Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC

Jul 10, 2007 - ESI-MS/MS methods, a total of 871 proteins were identified in a cytosolic protein preparation, which .... cooling, 8 μL of 2 mol L-1 i...
5 downloads 9 Views 436KB Size
Download PDF
Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC: An Alternative Approach to High-Resolution Peptide Separation for Shotgun Proteome Analysis Nathanae1 l Delmotte,† Maria Lasaosa,‡ Andreas Tholey,‡ Elmar Heinzle,‡ and Christian G. Huber*,† Department of Chemistry, Instrumental Analysis and Bioanalysis, and Biochemical Engineering, Saarland University, 66123 Saarbru ¨ cken, Germany Received July 10, 2007

A two-dimensional separation scheme for shotgun proteome analysis employing high-pH reversedphase HPLC in the first and low-pH ion-pair reversed-phase HPLC in the second dimension (RP × IPRP-HPLC) has been developed and evaluated. Compared to the classical strong cation exchange × ion-pair reversed-phase (SCX × IP-RP-HPLC) approach, the RP × IP-RP-HPLC system was characterized by a lower degree of orthogonality, which was, however, more than counterbalanced by higher separation efficiency, more homogeneous distribution of peptide elution, and easier experimental handling. Peptide fragment fingerprinting by electrospray-ionization tandem mass spectrometry (ESIMS/MS) was employed for peptide detection and identification. About 13% more peptides and 7% more proteins could be identified with the alternative approach in 30% less analysis time, enabling the analysis of the proteome of Corynebacterium glutamicum with a coverage of 24.9% (745 proteins). Combining the identification results both of the SCX- × IP-RP-HPLC-ESI-MS/MS and RP- × IP-RP-HPLCESI-MS/MS methods, a total of 871 proteins were identified in a cytosolic protein preparation, which represented 29.1% of all proteins annotated in the genome of C. glutamicum. Keywords: shotgun proteome analysis • multidimensional separation • high-pH reversed-phase chromatography • monolithic columns • low-pH ion-pair reversed-phase chromatography tandem mass spectrometry • Corynebacterium glutamicum

Introduction Proteomic analysis targets the global or at least large-scale characterization of the protein complement of whole organisms, tissues, cells, cell organelles, or body fluids.1,2 Analytical methods suitable for proteome analysis must be capable of identifying and quantifying thousands of proteins over several orders of magnitude, which requires the implementation of highly sophisticated separation and detection schemes.3,4 Because of generally insufficient efficiency of a single dimension of separation for such complex mixtures, several steps commonly need to be combined to multidimensional separation schemes to be able to adequately fractionate proteomic samples before identification of the individual components.5-16 Protein identification is typically performed by mass spectrometry (MS) after cleavage to peptides, employing the methods of peptide mass fingerprinting (PMF)17,18 or peptide fragment fingerprinting (PFF)19 in combination with database searches.20 Depending on the stage, at which protein or protein fragment separation is performed, two major strategies of proteome analysis have evolved: top-down proteomics,21-23 in which * To whom correspondence should be addressed. Tel: +49 681 302 2433. Fax: +49 681 302 2963. E-mail: [email protected]. † Instrumental Analysis and Bioanalysis. ‡ Biochemical Engineering. 10.1021/pr070424t CCC: $37.00

 2007 American Chemical Society

separation occurs at the intact protein level, and bottom-up or “shotgun” proteome analysis, where the proteins are digested immediately after their preparation and subsequent separation is conducted at the peptide fragment level.22,24 The most common chromatographic separation modes for peptides are strong cation-exchange (SCX),25 reversed-phase (RP),26 and ion-pair reversed-phase (IP-RP)27 high-performance liquid chromatography (HPLC). Although the terms RP- and IP-RP-HPLC are frequently used as synonyms, we here distinguish between the two modes. Hydrophobic stationary phases are employed in both modes, but RP-HPLC is based predominantly on solvophobic interactions28 between the peptides and the stationary phase, as opposed to IP-RP-HPLC, where the addition of amphiphilic ions to the mobile phase results in a combination of electrostatic and solvophobic retention.29 One of the most successful multidimensional separation strategies for bottom-up proteome analysis involves the combination of SCX- and IP-RP-HPLC. Employing a dual capillary column packed with a segment of a strong cation-exchange stationary phase followed by a second segment of a reversed-phase stationary phase, it was possible to perform fully automated, on-line multidimensional proteome analysis.30 Moreover, column-switching approaches have been realized successfully for automated SCX × IP-RP-HPLC fractionation.31-36 Alternatively, Journal of Proteome Research 2007, 6, 4363-4373

4363

Published on Web 10/09/2007

research articles off-line two-dimensional setups utilizing SCX- and IP-RP-HPLC have been devised which offer more flexibility in terms of elution conditions and sample concentrations in both dimensions at the cost of more laborious sample handling and transfer between the separation dimensions.37,38 The success of multidimensional separation schemes is determined by two major factors: orthogonality of the separation dimensions and chromatographic efficiency of the separation systems employed in the different dimensions.39 Recent reports have revealed that SCX as first-dimension separation mode might not be ideal, since the separation efficiency of ionexchange HPLC is generally rather limited and peptides of the same charge are clustering in relatively narrow elution windows rather than eluting continuously from an SCX column.40 Alternatively, RP- or IP-RP-HPLC using high-pH elution conditions has been combined with RP- or IP-RP-HPLC performed with acidic eluents.39-41 The orthogonality of high- and lowpH (IP-)RP-HPLC separations was considerably high, although not perfect, since long and hydrophobic peptides tended to show stronger retention in both separation dimensions.40,41 Although the principal applicability of such (IP-)RP- × (IP-)RP-HPLC systems for two-dimensional peptide separation has been demonstrated with relatively simple peptide mixtures such as tryptic digests of standard proteins,40,41 their potential for global proteome analysis has not been explored so far. Consequently, we aim in this communication at a detailed investigation of the benefits and detriments of two-dimensional RP × IP-RP-HPLC in comparison to SCX- × IP-RP-HPLC as separation scheme for shotgun proteome analysis. Peptide identification is achieved in both setups by electrospray ionization tandem mass spectrometry (ESI-MS/MS). Corynebacterium glutamicum serves not only as a well-characterized model system for proteome analysis but also has considerable biological relevance as a high-performance expression system for the industrial production of amino acids42 such as L-glutamate and L-lysine. Metabolic pathways of C. glutamicum are known,43 and the genome is available in databases.44 The annotation of around 3000 proteins permitted us to perform peptide and protein identification by MS/MS and database searches. The RP × IP-RP separation scheme described here is critically compared to the classical SCX × IP-RP approach in terms of peptide separation characteristics, dimension orthogonality, number of peptide and protein identifications, proteome coverage, and method complementarity.

Experimental Section Chemicals. Deionized water (18.2 MΩ cm) was prepared with a Purelab Ultra Genetic system (Elga, Griesheim, Germany). Tris-hydroxymethylaminomethane (g99.9%) was purchased from Carl Roth GmbH (Karlsruhe, Germany). The protease inhibitor cocktail tablets Mini complete were supplied by Roche Diagnostics GmbH (Penzberg, Germany). Yeast extract and tryptone were purchased from Difco Laboratories (Livonia, MI). Analytical reagent grade sodium dihydrogenphosphate-1-hydrate, acetic acid (analytical reagent grade), and triethylamine (>99%) were obtained from Merck KGaA (Darmstadt, Germany). Acetonitrile (E Chromasolv), ethylenediaminetetraacetic acid (∼99%), and 2-mercaptoethanol (>98%) were purchased from Sigma-Aldrich (Steinheim, Germany). Sodium chloride was supplied by Gru ¨ ssing GmbH (Filsum, Germany). Urea (g99.5%), o-phosphoric acid (85%), ammonium hydrogencarbonate (g99.5%), ammonium formate (>97%), iodacetic acid (g99.5%), formic acid (88-91%), heptafluorobutyric acid 4364

Journal of Proteome Research • Vol. 6, No. 11, 2007

Delmotte et al.

(g99.0%), and trifluoroacetic acid (g99.5%) were purchased from Fluka (Buchs, Switzerland). Sequencing grade modified trypsin was supplied by Promega (Madison, WI) and Slide-ALyzer dialysis cassettes by Perbio Science (Bonn, Germany). All other chemicals for medium preparation were purchased from Sigma GmbH (Steinheim, Germany) and Fluka (Buchs, Switzerland) and were of analytical grade. Cultivation of C. glutamicum and Protein Extraction. C. glutamicum wild-type strain ATCC 13032 (American Type and Culture Collection) cells were taken from -80 °C frozen stock culture and grown in agar plates at 30 °C. A single colony was taken for the first preculture and cells were grown for 8 h at 30 °C in 50 mL complex medium (5 g L-1 glucose, 5 g L-1 yeast extract, 10 g L-1 tryptone, 5 g L-1 sodium chloride) in 500 mL shake flasks. Cells were harvested by centrifugation (8800g, 2 min, 30 °C), washed twice with sterile 0.9% sodium chloride, and resuspended in 25 mL of minimal medium for the second preculture in 250 mL shake flasks. Main cultivation was performed in minimal medium at 30 °C and 230 rpm in an incubator shaker (Model Multitrom, Infors AG, Bottmingen, Switzerland), and cells were harvested at an optical density (OD660) of 14 measured as described previously,45 when the growth rate reached µ ) 0.37 h-1 corresponding to 4.9 g L-1 biomass dry weight. Minimal medium containing 15 g L-1 glucose was prepared by combination of the following solutions after autoclaving: (A) 1 g of NaCl, 55 mg of CaCl2 and MgSO4‚ H2O in 559 mL of deionized water; (B) 5 g of (NH4)2SO4 in 200 mL of deionized water; 16 g of K2HPO4 and 2 g of KH2PO4 in 100 mL of deionized water; (C) 20 mg of FeSO4‚7H2O, pH 1 with HCl; (D) 15 g of glucose in 100 mL of deionized water; (E) 20 mL of vitamin solution (2.5 mg of biotin, 5 mg of thiamin‚ HCl, 5 mg of Ca-panthotenate in 100 mL of deionized water); (F) 10 mL of 100× trace elements solution (200 mg L-1 FeCl3‚ 6H2O, 200 mg L-1 MnSO4‚H2O, 50 mg L-1 ZnSO4‚H2O, 20 mg L-1 CuCl2‚2H2O, 20 mg L-1 Na2B4O7‚1H2O, and 10 mg L-1 (NH4)6Mo7O24‚4H2O, and (G) 1 mL of DHB solution (300 mg of 3,4-dihydroxybenzoic acid and 0.5 mL of 6 mol L-1 NaOH in 9.5 mL of water). Cells were harvested by centrifugation at 4 °C for 5 min (8500 rpm, HFA 8.50 Highconic Rotor, Heraeus Biofuge Stratos Sorvall, VWR, Germany), washed twice with 1 mL of sterile 0.9% NaCl, and stored at -75 °C. Cell wall disruption was performed mechanically in a mixer mill (Model MM 301, Retsch GmbH, Haan, Germany). Approximately 200 mg of wet cell weight was resuspended 1:3 in lysis buffer (20 mmol L-1 TRIS, 5 mmol L-1 EDTA, pH 7.5 and protease inhibitor cocktail) at 4 °C. The sample was homogenized in the presence of 0.8 g of glass beads (0.25-0.50 mm external diameter) for 30 min. After that, the sample was centrifuged at 13000g for 5 min at 4 °C. Then, the supernatant was collected and the extraction procedure repeated twice. The resulting supernatant containing the cytoplasmic proteins was centrifuged for 30 min at 50000g and 4 °C. Protein concentration was determined by Bradford test (Bio-Rad Protein Assay, BioRad Laboratories GmbH, Munich, Germany), and the sample was stored in aliquots at -20 °C. Preparation of Tryptic Digests. Proteins were denatured by addition of 120 µL of 8 mol L-1 urea in 500 mmol L-1 ammonium hydrogen carbonate to 200 µL of cell lysate. The sample was incubated for 1 h at 37 °C under gentle agitation. The disulfide bridges were then reduced by addition of 12 µL of 300 mmol L-1 dithiothreitol, sample degassing under argon, and incubation for 2 h at 37 °C at 600 rpm. After sample cooling, 8 µL of 2 mol L-1 iodoacetic acid was added for

Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC

carboxymethylation, and the sample was incubated for 30 min at room temperature under light protection. Excess iodoacetic acid was eliminated with 16 µL of 1 mol L-1 2-mercaptoethanol and incubation for 20 min at room temperature. Finally, the sample was dialyzed for 16 h against 1 L of distilled water in a dialysis cassette (3500 MW cutoff membranes). Trypsin (1 µg for 50 µg protein) was dissolved for 30 min at 30 °C in 50 mmol L-1 acetic acid and added to the protein solution for 24 h digestion. Finally, the digestion was quenched by addition of trifluoroacetic acid (1.0% (v/v) end concentration). In order to ensure that no particulate material was injected onto the chromatographic column, the protein digest was centrifuged for 5 min at 13000 rpm. The supernatant was collected and split into aliquots. Analytical Setups for First-Dimension Separation. The chromatographic setup for SCX HPLC consisted of an analytical HPLC gradient pump (Model 1050, Agilent Waldbronn, Germany) and a manual injector (Model Rheodyne 7725, Rohnert Park, CA) with a 400 µL external loop. The UV absorbance was monitored at 214 nm with a spectrophotometer (Model Spectromonitor 3100, Milton Roy, Ivyland, PA). The flow cell had an optical pathway of 10 mm (volume: 14 µL). Eluents were degassed with helium. This setup was used to run a 250 × 4.0 mm ProPac SCX-10 column protected by a 100 × 4.0 mm guard column (Dionex, Idstein, Germany). The chromatographic setup utilized for RP-HPLC consisted of a low-pressure gradient pump (Rheos 2000, Flux Instruments, Basel, Switzerland), a degasser (Knauer GmbH, Berlin, Germany), and an injection system (Model 7725, Rheodyne, Rohnert Park, CA) with a 400 µL external loop. UV detection was monitored at 280 nm with a UV-detector equipped with a capillary detection cell (Model 433, Kontron AG, Zurich, Switzerland). The flow cell had an optical pathway of 5 mm and a volume of 1 µL. This setup was used to operate a 150 × 2.0 mm, 3 µm C18 Gemini column from Phenomenex (Aschaffenburg, Germany). After collection, the fractions were concentrated in a vacuum concentrator (Model 5301, Eppendorf AG, Hamburg, Germany). Second-Dimension Separation and Mass Spectrometry Data Acquisition. The second-dimension separation setup consisted of a 2D capillary/nano HPLC system (Model Ultimate, LCPackings, Amsterdam, The Netherlands), equipped with a low-pressure gradient micro-pump, a micro-column 10-port switching unit with loading pump, and a micro-autoinjector. Capillary preconcentration (10 × 0.2 mm) and capillary separation columns (60 × 0.1 mm) were polystyrene/divinylbenzene (PS-DVB) monoliths prepared according to the published procedure.46 Eluents were degassed with helium. In order to maximize trapping efficiency, preconcentration of the peptides on the 10 mm long preconcentration column was performed with a solvent containing 0.10% heptafluorobutyric acid, while analytical separations in the 60 mm long capillary columns was performed with a gradient of 0-20% acetonitrile in 0.05% trifluoroacetic acid. An ion-trap mass spectrometer (Model esquire HCT, Bruker Daltonics, Bremen, Germany) with a modified ESI-ion source (spray capillary: fused silica capillary, 0.090 mm o.d., 0.020 mm i.d.) was utilized as detector. The instrument was operated in data-dependent mode. MS/MS spectra were recorded in positive-ion mode with an electrospray voltage of 3500 V and fragmentation amplitude ramped from 0.5 to 3.0 V. The heated capillary temperature was set to 300 °C. The following mass spectrometric parameters were applied for automated peptide

research articles

identifications by data-dependent tandem mass spectrometry: mass range mode, ultra scan 50-3000 m/z; scan speed, 26000 m/z s-1; full scan, 450-1500 m/z; ion polarity, positive; trap drive, 93.2; octapole RF amplitude, 88.5 Vpp; lens 2, -36.1 V; capillary exit, 253.8 V; nebulizer gas, 20 psi; dry gas, 4 L min-1; high-voltage end-plate offset, -500 V; ICC target, 70000; maximum accumulation time, 200 ms; precursor ions auto MS(n), 3; MS averages, 5 spectra; MS/MS scan range, 200-2000 m/z; active exclusion, after 2 spectra for 0.50 min; MS/MS fragmentation amplitude, 1.5 V; smart fragmentation, on (30200%); absolute threshold MS/MS, 4500. Data Processing and Evaluation. Mass spectra processing was performed with Data Analysis 3.3 from Bruker Daltonics (Bremen, Germany). Database searches were performed against an in-house database containing the sequenced proteins of C. glutamicum ATCC 13032 Kitasato. The database (2993 entries) was downloaded from the Institute for Genomic Research (TIGR) under http://cmr.tigr.org/tigr-scripts/CMR/ CmrHomePage.cgi. The engine software was Mascot 2.1 based on the MOWSE algorithm (Matrix Science, London, UK).47 The following search parameters were applied: taxonomy, all entries; fixed modification, cysteine carboxymethylation; variable modification, methionine oxidation; enzyme, trypsin; peptide tolerance, ( 1.3 Da; MS/MS tolerance, ( 0.3 Da; maximum number of missed cleavages, 1. A protein was positively identified with a significance threshold of 0.05, meaning that random hits (so-called false positives) occurred with a frequency lower than 5%. The 95% significance level corresponded to a MOWSE score of 23. The ion score cutoff was set at 23. This value was already over the identification threshold for some peptides, leading to loss of peptides by the identification process. However, this strong cutoff value was utilized to ensure protein identifications based on reliable peptide hits. Proteome Analysis of C. glutamicum by SCX × IP-RPHPLC. Approximately 280 µg of a tryptic digest of C. glutamicum protein cell extract were injected onto a 250 × 4.0 mm ProPac SCX-10 column. Sample loading was performed at 1 mL min-1 with (A) 5 mmol L-1 NaH2PO4, pH 3.0, 20% ACN and elution with (B) 5 mmol L-1 NaH2PO4, pH 3.0, 20% ACN, 500 mmol L-1 NaCl. The gradient was 0-3% B in 9 min, 3-10% B in 8 min, and 10-100% B in 4 min. The shallow salt gradient of 1.7 mmol L-1 NaCl per min during the first 9 min was combined with increasing fraction volumes over the time in order to get higher homogeneity, in terms of peptide amount, between the collected fractions. Fractions were collected as follows: 250 µL fractions from 1.25 to 7.25 min, 500 µL fractions between 7.25 and 13.25 min, and 1000-µL fractions until the end of the chromatographic separation. The fractions were concentrated by evaporation to a final volume of 100 µL (2.5-, 5-, and 10-fold concentration for the 250-, 500-, and 1000-µL fractions, respectively). The second separation step was performed by loading and washing for 4 min 10 µL of the evaporated fractions on the 10 × 0.2 mm PS-DVB trap column with H2O + 0.10% heptafluorobutyric acid at 10 µL min-1. The switching valve was then commuted and a back flush elution over the 60 × 0.1 mm PSDVB analytical column was performed at 0.75 µL min-1. Mobile phases consisted of (A) 0.05% trifluoroacetic acid in water (pH 2.1) and (B) 0.05% trifluoroacetic acid in acetonitrile. Peptides were separated at room temperature by using a gradient from 0 to 20% B in 60 min, followed by isocratic conditions at 100% B for 3 min. A total of 44 fractions collected between 1.25 and Journal of Proteome Research • Vol. 6, No. 11, 2007 4365

research articles

Delmotte et al.

Figure 1. Strong cation-exchange (a) and reversed-phase (b) fractionation of a trypsinated lysate of C. glutamicum: (a) column, 100 + 250 × 4.0 mm i.d. ProPac SCX-10; mobile phase, (A) 5 mmol L-1 NaH2PO4, pH 3.0, 20% acetonitrile, (B) 0.50 mol L-1 NaCl in eluent (A); gradient, 0-3% B in 9.0 min, 3-10% B in 8 min, 10-100% B in 4 min; flow rate, 1 mL min-1; temperature, 25 °C; detection, UV at 214 nm; sample, 280 µg tryptic digest of a C. glutamicum protein cell extract; fractions, 24 × 0.25 min, 12 × 0.50 min, 8 × 1.00 min; (b) column, 150 × 2.0 mm i.d. 3 µm Gemini C18; mobile phase, (A) 72 mmol L-1 triethylamine titrated to pH 10.0 with acetic acid, (B) 72 mmol L-1 triethylamine and 52 mmol L-1 acetic acid in acetonitrile; gradient, 0-55% B in 55.0 min; flow rate, 200 µL min-1; temperature, 25 °C; detection, UV at 280 nm; sample, 280 µg tryptic digest of a C. glutamicum protein cell extract; fractions, 31 × 1.00 min.

21.25 min were analyzed in triplicate, leading to 132 HPLCMS/MS runs. Proteome Analysis of C. glutamicum by RP × IP-RP-HPLC. Approximately 280 µg of peptides from C. glutamicum were injected over a 150 × 2.0 mm Gemini C18 column. Sample loading was performed at 200 µL min-1 with (A) 72 mmol L-1 triethylamine titrated to pH 10.0 with acetic acid. Elution was performed with (B) 72 mmol L-1 triethylamine, 52 mmol L-1 acetic acid in acetonitrile. The gradient was 0-55% B in 55 min, followed by isocratic conditions at 100% B for 2 min. Twohundred µL fractions were collected every minute. Acetonitrile was removed by evaporating the fractions to a final volume of 20 µL (10-fold concentration). Fractions were finally reconstituted with 105 µL of 0.10% aqueous heptafluorobutyric acid. Thirty-one fractions collected between 14 and 45 min were analyzed in triplicate (total of 93 HPLC-MS/MS runs). Columns, mobile phase solvents, flow rates and gradients were identical to the one used for the SCX fraction analysis. Because the fractions did not contain salts, the switching valve was commuted directly after sample transfer (2.5 min).

Results and Discussion First-Dimension Peptide Separation by Strong CationExchange HPLC and High-pH Reversed-Phase HPLC. The primary goal of first dimension separation in shotgun proteome analysis is the homogeneous fractionation of a complex peptide mixture for subsequent peptide identification by IP-RP-HPLCMS/MS. The number of fractions taken and the fractionation interval usually represent a compromise between retaining the 4366

Journal of Proteome Research • Vol. 6, No. 11, 2007

separation achieved in the first dimension, dilution of the peptides in the obtained fractions, time required for their analysis in the second dimension, and carryover of peptides between adjacent fractions.48 Figure 1 compares the fractionation of two aliquots of the C. glutamicum lysate in a first dimension by SCX- and RPHPLC. Because of the very high complexity of the sample, no individual peaks could be recognized in the SCX separation (Figure 1a). Although a very shallow, two-step salt gradient was applied (1.7 and 4.4 mmol L-1 sodium chloride per minute, respectively), most of the peptides eluted during the first 9-min gradient segment within a narrow window of eluent strength, ranging from 0 to 15 mmol L-1 sodium chloride. This can be explained by the fact that the sample components were not individually separated but peptides with the same charge (1+, 2+, 3+, and 4+) eluted from the column as poorly resolved clusters in the order of increasing charge.25 The intense peak during the first 5 min of the chromatogram (maximum at about 2700 mAU) is most probably not only due to eluting peptides but also comes from strongly UV-absorbing components in the media used for sample preparation, which are only weakly retained on the SCX column. In order to achieve homogeneous fractions in terms of peptide amount, collection intervals were 0.25 min at the beginning of the fractionation (fractions nos. 1-24), whereas they were increased to 0.50 (nos. 25-36) and 1.0 (nos. 37-43) min in the later stages of the separation. A washing gradient ramped from 50 to 500 mmol L-1 sodium chloride was utilized to remove strongly adsorbed compounds from the ion-exchange stationary phase.

Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC

research articles

The same amount of sample material was injected into the RP-HPLC system operated with a single-segment gradient of 0-55% acetonitrile in 72 mmol L-1 triethylamine adjusted to pH 10.0 with acetic acid. As most silica-based RP columns are not suitable for operation at a pH above 8.0, we utilized a silicapolymer hybrid-type column material having extended stability up to pH 12. The column effluent was monitored at 280 nm because of strong UV absorption of the eluents at the low wavelengths (e.g., 214 nm) usually implemented for the detection of peptide bonds. The large peaks in the first 15 min (maximum at 980 mAU) of the chromatogram (Figure 1b) did not show any detectable peptides of m/z greater than 450 and are, therefore again most probably due to reagents utilized during sample preparation. Whereas they coeluted in the SCXseparation with early eluting peptides, they were clearly separated from the peptides in case of RP-HPLC. The RPchromatogram was significantly more structured than the SCX chromatogram, indicating a better separation performance of the RP-separation system. Fractions were collected with a constant periodicity of 1.0 min. Analogous to the SCX system, the RP-separation was concluded by a washing step of 100% acetonitrile. Second-Dimension Peptide Separation by IP-RP-HPLC at pH 2.1. All fractions collected after separation by SCX-HPLC were concentrated in a vacuum concentrator to a final volume of 100-125 µL. RP-HPLC fractions were evaporated almost to dryness and reconstituted with 100-125 µL 0.1% HFBA. This not only increased the portion of analytes transferred from the first to the second dimension, but also removed the acetonitrile contained in the eluents used for SCX- or RP-chromatography, which otherwise would seriously decrease the trapping and chromatographic efficiency of the second chromatographic dimension. Subsequently, 10-µL aliquots of the fractions were focused and, in the case of SCX-fractions, desalted on a short monolithic preconcentration column before transfer and separation in a 60 × 0.10 mm monolithic poly(styrene-divinylbenzene) capillary column. Typical reconstructed total ion current chromatograms for three consecutive SCX-fractions are illustrated in Figure 2a. Employing data-dependent MS/MS spectrum acquisition, the three most intense ions in a survey scan (Figure 2b) were subjected to collision-induced fragmentation and mass analysis (Figure 2c) for later peptide identification based on peptide fragment fingerprinting. The peptide identified in this example was IATGFIADHPHLLQAPPADDEQGR from pyruvate carboxylase. The chromatograms of three consecutive RP-fractions are depicted in Figure 2d. MS/MS spectrum acquisition and peptide identification were performed analogously to SCX fractions. Characterization of SCX-Separation in the First Dimension. Aliquots of the 44 fractions collected after SCX-HPLC were injected in triplicate into the second separation dimension. For each HPLC-MS/MS run, a database search was performed, leading to 132 identification reports. For the three replicates of each fraction, the reports were pooled and a nonredundant list of identified peptides was established. Then, the number of uniquely identified peptides, including identifications of the same peptide in different fractions, was plotted as a function of fraction number. The obtained distribution of identified peptides and proteins is illustrated in Figure 3a. More details about the number of identified unique peptides and proteins are discussed below in the section proteome coverage. The multimodal distribution of peptide hits showed maxima in

Figure 2. Separation and identification of peptides in three consecutive SCX or RP fractions of C. glutamicum. (a) Reconstructed total ion current chromatogram after SCX fractionation: columns, 10 mm × 0.20 mm i.d. monolithic PS-DVB preconcentration column, and 60 mm × 0.20 mm i.d. monolithic PS-DVB separation column; loading solvent for preconcentration, 0.10% aqueous heptafluorobutyric acid; trapping time, 4.0 min; loading flow rate, 10 µL min-1. Mobile phase for analytical separation: (A) 0.050% aqueous trifluoroacetic acid, (B) 0.050% trifluoroacetic acid in acetonitrile; gradient, 0-20% B in 60 min; flow rate, 0.750 µL min-1; temperature of preconcentration and separation column, 25 °C; detection, ESI-MS/MS; sample, tryptic peptides of C. glutamicum, fractions 27, 28, and 29 from SCXHPLC, 10 µL injected. (b) Mass spectrum of cation-exchange fraction 29 at 35.9 min. (c) Tandem mass spectrum of m/z 857.4. (d) Reconstructed total ion current chromatogram after RP fractionation, conditions as in (a).

fractions 1 and 25, and minima in fractions 16 and 44. The same data interpretation was performed at the protein level and a similar profile was obtained (see Figure 3a). Maxima of protein identifications were observed for fractions 7 and 27 and minima for fractions 18 and 44. The multimodal distribution can be attributed to the positive charge carried by the tryptic peptides. Singly and doubly charged peptides eluted at the very beginning of the separation, whereas peptides carrying three or more charges were more strongly retained on the column and eluted at higher salt concentrations. In order to verify this hypothesis, the number of positively charged residues was computed for each peptide Journal of Proteome Research • Vol. 6, No. 11, 2007 4367

research articles

Delmotte et al.

Figure 3. Peptide and protein hit distribution over (a) SCX and (b) RP fractions. Data correspond to identifications from three replicate injections in the second chromatographic dimension. Redundant hits within replicates have been eliminated, but redundant hits between fractions are incorporated.

Figure 4. Dependence of peptide retention in SXC fractionation on the number of positive charges carried. A zoom factor of 10 was applied to the number of singly charged peptides. Conditions as in Figure 1a.

in each fraction (Σ 6254 peptides). Because the separation was performed at pH 3.0, positive residues were the N-terminus of the peptides (pKa 9-11), as well as the three basic amino acids arginine (pKa 12.5), lysine (pKa 10.5), and histidine (pKa 6.0). Figure 4 illustrates the distribution of differently charged peptides over the SCX separation. Singly and doubly charged analytes eluted at the beginning of the separation process. The elution of most of the triply charged peptides started at fraction 19, when the majority of doubly charged peptides had already passed the column. This data clearly corroborates earlier observations that peptides carrying the same charge are eluting in clusters from SCX-columns.16 Acetonitrile was added to the eluents used for SCX-HPLC in order to suppress secondary hydrophobic interactions between peptides and the polymeric support material of the cation exchanger. In order to confirm the minor role of such interactions, the GRand AVerage of 4368

Journal of Proteome Research • Vol. 6, No. 11, 2007

hydropathY index (GRAVY index), representing the hydrophobic properties of each identified peptide, was computed using parameters published elsewhere.49 Then, arithmetic averages of GRAVY values were computed for each fraction and plotted (data not shown). Only a slight tendency of decreasing GRAVY values (between 0.1 and -0.5 ((0.1)) with increasing fraction number was observed. The results of these different data evaluations clearly show that the most important parameter influencing the retention of the peptides in SCX-HPLC was the number of basic functional groups present in the peptides, whereas hydrophobic properties played only a minor role. Characterization of RP-Separation in the First Dimension. The 31 fractions collected after reversed-phase separation at pH 10.0 were also injected in triplicate into the second separation dimension. As for the SCX fractions, a database search was performed for each HPLC-MS/MS run, and 93

Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC

Figure 5. Plot representing the average GRAVY value of the identified peptides as a function of retention in the RP-fractionation. Conditions as in Figure 1b.

identification reports were obtained. Redundant hits within replicates were filtered and a list of non-redundant identified peptides was established for each fraction. The distribution of peptide hits is plotted in Figure 3b. The peptide hit distribution observed with RP-HPLC was trapezoid. After an initial increase over 6 fractions, the number of peptide hits was relatively stable over 18 fractions, and finally decreased over 7 fractions. The same data interpretation was performed at the protein level and a very similar identification profile was obtained (see Figure 3b). In order to examine the influence of charge state of the peptides on retention in high-pH RP-HPLC, the number of negative charges was computed for each peptide. Because the first dimension of separation was performed at pH 10.0, peptides were carrying negative charges at the C-terminus (pKa 1.8-2.4) and at aspartic acid (pKa 3.9) and glutamic acid (pKa 4.3) residues. For each charge state (1- to 7-), the average retention time of the peptides in the first dimension was computed, which revealed that highly charged peptides eluted first. This observation suggests that electrostatic interactions are not important and that hydrophobicity determines retention, implying that the fractionation performed with the octadecyl phase at pH 10.0 is best described by a RP- and not an IP-RP-separation mechanism. To check this hypothesis, the GRAVY index of each identified peptide was computed. Then, for each fraction, the arithmetic average of the GRAVY values of all the detected peptides was calculated. Finally, these average GRAVY values were plotted as a function of the fraction number (Figure 5). A strong linear correlation was observed (R2 ) 0.9843) proving that hydrophobicity is indeed the determinant for elution of the peptides in high-pH RP-HPLC. Orthogonality of Two-Dimensional Separations. For both setups, the orthogonality between the two dimensions of separation was evaluated by plotting each peptide hit as a function of its retention time in both chromatographic dimensions.39 Because fractions were collected in relatively short time intervals (0.25 - 1.0 min), the retention time of a peptide in the first dimension of separation approximately corresponds to the midpoint of the collection time window of the fraction in which the peptide was identified. Such plots were computed for the classical SCX × IP-RP-HPLC and the RP × IP-RP-HPLC approach as shown in Figure 6. In the case of redundant identifications of peptides within the same fraction, the retention time in the map corresponds to the retention time of the peptide with the highest MOWSE score. Identical peptides identified in different fractions, on the other hand, were included into the map.

research articles

The distribution of peptides over the whole map is a clear indication of a high degree of orthogonality for the SCX × IPRP-HPLC approach (Figure 6a). The higher density of peptide hits in the first 7 min of the SCX separation in comparison to the following 6 min and the last 8 min is explained by the different collection intervals for the fractions (24 × 0.25 min, 12 × 0.50 min, and 8 × 1.00 min). As already discussed above, only few peptides were found in the SCX-separation in the elution window from 3 to 6 min due to the transition from doubly to triply charged peptides eluting from the column. In the RP × IP-RP-HPLC separation scheme employing basic and acidic elution conditions, peptide hits were also extensively dispersed over the map. However, peptides strongly retained in high-pH RP-HPLC also showed a tendency for longer retention in low-pH IP-RP-HPLC. Nevertheless, the peptide fractions obtained from the first dimension were generally spread over a more than 45-min elution window in the second dimension, allowing ample time for the recording of MS/MS spectra. Despite a lower degree of orthogonality, the RP × IPRP-HPLC setup permitted to identify significantly more unique peptides than the classical SCX × IP-RP-HPLC approach (see section proteome coverage below). Therefore, the fact that both dimensions are not fully orthogonal does not seem to be the most critical factor. The increase in terms of fraction homogeneity and sample simplification due to improved separation efficiency appears to be much more relevant than dimension orthogonality for improved peptide identification. Proteome Coverage Utilizing SCX × IP-RP- and RP × IPRP Methods. Proteome analysis of C. glutamicum employing the SCX × IP-RP-HPLC setup yielded 6254 peptide hits (44 fractions, three replicates in the second dimension). After removal of redundant peptide hits, a total of 2398 unique peptides were identified. With the RP × IP-RP-HPLC setup, 3124 peptide hits were obtained (31 fractions, three replicates in the second dimension) and after removal of redundancies, 2708 unique peptides were identified. Thus, in comparison with the SCX approach, approximately 13% more unique peptides were identified (2708 vs 2398) with approximately 50% less peptide hits (3124 vs 6254) in the RP × IP-RP-HPLC approach. Moreover, the total analysis time required for the peptide identifications in the RP × IP-RP-HPLC scheme (3 × 31 fractions of 80 min ) 124 h in the second dimension) was 30% less as compared to the SCX × IP-RP-HPLC scheme (3 × 44 fractions of 80 min ) 176 h in the second dimension). Each unique peptide was represented by 1.2 and 2.6 different hits with the RP × IP-RP-HPLC and SCX × IP-RP-HPLC method, respectively. The significant decrease in redundancies with the RP × IP-RP-HPLC approach in comparison to the SCX × IPRP-HPLC approach is explained by the higher efficiency of RPHPLC separation, resulting in narrower peptide peaks, and the analysis of less (31 vs 44) but larger (1.0-min vs 0.25-, 0.50-, and 1.0-min) fractions in the RP × IP-RP approach. However, the weak retention of low charged peptides (1+ and 2+) on the cation exchanger requires such a fractionation to avoid the collection of too complex fractions, which can be hardly analyzed because of strong ion suppression in ESI. The rapid elution of some peptides during the SCX separation is also illustrated by the fact that more than 45% of the unique identified peptides were already detected in the first 5 min of the separation. Proteome coverage can be defined as ratio between the number of detected proteins and the number of proteins annotated for a given genome, but not necessarily present in Journal of Proteome Research • Vol. 6, No. 11, 2007 4369

research articles

Delmotte et al.

Figure 6. Orthogonality map for the SCX × IP-RP-HPLC (a) and RP × IP-RP-HPLC (b) approaches. Each of the 6254 (a) and 3124 (b) peptide hits is represented as a cross.

the protein cell lysate. In the case of the present study on C. glutamicum, the number of potentially expressed proteins corresponds to the number of entries in the database (2993). At the protein level, the SCX × IP-RP-HPLC setup led to 4533 protein hits. After removal of redundant protein hits between the fractions, a total of 695 unique proteins were identified. The RP × IP-RP-HPLC approach led to the identification of a total number of 2580 proteins, which reduced to 745 unique proteins. This corresponds to 7% more identifications than with the classical SCX × IP-RP method. Both analytical methods gave relevant proteome coverage: 23.2% with the classical SCX × IP-RP-HPLC setup and 24.9% with the RP × IP-RP-HPLC setup. Considering that many proteins were not expressed in the cells, that mostly cytosolic proteins were isolated during sample preparation, and that proteins may be lost during sample preparation (extraction, dialysis), these values are very promising for further biologically relevant applications. Identification Confidence. No set of identified peptides in a PFF approach is free of false-positive hits. The presence of false-positive identifications is intrinsic to the method of identification and mostly results from the algorithms utilized to identify the peptides. Algorithms compare experimental MS/ MS spectra with theoretical fragmentation patterns and return the best peptide match for each MS/MS spectrum within the database.20,47 Because best matches are not necessarily correct 4370

Journal of Proteome Research • Vol. 6, No. 11, 2007

matches, a portion of the identifications are false peptide sequence assignments due to coincidental similarities in MS/ MS fragmentation patterns. In the present study, the identification confidence threshold for MASCOT was set to 95% by applying a MOWSE score cutoff of 23. However, the algorithm may always return random hits and the validity of the identifications needs to be checked a posteriori. To address this issue, all experimental mass lists were tested against a composite database created by concatenating the target protein sequences with decoy sequences of C. glutamicum obtained employing the criteria given by Elias and Gygi (target-decoy search strategy50). At the peptide level, false-positive identification rates of 1.5% and 2.3% were computed with the SCX × IP-RP-HPLC and with the RP × IP-RP-HPLC approaches, respectively. These values are in excellent agreement with the 95% identification confidence threshold set during the MASCOT searches and reveal that no systematic error occurred during the identification process. Complementarity of SCX × IP-RP-HPLC and RP × IP-RPHPLC Methods. Peptide identifications obtained with both 2DHPLC-MS/MS setups were compared by extracting a nonredundant list of identified peptides from these two identification reports. A total of 3659 unique peptides were identified and finally classified into three categories: peptides only identified with the SCX × IP-RP-HPLC setup, peptides only identified with

Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC

research articles

Figure 7. Number of peptides (a) and proteins (b) of C. glutamicum identified with a SCX × IP-RP-HPLC setup and an RP × IPRP-HPLC setup.

the RP × IP-RP-HPLC system, and peptides identified with both setups. The results are summarized in Figure 7. Peptides detected with only one single setup are representing 60.5% of all identified peptides. Only 39.5% of the peptides were detected with both setups and at first sight both methods are strongly complementary. However, the identification of Σ 3659 peptides was performed with six instead of three analyses in the second dimension. The additional identifications are therefore due to both additional analyses in the IP-RP-HPLC-ESI-MS/MS system and the complementarity of both methods of separation. The increase in peptide identifications was estimated from 6-fold analyses of selected fractions. By three from a total of six replicates approximately 80% of all identified peptides were found. It is then possible to compute the theoretical number of peptides identified by performing six replicate injections in the classical SCX × IP-RP-HPLC setup (2398/0.8 ≈ 2998 peptides). Accordingly, the gain in terms of peptide identification due to the complementarity of both setups was evaluated to approximately 22%, which is still significant. This signifies that the analysis in triplicate of one sample with two different setups is more advantageous than 6-fold repetition using a single setup. Upon combining both datasets at the protein level, 871 unique proteins were identified (Figure 7b, the complete protein list is available as Supporting Information). This value corresponds to 29.1% of the proteome of C. glutamicum. The combination of the RP × IP-RP-HPLC setup with the classical SCX × IP-RP-HPLC setup permitted to increase the amount of unique identified proteins by 25% (871 vs 695). 126 proteins were just identified with SCX × IP-RP-HPLC and 176 proteins were just identified with RP × IP-RP-HPLC. The number of proteins identified with both setups is rather high (65.3%) in comparison with the amount of peptides identified with both setups (39.5%). This is explained by the fact that different peptides may lead to the identification of the same protein. Consequently, the number of proteins identified with more than one peptide significantly increases by combining both setups. In the SCX × IP-RP-HPLC and RP × IP-RP-HPLC approaches, 414 and 468 proteins were identified by more than one peptide, respectively.51 Hence, the alternative separation approach permitted to identify 13% more proteins than the

Figure 8. Two-dimensional plots representing isoelectric point (pI) and molecular mass (Mr) (a) of the proteins identified in this study and (b) annotated for C. glutamicum. The distribution of codon adaptation indexes for the whole genome of C. glutamicum and the proteins identified is shown in (c).

classical approach under such criteria of data validation. By combining both methods, 585 proteins were identified by at least two peptides, which corresponds to a gain of 41% as compared to a single analysis based on the SCX × IP-RP-HPLC setup. Hence, the combination of both methods of analysis does not only increase the amount of identified proteins but also leads to significantly more confident identifications. Suitability of the 2D-HPLC Setups for Proteome Analysis. The aim of this work was to develop and evaluate chromatography-based, multidimensional separation systems for proteome analysis. In order to illustrate that the 2D-HPLC systems Journal of Proteome Research • Vol. 6, No. 11, 2007 4371

research articles described in this work are not discriminating against proteins of particular isoelectric point or molecular mass, the proteins identified with both 2D-HPLC setups were plotted in a molecular mass vs isoelectric point diagram. A similar diagram was also plotted for all proteins annotated in C. glutamicum. Figure 8a shows that proteins are well represented over a molecular mass range of a few thousand to several hundred thousands and that isoelectric points ranged from 3 to 12.5. The observed protein properties correlate well with the molecular mass and isoelectric point ranges predicted for the proteome of C. glutamicum (Figure 8b). It has been suggested to utilize the codon bias values of genes as a measure of expression of the corresponding proteins, since highly expressed proteins generally show large codon bias levels.52 The distributions of codon bias values for the entire genome of C. glutamicum and for the proteins identified in this study are illustrated in Figure 8c. Not surprisingly, the distribution for the identified proteins shows a shift toward higher codon bias values, indicating that highly abundant proteins are preferentially detected in the HPLC-ESI-MS/MS method. For codon adaptation indexes higher than 0.5, between 75 and 90% of all proteins were detectable. However, 44 of the 558 proteins with the lowest codon bias values between 0.1 and 0.2 were identified, demonstrating that at least 7.9% of the low expressed proteins were detectable with our setup. The proteins identified in this study cover the main biochemical pathways in C. glutamicum, including the majority of enzymes involved in central metabolic pathways, such as glycolysis, oxidative decarboxylation, tricarboxylic acid cycle (Krebs cycle), as well as the pentose phosphate pathway responsible for the production of NADPH, which is an important cofactor for various biosynthetic pathways. Furthermore, enzymes involved in the biosynthesis of the two main biotechnological products of this organism, L-glutamate and L-lysine, could be identified. The knowledge of the expression level of these enzymes is a necessary prerequisite for the understanding of the complex metabolic/anabolic networks in the cells and for the conception of modified organisms optimized for the biosynthesis of high value products. Therefore, the high coverage of the proteome achieved by the application of 2D-HPLCESI-MS/MS provides a valuable tool for further studies in this field, e.g., the differential (quantitative) analysis of proteome wide changes in expression levels under different growth conditions or as response to genetic modification of the organism.

Conclusions First-dimension separation by RP-HPLC at basic pH facilitates the fractionation of tryptic peptides generated from a bacterial proteome with higher efficiency and homogeneity as compared to classical SCX-HPLC. This allows not only a significant increase both in the number of unique peptides and proteins identified but also a substantial decrease in total analysis time, i.e., 124 vs 176 h for triplicate analysis of the collected fractions in our setup, because a smaller number of fractions need to be collected for an equivalent number of unique peptide identifications. Moreover, RP-HPLC offers some advantages in terms of practical handling and routine operation, such as the use of volatile hydro-organic mobile phases allowing easy evaporation of solvent in order to concentrate fractions. Contrarily, SCX-HPLC utilizes buffered, involatile salt solutions, which are more laborious to prepare and often a source of instrument corrosion and system clogging. 4372

Journal of Proteome Research • Vol. 6, No. 11, 2007

Delmotte et al.

The number of positive charges carried by the peptides is the determinant for retention of peptides in first-dimension SCX-HPLC. The addition of acetonitrile to the eluents facilitates suppressing secondary interactions such as hydrophobic interactions between the peptides and the cation exchanger. On the other hand, first-dimension RP-HPLC separation using high-pH eluents rests upon a strong dependence of retention on the hydrophobic properties of the peptides. Due to deprotonation of the carboxylic groups, ammonium groups, and other acidic functionalities such as phenolic hydroxyl groups at high pH as opposed to their protonation at low pH, the hydrophobic properties of the peptides change substantially as a function of pH, forming a basis for two-dimensional separations on RP-phase stationary phases under basic and acidic elution conditions. Supplementary selectivity in seconddimension IP-RP-HPLC is achieved through the addition of amphiphilic ions to the eluent that effect additional electrostatic interactions. The orthogonality between the two dimensions is higher in the SCX × IP-RP-HPLC than in the RP × IPRP-HPLC approach. However, more homogeneous peptide distribution and better separation efficiency achieved with the RP × IP-RP-HPLC setup permitted the identification of significantly more peptides than with the classical SCX × IP-RP-HPLC setup. The approaches are complementary and in due consequence, application of both setups can identify more peptides than replicate injections performed with a single setup. Both setups flavor the detection of highly abundant proteins but are not discriminating against pI and molecular mass of the proteins.

Acknowledgment. We thank Judith Becker for technical support with cultivation of C. glutamicum and Christian Schley for helpful discussions. Supporting Information Available: Complete protein lists. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Lottspeich, F. Angew. Chem., Int. Ed. 1999, 38, 2476-92. (2) Wilkins, M. R.; Williams, K. L.; Hochstrasser, D. F. Proteome research: New frontiers in functional genomics; Springer: Berlin, 1997. (3) Marko-Varga, G. Proteomics and Peptidomics. New Technology Platforms Elucidating Biology; Elsevier: Amsterdam, 2005. (4) Speicher, D. W. Proteome Analysis: Interpreting the Genome; Elsevier: Amsterdam, 2004. (5) Geng, M.; Ji, J.; Regnier, F. E. J. Chromatogr. A 2000, 870, 295313. (6) McDonald, W. H.; Ohi, R.; Miyamoto, D. T.; Mitchison, T.; Yates, J. R., III. Int. J. Mass Spectrometry 2002, 219 245-51. (7) Wu, S. L.; Choudhary, G.; Ramstrom, M.; Bergquist, J.; Hancock, W. S. J. Proteome Res. 2003, 2, 383-93. (8) Gygi, S. P.; Rist, B.; Griffin, T. J.; Eng, J.; Aebersold, R. J. Proteome Res. 2002, 1, 47-54. (9) Wang, H.; Hanash, S. J. Chromatogr. B 2003, 787, 11-8. (10) Sickmann, A.; Reinders, J.; Wagner, Y.; Joppich, C.; Zahedi, R.; Meyer, H. E.; Schonfisch, B.; Perschil, I.; Chacinska, A.; Guiard, B.; Rehling, P.; Pfanner, N.; Meisinger, C. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 13207-12. (11) Shen, Y.; Jacobs, J. M.; Camp, D. G. 2nd; Fang, R.; Moore, R. J.; Smith, R. D.; Xiao, W.; Davis, R. W.; Tompkins, R. G. Anal. Chem. 2004, 76, 1134-44. (12) Evans, C. R.; Jorgenson, J. W. Anal. Bioanal. Chem. 2004, 378, 1952-61. (13) Huber, C. G.; Walcher, W. Timperio A. M.; Troiani, S.; Porceddu, E.; Zolla, L. Proteomics 2004, 4, 3909-20. (14) Chen, H. S.; Rejtar, T.; Andreev, V.; Moskovets, E.; Karger, B. L. Anal. Chem. 2005, 77, 2323-31. (15) Yang, Z.; Harris, L. E.; Palmer-Toy, D. E.; Hancock, W. S. Clin. Chem. 2006, 52, 1897-905.

research articles

Two-Dimensional Reversed-Phase × Ion-Pair Reversed-Phase HPLC (16) Schley, C.; Altmeyer, M.; Mu ¨ ller, R.; Swart, R.; Huber, C. G. J. Proteome Res. 2006, 5, 2760-68. (17) Pappin, D. J.; Hojrup, P.; Bleasby, A. J. Curr. Biol. 1993, 3, 32732. (18) Mann, M.; Hojrup, P.; Roepstorff, P. Biol. Mass Spectrom. 1993, 22, 338-45. (19) Mann, M.; Wilm, M. Anal. Chem. 1994, 66, 4390-9. (20) Eng, J. K.; McCormack, A. L.; Yates, J. R., III. J. Am. Soc. Mass Spectrom. 1994, 5, 976-89. (21) Kelleher, N. L. Anal. Chem. 2004, 76, 197A-203A. (22) Bogdanov, B.; Smith, R. D. Mass Spectrom. Rev. 2005, 24, 168200. (23) Wu, S. L.; Jardine, I.; Hancock, W. S.; Karger, B. L. Rapid Commun. Mass Spectrom. 2004, 18, 2201-7. (24) Wolters, D. A.; Washburn, M. P.; Yates, J. R. Anal. Chem. 2001, 73, 5683-90. (25) Alpert, A. J.; Andrews, P. C. J. Chromatogr. 1988, 443, 85-96. (26) Molnar, I.; Horvath, C. J. Chromatogr. 1977, 142, 623-40. (27) Horvath, C.; Melander, W.; Molnar, I.; Molnar, P. Anal. Chem. 1977, 49, 2295-305. (28) Horvath, C.; Melander, W.; Molnar, I. J. Chromatogr. 1976, 125, 129-56. (29) Bartha, A.; Stahlberg, J. J. Chromatogr. A 1994, 668, 255-84. (30) Washburn, M. P.; Wolters, D.; Yates, J. R., III. Nat. Biotech. 2001, 19 242-47. (31) Davis, M. T.; Beierle, J.; Bures, E. T.; McGinley, M. D.; Mort, J.; Robinson, J. H.; Spahr, C. S.; Yu, W.; Luethy, R.; Patterson, S. D. J. Chromatogr. B 2001, 752, 281-91. (32) Wagner, K.; Miliotis, T.; Marko-Varga, G.; Bischoff, R.; Unger, K. K. Anal. Chem. 2002, 74, 809-20. (33) Mitulovic, G.; Stingl, C.; Smoluch, M.; Swart, R.; Chervet, J. P.; Steinmacher, I.; Gerner, C.; Mechtler, K. Proteomics 2004, 4, 2545-57. (34) Chelius, D.; Zhang, T.; Wang, G.; Shen, R.-F. Anal. Chem. 2003, 75, 6658-65. (35) Wienkoop, S.; Glinski, M.; Tanaka, N.; Tolstikov, V.; Fiehn, O.; Weckwerth, W. Rapid Commun. Mass Spectrom. 2004, 18, 64350.

(36) Tschappat, V.; Varesio, E.; Signor, L.; Hopfgartner, G. J. Sep. Sci. 2005, 28, 1704-11. (37) Wagner, Y.; Sickmann, A.; Meyer, H. E.; Daum, G. J. Am. Soc. Mass Spectrom. 2003, 14, 1003-11. (38) Hu, S.; Loo, J. A.; Wong, D. T. Proteomics 2006, 6, 6326-53. (39) Gilar, M.; Olivova, P.; Daly, A. E.; Gebler, J. C. Anal. Chem. 2005, 77, 6426-34. (40) Gilar, M.; Olivova, P.; Daly, A. E.; Gebler, J. C. J. Sep. Sci. 2005, 28, 1694-703. (41) Toll, H.; Oberacher, H.; Swart, R.; Huber, C. G. J. Chromatogr. A 2005, 1079, 274-86. (42) Hermann, T. J. Biotechnol. 2003, 104, 155-72. (43) Sahm, H.; Eggeling, L.; de Graaf, A. A. Biol. Chem. 2000, 381, 899910. (44) Kalinowski, J.; Bathe, B.; Bartels, D.; Bischoff, N.; Bott, M.; Burkovski, A.; Dusch, N.; Eggeling, L.; Eikmanns, B. J.; Gaigalat, L.; Goesmann, A.; Hartmann, M.; Huthmacher, K.; Kramer, R.; Linke, B.; McHardy, A. C.; Meyer, F.; Mockel, B.; Pfefferle, W.; Puhler, A.; Rey, D. A.; Ruckert, C.; Rupp, O.; Sahm, H.; Wendisch, V. F.; Wiegrabe, I.; Tauch, A. J. Biotechnol. 2003, 104, 5-25. (45) Kiefer, P.; Heinzle, E.; Zelder, O.; Wittmann, C. Appl. Environ. Microbiol. 2004, 70, 229-39. (46) Premstaller, A.; Oberacher, H.; Huber, C. G. Anal. Chem. 2000, 72, 4386-93. (47) Perkins, D. N.; Pappin, D. J.; Creasy, D. M.; Cottrell, J. S. Electrophoresis 1999, 20, 3551-67. (48) Schley, C.; Swart, R.; Huber, C. G. J. Chromatogr. A 2006, 1136, 210-20. (49) Kyte, J.; Doolittle, R. F. J. Mol. Biol. 1982, 157, 105-32. (50) Elias, J. E.; Gygi, S. P. Nat. Methods 2007, 4, 207-14. (51) Bradshaw, R. A.; Burlingame, A. L.; Carr, S.; Aebersold, R. Mol. Cell Proteomics 2006, 5, 787-8. (52) Gygi, S. P.; Corthals, G. L.; Zhang, Y.; Rochon, Y.; Aebersold, R. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9390-5.

PR070424T

Journal of Proteome Research • Vol. 6, No. 11, 2007 4373