Quantitative Analysis of Bacterial and Mammalian Proteomes Using a

tides) from Deinococcus radiodurans as well as from mouse B16 melanoma cells. D. radiodurans were cul- tured in both natural isotopic abundance and 15...
5 downloads 0 Views 131KB Size
Download PDF
Anal. Chem. 2001, 73, 2132-2139

Quantitative Analysis of Bacterial and Mammalian Proteomes Using a Combination of Cysteine Affinity Tags and 15N-Metabolic Labeling Thomas P. Conrads,† Kim Alving,† Timothy D. Veenstra,† Mikhail E. Belov,† Gordon A. Anderson,† David J. Anderson,† Mary S. Lipton,† Lijliana Pasˇa-Tolic´,† Harold R. Udseth,† William B. Chrisler,‡ Brian D. Thrall,‡ and Richard D. Smith*,†

Environmental and Molecular Sciences Laboratory and Molecular Biosciences Department, Pacific Northwest National Laboratory, P.O. Box 999, MSIN: K8-98, Richland, Washington 99352

We describe the combined use of 15N-metabolic labeling and a cysteine-reactive biotin affinity tag to isolate and quantitate cysteine-containing polypeptides (Cys-polypeptides) from Deinococcus radiodurans as well as from mouse B16 melanoma cells. D. radiodurans were cultured in both natural isotopic abundance and 15Nenriched media. Equal numbers of cells from both cultures were combined and the soluble proteins extracted. This mixture of isotopically distinct proteins was derivatized using a commercially available cysteine-reactive reagent that contains a biotin group. Following trypsin digestion, the resulting modified peptides were isolated using immobilized avidin. The mixture was analyzed by capillary reversed-phase liquid chromatography (LC) online with ion trap mass spectrometry (MS) as well as Fourier transform ion cyclotron resonance (FTICR) MS. The resulting spectra contain numerous pairs of Cyspolypeptides whose mass difference corresponds to the number of nitrogen atoms present in each of the peptides. Designation of Cys-polypeptide pairs is also facilitated by the distinctive isotopic distribution of the 15N-labeled peptides versus their 14N-labeled counterparts. Studies with mouse B16 cells maintained in culture allowed the observation of hundreds of isotopically distinct pairs of peptides by LC-FTICR analysis. The ratios of the areas of the pairs of isotopically distinct peptides showed the expected 1:1 labeling of the 14N and 15N versions of each peptide. An additional benefit from the present strategy is that the 15N-labeled peptides do not display significant isotope-dependent chromatographic shifts from their 14Nlabeled counterparts, therefore improving the precision for quantitating peptide abundances. The methodology presented offers an alternate, cost-effective strategy for conducting global, quantitative proteomic measurements. Proteomics, the analysis of the entire complement of proteins expressed by a cell, tissue, or organism, is viewed as a necessary complement to genomics. While genomics provides the blueprint describing the proteins an organism may express, a primary goal † ‡

Environmental and Molecular Sciences Laboratory. Molecular Biosciences Department.

2132 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

of proteomics is to monitor the changes in protein abundances in response to either temporal or environmental changes. It is also desirable to precisely measure changes in protein abundances in a high-throughput manner so that the effects of many “perturbations” upon, or changes to, a cell type, tissue, or organ can be studied in a reasonable time period. The conventional method of measuring changes in protein expression levels utilizes two-dimensional polyacrylamide gel electrophoresis1 (2-D PAGE) and generally involves a comparison of protein spot intensities.2-4 This strategy suffers from limitations in protein coverage, sensitivity, dynamic range, and precision of the measurements.5,6 It is also difficult to compare 2-D PAGE results from different laboratories despite increased reproducibility through standardization and automation.7,8 Even with more than a decade of significant effort, this technique has not been shown to be amenable to high-throughput automation, and therefore, it is unlikely to meet the throughput demands of future proteomic studies in which the results of many perturbations will need to be compared. Efforts in our laboratory aim to establish high-throughput proteomic methods that allow for multiple perturbations of a cell system to be analyzed in a short period of time. The general strategy taken employs a high-resolution separation technique such as capillary isoelectric focusing (CIEF) for intact proteins, or capillary reversed-phase liquid chromatography (LC) for peptides, coupled on-line with Fourier transform ion cyclotron resonance mass spectrometry (FTICR).9,10 To measure differences (1) O’Farrell, P. H. J. Biol. Chem. 1975, 250, 4007-4021. (2) Newsholme, S. J.; Maleeff, B. F.; Steiner, S.; Anderson, N. L.; Schwartz, L. W. Electrophoresis 2000, 21, 2122-2128. (3) Yan, J. X.; Sanchez, J. C.; Tonella, L.; Williams, K. L.; Hochstrasser, D. F. Electrophoresis 1999, 20, 738-742. (4) Patton, W. F. J. Chromatogr. B: Biomed. Sci. Appl. 1999, 722, 203-223. (5) Gygi, S. P.; Corthals, G. L.; Zhang, Y.; Rochon, Y.; Aebersold, R. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9390-9395. (6) Hoving, S.; Voshol, H.; van Oostrum, J. Electrophoresis 2000, 21, 26172621. (7) Celis, J. E.; Gromov, P. Curr. Opin. Biotechnol. 1999, 10, 16-21. (8) Patterson, S. D. Curr. Opin. Biotechnol. 2000, 11, 413-418. (9) Severs, J. C.; Hofstadler, S. A.; Zhao, Z.; Senh, R. T.; Smith, R. D. Electrophoresis 1996, 17, 1808-1817. (10) Shen, Y.; Zhao, R.; Belov, M. E.; Conrads, T. P.; Anderson, G. A.; Tang, K.; Pasˇa-Tolic´, L.; Veenstra, T. D.; Lipton, M. S.; Smith, R. D. Submitted to Anal. Chem. 10.1021/ac001487x CCC: $20.00

© 2001 American Chemical Society Published on Web 03/13/2001

in relative protein abundances, two isotopically distinct versions of a proteome are combined and analyzed. Our initial demonstration of isotopic labeling strategies for whole proteomes involved the analysis of intact proteins using CIEF-FTICR11 to examine the cadmium (Cd2+) stress response in Escherichia coli.12 In these studies, E. coli was grown in both normal (i.e., natural isotopic abundance) and rare-isotope (13C, 15N)-depleted media. To measure changes in the relative expression of proteins, aliquots were removed from the unstressed (normal medium) and stressed (depleted medium) cultures at different time intervals after Cd2+ addition, and the cells were mixed prior to sample processing and analyzed by CIEF-FTICR. Combining the cells at this early time point eliminates the experimental variables associated with cell lysis, separation, and MS analysis. Other demonstrations of stable isotope labeling have utilized cells that were cultured in 15Nenriched medium and combined with cells cultured in normal medium.13,14 The proteins extracted from these combined populations are typically proteolytically digested and analyzed at the peptide level by capillary LC-MS or LC-MS/MS. Another approach for high-throughput proteome quantitation, recently reported by Aebersold and co-workers, introduces the isotopic label after the proteins are harvested from the proteomes being studied.15 This approach, which employs isotope-coded affinity tags (ICAT), involves the affinity selection of cysteinecontaining polypeptides (Cys-polypeptides) after proteins are modified with a Cys-specific reagent that contains a biotin group along with a linker arm that connects the thiol-reactive group to the biotin moiety. The reagent is used in both a normal and a “heavy” isotopic version, where the heavy version of the ICAT reagent differs by eight hydrogen atoms in the linker arm that have been substituted by eight deuterium atoms. Derivatization of two distinct proteomes with the light and heavy versions of the ICAT reagent provides the basis for proteome quantitation while providing an additional Cys-constraint that can be used to increase the confidence of peptide identification. In this work, we have combined 15N-metabolic labeling with a commercially available Cys-affinity tag to derivatize and isolate Cys-polypeptides that allows for the quantitation of relative peptide abundances from two separate proteomes. In the following, we present our initial results using the highly radioresistant bacterium Deinococcus radiodurans. After equal numbers of cells grown in either normal or 15N-enriched media were combined, the extracted proteins were derivatized with (+)-biotinyliodoacetamidyl-3,6dioxaoctanediamine (iodoacetyl-PEO-biotin), digested with trypsin, and the Cys-polypeptides isolated using immobilized avidin chromatography. The resulting Cys-polypeptide mixture was analyzed by capillary LC-FTICR. The same technique was used to demonstrate the ability to conduct quantitative measurements of a mammalian proteome (using mouse B16 melanoma cells) maintained in culture. The results show this to be an effective method (11) Pasˇa-Tolic´, L.; Jensen, P. K.; Anderson, G. A.; Lipton, M. S.; Peden, K. K.; Martinovic, S.; Tolic, N.; Bruce, J. E.; Smith, R. D. J. Am. Chem. Soc. 1999, 121, 7949-7950. (12) Ferianc, P.; Farewell, A.; Nystrom, T. Microbiology 1998, 144, 1045-1050. (13) Gao, H.; Shen, Y.; Veenstra, T. D.; Harkewicz, R.; Anderson, G. A.; Bruce, J. E.; Pasˇa-Tolic´, L.; Smith, R. D. J. Microcolumn Sep. 2000, 12, 383-390. (14) Oda, Y.; Huang, K.; Cross, F. R.; Cowburn, D.; Chait, B. T. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6591-6596. (15) Gygi, S. P.; Rist, B.; Gerber, S. A.; Turecek, F.; Gelb, M. H.; Aebersold, R. Nature Biotechnol. 1999, 17, 994-999.

to determine relative Cys-polypeptide abundances using readily available reagents. In addition to a Cys-constraint, the combined use of a Cys-affinity tag and 15N-metabolic labeling in conjunction with high-resolution FTICR also provides an identification constraint based on the number of nitrogens present in the peptide. EXPERIMENTAL SECTION Growth of D. radiodurans. D. radiodurans R1 were cultured in Celtone-U and Celtone-N (>98% 15N) (Martek Biosciences, Columbia, MD). Both cultures were grown at 30 °C with shaking at 225 rpm and were combined at an optical density (OD600 nm) of 1.0. The cells were harvested by centrifugation at 10000g for 1 min. The cells were resuspended in 500 µL of phosphate-buffered saline (PBS) (pH 7.2) and lysed by bead beating in the presence of 0.1-mm acid zirconium beads for three cycles of 60 s at 5000 rpm. Between each cycle of bead beating, the samples were incubated on ice for 5 min. The cell lysate was recovered and centrifuged at 10000g for 10 min to remove cell debris. Growth of B16 Mouse Cells. Murine B16 melanoma cells (ATCC 30-2002) were routinely maintained in Dulbecco’s modified Eagles media supplemented with 10% fetal bovine serum, 2 mM L-glutamine, and 100 units/mL penicillin and streptomycin in a humidified atmosphere of 5% CO2 in air at 37 °C. For proteome analysis, the cells were plated on six-well culture plates at a concentration of 104 cells/well and following attachment the medium was changed to Celtone M Natural or Celtone M-N (>98% 15N) (Martek Biosciences Corp. Catalog No. 53998), both of which were supplemented with 10% heat-inactivated fetal bovine serum and 100 units/mL penicillin and streptomycin. The cells were fed with fresh medium daily and allowed to grow to an estimated 85% confluence (4 days). Daily analysis of the number of viable cells by trypan blue exclusion and hemacytometer counts demonstrated that there were no significant differences in proliferation rates or viability between the 15N-labeled and natural mediums (data not shown). For protein extraction, the cells were washed twice with PBS and trypsinized and equal numbers of cells lysed by resuspending in buffer containing 50 mM Tris-HCl (pH 7.4), 100 mM sodium chloride, 50 mM sodium fluoride, and 1% Triton X-100. Labeling and Affinity Isolation of Cys-polypeptides. To evaluate the efficiency of the labeling and extraction of Cyspolypeptides from the normal and 15N-labeled cell cultures, the soluble protein extracts obtained from the D. radiodurans and B16 cells were split into two separate, but identical, aliquots. The procedure to label and affinity isolate the Cys-polypeptides using iodoacetyl PEO-biotin (Pierce, Rockford, IL) is similar to that used for the ICAT method.15 Briefly, the soluble proteins extracted from D. radiodurans, or mouse B16 cells, were desalted into 50 mM Tris-HCl, 5 mM EDTA, pH 8.4, and the protein concentration was measured using the Biuret assay. Guanidine hydrochloride was added to a final concentration of 6 M and boiled. The samples were reduced by addition of tributylphosphine to 5 mM followed by a 1-h incubation at 37 °C. Iodoacetyl-PEO-biotin was added to the samples to a 5-fold excess of the calculated approximate number of cysteine residues (assuming 1 mg of protein equals 30 nM and ∼6 Cys residues/protein). The reaction mixture was incubated with stirring for 90 min in the dark. The PEO-biotinderivatized samples were desalted into 100 mM NH4HCO3, 5 mM EDTA, pH8.4, and digested with trypsin (1:50 enzyme-to-protein Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2133

ratio) overnight at 37 °C. After the digestion was complete, phenylmethanesulfonyl fluoride was added to a final concentration of 1 mM. A packed avidin column (2-mL bed volume) was prepared and equilibrated with 2× PBS (pH 7.2). The column was blocked with 2 mM biotin in PBS (pH 7.2), reversible binding sites were stripped by washing the column with 0.1 M glycine (pH 2.8), and the column was reequilibrated with 2× PBS (pH 7.2). After the PEO-biotin-labeled peptide mixtures were boiled for 5 min, they were loaded onto the avidin column. The avidin column was incubated at room temperature for 30 min. After the column was washed with 5 bed volumes of 2× PBS (pH 7.2) followed by 5 bed volumes of 50 mM NH4HCO3 (pH 8.4), the bound Cys-polypeptides were eluted using 30% acetonitrile, 0.4% trifluoroacetic acid. Analysis of Cys-polypeptides. The Cys-polypeptide mixture was analyzed by capillary LC on-line with both an LCQ (ion trap) (Finnigan-MAT, San Jose, CA) and an FTICR mass spectrometer. The sample was loaded onto a 60-cm capillary column (150 µm i.d. × 360 µm o.d., Polymicro Technologies, Phoenix, AZ) packed with C18 5-µm-diameter particles (PoroS 20R2, Perspective Biosystems, Framingham, MA). A solvent gradient was used to elute the peptides using 0.4% acetic acid in water (solvent A) and 0.4% acetic acid in 80% acetonitrile (solvent B). The peptides were eluted using a linear gradient of 0-100% solvent B over 80 min. Solvents were delivered to the capillary column using two Shimadzu LC-10AD pumps controlled by a Shimadzu 10A system controller (Shimadzu Scientific Instruments, Inc., Columbia, MD) using an LC Packings Accurate microflow processor splitter (LC Packings, San Francisco, CA) resulting in a capillary column flow rate of ∼1 µL/min. To obtain amino acid sequence information, the LCQ-MS was operated in the data-dependent MS/MS mode, (a full MS scan followed by three MS/MS scans), where the three most intense ions are selected dynamically from the most recent MS scan and dissociated in the subsequent MS/MS scans. For FTICR analysis, the LC capillary was coupled on-line to either a 3.5-16,17 or 11.5-T FTICR mass spectrometer, both designed and constructed in our laboratory. For both instruments, the output of the LC was coupled to an atmospheric ESI source comprising a stainless steel heated capillary/electrodynamic ion funnel assembly followed by a collisional focusing quadrupole.18 As for the instrument incorporated with the 11.5-T superconducting magnet, ions transmitted through a 45-cm collisional quadrupole were externally trapped in accumulation quadrupoles Q1A and Q1B (47 and 10.2 cm long, respectively), situated between the two conductance limits C1 and C2, resulting in collisional relaxation of their translational energy. Trapping in a linear ion trap, i.e., dynamic application of dc potentials at the entrance and exit conductance limits of a quadru-/hexa-/octopole to trap and subsequently eject ions in z-direction, has previously been described.19,20-26 A 0.3-V trapping well was applied on the quad(16) Belov, M. E.; Gorshkov, M. V.; Anderson, G. A.; Udseth, H. R.; Smith R. D. Anal. Chem. 2000, 72, 2271-2276. (17) Belov, M. E.; Nikolaev, E. N.; Anderson, G. A.; Auberry, K. J.; Harkewicz, R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2001, 12, 38-48. (18) Belov M. E.; Gorshkov M. V.; Udseth H. R.; Anderson G. A.; Tolmachev A. V.; Prior D. C.; Harkewicz R.; Smith R. D. J. Am. Soc. Mass Spectrom. 2000, 11, 19-23. (19) Douglas, D. J. U.S. Patent 5179278, 2000. (20) Senko, M. W.; Hendrickson, C. L.; Emmett, M. R.; Shi, S. D-H.; Marshall, A. G.; J. Am. Soc. Mass Spectrom. 1997, 8, 970-976.

2134

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

rupole accumulation trap to minimize space charge-induced discrimination effects.23-25 The pressure in the quadrupole accumulation region (Q1B) is 5.5 × 10-5 Torr. The conductance limits C1 and C2 both have 3-mm orifices. The quadrupole inscribed radius is 4.13 mm. The dc potentials applied for accumulation in the Q1B quadrupole were as follows: C1, 2.5 V, Q1A, 2.2 V, Q1B, 1.6 V, and C2, 4 V. C2 was changed to -5 V for ejection of ions from the quadrupole trap. The transfer time from Q1B through the ion guiding quadrupoles Q2 and Q3 to the ICR cell was 1.9 ms. The quadrupole rf potentials were, for Q0, Q1B, Q2, and Q3, 1 MHz 400 Vpp, and for Q1A, 543 kHz 400 Vpp. Following the quadrupole accumulation, the trapped ions were ejected from the accumulation quadrupoles into a quadrupole ion guide operating in a pressure region of ∼10-8 Torr and then trapped in an FTICR cell using dynamic trapping.20 A total of 256K data sets in the m/z range of 366-3000 Da were acquired following each accumulation event. For on-line LC-FTICR analysis using the 3.5-T FTICR, the ESI source is followed by an interface for external selective ion accumulation, an electrostatic ion guide, and a cylindrical dualcell combination. The external accumulation interface incorporates three quadrupoles, referred to as “ion guiding,” “selection”, and “accumulation” quadrupoles, respectively.18,26 The accumulation quadrupole was segmented to provide an axial electric field for prompt ion ejection. In this work, broadband ion accumulation (i.e., no ion preselection in the selection quadrupole prior to external ion accumulation) was employed. To maximize the duty cycle for the FTICR experiment, the accumulation quadrupole exit plate was maintained at high trapping potential to continuously accumulate ions from the ESI source, except for a short period of time (∼400 µs) required to eject ions to the FTICR cell. Given the duration of one scan of ∼5 s, the effective duty cycle for ion accumulation throughout the separation was close to 100%. RESULTS AND DISCUSSION The ability to rapidly make sensitive and quantitative measurements of large numbers of proteins would be of broad utility in biological research. Our laboratory is presently developing highthroughput quantitative methods for proteomics based upon the use of stable isotope-labeling strategies for quantitation and rapid analysis using high-resolution separations (i.e., CIEF or capillary LC) coupled with very high performance mass spectrometry. Significant recent progress has been made in the area of protein labeling to allow measurements of relative protein abundances between different proteomes. The ICAT strategy, developed by Aebersold and co-workers, shows considerable promise since it incorporates a stable isotope label after extracting the (21) Welling, M.; Scheussler, H. A.; Thompson, R. I.; Walther, H. Int. J. Mass Spectrom. Ion Processes 1998, 172, 95-114. (22) Campbell, J. M.; Collings, B. A.; Douglas, D. J. Rapid Commun. Mass Spectrom. 1998, 12, 1463-74. (23) Belov M. E.; Nikolaev, B. N.; Harkewicz, R.; Masselon, C. D.; Alving, K.; Smith, R. D. Submitted to Int. J. Mass Spectrom. (24) Tolmachev, A. V.; Udseth, H. R.; Smith, R. D. Rapid Commun. Mass Spectrom. 2000, 14, 1907-1913. (25) Alving, K.; Belov, M. E.; Pasˇa-Tolic´, L.; Conrads T. P.; Veenstra T. D.; Anderson, G. A.; Smith, R. D., manuscript in preparation. (26) Belov, M. E.; Nikolaev, E. N.; Anderson, G. A.; Udseth, H. R.; Conrads, T. P.; Veenstra, T. D.; Masselon, C. D.; Gorshkov, M. V.; Smith, R. D. Anal. Chem. 2001, 73, 253-261.

Figure 1. Capillary LC-MS analysis of cysteine-containing polypeptides (Cys-polypeptides) isolated from a combined culture of D. radiodurans grown in normal and 15N-enriched media. After combining the two cultures of cells and extracting the soluble proteins, the Cyspolypeptides were derivatized and affinity isolated using PEO-biotin and immobilized avidin. The example full-scan MS spectrum clearly shows five pairs of Cys-polypeptides with each pair representing two peptides of identical sequence whose m/z ratio differs based on the number of nitrogen atoms in the peptides.

protein sample, enabling proteomes from any conceivable source to be quantitatively compared.15 One limitation of the ICAT approach is the very scarce availability of the necessary reagents. A second limitation arises from the significant chromatographic shift that results from an apparent isotopic effect arising from the isotopic inequivalence between the light and heavy ICAT reagent. In the following, we present an alternative approach combining metabolic labeling of proteins with a readily commercially available cysteine affinity tag that provides an effective means to quantitate Cys-polypeptides in a manner analogous to the ICAT strategy. For the initial demonstration of this labeling strategy, a sample of D. radiodurans was prepared that contained equal numbers of cells cultured in normal isotopic abundance and 15N-enriched media and labeled with iodoacetyl-PEO-biotin. Iodoacetyl-PEObiotin provides the features of the light ICAT reagent: a Cysspecific reactive group and a biotin affinity tag to isolate the derivatized peptides using immobilized avidin.15 Typical mass spectra obtained using an ion trap mass spectrometer operated in the MS mode is shown in Figure 1. At least five pairs of peptide ions are evident in this spectrum where the difference in the spacing between the pairs depends on the number of nitrogen atoms in the peptides. As expected, the spacing between the pairs increases in proportion to peptide mass. Calculations based on the number of nitrogen atoms in the individual amino acids show that a given random 10-residue peptide will have an average of 13.7 nitrogen atoms. The fact that the mass shift for the heavy version of the PEO-labeled peptide depends on the number of nitrogen atoms can be both a disadvantage or an advantage, depending on the quality of the mass spectrometric data. Clearly, once the peptide pairs are correlated, the resulting knowledge regarding the nitrogen content of the peptide can assist in identification. Since the exact mass shift cannot be predicted for an unknown peptide, assignment can become problematic using

Figure 2. Identification of Cys-polypeptide pairs observed from a combined culture of D. radiodurans grown in normal and 15N-enriched media. The MS/MS spectrum of the normal isotopic abundance peptide (A) is very similar to the 15N-enriched version of the same peptide (B). SEQUEST searching of the MS/MS spectra using a D. radiodurans FASTA database identified the peptide sequence as HETGGLVFFEPILDACR from arginase (DR0651).

conventional mass spectrometers if the spectra are highly complex (i.e., several peaks appear in the m/z range of interest). However, the much higher resolution and mass measurement accuracy provided by FTICR allows the mass differences between the peaks to be accurately determined and the peptide pairs assigned with high confidence. Indeed, efforts in our laboratory, to be described elsewhere, have resulted in software for the automated assignment of peptide pairs, as a component of software for proteome-wide quantitative measurements. If only low-resolution or low-mass measurement accuracy (e.g., ion trap) data is available, identification of peptide pairs is still possible based upon the use of tandem MS, but at the cost of considerably lower throughput than possible with FTICR. The identification of peptide pairs using conventional MS/MS technology exploits the similar fragmentation patterns of the pairs of peptides. To illustrate this approach, the same sample was analyzed in a data-dependent MS/MS mode to investigate the fragmentation of the 14N- and 15N-labeled Cys-polypeptides. The MS/MS spectra of a single pair of peptides is shown in Figure 2. Both spectra show that the fragmentation patterns for the peptides are quite similar, with the peaks arising from the 15N-labeled peptide offset from its normal isotopic abundance counterpart by the number of nitrogens present in the remaining peptide fragment. The peptide sequence was identified as HETGGLVFFEPILDACR from arginase (DR0651) by searching a D. Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2135

Figure 3. Examples of Cys-polypeptides observed in the LC-FTICR analysis of peptides isolated from a combined culture of D. radiodurans grown in normal and 15N-enriched media. Pairs of related peptides can be detected based on the distinctive isotopic distributions of the 15N-labeled peptide and its normal isotopic abundance partner. (A) Total ion chromatogram representing the total FTICR signal in the various scans during the capillary separation of the Cys-polypeptides.

radiodurans FASTA protein database (downloaded from http:// www.genome.ad.jp/kegg/) with the program SEQUEST.

The affinity-isolated Cys-polypeptide sample from D. radiodurans was also analyzed by capillary LC-MS coupled on-line with the 3.5 T-FTICR. The high resolution of FTICR simplifies the assignment of peptides pairs due to the distinctive isotopic distribution of the 15N-labeled peptides. Portions of several typical spectra observed in the LC-FTICR analysis are shown in Figure 3. More than 600 pairs of Cys-polypeptides were observed from a total of 4906 Cys-peptides predicted from the genome sequence between 500 and 5000 Da. The average ratio of the areas under the differentially isotopically labeled versions of each peptide was 1.12, corresponding to the expected ratio as the sample contained approximately equal numbers of cells from the cultures grown in either normal isotopic abundance or 15N-labeled media. That the ratio of 14N-labeled peaks to 15N-labeled peaks is not 1.00 suggests a ∼12% excess presence of 14N-labeled cells in the initial sample preparation rather than any intrisic differential reactivity of 15Nlabeled peptides (a secondary isotope effect would not a priori be expected to lead to such a large difference in chemical reactivity at the cysteinyl residues). The peptide pairs observed in the LC-FTICR analysis of D. radiodurans are displayed in a 2-D format in Figure 4. This 2-D display presents all of the peptides observed in the capillary LCFTICR analysis of the Cys-polypeptides isolated from the mixed D. radiodurans sample. In contrast to 2-D PAGE, this 2-D display shows the peptide “spots” based on their molecular mass and LC elution order instead of molecular mass and isoelectric point. As seen in Figure 4, hundreds of pairs of peptide masses were observed in this single analysis. The strategy of 15N-metabolic labeling combined with isolation of the Cys-polypeptides was extended to mouse B16 melanoma cells. Again two separate cultures of cells were grown in normal isotopic abundance and 15N-enriched media and equal numbers of cells from both cultures combined. After the proteins were extracted from this combined sample, the Cys residues were

Figure 4. Two-dimensional display of Cys-polypeptides observed in the LC-FTICR analysis of a peptides isolated from a combined culture of D. radiodurans grown in normal and 15N-enriched media. Hundreds of pairs of isotopically distinct Cys-polypeptides were observed in this single analysis. The inset shows two Cys-polypeptide pairs. 2136 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

Figure 5. Examples of Cys-polypeptides observed in the LC-FTICR analysis of peptides isolated from a combined culture of mouse B16 cells grown in normal and 15N-enriched media. Pairs of related peptides can be detected based on the distinctive isotopic distributions of the 15N-labeled peptide and its normal isotopic abundance partner. (A) Total ion chromatogram representing the total ion current detected by the FTICR in the various scans during the capillary separation of the Cys-polypeptides.

modified with PEO-biotin and affinity isolated with immobilized avidin. The sample was then analyzed by LC-FTICR using the 11.5-T FTICR instrument developed in our laboratory. As with D. radiodurans, pairs of differentially labeled Cys-polypeptides were observed whose m/z ratio differed based on the number of nitrogen atoms in the peptide. Portions of several typical spectra are shown in Figure 5. A key observation was that the labeling of the proteins grown in the 15N-enriched medium was effectively complete based on the observed isotopic distributions. Surprisingly, the addition of 10% FBS to the cell culture medium did not significantly affect the metabolic labeling of these cells. This represents, to the best of our knowledge, the first proteome analysis demonstrating the use of 15N isotopic labeling of mammalian cells in culture to provide a “reference proteome” for quantitative measurements of relative peptide abundances. While the use of the PEO-biotin affinity tag to isolate only Cyspolypeptides from the mouse sample does significantly reduce the complexity of the mixture, this proteome-wide strategy still results in samples with a formidable number of peptides. The use of FTICR, combined with an effective on-line separation, provides a means for addressing this complexity. As shown in Figure 6 for a single scan from the LC-FTICR analysis of the mouse B16 sample, many low-intensity pairs of Cys-polypeptides can be readily detected in spectra containing very intense peaks. As with the case of intense peaks, the distinct isotopic distribution of the 15Nlabeled peaks provides an additional marker that assists verification of assignments for pairs of low-intensity Cys-polypeptides.

Figure 6. Examples showing the high density of information obtained by combining high-resolution capillary LC separations with the high dynamic range and sensitivity of FTICR. Low-abundance pairs of Cys-polypeptides (A) are readily seen in a spectrum containing more highly abundant species (B and C).

The peptide pairs observed in the LC-FTICR analysis of the mouse B16 cells are displayed in a 2-D format in Figure 7. As seen in Figure 4 for the analysis of the D. radiodurans sample, hundreds of pairs of peptide masses were observed in this single analysis. CONCLUSIONS There have been tremendous recent strides in developing methods to make quantitative measurements of differences in relative protein abundances between distinct proteomes using isotope-labeling strategies. These strategies can be classified as either metabolic labeling in which the proteins are isotopically labeled during translation11,13,14 or postextraction labeling in which an isotopic label is incorporated by a chemical modification of the protein.15 The ICAT strategy, which incorporates the isotopic label after protein extraction, has generated a large amount of interest due to its universal applicability and ability to simplify proteome analysis by isolating only Cys-polypeptides.15 However, the ICAT approach is limited both by the scarcity of the reagent and more seriously so by a significant chromatographic shift arising from the isotopic inequivalence of the ICAT reagents that necessarily complicates quantitative measurements of peptide abundances. The use of metabolic labeling and a Cys-affinity tag provides reliable quantitative results based on the analysis of the pairs observed in the D. radiodurans and mouse B16 melanoma cell samples analyzed above. Our results showing equal abundances Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2137

Figure 7. Two-dimensional display of Cys-polypeptides observed in the LC-FTICR analysis of a peptides isolated from a combined culture of mouse B16 cells grown in normal and 15N-enriched media. Hundreds of pairs of isotopically distinct Cys-polypeptides were observed in this single analysis. The inset shows a subset of Cys-polypeptide illustrating the presence of pairs of Cys-polypeptides whose mass difference is based on the number of nitrogens present in the peptide.

of hundreds of pairs of peptides from combined cell cultures containing equal cell numbers confirm the equivalent reactivity of both versions of the peptides with the Cys-affinity tag. Equally important for peptide identification is the similarity of the MS/ MS spectra for both the natural isotopic abundance and 15N-labeled versions of the peptide which allows redundant identification of each peptide pair. In combination with more sensitive FTICR instrumentation, and the more accurate mass measurements provided, the implementation of effective software promises to greatly enhance the speed and effectiveness of proteomic applications. The success of MS/MS methods has shown that sequence information from a single tryptic peptide provides a basis for confident protein identification. If the molecular mass of a single peptide could be measured with high enough mass measurement accuracy (MMA), such that its mass was unique among all of the possible peptides predicted from a genome, it could be used as an accurate mass tag (AMT) for unambiguous protein identification.27 The generation of such AMTs provides a basis for the tryptic fragments generated from an entire proteome or complex protein mixture, for example, to be analyzed and the proteins identified with much greater speed and sensitivity. One issue with this approach is the degree of MMA necessary that allows for peptide mass measurements to function as AMTs for unambiguous protein identification. With the very high MMA that can be achieved using FTICR, it is possible to use a single peptide AMT for protein identification.27 Conventional MS/MS approaches can then be used to validate AMTs assigned on the basis of accurate masses obtained from FTICR. Initial validation of the AMTs will (27) Conrads, T. P.; Anderson, G. A.; Veenstra, T. D.; Pasˇa-Tolic´, L.; Smith, R. D. Anal. Chem. 2000, 72, 3349-3354.

2138 Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

enable subsequent proteome measurements to be based almost exclusively on the use of AMTs, thereby obviating most MS/MS measurements and substantially increasing throughput. We are currently developing software methodologies to correlate predicted tryptic fragment molecular masses from SEQUEST identifications (that arise from data from conventional ion trap MS/ MS experiments) to validate AMTs assigned on the basis of accurate mass measurements that will allow more confident protein identification on a proteome-wide basis. The strategy presented here combines metabolic labeling with the use of a Cys-affinity tag to provide quantitative proteome measurements similar to those provided by the ICAT approach. This strategy offers many features of the ICAT approach including the presence of a Cys-constraint to aid identification, the reduction of the sample complexity, and the compatibility to use avidin isolation of peptides prior to capillary LC-MS. The ability to combine proteomes at the cellular level, instead of after the derivatization of proteins separately extracted from distinct cell populations, removes several variables related to sample handling and protein modification, affording more precise measurements of differences in relative peptide abundances. The Cys-specific reagent, iodoactyl-PEO-biotin, used to modify and isolate the Cyspolypeptides is inexpensive and commercially available, as are all of the components necessary for metabolic labeling of the proteomes. In addition to the benefit of the Cys-constraint, the present strategy of using 15N-labeling also serves to improve quantitation of peptide abundances as the chromatographic shift of the isotopic pairs of peptides is significantly reduced compared to that observed with ICAT labeled peptide pairs.

ACKNOWLEDGMENT Portions of this work were supported by the Office of Biological and Environmental Research, United States Department of Energy, and the National Cancer Institute through Grant CA86340, and the National Institute of Neurological Disorders and Stroke. Pacific Northwest National Laboratory is operated by the Battelle

Memorial Institute for the United States Department of Energy through Contract DE-ACO6-76RLO 1830. Received for review December 19, 2000. Accepted February 13, 2001. AC001487X

Analytical Chemistry, Vol. 73, No. 9, May 1, 2001

2139