a phosphoproteome map of C. glutamicum, wechose a method for metabolic labeling and subsequentautoradiography of proteins separated by 2-DE. Untilnow, for this approach solely the use of the high-energyradionuclide [32P] (1.7 MeV) has been described [18, 19].In this study we demonstrate the use of [33P], whose lowerenergy (0.2 MeV) renders it much safer to handle anddiminishes its detrimental impact on the cells. A furtheradvantage of [33P] is the relatively long half-life of 25 dayscompared to the 14 days of [32P].

A prerequisite for in vivo labeling approaches is theuptake of the labeled compound. While ATP, which isoften applied in in vitro experiments, is not transportedinto the cell, phosphate is taken up and used for the cel-lular metabolism. Unfortunately, the radioactive phos-phate is not only incorporated into phosphoproteins, butalso into nucleic acids, which can result in a severe back-ground, making the identification of distinct spots difficultor even impossible (compare [20]). However, this problemcan be circumvented by treatment of the protein samples

with DNase I/RNAse A. The Coomassie blue-stainedmastergel of cytoplasmic proteins between pI 4–7 ofC. glutamicum contains approximately 540 protein spots(Fig. 1). After metabolic labeling and subsequent 2-DE, atleast 59 protein spots could be reproducibly detected inthree independent experiments (Fig. 2). The majority ofthem (54 spots, representing 41 different proteins) wasidentified via MALDI-TOF-MS (Table 1). For 5 spotsdetected by [33P]-labeling, no corresponding protein spotin a Coomassie- or silver-stained or even a [35S]-labeled2-D gel could be localized, indicating the much highersensitivity of this detection method.

Since we are interested in regulatory processes regardingthe nitrogen control, we wanted to analyze the influenceof nitrogen deprivation on the phosphoproteome of C.glutamicum. With the labeling method described, phos-phorylation or dephosphorylation processes happeningin the first minutes after the onset of a specific stress can-not be visualized, due to the necessity of a long timeframe of in vivo labeling in order to achieve sufficientuptake of radioactive phosphate into the cells. Therefore,additionally a method of immunostaining was applied inthis study.



Figure 2. Autoradiography of [33P]-phosphoric acid-labelled proteins ofC. glutamicum. 300 000 cpmof cytoplasmic proteins were separated by 2-DE in the range of pI 4–7 andMr 5–100 kDa. Radioactivephosphoproteins were detected by scanning the dried gel after exposition to Fuji imaging plates for3 days. pI values are indicated across the bottom. Molecular weight markers are indicated at the leftand pI values at the top of the panel.

3.2 Immunostaining of phosphorylated proteins

In bacteria, proteins are often phosphorylated on histi-dine, glutamic acid, and aspartic acid residues [21], butthese N- and acylphosphates are relatively unstable invitro and in vivo [22]. Consequently, no antisera directedagainst these phosphoamino acids are commerciallyavailable. Furthermore, it is questionable, if phosphoryla-tion of these amino acids would survive the lengthy pro-cedure of 2-D separation. In contrast, the O-phosphatespSer, pThr, and pTyr are stable in the absence of specificprotein O-phosphatases and monoclonal antibodies di-rected against these phosphoamino acids are available.Applying these antibodies to C. glutamicum, we at-tempted to analyze to what extent protein O-phosphoryl-ation exists in this organism. Although hints for tyrosinephosphorylation in bacteria have been found [21] and thegenome sequence of C. glutamicum encodes one tyro-sine phosphatase, no pTyr-phosphoproteins were visual-ized when probing the C. glutamicum proteome with anti-pTyr antibodies. Approximately 20 phosphoprotein spotswere detected with anti-pThr antibodies (data not shown).Most phosphoprotein spots were detected using the anti-

phosphoserine sampler kit (Biomol), with monoclonal IgMantibodies of five different mouse clones, being specificfor epitopes of sequence motifs containing pSer. Thisratio of seryl-, threonyl-, and tyrosyl-phosphoproteins isin agreement with data obtained for eukaryotes, where ithas been shown that most frequently serine residues arephosphorylated, indicated by a phosphoamino acid con-tent (pSer:pThr:pTyr) being 1800:200:1 [23].

By immunostaining with pSer-specific monoclonal anti-bodies, more than 90 protein spots of the electroblottedcytoplasmic fraction separated between pI 4–7 were vis-ualized. However, these results were not fully reproduci-ble in three independent experiments. Therefore, onlyspots detected in all experiments and with known identityare labelled in Fig. 3. Similar to the results of radiolabeling,for at least 20 immunodetected spots no correspondingprotein spot in a Coomassie- or silver-stained or even a[35S]-labeled 2-D gel was found, due to the high sensitivityof as little as a few femtomole of epitope detectable byimmunostaining. This makes immunodecoration in com-bination with chemiluminescence detection the most sen-sitive method.



Table 1. Identified proteins detected by labelling with [33P]-orthophosphate and/or with anti-pSer antibodies

Spot ORFa) NCgl Gene Putative protein function [33P] pSer Net-Phosb) Pho-Baseb) BRENDAb) dbc)

P1 0659 pyc Pyruvate carboxylase � � � � � PubMed

P2 0471 rpoB �-Subunit RNA polymerase � � � � � PubMed

P3a 1547 carB Carbamoyl-phosphate synthase � � � � � PubMed

P3b 2915 – Leucine-tRNA ligase � � � � � –

P4 1547 carB Carbamoyl-phosphate synthase � � � � � PubMed

P5 1482 acn Aconitase � � � � � PubMed

P6 1053 bipA GTPase � � � � � PubMed

P7 1053 bipA GTPase � � � � � PubMed

P8 0478 fusA Elongation factor G � � � � � PubMed



P11a 1094 metE Methionine synthase � � � � � –

1512 tkt Transketolase � � � � � –

P11b 1512 tkt Transketolase � � � � � –

P11c 1094 metE Methionine synthase � � � � � –

P13 0634 icd Isocitrate dehydrogenase � � � � � PubMed

P14a 2702 dnaK DnaK � � � � � PubMed

P14b 2702 dnaK DnaK � � � � � PubMed

P15 1304 rpsA Ribosomal protein S1 � � � � � PubMed

P16 0670 accBC Acyl CoA carboxylase � ?e) � � � –

P17 0670 accBC Acyl CoA carboxylase � � � � � –

P18 2621 groEL2 GroEL 2 � � � � � PubMed

P19a 2077 murC UDP-N-acetylmuramoyl-L-alanineligase

� � � � � –

P19b 2133 glnA Glutamine synthetase I � � � � � PubMed

P20 1163 atpA ATP synthase (� chain) � ?e) � � � –

P21 2480 cat 1 Succinyl-CoA:CoA transferase � � � � � –

P22 0967 fum Fumarase � ?e) � � � –

P23a 0480 tuf Elongation factor TU � � � � � PubMed

P23b 0625 metY O-Acetylhomoserine sulfhydrylase � � � � � –

P24 0480 tuf Elongation factor TU � � � � � PubMed



P27 2521 poxB Pyruvate oxidase � � � � � –

P28 0935 eno Enolase � � � � � PubMed

P29 1224 ilvC Ketol-acid reductoisomerase � � � � � –

P30 1224 ilvC Ketol-acid reductoisomerase � � � � � –

2558 – Hyp. Glucose kinase � ?e) n.b.d) n.b.d) n.b.d) n.b.d)

P31 2008 pyk Pyruvate kinase � � � � � PubMed

P32 0355 lpd Lipoamide dehydrogenase � � � � � –

P33 0954 glyA Glycine hydroxymethyl-transferase � � � � � –

P34 2210 dnaJ2 DnaJ2 � � � � � –

P35 2008 pyk Pyruvate kinase � � � � � PubMed



Table 1. Continued

Spot ORFa) NCgl Gene Putative protein function [33P] pSer Net-Phosb) Pho-Baseb) BRENDAb) dbc)

P36 0795 gltA Citrate synthase � � � � � –

P37a 1526 gap Gap-DH � � � � � PubMed

P37b 1526 gap Gap-DH � � � � � PubMed

P38 2449 – Hyp. 35k protein � � � n.b.d) n.b.d) n.b.d)

P39 2810 ldh Lactate dehydrogenase � � � � � –

P40 2673 fda Fructose bisphosphate aldolase � � � � � PubMed

P41 1949 tfs Elongation factor Ts � � � � � PubMed

P42 2473 cysK Cysteine synthase � ?e) � � � –

P43 1557 efp Elongation factor P � � � � � –

P44 2607 ppa Inorganic pyrophosphatase � � � � � PubMed

P47 0730 aroA 5-Enolpyruvylshikimate 3-phosphatesynthase

� � � � � –

P48 2774 fadD Acyl CoA synthetase � � � � –

P49 1165 atpD ATP synthase (�-chain) � � � � –

P50 2774 fadD Acyl CoA synthetase � � � � –

a) Nomenclature according to NCBI genome annotationb) A plus or minus indicates whether information concerning a phosphorylation were obtained.c) Database entry of phosphorylationd) Not determinede) No definite detectionIndicated are the corresponding NCgl ORF numbers, gene names, putative functions of the gene products, the detectionmethod and whether there are predicted phosphorylation sites. The spot numbers correspond to those given in Figs. 1–3.

3.3 Identification of phosphorylated proteins

In total, 54 proteins spots visualized by radioactive label-ing and/or immunostaining were identified after in-geldigestion and subsequent MALDI-TOF-MS analysis(Table 1), representing 41 different proteins. Searcheswithin the NetPhos WWW server (http://cbs.dtu.dk/ser-vices/NetPhos) predicted that all identified proteins con-tained at least three serine phosphorylation sites. Furtherdatabase searches with PhosphoBase (http://cbs.dtu.dk/databases/PhosphoBase), BRENDA (http://brenda.uni-koeln.de), and PubMed (www.ncbi.nlm.nih.gov/PubMed/) were carried out. For the majority of the identified pro-teins a putative modification by phosphorylation waseither predicted by the mentioned databases or hasbeen described by other groups (Table 1) [24].

Interestingly, for several proteins two or more spots differ-ing in their pI were found (Figs. 2 and 3), for example, theelongation factors EF-Tu and EF-G were occurring in four(P23a, P24–26) and three (P8–10) spots, respectively.Three proteins occurred twice with differing pI, namelyGTPase (P6, P7), ACCase (P16, P17) and ketol-acidreductoisomerase (P29, P30). Three proteins were found

in two spots with different pI and different molecularmass, namely Gap-DH (P37a, P37b), acyl-CoA synthe-tase (P48, P50), and pyruvate kinase (P31, P35). Proteinsappearing in multiple spots might either indicate differen-tially phosphorylated, or otherwise covalently modified,forms of the same protein (spots with differing pI, butnearly identical molecular mass) or proteolytic processingof the protein (spots with different molecular masses). Ofcourse, degradation as well as artificial chemical modifi-cations of the protein during sample preparation has to beconsidered.

Most of the proteins identified are enzymes acting in thecells central metabolic pathways such as glycolysis (eno-lase, fructose-1,6-bisphosphate aldolase, Gap dehydro-genase, pyruvate decarboxylase, pyruvate kinase), Krebscycle (aconitase, citrate synthase, isocitrate dehydrogen-ase) and fatty-acid metabolism (acyl CoA synthetase,acyl CoA carboxylase, succinyl-CoA:CoA-transferase).Also chaperones (DnaK, GroEL2, DnaJ2) as well as com-ponents of the protein synthesis machinery (translationelongation factors and ribosomal proteins) were identi-fied. We were not able to identify regulatory proteins or



Figure 3. Western blot of serine-phosphorylated proteins of C. glutamicum. 300 �g cytoplasmicproteins were separated by 2-DE in the range of pI 4–7 and Mr 5–100 kDa, electroblotted onto PVDFmembranes and probed with anti-phosphoserine specific monoclonal antibodies. pI values areindicated across the bottom. Molecular weight markers are indicated at the left and pI values atthe top of the panel.

signal transduction molecules, which are mainly low-abundance proteins and thus difficult to detect with the2-DE methods applied. Surprisingly, such proteins couldnot even be identified in a recent study upon selectiveenrichment of phosphopeptides [24].

3.4 Comparison of radiolabeling andimmunostaining

By radiolabeling and immunostaining, around 60 and 90phosphoprotein spots were detected, respectively; 31 ofthem were visualized with both methods, verifying bothapproaches. Each of the two methods has its advan-tages, as well as its drawbacks. Immunostaining follow-ing the protocol described is more sensitive than radiola-belling with [33P], allowing detection of as little as a fewfemtomole of epitope. Therefore, a higher number ofphosphoprotein spots were visualized using this tech-nique. Additionally, as mentioned by Kaufmann and co-workers [16], also constitutively phosphorylated proteinswith therefore slow phosphate turnover rates can bedetected. However, some proteins were solely detected

by radiolabelling. This might be due to a considerabledrawback of immunostaining: the insufficient recognitionof certain phosphorylated proteins by anti-phosphoaminoacid antibodies due to steric hindrance of the recognitionsite [16]. Another explanation is that the proteins solelydetected by metabolic radiolabeling are phospho-en-zyme intermediates, as it has been described for exampleby Solow and co-workers [25] for ATPases and phospho-glucomutases, the latter forming phosphoseryl enzymeintermediates during their catalytic cycles. An examplederived from this study is the detection of pyrophospha-tase (P44) by radiolabeling, but not by immunostaining.Also the proteins carbamoyl-phosphate synthase (P3),leucine-tRNA ligase (P3a), transketolase (P1b) and UDP-N-acetylmuramoyl-L-alanine ligase (P19a) were exclu-sively detected by metabolic radiolabelling and mighttherefore form phospho-enzyme intermediates, as theycatalyze reactions involving transfer of phosphogroups.Finally, the observed differences could be due to thefact, that [33P]-labelling approach theoretically allowsdetection of proteins containing acyl-, N-, and O-phos-phates, while immunodetection as performed in thisstudy is restricted to the latter type of protein phosphoryl-



ation. However, as discussed above, it is questionable ifN- and acylphosphates are stable enough to endure theentire 2-D procedure.

For the subsequent identification of detected phospho-proteins by MS, the method of radiolabelling shows theadvantage of a very reliable allocation of the detectedphosphoprotein spots within the Coomassie blue-stainedmastergel, allowing precise excision of the spot from the2-D gel. In contrast, a severe drawback of immunostain-ing is the inability to identify proteins after the Westernblot procedure, e.g., proteins electroblotted onto PVDFmembranes. Reasons might be the low protein concen-tration on the membrane and/or an insufficient extractionof peptide digest even in presence of 1% octyl-glucopyr-anosid, caused by the peptides adhering to the mem-brane and thus making it impossible to acquire sufficientMS spectra. However, immunostaining seems to be themethod of choice for analyzing changes in the cells phos-phoprotein status within short time frames, as labellingtimes required for sufficient [33P]-labelling of proteins areprohibitive for that kind of analyses. In this study it wasnecessary to perform the labelling for 45 min in order toobtain sufficient signal intensity upon autoradiography. Inother studies labelling was even performed for two gen-eration times (Saccharomyces cerevisiae [28]) or 20 h(HeLa 229 cells [20]).

In summary, the labeling methods described in this studyhave their advantages as well as their limitations. There-fore, a combination of both is required to obtain a com-prehensive and reliable data set about the phosphopro-teome, as it has also been suggested by Kaufmann andco-workers [16] for mammalian cells.

3.5 Estimated quantity of the C. glutamicumphosphoproteome

About 30% of eukaryotic proteins are supposed toundergo phosphorylation or dephosphorylation pro-cesses, with up to 5% of the genome encoding either pro-tein kinases or phosphatases [26]. For prokaryotes com-parable studies do not exist. Interestingly, within the pre-dicted 3745 genes of C. glutamicum there are only fourserine/threonine kinases, one serine/threonine phospha-tase, and one tyrosine phosphatase encoded, represent-ing less than 0.2% of the genome sequence. This numberindicates that at least in this bacterium protein phospho-rylation processes are less frequent than in eukaryotes.Given that approx. 1900 proteins of C. glutamicum showa pI within the range covered in this study (pI 4–7), with 35of them being smaller than 5 kDa and 58 of them biggerthan 100 kDa, the number of proteins detectable with the2-DE protocol applied can be estimated as 1800. Since

27% of the corynebacterial proteins are membrane pro-teins, but only the cytoplasmic proteins were consideredin this study, around 1300 proteins remain as being ana-lyzable. In this study we detected �120 phosphoproteinspots (60 spots using [33P]-labelling and 90 spots usingimmunodetection with 31 spots detected by both meth-ods), representing at least 41 different proteins. We donot know to what extent the detection of low-abundancephosphoproteins failed due to the detection limit inherentto the applied methodology. However, we can state thatC. glutamicum contains at least 41 proteins which canbecome phosphorylated, representing 3% of the proteinswithin the detection window described above.

3.6 Influence of metabolic conditions

In order to investigate the physiological function of proteinphosphorylation, the adaptation of C. glutamicum cells tostarvation conditions was studied. Neither by radioactivelabelling nor by immunostaining differences between thephosphoproteomes of nitrogen-starved cells and of cellsgrown in nitrogen-rich medium were observed. A possibleexplanation might be that phosphorylation processes arenot important for the physiological response of C. gluta-micum to nitrogen deprivation. Regulation via co- or post-translational modifications elucidated so far depends onadenylylation and uridylylation of different proteins [11,14, 15]. Other explanations might be that the involvedphosphoproteins are membrane proteins, which werenot covered in this study, or that their pI and/or Mr do notlie within the tested range of pI 4–7 and 5–100 kDa,respectively. Also if the phosphoproteins important forN-control are proteins with low abundance, as it is usualfor regulatory elements, they could not be identified withthe methods applied. In addition, phosphorylation pro-cesses important for N-regulation might occur on histi-dine, aspartic acid or glutamic acid residues and we donot know if the phosphorylated forms of these aminoacids remain stable during the standard protein prepara-tion methods and the 2-DE protocol applied in the courseof this study.

4 Concluding remarks

In this study we showed the detection of at least 120 cyto-plasmic protein spots, containing phosphorylated pro-teins and identified many of them with MALDI-TOF-MS.While indications for several phosphoproteins were foundfor B. subtilis, recently [27], to our knowledge this is thefirst systematic study of phosphorylated proteins in bac-teria on a global scale. The challenge of future studies willbe not only the visualization and identification of phos-



phorylated proteins, but also the localization of the corre-sponding phosphorylation sites as well as a quantificationof protein phosphorylation. Assignment of the phospho-rylated peptides within the MS spectra of the identifiedproteins failed in this study. This is not astonishing, sincephosphopeptides are, especially in a digest, not observedas intense peaks, provoked by ionic suppression causedby the nonphosphorylated peptides. Additionally, duringthe desalting procedure with C18 columns, the hydrophilicphosphopeptides are easily lost [23]. Thus, to assignphosphorylation sites, an enrichment of phosphoproteinsor -peptides via immunoprecipitation or immobilizedmetal ion-affinity chromatography, respectively, has tobe carried out.

The authors would like to thank B. Walter and M. Nicko-laus for their excellent technical assistance. We are espe-cially grateful to R. Krämer, H. Sahm, and W. Pfefferle forcontinuous support. The work in the authors laboratoriesis supported by the Bundesministerium für Bildung undForschung.

Received November 11, 2002

5 References

[1] Shi, L., Potts, M., Kennelly, P. J., FEMS Microbiol. Rev. 1998,22, 229–253.

[2] Kennelly, P. J., FEMS Microbiol. Lett. 2002, 206, 1–8.[3] Kinoshita, S., Udaka, S., Shimono, M., J. Gen. Appl. Micro-

biol. 1957, 3, 193–205.[4] Gutmann, M., Hoischen, C., Krämer, R., Biochim. Biophys.

Acta 1992, 1112, 115–123.[5] Leuchtenberger, W., in: Roehr, M. (Ed.), Biotechnology, Vol.

VI, VCH, Weinheim 1996.[6] Hermann, T., Wersch, G., Uhlemann, E.-M., Schmid, R., Bur-

kovski, A., Electrophoresis 1998, 19, 3217–3221.[7] Hermann, T., Finkemeier, M., Pfefferle, W., Wersch, G., Krä-

mer, R., Burkovski, A., Electrophoresis 2000, 21, 654–659.

[8] Hermann, T., Pfefferle, W., Baumann, C., Busker, E., Schaf-fer, S., Bott, M., Sahm, H., Dusch, N., Kalinowski, J., Pühler,A., Bendt, A. K., Krämer, R., Burkovski, A., Electrophoresis2001, 22, 1712–1723.

[9] Hodgson, J., Nature Biotechnol. 1998, 16, 715.[10] Schaffer, S., Weil, B., Nguyen, V. N., Dongmann, G., Günther,

K., Nickolaus, M., Hermann, T., Bott, M., Electrophoresis2001, 22, 4404–4422.

[11] Jakoby, M., Krämer, R., Burkovski, A., FEMSMicrobiol. Lett.1999, 173, 303–310.

[12] Schmid, R., Uhlemann, E.-M., Nolden, L., Wersch, G.,Hecker, R., Hermann, T., Marx, A., Burkovski, A., FEMSMicrobiol. Lett. 2000, 187, 83–88.

[13] Jakoby, M., Nolden, L., Meier-Wagner, J., Krämer, R., Bur-kovski, A., Mol. Microbiol. 2000, 37, 964–977.

[14] Nolden, L., Farwick, M., Krämer, R., Burkovski, A., FEMSMicrobiol. Lett. 2001, 201, 91–98.

[15] Nolden, L., Ngouoto-Nkili, C.-E., Bendt, A. K., Krämer, R.,Burkovski, A., Mol. Microbiol. 2001, 42, 1281–1295.

[16] Kaufmann, H., Bailey, J. E., Fussenegger, M., Proteomics2001, 1, 194–199.

[17] Keilhauer, C., Eggeling, L., Sahm, H., J. Bacteriol. 1993, 175,5595–5603.

[18] Djafarzadeh, S., Niggli, V., Exp. Cell. Res. 1997, 236, 427–435.

[19] Immler, D., Gremm, D., Kirsch, D., Spengler, B., Presek, P.,Meyer, H. E., Electrophoresis 1998, 19, 1015–1023.

[20] Larsen, M. R., Sørensen, G. L., Fey, S. J., Larsen, P. M.,Roepstorff, P., Proteomics 2001, 1, 223–238.

[21] Kennelly, P. J., Potts, M., Front. Biosci. 1999, 15, 372–385.[22] Sickmann, A., Meyer, H. E., Proteomics 2001, 1, 200–206.[23] Mann, M., Ong, S.-E., Grønborg, M., Stehen, H., Jensen,

O. N., Pandey, A., Trends Biotechnol. 2002, 20, 261–268.[24] Zhou, H., Watts, J. D., Aebersold, R., Nat. Biotechnol. 2001,

19, 375–378.[25] Solow, B., Bischoff, K. M., Zylka, M. J., Kennelly, P. J., Prot.

Sci. 1998, 7, 105–111.[26] Blom, N., Kreegipuu, A., Brunkak, S., Nucleic Acids Res.

1998, 26, 382–386.[27] Büttner, K., Bernhardt, J., Scharf, C., Schmid, R., Mäder, U.,

Eymann, C., Antelmann, H., Völker, A., Völker, U., Hecker,M., Electrophoresis 2001, 22, 2908–2935.

[28] Alms, G. R., Sanz, P., Carlson, M., Haystead, T. A., EMBO J.1999,18, 4157–4168.


Towards a phosphoproteome map ofCorynebacterium glutamicum

Documents