Identification and Quantification of Basic and Acidic ... - ACS Publications


Identification and Quantification of Basic and Acidic...

9 downloads 94 Views 546KB Size

Identification and Quantification of Basic and Acidic Proteins Using Solution-Based Two-Dimensional Protein Fractionation and Label-Free or 18O-Labeling Mass Spectrometry Wells W. Wu,† Guanghui Wang,† Ming-Jiun Yu,‡ Mark A. Knepper,‡ and Rong-Fong Shen*,† Proteomics Core Facility, and Laboratory of Kidney and Electrolyte Metabolism, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland 20892 Received November 22, 2006

Reduction in sample complexity enables more thorough proteomic analysis using mass spectrometry (MS). A solution-based two-dimensional (2D) protein fractionation system, ProteomeLab PF 2D, has recently become available for sample fractionation and complexity reduction. PF 2D resolves proteins by isoelectric point (pI) and hydrophobicity in the first and second dimensions, respectively. It offers distinctive advantages over 2D gel electrophoresis with respects to automation of the fractionation processes and characterization of proteins having extreme pIs. Besides fractionation, PF 2D is equipped with built-in UV detectors intended for relative quantification of proteins in contrasting samples using its software tools. In this study, we utilized PF 2D for the identification of basic and acidic proteins in mammalian cells, which are generally under-characterized. In addition, mass spectrometric methods (label-free and 18O-labeling) were employed to complement protein quantification based on UV absorbance. Our studies indicate that the selection of chromatographic fractions could impact protein identification and that the UV-based quantification for contrasting complex proteomes is constrained by coelution or partial coelution of proteins. In contrast, the quantification post PF 2D chromatography based on label-free or 18O-labeling mass spectrometry provides an alternative platform for basic/acidic protein identification and quantification. With the use of HCT116 colon carcinoma cells, a total of 305 basic and 183 acidic proteins was identified. Quantitative proteomics revealed that 17 of these proteins were differentially expressed in HCT116 p53-/- cells. Keywords: basic and acidic proteins • solution-based two-dimensional protein fractionation • label-free 18O-labeling • mass spectrometry

Introduction Two-dimensional (2D) gel electrophoresis has been a popular and widely practiced method for protein separation. Over the past several years, this labor-intensive technique has been complemented by liquid chromatography (LC)-based methods, particularly in the area of high-throughput proteomic research using mass spectrometry (MS). Among the compelling causes driving this trend are issues concerning the reproducibility,1 poor representation of low-abundance, highly acidic/basic, high molecular weight, or highly hydrophobic proteins, as well as difficulties in automation of the gel-based techniques.2 LCMS methods such as MudPIT2 is an alternative to the 2D-gel MALDI-TOF approach, relying on LC to reduce sample peptide complexity prior to MS analysis. In a typical 2D LC-MS * All correspondence should be addressed to Rong-Fong Shen, Ph.D., National Heart, Lung, and Blood Institute, NIH, Building 10, Rm 8C103C, 10 Center Dr., Bethesda, MD 20892-1597. Tel: 301-594-1060. Fax:301-4022113. E-mail: [email protected]. † Proteomics Core Facility, National Heart, Lung, and Blood Institute, National Institutes of Health. ‡ Laboratory of Kidney and Electrolyte Metabolism, National Heart, Lung, and Blood Institute, National Institutes of Health. 10.1021/pr060621c Not subject to U.S. Copyright. Publ. 2007 Am. Chem. Soc.

method, proteins are digested into peptides which are then separated first by strong cation exchange, followed by reversedphase column chromatography and MS analysis.2 A new fractionation system, the ProteomeLab PF 2D (Beckman Coulter, Fullerton, CA), resolves proteins according to isoelectric point (pI) in the first dimension (chromatofocusing) and hydrophobicity in the second dimension.3 The system is equipped with two UV detectors and two fractional collectors, and often used for off-line sample complexity reduction at the protein level. Compared to 2D gel electrophoresis, PF 2D offers (i) consistent reproducibility,3,4 as the same columns could be used for multiplexed samples; (ii) improved detection of lower abundance proteins, as up to 5 mg of proteins could be loaded (as opposed to the typically 100-500 µg of proteins for an immobilized pH gradient (IPG) strip for 2D gel analysis);5 (iii) the opportunity to analyze proteins with extreme pIs; and (iv) easier automation with other liquid handling systems. The fractionation capability of PF 2D has been demonstrated by several groups of investigators.6-8 Most studies focused on proteins within the resolving range of chromatofocusing (pI 8.5-4.0). However, basic proteins make up a significant portion of the eukaryotic proteomes. The proteome map of SaccharoJournal of Proteome Research 2007, 6, 2447-2459

2447

Published on Web 05/17/2007

research articles myces cerevisiae suggests that at least 1900 (30.1%) proteins encoded by the yeast genome have a pI between 9 and 12.9 With the use of the PF 2D system, it has been reported that ∼41% of mammalian proteins were recovered in fractions with pI > 76 and ∼23% in fractions with pI > 9.10 when the flowthrough and salt-eluted fractions are collected, it should be possible to analyze those highly basic/acidic proteins. Despite this obvious advantage, the characterization of basic and acidic proteins using the PF 2D system has not been fully explored.10-12 In addition to protein fractionation, differential profiling of proteomes could be attempted using PF 2D’s built-in UV detectors.3,6,13 However, protein quantification based on UV absorbance during chromatography, though technically simple, has certain limitations. In particular, poorly resolved or coeluted peaks, just like overlapped spots in 2D gels, could render quantitative comparison rather inaccurate.9,14 In routine 2Dgel analysis, acidic and basic proteins have been largely undercharacterized, as they either do not enter IPG strips or are poorly resolved.15 Since proteome analysis is incomplete without characterizing these proteins, there has been an increasing interest in their identification and quantification. In this study, we used PF 2D to identify acidic and basic proteins in colon carcinoma cells and applied MS methods to quantify their differential expression in two closely related cell lines (HCT116 and HCT116 p53-/-). The combination of PF 2D fractionation and quantification by the label-free16 and 18Olabeling17 mass spectrometry enables quantitative proteomics to be conducted for these subsets of mammalian proteins.

Materials and Methods Cell Culture and Sample Preparation. HCT116 and HCT116 p53 -/- cell lines (gifts from Dr. Bert Vogelstein of the Johns Hopkins University) were cultured in McCoy’s 5A media (Gibco) supplemented with 10% fetal bovine serum and 1% antibiotic-antimycotic (Gibco) in a humidified incubator (37 °C, 5% CO2). Upon reaching 80% of cell confluence, the cells were harvested by brief trypsinization. The cells were pelleted at 1200 rpm (306g) and washed six times with PBS. The cells were then lysed, following the PF 2D Use Protocol provided by Beckman Coulter.13 Liquid Chromatography Using Proteome PF 2D. All buffers and columns used in this study were purchased from Beckman Coulter. The ProteomeLab PF 2D Chemistry Kit includes a chromatofocusing HPCF column (first dimension), a nonporous reversed-phase HPRP column (second dimension), and a start buffer (pH 8.5) and an elution buffer (pH 4.0). The first dimension separation (chromatofocusing) was performed under ambient temperature at a flow rate of 0.2 mL/min. Protein peaks in the eluate were monitored by absorbance at 280 nm. The column was first equilibrated with 30 column vol of the start buffer. Up to 5.0 mg of protein sample, prepared in start buffer, could be loaded. The column was washed with the start buffer for 20 min to elute the unbound proteins (pI > 8.5), which were collected in 10 fractions. The pH gradient was then generated by replacing the start buffer with the elution buffer (pH 4.0), and the elution continued until the eluate reached pH 4.0. The column was then washed with 10 column vol of 1 M NaCl to elute tightly bound proteins, followed by 10 column vol of water until the absorbance at 280 nm reached baseline. The first dimension fractions were collected in 96 well plates (every 0.3 pH units during pH gradient or every 5 min during the other stages of the run, before pH gradient and during salt washing). The second dimension separation was performed at 2448

Journal of Proteome Research • Vol. 6, No. 7, 2007

Wu et al.

50 °C at a flow rate of 0.75 mL/min. The absorbance of proteins in the eluate was monitored at UV 214 nm. The mobile phase consists of two solvents: solvent A (0.1% TFA in water) and solvent B (0.08% TFA in acetonitrile). The column was first equilibrated with 10 column vol of 100% solvent A prior to each sample injection. Two minutes after sample injection (200 µL), bound proteins were eluted with a gradient of 0-100% solvent B for 30 min, followed by 100% solvent B for 4 min. The column was then re-equilibrated with 100% solvent A. The second dimension fractions were collected in 96 well plates every 0.5 min. Protein Digestion and 18O-Labeling. Fractions from the second dimension were dried in a SpeedVac and reconstituted in 50 µL of ammonium bicarbonate (50 mM, pH 8.1). The samples were reduced with 25 µL of a 10 mM DTT at 60 °C for 1 h, alkylated with 25 µL of a 55 mM iodoacetamide solution at room temperature in the dark for 0.5 h, and then trypsinized at 37° for 16-18 h using a protein/trypsin ratio of ∼50. The digest was then split into two aliquots: one for label-free analysis (acidified to pH < 2) using nanospray LC-MS/MS and the other for 18O-labeling experiment. The aliquot for 18Olabeling was mixed with 5 µL of unwashed immobilized trypsin (Poroszyme from Applied Biosystems), and the mixture was completely dried in a SpeedVac for 1 h at room temperature. The dried samples were then resuspended in 50 µL of H216O/ acetonitrile (4:1) for HCT116 cell extract or 50 µL of H218O/ acetonitrile (4:1) for HCT116 p53 -/- cell extract. Both H216O and H218O contained 50 mM ammonium bicarbonate for maintenance of pH. The samples were continuously agitated at 1200 rpm (to suspend the immobilized trypsin) for 3-5 h at room temperature. The enzyme-catalyzed labeling was stopped by adding 0.5 µL of concentrated formic acid, followed by brief vortexing. The immobilized trypsin was pelleted by centrifugation at 12,000 rpm for 5 min and the supernatant carefully transferred to a clean tube. An equal volume (30 µL) of the 16O and 18O samples was combined and placed in a SpeedVac to near dryness. The sample was reconstituted with 10 µL of 0.1% TFA, and then ziptip spotted onto a MALDI target plate using 5-10 µL of R-cyano-4-hydroxycinnamic acid (4 mg/mL in 60/ 40 ACN/water containing 0.1% TFA). Mass Spectrometry. Nanospray LC-MS/MS analyses of label-free samples were carried out using a linear ion trap LTQ (Thermo Finnigan, San Jose, CA), as previously described.16 Briefly, peptides were first loaded onto a trap cartridge (Agilent, Palo Alto, CA) at a flow rate of 2 µL/min. Trapped peptides were then eluted onto a reversed-phase PicoFrit column (New Objective, Woburn, MA) using a linear gradient of acetonitrile (0-60%) containing 0.1% formic acid. The duration of the gradient was 20 min at a flow rate of 0.25 µL/min, which was followed by 80% acetonitrile washing for 5 min. The eluted peptides from the PicoFrit column were sprayed into an LTQ mass spectrometer equipped with a nanospray ion source. The data-dependent acquisition mode was enabled, and each survey MS scan was followed by five MS/MS scans with dynamic exclusion option on. The spray voltage and ion transfer tube temperature were set at 1.8 kV and 160 °C, respectively. The normalized collision energy was set at 35%. With the use of ABI 4700 MALDI TOF/TOF (Applied Biosystems, Foster City, CA), the analysis of 18O-labeled samples was carried out as previously described14 with the maximum number of precursors per fraction, S/N filter, and fraction-to-fraction precursor mass tolerance set at 10, 30, and 200 ppm, respectively.

research articles

Identification and Quantification of Basic and Acidic Proteins

Figure 1. First-dimension chromatogram of HCT116. (a) Dotted line, pH gradient; (b) solid line, UV 280 absorbance; (c) vertical line, collection fractions.

Database Searching and Quantification by Label-Free and O-Labeling Methods. For nanospray LC-MS/MS data, SEQUEST/Bioworks 3.1 was used to match MS/MS spectra to peptides in the Swiss-Prot human database. Spectra/peptide matches were considered significant if they had a normalized difference in cross-correlation scores (∆Cn) of at least 0.1 and minimum cross-correlation scores (XCorr) of 2.0 for +1, 2.5 for +2, and 3.5 for +3 charged ions. For MALDI TOF/TOF data, Mascot 2.0 was used to match MS/MS spectra to peptides in human Swiss-Prot database. Spectra/peptide matches were considered significant if individual ion score (based on MS/ MS) is g95% confidence interval. Proteins were identified on the basis of having at least two peptides in at least a given fraction. The quantification of label-free samples was achieved using QUOIL label-free algorithm developed in-house.16 Essentially, individual second dimension fractions were analyzed by LC-MS/MS. Label-free quantification was performed at the peptide level, where peptide identifications from individual second dimension fractions were assembled into a master list and a ratio of reconstructed chromatogram peak area between the contrasting fractions for each identified peptide was calculated. The median of the computed peptide ratios represents the relative quantity of the protein in the two samples. Quantification of 18O-labeled samples was achieved using GPS Explorer and a mass difference of 4 between isotope pairs. The median of peptide ratios in multiple fractions of the same protein represents the protein expression ratio between the two contrasting samples. RT-PCR and Quantitative Real Time PCR Methods. Reverse transcription-polymerase chain reaction (RT-PCR) was performed as previously described.18 Briefly, cell pellets were collected as described above and total RNA was extracted using the TRIzol reagent (Invitrogen, Carlsbad, CA). First-strand cDNA was produced from the total RNA (5 µg), using an oligodT primer and SuperScript reagents (Invitrogen) in a reaction volume of 20 µL. PCR products were generated from 0.5 µL of the first-strand cDNA mixture using gene-specific primers (Supplemental Table 1 in Supporting Information) and Immo18

Mix (Bioline USA, Inc., Randolph, MA). The PCR reaction was performed in a Peltier Thermal Cycler PCT-200 (MJ Research, Waltham, MA) with the following conditions: activation of DNA polymerase at 95 °C for 7 min, 30 cycles of amplification (94 °C, 30 s; 58 °C, 30 s; and 72 °C, 30 s), and final extension at 72 °C for 10 min. The RT-PCR products were resolved using 1.5% agarose gel electrophoresis, stained with ethidium bromide, and visualized using the Gel Logic 100 imaging system (Kodak, New Haven, CT). Real time quantitative PCR was performed as previously described19 using an ABI Prism 7900HT system. Briefly, the amplification was carried out using 0.5 µL of cDNA, genespecific primers, Quantitect SYBR green PCR kit (Qiagen, Valencia, CA), and default thermal profile according to the manufacturer’s protocol. Typical threshold cycle (CT) values for these experiments were in the range of 20-25. The relative quantification of gene expression was determined using the comparative CT method.20 All experiments were performed in duplicate, and the relative quantification values were normalized with actin as a control.

Results Chromatofocusing of Proteins. A typical first-dimension chromatogram of HCT116 cell lysate is presented in Figure 1. The broad peak eluted before time 50 min is composed of a mixture of proteins, presumably with pI g 8.5, which do not bind to the HPCF column. The peaks eluted between 50 and 135 min are resolved by chromatofocusing (pH 8.5-4), while those eluted after 135 min are, in theory, proteins having pI e 4 or bound tightly to the column eluted by the high salt buffer. It should be noted that the broad peaks outside the pH gradient region may also contain non-protein constituents present in the lysis, start, or elution buffer that absorb at 280 nm. Basic Proteins. The unbound basic proteins were collected in the first nine fractions. The elution profile of these proteins is nonsymmetric with a slight tailing, suggesting that there is some degree of nonspecific interaction between the matrix and Journal of Proteome Research • Vol. 6, No. 7, 2007 2449

research articles

Wu et al.

Figure 2. Reversed-phase second-dimension chromatograms of the unretained basic proteins with pI g 8.5 of HCT116. (a) Fraction nos. 2-7; (b) fraction nos. 4-7.

proteins. Under an ideal situation, any of the nine fractions would be suitable for protein analysis or quantitative comparison between duplex samples. Because of the asymmetric chromatographic elution, we first determined the composition of each fraction in order to select the appropriate fraction(s) for identification and quantitative comparison. Since fraction no. 1 consists of the eluate around the void volume of the HPCF column (∼4.3 min at 0.2 mL/min flow rate) and fractions 8 and 9 have only trace amount of proteins (determined by nanospray LTQ analysis, data not shown); only fractions nos. 2-7 were further characterized. The reversed-phase (RP) chromatograms of fractions nos. 2-7 are illustrated in Figure 2a, 2450

Journal of Proteome Research • Vol. 6, No. 7, 2007

and the zoom-in chromatograms for fractions nos. 4-7 are shown in Figure 2b. The UV absorbance spectra in Figure 2 are highly similar with respect to the number of peaks and their retention time, indicating the bulk of the proteins are largely similar, albeit at different concentrations. This is consistent with the observed trend of decreasing intensity (area) with increasing fraction number seen in Figure 1. Fractions nos. 2 and 3 contain the majority of proteins, as reflected by higher intensity and less resolution of the peaks. Proteins in the collected fractions of the reversed-phase chromatography for each of the fractions (nos. 2-4) were profiled using nanospray LTQ tandem mass spectrometry. The

research articles

Identification and Quantification of Basic and Acidic Proteins

proteins identified were compiled in multiconsensus reports using Bioworks 3.1. The box and whisker graphs showing the number of identified peptides for proteins from fraction no. 3 or fraction no. 4 versus that from fraction no. 2 are shown in Figure 3, panels a and b, respectively. For illustration, 30 peptides of HNRPM_HUMAN (human heterogeneous nuclear ribonucleoprotein M, indicated by arrows at the right) were identified in fraction no. 2, while 23 and 11 peptides of the same protein were identified in fraction nos. 3 and 4, respectively. Since the number of peptides of a protein detected by MS reflects its concentration in respective samples semiquantitatively,21 it suggests that HNRPM_HUMAN is more abundant or dominant in fraction no. 2 than in fraction no. 3. For convenience of discussion, proteins were arbitrarily classified into 4 categories: highly abundant (I, g16 peptides identified), abundant (II, 8-15 peptides), medium abundant (III, 4-7 peptides), or low abundant (IV, e 3 peptides identified). As shown in Figure 3a,b, for the highly abundant proteins, a greater number of peptides of the same protein was identified in fraction no. 2 than in fraction no. 3 (with a few exceptions) or no. 4 (no exception). For abundant proteins, about threequarters of them have a greater number of peptides identified in fraction no. 2 than in fraction no. 3, while all in fraction no. 2 have more peptides identified than those in fraction no. 4. The medium abundant proteins exhibit a similar trend, but with more peptides of the same protein found in either fraction no. 3 or 4, than in fraction no. 2. For the low-abundance proteins, the number of peptides identified from fractions nos. 3 and 4 tend to be higher than that from fraction no. 2. Several proteins undetected in fraction no. 2 (zero number of peptides in x-axis) actually appear quite abundant in fractions no. 3 or 4. Thus, these results indicate that the choice of fractions in the alkaline pI region could impact protein identification and subsequent quantification. Table 1 lists some highly basic proteins (pI g 10) identified from the flow-through fractions. The theoretical pIs were calculated, using an online program (Compute pI/Mw of Expasy, http://us.expasy.org), from the pKa of amino acids. It is interesting, though not surprising, that histones, hnRNPs, and ribosomal proteins make up the majority of the proteins in this high pI region, consistent with their function as DNAor RNA-binding proteins. The complete list of the 305 proteins identified in the flow-through region is presented in Supplemental Table 2 in Supporting Information, which gives the name, the pI, and the number of peptides identified for each protein in fractions 2-4. Figure 3c is a pie chart showing the pI distribution among the proteins identified. About 60% of the proteins fall within the expected pI region (i.e., pI g 8.5). However, the remaining 40% have pIs lower than the cutoff point (pI ) 8.5), suggesting that these proteins might be posttranslationally modified/processed, or that the pI-dependent binding of these proteins to the column might be perturbed. It should be mentioned that the calculation of pIs does not take into consideration the effect of protein tertiary structure on the apparent pI. Acidic Proteins. Five fractions (fractions 29-33) were collected from the salt washing step following chromatofocusing (Figure 1). Figure 4a shows the reversed-phase chromatograms of fractions nos. 29-33, while Figure 4b displays the zoom-in view of the chromatograms for fractions nos. 30-33. The absorbance spectra exhibit a trend of decreasing intensity (protein concentration) with increasing fraction number, similar to that seen in Figure 2. Fractions 31-33 are near the lower

right shoulder of the eluted peak, and nanospray LTQ analysis of proteins in their second-dimension fractions yielded only trace amounts of proteins (data not shown). The proteins in the second-dimension fractions of 29 or 30 were profiled, and the results were compiled. The abundance comparison (Figure 5a) shows a similar pattern to those of the basic proteins. Table 2 lists a number of highly acidic proteins (pI e 5) identified, and the complete list of the 183 proteins identified is presented in Supplemental Table 3 in Supporting Information. The number of acidic proteins is ∼60% that of the basic proteins. The pI distribution of these acidic proteins is shown in the pie chart (Figure 5b). Interestingly, no protein identified has a theoretical pI lower than 4.0. Nearly 17% of the proteins have a pI between 4.0 and 5.0, while a vast majority (83%) have a pI above 5.0 (7% with pI g 7.0). The result suggests that posttranslational modification or protein processing might have shifted the pIs of some of these proteins. Alternatively, pIindependent interaction such as hydrophobic interaction between the matrix and proteins might prevent their elution by the pH 4.0 buffer. Quantification Using Label-Free and 18O-Labeling. To investigate differential expression of acidic/basic proteins between two closely related colon carcinoma cell lines, HCT116 and HCT116 p53-/-, the profiling of proteins in corresponding fractions using UV absorbance at 214 nm was initially attempted. A typical UV profiling using fraction no. 3 of the first dimension from each of the cell lysates is shown in Figure 6. The horizontal lines denote the fractions from which 12 of the abundant proteins were identified through MS analysis. The low resolution of protein peaks and the multi-fraction spreading of the same protein suggested that such profiling would not be able to correlate protein quantities, based on UV absorbance, with protein identities determined by MS analysis. Mass spectrometric methods using the label-free or 18O-labeling approach was therefore chosen. Since protein identification could be fraction-biased, as seen in Figures 3 and 5, we decided to use an equal aliquot of fractions nos. 2-7 and fractions 29 and 30 for basic and acid proteins, respectively. Table 3 is a list of proteins whose abundance varies ( 50% or more between the two cell lines. The overlap between the two methods is 18% (2/11) and 17% (1/6) for basic and acidic proteins, respectively. Despite the presence of many abundant proteins, 9 non-nucleic acid-binding proteins were identified by the combined approaches. To validate this result, RT-PCR was carried out to ensure that mRNAs are expressed in both cell lines. As shown in Figure 7, the transcripts of all 17 proteins in Table 3 were detectable in both cell lines after 30-cycle amplification. The intensities of these amplified products however varied significantly. Clearly, a reliable quantification of the transcripts necessitates the adjustment of the amount of input cDNA and/or the number of amplification cycles. This prompted us to employ real-time PCR for quantification. As shown in the last column of Table 3, the trend of changes in the amount of mRNAs correlated reasonably well with the changes determined by quantitative proteomic methods.

Discussion The convenience of automatically collecting flow-through and salt washing fractions using the PF 2D system provides an opportunity to identify highly basic and acidic proteins in mammalian cells. The fact that 40% of the identified proteins in the flow-through fractions have a pI value less than 8.5 suggests that some mature proteins within the cells may differ Journal of Proteome Research • Vol. 6, No. 7, 2007 2451

research articles

Wu et al.

Figure 3. The number of peptides identified in fraction no. 3 (a) and fraction no. 4 (b) versus the number of peptides identifieda in fraction no. 2. [Section nos. I, II, III, and IV arbitrarily represent highly abundant, abundant, medium abundant, and low-abundance proteins, respectively.] (c) The pie chart of theoretical pI of the proteins reported. 2452

Journal of Proteome Research • Vol. 6, No. 7, 2007

research articles

Identification and Quantification of Basic and Acidic Proteins Table 1. Proteins Identified with pI g 10a

a

accession number

protein name

theoretical pI

MW

P62314 O60783 P62805 O75494 P14678 Q86V81 Q93077 P62269 P20670 Q99878 P16104 P62857 P62899 P62277 P62829 P62266 P46778 P62841 P62318 O60814 P06899 P62280 P39019 P22090 P15880 P62249 P18621 P22087 P62701 P46783 P62244 P62851 P27635 P63173 P62081 P62263 P38159 P37108 Q9Y4Z0 Q9BY77

Small nuclear ribonucleoprotein Sm D1 Mitochondrial 28S ribosomal protein S14 Histone H4 FUS-interacting serine-arginine-rich protein 1 Small nuclear ribonucleoprotein-associated proteins B and B′ THO complex subunit 4 Histone H2A type 1-C 40S ribosomal protein S18 Histone H2A type 2-A Histone H2A type 1-J Histone H2A.x 40S ribosomal protein S28 60S ribosomal protein L31 40S ribosomal protein S13 60S ribosomal protein L23 40S ribosomal protein S23 60S ribosomal protein L21 40S ribosomal protein S15 Small nuclear ribonucleoprotein Sm D3 Histone H2B type 1-K Histone H2B type 1-J 40S ribosomal protein S11 40S ribosomal protein S19 40S ribosomal protein S4, Y isoform 1 40S ribosomal protein S2 40S ribosomal protein S16 60S ribosomal protein L17 34 kDa nucleolar scleroderma antigen 40S ribosomal protein S4, X isoform 40S ribosomal protein S10 40S ribosomal protein S15a 40S ribosomal protein S25 60S ribosomal protein L10 60S ribosomal protein L38 40S ribosomal protein S7 40S ribosomal protein S14 Heterogeneous nuclear ribonucleoprotein G Signal recognition particle 14 kDa protein U6 snRNA-associated Sm-like protein LSm4 DNA polymerase delta interaction protein 3

11.56 11.42 11.36 11.26 11.20 11.15 11.05 10.99 10.90 10.88 10.74 10.70 10.54 10.53 10.51 10.50 10.49 10.39 10.33 10.32 10.32 10.31 10.31 10.25 10.25 10.21 10.18 10.18 10.16 10.15 10.14 10.12 10.11 10.10 10.09 10.08 10.06 10.05 10.02 10.00

13281.6 15138.7 11236.2 31300.5 24610.1 26756.7 13974.3 17718.7 13964.3 13805.1 15013.4 7841.0 14462.9 17091.1 14865.4 15676.4 18433.6 16908.9 13916.3 13759.0 13773.0 18299.5 15929.3 29324.4 31324.4 16314.1 21265.8 33784.2 29466.5 18897.8 14708.3 13742.1 24445.7 8086.7 22126.9 16141.5 42331.9 14543.9 15349.7 46089.3

Note: Theoretical pI and MW are obtained from Expasy’s Compute pI/MW tool.

in sequences from those directly translated from mRNAs, likely as a result of post-translational modification and/or processing. Although beyond the focus of this study, it would be interesting to see in the future whether this hypothesis can be substantiated by determining the apparent molecular weight and pI of each protein identified. While post-translational modification and processing may be plausible explanations, the possibility that the apparent pI of a protein in 3D structure could significantly deviate from that calculated from its amino acids cannot be excluded. Nor can one rule out that imperfect pH gradient formation in the column may also alter the affinities of proteins, in particular those having a pI around the cutoff point (pI 8.5). The same argument is probably applicable to those tightly retained by the column not eluted by pH 4.0 buffer. Surprisingly, none of those tightly bound protein has a pI less than 4.0. To rationalize this somewhat unexpected observation, the number of proteins having a pI e 4.0 among the 14 261 entries in Swiss-Prot human database was surveyed using the Compute pI/Mw software (Expasy). Interestingly, only 21 proteins (0.15%) in the database have a theoretical pI e 4.0 (Supplemental Table 4 in Supporting Information). The fact that none of these proteins were detected suggests that they are either not very abundant or their tryptic peptides are too acidic to be retained by the reversed-phase column, therefore, not detected during MS analysis. Our observations are consis-

tent with two earlier reports showing that proteins eluted from the lower pH gradient consistently had a poor correlation with the expected pI range10 and that the pIs of proteins eluted by salt washing ranged from 7.3 to 5.2.11 Under recommended PF 2D operation conditions, ∼50-150 proteins were typically found in each of the pI fractions,3,6,22 while ∼1-20 proteins were detected in each reversed-phase fraction.11 Although the effect of peak coelution on quantification of complex proteomes (e.g., whole cell lysates, serum, or urine samples) has yet to be addressed for the flow-through fractions, such effect has certainly been noted in fractions within the pH gradient region.4,6,23 The second-dimension chromatograms for those acidic and basic protein fractions also contain many poorly resolved peaks (Figures 2 and 4). In a recent study on drug-treated glioma cells using PF 2D, comparative evaluation based on UV absorbance was performed only on those chromatographically distinctive peaks (both sides of the peak reaching the baseline13), suggesting that accurate quantification using UV-based approach is more reliable when one single protein is present in a peak. For proteins in the flowthrough and salt washing fractions in our study, quantification using UV absorbance is technically impractical. Not only are most protein peaks at best nominally resolved, but many proteins are also present in multiple fractions (Figures 2, 4, and 6). The correlation between protein quantity (determined by Journal of Proteome Research • Vol. 6, No. 7, 2007 2453

research articles

Wu et al.

Figure 4. Reversed-phase second-dimension chromatograms of the acidic proteins with pI e 4.0 of HCT116. (a) Fraction nos. 29-33; (b) fraction nos. 30-33.

UV absorbance) and protein identity (determined by mass spectrometry) will be hard to delineate when more than one protein are in the same fraction. This together with the concerns on the sensitivity of UV 214 nm for protein detection (10s-100s ng level24) as well as the potential masking effects by abundant nucleic acid-binding proteins prompted us to use mass spectrometric methods for quantifying differentially expressed proteins in the two colon carcinoma cell lines. 2454

Journal of Proteome Research • Vol. 6, No. 7, 2007

We have shown that MS identification (and subsequent quantification) of proteins could be affected by the selection of the first-dimension fractions (Figures 3 and 5). This might be attributed to one or the combination of the following: (1) ionization suppression by the coeluted abundant peptides, favoring detection of abundant peptides; (2) ion accumulation within the ion trap with abundant peptides precluding detection of otherwise detectable low-abundance ones;25 (3) imper-

Identification and Quantification of Basic and Acidic Proteins

research articles

Figure 5. The number of peptides identified in fraction no. 30 (a) versus the number of peptides identified in fraction no. 29. (b) The pie chart of theoretical pI of the proteins reported.

fect chromatography causing abundant peptides not confined to consecutive scans during LC-MS;26 and (4) nonideal binding caused by transient local pI variation when unbound proteins (pI g 8.5) are being eluted (see the rise and dip in the pI value in the flow-through region in Figure 1; also reported by others11,12). For quantification, we figured that the logical choice was a combination of an equal aliquot from fractions nos. 2-7 or 29 and 30. The combined sample should contain all proteins in the region of interest and could dilute out concentrations of abundant proteins, thus, minimizing the ion suppressive effect during mass analysis. In total, 17 basic/acidic proteins were identified to be differentially expressed between HCT116 and HCT116 p53-/cells by the combined label-free/nanospray and 18O-labeling/ MALDI-TOF methods. The complementary nature of electrospray and MALDI for protein profiling has been well-documented.27,28 It has been reported that only 5 out of 35 upregulated proteins involved in p53-induced apoptosis were

commonly identified by both electrospray and MALDI.29 Likewise, different labeling methods (such as CyDyes, cICAT, and iTRAQ) analyzed on MALDI platform also yield complementary sets of identified proteins.14 Thus, the low overlap is likely due to the different labeling schemes and/or ionization modes, and the degree of overlapping (18% and 17% for basic and acidic proteins, respectively) determined here is in line with other studies. The combined result nevertheless provides a more comprehensive quantification. Obviously, the presence of many abundant proteins such as histones and ribosomal proteins in the flow-through fractions might have affected the outcome. The fact that more than 50% (9/17) of the proteins identified are unrelated to the predominant nucleic acid-binding proteins suggests that label-free and 18O-labeling mass spectrometric methods employed here have the potential to unveil quantitative differences of basic/acid proteins in complex proteomes. This is supported by the observation that five of the proteins in Table 3 were previously identified from unfractionated Journal of Proteome Research • Vol. 6, No. 7, 2007 2455

research articles

Wu et al.

Table 2. Proteins Identified with pI e 5a

a

accession number

protein name

theoretical pI

MW

O75506 Q01105 Q8NI22 P05387 P60660 P07951 P19105 P67936 P49006 P06753 P55036 O00273 P31947 P07237 Q15370 Q14444 P54727 P24844 Q9UMX5 P10599 P13693 Q9C005 Q969G3 Q9UKK9 P20674 O76070 Q15121 P55327 P08779 Q9UNZ2 P06576

Heat shock factor-binding protein 1 Protein SET Multiple coagulation factor deficiency protein 2 60S acidic ribosomal protein P2 Myosin light polypeptide 6 Tropomyosin beta chain Myosin regulatory light chain 2, nonsarcomeric Tropomyosin alpha-4 chain MARCKS-related protein Tropomyosin alpha-3 chain 26S proteasome non-ATPase regulatory subunit 4 DNA fragmentation factor subunit alpha 14-3-3 protein sigma Protein disulfide-isomerase [Precursor] Transcription elongation factor B polypeptide 2 GPI-anchored protein p137 Excision repair protein RAD23 homolog B Myosin regulatory light chain 2, smooth muscle isoform Neudesin [Precursor] Thioredoxin Translationally controlled tumor protein Dpy-30-like protein BRG1-associated factor 57 ADP-sugar pyrophosphatase Cytochrome c oxidase polypeptide Va, mitochondrial [Precursor] Gamma-synuclein Astrocytic phosphoprotein PEA-15 Tumor protein D52 Keratin, type I cytoskeletal 16 NSFL1 cofactor p47 ATP synthase subunit beta, mitochondrial

4.17 4.23 4.31 4.42 4.56 4.66 4.67 4.67 4.68 4.68 4.68 4.68 4.68 4.69 4.73 4.76 4.79 4.80 4.81 4.82 4.84 4.84 4.85 4.87 4.88 4.89 4.93 4.94 4.98 4.99 5.00

8543.6 33488.9 13524.8 11664.9 16798.9 32850.7 19662.9 28390.6 19397.6 32818.8 40736.7 36521.9 27774.1 55294.0 13132.8 72751.7 43171.2 19696.0 15684.5 11606.3 19595.3 11249.7 46649.4 24327.6 12513.2 13330.8 15040.1 19863.1 51136.6 40572.8 51769.3

Note: Theoretical pI and MW are obtained from Expasy’s Compute pI/MW tool.

lysates of the same cell lines in two separate experiments.14,16 The greater number of differentially expressed basic/acidic proteins identified with PF 2D fractionation reiterates the importance of reducing sample complexity on biomarker discovery. The expression of mRNA for all differentially expressed proteins determined by mass spectrometric methods was confirmed by RT-PCR. For the majority of the proteins (11 out of 17), the relative changes obtained by real time RT-PCR are consistent with those by proteomic determination. It has been well-established that the degree of mRNA change may not necessarily positively correlate with that of protein change in the cells.30 Studies on the yeast proteome have demonstrated instances of poor,31 weakly positive,32 and good33 correlation between mRNA expression and protein expression, depending on the criteria and statistical methods used. Using cell lines at distinctive hematopoietic differentiation stages and a mouse model responding to drug treatment, Tian et al.34 reported a 40% correlation between mRNA and protein levels. Similarly, Johnson et al.35 reported a 48% (23 out of 48) concordant RNA/ protein expression ratio using ICAT and cDNA microarray quantitation on proteins and mRNAs of a mouse p53 +/+ cortical neuronal cells. In our study, of the 17 protein/mRNA pairs analyzed, 11 (65%) exhibit a change in mRNA that was consistent with the general trend in protein expression, and 6 exhibit discordance in the two determinations. While the degrees of change by the two methods was somewhat varied, our results suggest that quantification using mass spectrometric methods for basic/acidic proteins could be accomplished with the aid of fractionation. It should be pointed out that the primary goal of this study is to implement mass spectrometrybased quantification methods to PF 2D fractions. Our analysis 2456

Journal of Proteome Research • Vol. 6, No. 7, 2007

might have revealed subsets of differentially expressed acidic and basic proteins in the two closely related cell lines. No systematic attempts however were made to optimize conditions in order to identify the maximal number of proteins that have changed the level of expression in the two cell lines. The significance of protein changes and their relationship with p53 deficiency remain to be elucidated. Of the two mass spectrometric quantification methods used, the label-free/nanospray approach does not involve any chemical labeling, but its throughput is limited by LC running time, typically >45 min per run. The application of this method to PF 2D fractionated samples becomes challenging due to the number of samples generated. In contrast, the 18O-labeling/ MALDI method offers a higher throughput, but is less sensitive. A faster LC in an LC-MS setup with sensitivity comparable to that of current nano LC-MS could be developed if PF 2D fractionation is to be efficiently coupled with LC-based mass spectrometric quantification. The merit as well as simplicity of UV-based quantification offered by PF 2D should also not be discounted. Samples to be compared that are low in complexity, for example, proteins from a narrow pI range, secretion from special tissues/organelles, or immuno-precipitates, and so forth, are more likely to yield differentiable peaks using the UV-based quantification. Since PF 2D is solutionbased and fully automatic, the implementation of an additional dimension (e.g., using a composite RP/SCX/RP column) may allow sample complexity to be reduced to a point where UV quantification becomes practical. For contrasting samples, the current protocol calls for separate sample injection for both the first and second dimensions. It is conceivable that a higher throughput could be achieved by adopting isotope labeling of proteins (e.g., ICAT, SILAC approaches, etc) so that samples

Identification and Quantification of Basic and Acidic Proteins

research articles

Figure 6. A typical differential profiling of fraction no. 3 from both cell lines and the 12 highly abundant proteins identified.

Figure 7. RT-PCR confirmation of mRNA expression of the reported proteins. [Note: + and - denote HCT 116 p53 +/+ and HCT 116 p53 -/-, respectively.]

could be combined and injected in a single run, from protein separation by PF 2D to MS analysis. Such an approach could potentially improve the overall throughput and data quality,

as it eliminates a critical variable caused by retention time drifting when contrasting samples are profiled separately in two dimensions. Journal of Proteome Research • Vol. 6, No. 7, 2007 2457

research articles

Wu et al.

Table 3. Quantitative Protein Ratios between HCT116 and HCT116 p53 -/- Cells (HCT116 p53-/- over HCT116) Using Label-Free, 18O-Labeling, and Real Time RT-PCR

theoretical pl

label-free ratio (no. of matched peptides)

(a) Basic Proteins YBOX1_HUMAN Nuclease sensitive element binding protein 1 9.87 1.96 (5) PTBP1_HUMAN Polypyrimidine tract-binding protein 1 9.22 2.01 (2) ATPA_HUMAN ATP synthase alpha chain 8.28 0.47 (4) THIL_HUMAN Acetyl-CoA acetyltransferase 8.16 1.50 (2) RT22_HUMAN Mitochondrial 28S ribosomal protein S22 7.70 0.50 (2) SERC_HUMAN Phosphoserine aminotransferase 7.56 1.50 (2) FUS_HUMAN RNA-binding protein FUS 9.40 0.36 (6) RS3A_HUMAN 40S ribosomal protein S3a 9.75 2.02 (3) RL31_HUMAN 60S ribosomal protein L31 10.54 1.54 (3) NSDHL_HUMAN Sterol-4-alpha-carboxylate 3-dehydrogenase 8.16 1.61 (2) RL10_HUMAN 60S ribosomal protein L10 10.11 (b) Acidic Proteins IF4B_HUMAN Eukaryotic translation initiation factor 4B 5.49 2.55 (3) GRP78_HUMAN 78 kDa glucose-regulated protein precursor 5.01 2.33 (7) MAP4_HUMAN Microtubule-associated protein 4 5.32 1.79 (5) NSF1C_HUMAN NSFL1 cofactor p47 4.99 1.50 (3) HS90B_HUMAN Heat shock protein HSP 90-beta 4.97 SUMO2_HUMAN Protein SMT3B 5.32 a

18O-labeling

ratio (no. of matched peptides

1.65 (3) 1.89 (4)

ratio reportedb,c

1.90b

0.10c

3.90 (2) 3.02 (3) 1.54c 1.59 (2) 0.30 (2)

1.85,b 2.02b 0.33c

confirmeda

real time RT-PCR ratio

+ + + + + + + + + + +

2.13 2.15 0.60 1.22 1.16 0.47 1.35 1.50 1.69 1.34 0.89

+ + + + + +

3.39 1.46 0.73 1.38 2.90 1.21

Protein identifications confirmed by RT-PCR. b Identified by either iTRAQ or clCAT in reference 14. c Identified by label-free method in reference 16.

PF 2D offers an integrated two-dimensional fractionation at the protein level. When coupled with a label-free MS analysis, using reversed phase chromatography at peptide level, the fractionation is extended to three-dimensional (IEF and RP separations, both at the protein level, followed by RP at the peptide level). The coupling of first dimension fractionation of PF 2D with MudPIT analysis, which is another threedimensional fractionation (IEF at the protein level; SCX and RP at the peptide level), has been reported by Chen et al.10 We had also conducted such a combination of analysis using basic proteins collected from the first-dimensional fractionations of PF 2D. When comparable amounts of proteins were used, our results indicated that the combined method (first dimension of PF 2D followed by MudPIT analysis) identified fewer number of proteins than that identified using the current (2D with PF2D + 1D with RP) method (86 vs 110 proteins, see Supplement Table 5 in Supporting Information). All together, the two approaches identified 138 proteins, of which 42% are common to both methods, 38% are uniquely identified by the approach described herein, and 20% uniquely identified by MudPIT analysis using the IEF fractions from PF 2D. Among the 10 differentially proteins reported in Table 3, four were not identified using MudPIT. In general, the number of peptides identified for each protein (e.g., those in Figure 6) is higher using the current approach (data not shown). These results reiterate that reduction in sample complexity as well as enrichment of target proteins are beneficial for maximal protein identification by MS analysis. It is likely that the observed differences in the numbers of proteins identified might result from differences in sample preparation steps needed prior to mass analysis. The presence of 6 M urea, polybuffer 74, and n-octyl β-D-glucopyranoside in the buffer used for PF 2D first-dimensional fractionation36 calls for additional sample preparation steps before carrying out the MudPIT analysis. TCA precipitation used to pellet proteins from chromatofocusing buffer, as employed in the aforementioned MudPIT study10 or in generic MudPIT methods,37,38 may result in loss of proteins due to incomplete 2458

Journal of Proteome Research • Vol. 6, No. 7, 2007

precipitation. The use of urea to solubilize TCA-precipitated proteins before tryptic digestion also necessitates desalting prior to strong cation exchange chromatography.37,38 This could potentially lead to less-than-complete recovery of peptides. In contrast, protein fractions from the first dimension of PF 2D can directly be injected into the second-dimension (RP) column, yielding essentially salt-free samples in a fully automated fashion. The coupling of IEF/MudPIT approach nevertheless offers a distinct advantage of better resolving peptides by fractionation with chemically distinctive matrixes (SCX and RP). The coupling of PF 2D and MudPIT, though timeconsuming, could potentially provide an even better analysis, as together they offer four-dimensional fractionation of proteins and peptides, that is, IEF and RP at the protein level; and SCX and RP at the peptide level.

Acknowledgment. This research was supported by the Intramural Research Program of the National Heart, Lung, and Blood Institute, National Institutes of Health. Note Added after ASAP Publication. This article was published ASAP on May 17, 2007. Figure 6 was incorrect and the proper changes have been made. The correct version was published on June 18, 2007.

Supporting Information Available: Tables of RT-PCR primer sequences used to examine mRNA expression of the proteins identified, proteins identified in the flow-through (basic) region, proteins identified in the salt wash (acidic) region, entries in Swiss-Prot human database, and the results between PF 2D/label-free approach and PF 2D first-dimension MudPIT approach. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Lilley, K. S.; Razzaq, A.; Dupree, P. Curr. Opin. Chem. Biol. 2002, 6, 46-50. (2) Washburn, M. P.; Wolters, D.; Yates, J. R., III. Nat. Biotechnol. 2001, 19, 242-47.

research articles

Identification and Quantification of Basic and Acidic Proteins (3) Wang, Y.; Wu, R.; Cho, K. R.; Shedden, K. A.; Barder, T. J.; Lubman, D. M. Mol. Cell. Proteomics 2006, 5, 43-52. (4) Soldi, M.; Sarto, C.; Valsecchi, C.; Magni, F.; Proserpio, V.; Ticozzi, D.; Mocarelli, P. Proteomics 2005, 5, 2641-47. (5) Gorg, A.; Weiss, W.; Dunn, M. J. Proteomics 2004, 4, 3665-85. (6) Yan, F.; Subramanian, B.; Nakeff, A.; Barder, T. J.; Parus, S. J.; Lubman, D. M. Anal. Chem. 2003, 75, 2299-308. (7) Zheng, S.; Schneider, K. A.; Barder, T. J.; Lubman, D. M. BioTechniques 2003, 35, 1202-12. (8) Hamler, R. L.; Zhu, K.; Buchanan, N. S.; Kreunin, P.; Kachman, M. T.; Miller, F. R.; Lubman, D. M. Proteomics 2004, 4, 562-77. (9) Wildgruber, R.; Harder, A.; Obermaier, C.; Boguth, G.; Weiss, W.; Fey, S. J.; Larsen, P. M.; Gorg, A. Electrophoresis 2000, 21, 261016. (10) Chen, E. I.; Hewel, J.; Felding-Habermann, B.; Yates, J. R., III. Mol. Cell. Proteomics 2006, 5, 53-56. (11) Sheng, S.; Chen, D.; Van Eyk, J. E. Mol. Cell. Proteomics 2006, 5, 26-34. (12) Shin, Y. K.; Lee, H. J.; Lee, J. S.; Paik, Y. K. Proteomics 2006, 6, 1143-50. (13) Billecke, C.; Malik, I.; Movsisyan, A.; Sulghani, S.; Sharif, A.; Mikkelsen, T.; Farrell, N. P.; Bogler, O. Mol. Cell. Proteomics 2006, 5, 35-42. (14) Wu, W. W.; Wang, G.; Baek, S. J.; Shen, R. F. J. Proteome Res. 2006, 5, 651-58. (15) Ong, S. E.; Pandey, A. Biomol. Eng. 2001, 18, 195-205. (16) Wang, G.; Wu, W. W.; Zeng, W.; Chou, C. L.; Shen, R. F. J. Proteome Res. 2006, 5, 1214-23. (17) Heller, M.; Mattou, H.; Menzel, C.; Yao, X. J. Am. Soc. Mass Spectrom. 2003, 14, 704-18. (18) Yu, M. J.; Pisitkun, T.; Wang, G.; Shen, R. F.; Knepper, M. A. Mol. Cell. Proteomics 2006, 5, 2131-45. (19) Fenton, R. A.; Chou, C. L.; Ageloff, S.; Brandt, W.; Stokes, J. B.; Knepper, M. A. Am. J. Physiol., Renal Physiol. 2003, 285, F143F151. (20) Livak, K. J.; Schmittgen, T. D. Methods 2001, 25, 402-8. (21) Liu, H.; Sadygov, R. G.; Yates, J. R., III. Anal. Chem. 2004, 76, 4193-201. (22) Lubman, D. M.; Kachman, M. T.; Wang, H.; Gong, S.; Yan, F.; Hamler, R. L.; O’Neil, K. A.; Zhu, K.; Buchanan, N. S.; Barder, T.

(23) (24) (25) (26) (27) (28) (29) (30) (31) (32)

(33) (34)

(35)

(36) (37) (38)

J. J. Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2002, 782, 183-96. Kachman, M. T.; Wang, H.; Schwartz, D. R.; Cho, K. R.; Lubman, D. M. Anal. Chem. 2002, 74, 1779-91. Dauly, C.; Perlman, D. H.; Costello, C. E.; McComb, M. E. J. Proteome. Res. 2006, 5, 1688-700. Smith, R. D.; Tang, K.; Shen, Y. Mol. Biosyst. 2006, 2, 221-30. Li, X. J.; Zhang, H.; Ranish, J. A.; Aebersold, R. Anal. Chem. 2003, 75, 6648-57. Bodnar, W. M.; Blackburn, R. K.; Krise, J. M.; Moseley, M. A. J. Am. Soc. Mass Spectrom. 2003, 14, 971-79. Zhang, N.; Li, N.; Li, L. J. Proteome Res. 2004, 3, 719-27. Gu, S.; Liu, Z.; Pan, S.; Jiang, Z.; Lu, H.; Amit, O.; Bradbury, E. M.; Hu, C. A.; Chen, X. Mol. Cell. Proteomics 2004, 3, 998-1008. Greenbaum, D.; Colangelo, C.; Williams, K.; Gerstein, M. Genome Biology 2003, 4, 117. Gygi, S. P.; Rochon, Y.; Franza, B. R.; Aebersold, R. Mol. Cell Biol. 1999, 19, 1720-30. Washburn, M. P.; Koller, A.; Oshiro, G.; Ulaszek, R. R.; Plouffe, D.; Deciu, C.; Winzeler, E.; Yates, J. R., III. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 3107-12. Futcher, B.; Latter, G. I.; Monardo, P.; McLaughlin, C. S.; Garrels, J. I. Mol. Cell Biol. 1999, 19, 7357-68. Tian, Q.; Stepaniants, S. B.; Mao, M.; Weng, L.; Feetham, M. C.; Doyle, M. J.; Yi, E. C.; Dai, H.; Thorsson, V.; Eng, J.; Goodlett, D.; Berger, J. P.; Gunter, B.; Linseley, P. S.; Stoughton, R. B.; Aebersold, R.; Collins, S. J.; Hanlon, W. A.; Hood, L. E. Mol. Cell. Proteomics 2004, 3, 960-69. Johnson, M. D.; Yu, L. R.; Conrads, T. P.; Kinoshita, Y.; Uo, T.; Matthews, J. D.; Lee, S. W.; Smith, R. D.; Veenstra, T. D.; Morrison, R. S. J. Biol. Chem. 2004, 279, 26685-97. Levreri, I.; Musante, L.; Petretto, A.; Bruschi, M.; Candiano, G.; Melioli, G. Clin. Chem. Lab Med. 2005, 43, 1327-33. Delahunty, C.; Yates, J. R., III. Methods 2005, 35, 248-55. Florens, L.; Washburn, M. P. Methods Mol. Biol. 2006, 328, 159-75.

PR060621C

Journal of Proteome Research • Vol. 6, No. 7, 2007 2459