Comparative Study of Three Proteomic Quantitative Methods, DIGE, cICAT, and iTRAQ, Using 2D Gel- or LC-MALDI TOF/TOF Wells W. Wu,† Guanghui Wang,† Seung Joon Baek,‡ and Rong-Fong Shen*,† Proteomics Core Facility, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland 20892, and Department of Pathobiology, College of Veterinary Medicine, The University of Tennessee, Knoxville, Tennessee 37996 Received November 15, 2005
A comparative study on the three quantitative methods frequently used in proteomics, 2D DIGE (difference gel electrophoresis), cICAT (cleavable isotope-coded affinity tags) and iTRAQ (isobaric tags for relative and absolute quantification), was carried out. DIGE and cICAT are familiar techniques used in gel- and LC-based quantitative proteomics, respectively. iTRAQ is a new LC-based technique which is gradually gaining in popularity. A systematic comparison among these quantitative methods has not been reported. In this study, we conducted well-designed comparisons using a six-protein mixture, a reconstituted protein mixture (BSA spiked into human plasma devoid of six abundant proteins), and complex HCT-116 cell lysates as the samples. All three techniques yielded quantitative results with reasonable accuracy when the six-protein or the reconstituted protein mixture was used. In DIGE, accurate quantification was sometimes compromised due to comigration or partial comigration of proteins. The iTRAQ method is more susceptible to errors in precursor ion isolation, which could be manifested with increasing sample complexity. The quantification sensitivity of each method was estimated by the number of peptides detected for each protein. In this regard, the global-tagging iTRAQ technique was more sensitive than the cysteine-specific cICAT method, which in turn was as sensitive as, if not more sensitive than, the DIGE technique. Protein profiling on HCT-116 and HCT-116 p53 -/cell lysates displayed limited overlapping among proteins identified by the three methods, suggesting the complementary nature of these methods. Keywords: protein quantification • DIGE • cICAT • iTRAQ • 2D gel • LC-MALDI TOF/TOF
Introduction One main objective of proteomic research is the systematic identification and quantification of proteins expressed in a biological system. The standard approaches to proteomics have been one-/two-dimensional gel electrophoresis or liquid chromatography (LC) followed by mass spectrometry. For the gelbased methods, proteins are separated by pI and/or molecular weights, detected by staining, and quantified according to staining intensities. Proteins are then cut out of the gel, digested, and identified by mass spectrometry. For the LCbased methods, proteins or peptides are separated by LC columns, and then detected, identified, and quantified by mass spectrometry. The goal of earlier proteomics research was merely the identification and characterization of proteins. The goal has now extended to quantitative and comparative studies, thanks to recent advances in chromatography, mass spectrometry, and bioinformatics. * To whom correspondence should be addressed. Bldg. 10, Rm. 8C213, National Institutes of Health, Bethesda, MD 20892. Tel: (301) 594-1060. Fax: (301) 402-2113. E-mail:
[email protected]. † National Institutes of Health. ‡ The University of Tennessee. 10.1021/pr050405o CCC: $33.50
2006 American Chemical Society
The gel-based method has been the method of choice for decades, and is still widely practiced. A number of gel-based methods are commonly employed in quantitative proteomics. The first and oldest method is based on statistical analysis via one of the powerful software packages that merge and compare a number of replicate sets of gels for control and treated samples. This approach requires several replicate runs to overcome variations in running gels, and therefore is very laborious and prone to experimental errors.1 An alternative to the above protocol is difference gel electrophoresis (DIGE). In the DIGE method, CyDye fluors that are spectrally resolvable (e.g., Cy2, Cy3, and Cy5) and matched for mass and charge are used to covalently modify the - amino group of lysines in proteins via an amide linkage. Consequently, the same protein labeled with any of the fluors will migrate to nearly the same position on a 2D gel. In a typical protocol, the control and treated samples are separately labeled using different dyes (e.g., Cy3 and Cy5, respectively), while a mixture consisting of an equal amount of the control and treated samples is labeled with Cy2. The labeled samples are combined and run in a single 2D gel to allow better spot matching and minimize gel-to-gel variations. Therefore, in principle, a single gel would suffice Journal of Proteome Research 2006, 5, 651-658
651
Published on Web 02/07/2006
research articles for full quantitative analysis. The DIGE method also allows multiplexing for up to three separate protein mixtures on the same 2D gel if the third protein sample is labeled with Cy2. Several papers provide a good review of the DIGE technique.2,3 Other 2D gel methods for quantitative proteomics employ stable isotope tagging of the Cys residues with light- and heavyacrylamide or -vinylpyridine.1 In the past 3 to 4 years, the labor intensive gel-based techniques have been challenged and complemented by LC-based methods, particularly in the area of high-throughput proteomic research. Some of the reasons behind this trend include issues related to reproducibility,4 poor representation of low abundant proteins,5 highly acidic/basic proteins, or proteins with extreme size or hydrophobicity,6 as well as difficulties in automation of the gel-based techniques.3 The LC-based methods offer flexibility of choosing a wide range of stationary and mobile phases to resolve complex biological samples at the protein or peptide level.7,8 In the multidimensional LC approach, proteins are usually digested into peptides prior to separation first by cation exchange, and then with C18 reversed-phase column chromatography. The advantage of such an approach is that the resolved peptides from the C18 column can be directly introduced into a mass spectrometer. Chemical tagging (usually stable isotope labeling) of proteins/peptides also allows relative quantification of protein samples by liquid chromatography/mass spectrometry (LC-MS) analyses. The prototypical stable-isotope labeling for quantitative proteomics was isotope-coded affinity tags (ICAT) technology.9 Essentially, proteins from the two states to be compared are labeled at cysteine residues with light and heavy, respectively, tags carrying a biotin moiety. The labeled proteins are then mixed and digested. After a cation exchange chromatography step to remove excess reagents, the mixed peptides are affinity purified using immobilized avidin. Peaks corresponding to the same peptide are identified as doublets in mass spectra due to the mass difference between light and heavy isotopes. The peak intensities of the peptides correlate directly with the relative abundance of the proteins in the two states. A number of limitations to the prototypical ICAT technique has been noted in the literature, including missed identification of proteins with few or no cysteine residue, lost information for post-translational modifications, differential reversed-phase elution of identical peptides labeled with the hydrogen/ deuterium isotope pairs, and complicated interpretation of tandem mass spectrometry (MS/MS) spectra due to the addition of the biotin group.10,11 Many, but not all of the aforementioned problems have been solved by the new cleavable ICAT (cICAT) reagent that employs 13C isotopes and an acid-cleavable biotin group.12,13 Besides ICAT, other stable isotope coding techniques applied in quantitative proteomics have been reported and reviewed.11 Recently a new quantitative method, isobaric tags for relative and absolute quantitation (iTRAQ), was developed. This technology employs a 4-plex set of amine reactive isobaric tags to derivatize peptides at the N-terminus and the lysine side chains, thereby labeling all peptides in a digest mixture.14 In MS, peptides labeled with any of the isotopic tags are indistinguishable (isobaric). Upon fragmentation in MSMS, signature ions (m/z from 114 to 117) are produced, which provide quantitative information upon integration of the peak areas.14 In addition to the stable isotope labeling approach, several researchers have developed alternative LC-based quantification strategies without any labeling.15,16 Among the different techniques discussed above, DIGE and ICAT are the two most commonly practiced techniques in gel652
Journal of Proteome Research • Vol. 5, No. 3, 2006
Wu et al.
based and LC-based quantitative proteomics, respectively. iTRAQ is a relatively new technique, but is gaining in popularity as an alternative to ICAT. Although several recent publications17,18,19 focused on two-method comparison, a head-to-head comparison of the three methods has not been reported. Such a comparison could provide useful information regarding intrinsic merits and constraints of these methods. In this study, we conducted carefully designed experiments to compare these three quantitative methods, with respect to accuracy and sensitivity toward samples of various complexity.
Materials and Methods Preparation of Protein Samples. A six-protein mixture consisting of bovine serum albumin (22 µg), β-galactosidase (38 µg), R-lactalbumin (10 µg), β-lactoglobulin (24 µg), lysozyme (10 µg), and apotransferrin (25 µg) was from Applied Biosystems. For each technique studied, one tube containing the mixture was reconstituted with the solvent appropriate for the technique. The content was then split into two equal aliquots for subsequent differential labeling. Human plasma was obtained from a healthy volunteer. A polyclonal antibody affinity column (Agilent) was used to remove six major proteins (i.e., albumin, IgG, IgA, antitrypsin, transferrin, haptoglobin) from the plasma, according to the manufacturer’s protocol. The flowthrough portion (the plasma devoid of the 6 major proteins, referred hereafter as depleted plasma) was collected and analyzed with a Finnigan ProteomeX ion trap mass spectrometer to confirm the absence of these six abundant proteins. The buffer of the depleted plasma was exchanged, using Millipore Centricon YM-3 (MW cutoff at 3000 Da), to those appropriate for the respective techniques. Total protein concentration was determined using the Bradford assay, with bovine serum albumin (BSA) as the standard. The depleted plasma was not used as a negative control. Rather, various amounts of BSA were added into the depleted plasma to obtain two sets of reconstituted plasma samples (1 or 3 pmol BSA in 7 µg plasma proteins; 10 or 30 pmol BSA in 7 µg plasma proteins). The same levels of BSA were also added to the respective method buffers as baseline samples. Two HCT-116 cell lines, HCT-116 and HCT-116 p53-/-, were cultured in 150 × 20 mm dishes to around 80% cell confluence. The cells were washed three times with PBS and scraped into a 1× protease inhibitor cocktail solution (0.1 M tris/pH 8.0 containing 10 µg/mL aprotinin, 10 µg/mL leupeptin, and 0.2 mM phenyl methyl sulfonyl fluoride). The cells were sonicated twice (10 s each) and centrifuged at 13 000 × g for 20 min. The supernatant was saved as the cell lysate. The buffer of cell lysates was exchanged to triethylammonium bicarbonate (a volatile buffer) using Millipore Microcon YM-10, and the sample was evaporated in a speedvac to near dryness. The proteins were redissolved in buffers appropriate for respective techniques. For all of the three labeling methods, 75 µg of lysate proteins from the wild type (HCT116) or the mutant (HCT-116 p53 -/-) cells were used. DIGE and 2D-SDS-PAGE. All chemicals and instruments used in DIGE and 2D-SDS-PAGE were purchased from Amersham (now GE Healthcare). Samples were separately labeled with different CyDyes (Cy2, Cy3, and Cy5) unless otherwise stated. The ratio of protein to CyDye was maintained at 50 µg/ 400 pmol. The CyDye labeling only labels approximately 1-2% of the lysine residues. The labeled mixtures were combined and loaded onto pH 3-10 NL 24-cm Immobiline Dry Strips (IPG) for first dimension separation by isoelectric focusing (IEF) for 24 h. The strips were then loaded onto 10-20% gradient or
Three Proteomic Quantitative Methods
12.5% polyacrylamide gels. The gels were run overnight and then scanned using a Typhoon scanner for the CyDye fluors, followed by Sypro Rudy staining overnight. An Ettan Spot Handling Workstation was used for automated spot picking, in-gel digestion, and spotting onto matrix assisted laser desorption ionization (MALDI) target plates. Spot image analyses and spot selection were carried out using the Progenesis Discovery 2005 software (Nonlinear, UK). cICAT and iTRAQ. The cICAT and iTRAQ reagent kits were obtained from Applied Biosystems (ABI). Samples were treated essentially as recommended in the standard manufacturer’s protocols, except that, in the iTRAQ protocol the final eluates were concentrated 10-fold to enrich the peptide concentration and reduce acetonitrile content (originally 25%). To reduce sample complexity, a strong cation exchange (SCX) chromatography, using an off-line SCX cartridge and buffers supplied by ABI, was carried out for cell lysate-derived peptides when the cICAT (6 fractions) or iTRAQ (10 fractions) method was used. LC-MALDI-TOF/TOF and Database Search. Peptide separation was performed on a Famos/Switchos/Ultimate chromatography system (Dionex/LC Packings) equipped with a Probot (MALDI-plate spotting device). Individual SCX and/or acid cleaved avidin-eluted fractions were injected and captured onto a trap column (PepMap C18, 5 µm, 100 A, 300 µm i.d. × 5 mm) at 10 µL/min. Peptide separation of the BSA-spiked depleted human plasma or HCT-116 cell lysates was achieved on an analytical nano-column (PepMap C18, 3 µm, 100 A, 75 µm i.d. × 15 cm) using a gradient of 5 to 60% solvent B in A over 90 min (solvent A: 100% water, 0.1% TFA; solvent B: 80% ACN/ 20% water, 0.1% TFA) (TFA: trifluoroacetic acid), 60 to 95% solvent B in A for 1 min, and then 95% solvent B for 19 min at a flow rate of 0.16 µL/min. The 6-protein mixture was analyzed using a shorter gradient (5 to 60% solvent B over 30 min), with formic acid, instead of TFA, as the mobile phase modifier. The high performance liquid chromatography (HPLC) eluant was supplemented with 5 mg/mL R-cyano-4-hydroxycinnamic acid (in 50/50 ACN/water containing 0.1% TFA) from a syringe pump at a flow rate of 1 µL/min, and spotted directly onto the ABI 4700 576-well target plates using the Probot. MALDI-TOF/ TOF data were acquired in batch mode using an ABI 4700 Proteomics Analyzer. The number of precursor peaks for MS/ MS acquisition per spot and S/N filter were set at 5 and 20, respectively, for DIGE samples; but 3 and 20, respectively, for the six-protein mixture. The maximum number of peaks per spot, S/N filter, and spot-to-spot precursor mass tolerance were 10, 40, 200 ppm and 20, 100, 200 ppm, respectively, for cICAT and iTRAQ samples. The database search criterion was based on individual ion score (based on MS/MS) g 95% confidence interval (p < 0.05) for unknown proteins searched against Swiss-Prot. The differential expression ratios for the proteins which passed such database search criterion are listed in Table 1.
Results and Discussion Quantification of a Six-Protein Mixture. All three techniques were used to quantify a 6-protein sample equally aliquoted into two tubes for labeling. As shown in Table 1a, all three methods yielded reasonably good accuracy (between 90% and 112% of expected values) with a variation of less than 15%. The accuracy is defined as the acquired ratio divided by the theoretical ratio, and the relative standard deviation is defined as standard deviation divided by the average of the ratios for that protein.
research articles A semiquantitative approach based on the number of peptides detected for each protein has recently been reported.16 The approach assumes that the more abundant a protein is in the sample, the greater the chance that it can be detected. The larger number of peptides detected (Table 1a) by iTRAQ suggests that iTRAQ is more sensitive than cICAT. The superior sensitivity of iTRAQ over cICAT may be attributed to two factors. First, the tagging by iTRAQ reagent is global, while that by cICAT is cysteine-specific. It has been reported that the avidin affinity purification steps for biotinated peptides used in the cICAT protocol recover approximately 10% of the original peptides.2 Furthermore, not all cysteine-containing peptides with a biotin group could be captured.20 Second, earlier studies have shown that peptides that end with an arginine tend to give a much stronger signal in MALDI-MS analysis than those that end with a lysine.21-23 The basicity of the guanidino functionality of the arginine side chain possibly accounted for better ionization,21 as the conversion of a lysine to a homoarginine also improves ionization.22,23 The iTRAQ tags contain a moderately strong basic group (N-methylpiperazine), which likely promotes ionization of lysine-containing peptides in a way similar to the effect due to a guanidino group. Table 1a also illustrates a distinctive difference between cICAT and iTRAQ in that the number of peptides identified by cICAT is cysteine abundance-dependent, while that by iTRAQ is not. This is further supported by the correlation coefficients between the number of cysteine residues and the peptides identified, as shown in Figure 1. This observation is consistent with those reported in two previous studies, showing that proteins with fewer than 8 cysteines were less likely to be detected (quantified) by the cICAT approach12 and that the order of abundance among 491 microsomal proteins identified could be attributed to their cysteine content.24 In the latter study,24 a significant percentage of the identified proteins which contain between one and four cysteine residues are abundant proteins, while those low abundant proteins identified contain an average of 12-17 cysteine residues. Our results suggest that global tagging by iTRAQ followed by fractionation, to reduce sample complexity prior to quantitative analyses, was superior to the cICAT approach for those which are not particularly cysteine-rich. Since the total number of peptides in most proteins is much higher than that of cysteine-containing peptides, the presence of more cysteines did not seem to influence the observed superior sensitivity of iTRAQ over cICAT. The semiquantitative analysis based on the number of peptides detected was not applicable to the DIGE analysis, as the number of precursors selected for MS/MS in this gel-based method was set at 3, instead of 5 for both cICAT and iTRAQ methods. It should also be pointed out that although the same amount of protein (129 µg per tube) was labeled, the actual amounts of peptides introduced into the mass spectrometer were different in the three methods. For the cICAT or iTRAQ sample, only a small fraction of the peptides from labeled proteins was actually analyzed by the LC-MALDI. The mass limitation was primarily due to the capacities of the trap and analytical columns used in the LC-MALDI set up. The capacities of the trap column (PepMap C18, 5 µm, 100 A, 300 µm i.d. × 5 mm) and analytical column (PepMap C18, 3 µm, 100 A, 75 µm i.d. × 15 cm) are about 1-10 µg and 1-10 pmol protein digest, respectively (the 1-10 pmol protein digest is equivalent to 0.06 to 0.6 µg protein digest for BSA). Therefore, it is estimated that less than one-tenth the amount of proteins (∼10 µg vs ∼129 µg) was actually introduced into the mass specJournal of Proteome Research • Vol. 5, No. 3, 2006 653
research articles
Wu et al.
Table 1. Quantification Ratios of Six Protein Mixture, Reconstituted Protein Mixture, and HCT-116 Cell Lysates (a) six protein mixture method
iTRAQ cICAT DIGE
BSA
no. of matched peptides accuracy %a relative SDb no. of matched peptides accuracy %a relative SDb no. of matched peptides accuracy %a
R-lactalbumin
β-galactosidase
β-lactoglobulin
lysozyme
apotransferrin
5 14 93.0% 12.7% 1 94.0% n/a 3 105.0%
8 7 96.0% 14.5% 1 111.0% n/a 3 90.1%
38 24 93.0% 8.8% 8 98.0% 4.3% 3 99.0%
no. of cysteine 16 8 22 3 97.0% 112.0% 10.9% 13.3% 2 1 100.0% 91.0% 1.0% n/a 3 3 109.0% 95.2%
35 16 98.0% 8.9% 6 93.0% 9.7% 3 92.6%
(b) Reconstituted protein mixture method
iTRAQ cICAT DIGE
no. of matched peptides accuracy %a relative SDb no. of matched peptides accuracy %a relative SDb no. of matched peptides accuracy %a
1 or 3 pmol BSA in method buffer
1 or 3 pmol BSA in depleted plasma
10 or 30 pmol BSA in method buffer
10 or 30 pmol BSA in depleted plasma
10 92.0% 19.1% 2 121.7% 28.2% 2 83.4%
5 99.7% 14.8% 1 86.3% n/a 2 85.8%
33 93.7% 30.0% 14 92.7% 20.9% 5 92.3%
23 85.0% 27.4% 7 87.0% 14.0% 4 80.9%
(c) HCT-116 cell lysatesc iTRAQ
cICAT
654
protein name
ratio
no. of Cd
MW
pI
(O43707) Alpha-actinin 4 (P09972) Fructose-bisphosphate aldolase C (P07355) Annexin A2 (P21333) Filamin A (P04406) Glyceraldehyde-3-phosphate dehydrogenase (Q14697) Neutral alpha-glucosidase AB (Q14697) Neutral alpha-glucosidase AB (P62805) Histone H4. (P52926) High mobility group protein HMGI-C (P08238) Heat shock protein HSP 90-beta (P08107) Heat shock 70 kDa protein 1 (P14618) Pyruvate kinase, isozymes M1/M2 (P09382) Galectin-1 (P19338) Nucleolin (P19338) Nucleolin (P07237) Protein disulfide-isomerase (P30101) Protein disulfide-isomerase A3 (Q06830) Peroxiredoxin 1 (Q06830) Peroxiredoxin 1 (P35998) 26S protease regulatory subunit 7 (P35998) 26S protease regulatory subunit 7 (P18621) 60S ribosomal protein L17 (P62081) 40S ribosomal protein S7 (P46781) 40S ribosomal protein S9 (Q9BQE3) Tubulin alpha-6 chain (P60174) Triosephosphate isomerase (P60174) Triosephosphate isomerase (P67809) Nuclease sensitive element binding protein 1 (P67809) Nuclease sensitive element binding protein 1 (P67809) Nuclease sensitive element binding protein 1
1.54 0.44 0.67 1.75 0.63 0.51 0.60 1.63 1.65 1.85 1.55 0.46 1.51 0.65 0.64 1.51 1.69 0.63 0.65 0.37 0.67 1.50 1.65 1.50 1.60 2.14 1.50 2.48 1.66 1.56
8 7 4 48 3 7 7 0 0 6 5 10 6 1 1 6 7 4 4 7 7 4 0 1 12 5 5 0 0 0
113,658 42,792 43,388 305,993 39,928 111,209 111,209 12,959 13,842 94,309 77,588 63,705 16,148 89,182 89,182 64,319 64,130 25,162 25,162 53,549 53,549 25,183 25,716 25,374 53,297 29,778 29,778 38,352 38,352 38,352
5.3 6.5 7.6 5.7 8.6 5.7 5.7 11.4 10.6 5.0 5.5 7.9 5.3 4.6 4.6 4.8 6.0 8.3 8.3 5.7 5.7 10.2 10.1 10.7 5.0 6.5 6.5 9.9 9.9 9.9
protein name
ratio
no. of Cd
MW
pI
(P02765) Alpha-2-HS-glycoprotein (P02765) Alpha-2-HS-glycoprotein (P36551) Coproporphyrinogen III oxidase, mitochondrial (P09382) Galectin-1 (P09382) Galectin-1 (Q99832) T-complex protein 1, eta subunit (P20333) Tumor necrosis factor receptor superfamily member 1B (P11387) DNA topoisomerase I (P60174) Triosephosphate isomerase (P08238) Heat shock protein HSP 90-beta
0.65 0.65 0.38 1.60 1.58 1.52 0.08 1.61 1.68 2.02
11 11 9 6 6 9 28 8 5 6
39,300 39,300 50,143 16,148 16,148 59,329 48,260 90,669 29,778 94,309
5.4 5.4 8.6 5.3 5.3 7.6 5.9 9.3 6.5 5.0
Journal of Proteome Research • Vol. 5, No. 3, 2006
research articles
Three Proteomic Quantitative Methods Table 1 (Continued) DIGE
protein name
ratio
no. of Cd
MW
pI
(Q92945) Far upstream element binding protein 2 (P31943) Heterogeneous nuclear ribonucleoprotein H (P29401) Transketolase (P13639) Elongation factor 2 (P12429) Annexin A3 (P60174) Triosephosphate isomerase
0.58 0.52 0.70 1.60 4.01 1.60
7 5 12 17 3 5
73,063 49,352 68,519 96,115 36,393 26,807
8.0 5.9 7.6 6.4 5.6 6.5
a Accuracy % is defined as the acquired ratio divided by the theoretical ratio. b Relative standard deviation is defined as standard deviation divided by the average of the ratios for that protein. c replicate identifications of the same protein are from different salt steps d no. of C represents the numbers of cysteine.
Figure 1. Cysteine-abundance bias based on the correlation between the number of cysteine and the number of peptides detected in the 6 protein mixture in (a) cICAT (filled circles and solid line), but not in (b) iTRAQ (open circles and dotted line).
trometer in the two LC-MALDI methods (cICAT and iTRAQ). The precise estimation is difficult because the extent of mass transfer of these two LC-MALDI platforms is largely unknown. For gel samples, all proteins were loaded onto the IPG strip. The mass transfer of proteins from IPG strip to a 2D gel, and from gel spots of variable sizes picked with a fix-sized picker head could only be estimated. Similarly, the mass transfer of peptides from gel matrix to microtiter plate and from microtiter plate to MALDI target plate vary among different experiments. For LC-MALDI methods, peptides from any given protein are distributed across the plate in multiple spots. The MS/MS spectra for the precursor ions are acquired from each spot where their MS signals are most intense in the LC elution profile. A quantitative evaluation of all the aforementioned mass transfer efficiencies (or the extents of sample loss) is difficult and beyond the scope of this study. Quantification for BSA-Spiked Plasma Protein Mixture. We next explored the sensitivity and accuracy of the three different methods using a moderately complex sample, i.e., BSA spiked into depleted plasma. Various amounts of BSA were added to 7 µg of plasma proteins to obtain two sets of samplessone contains 1 or 3 pmol BSA; and the other contains 10 or 30 pmol BSA. The same amounts of BSA were also spiked into buffers for the respective methods to form baseline samples. On the basis of the aforementioned column capacities of the trap and analytical columns (∼1-10 µg and ∼1-10 pmol protein digest, respectively), comparable amounts of proteins (
