Evaluation of Comprehensive Multidimensional Separations Using Reversed-Phase, Reversed-Phase Liquid Chromatography/Mass Spectrometry for Shotgun Proteomics Tatsuji Nakamura, Junro Kuromitsu, and Yoshiya Oda* Eisai Company, Ltd., Laboratory of Core Technology, Tokodai 5-1-3, Tsukuba, Ibaraki 300-2635, Japan Received September 10, 2007
Two-dimensional liquid-chromatographic (LC) separation followed by mass spectrometric (MS) analysis was examined for the identification of peptides in complex mixtures as an alternative to widely used two-dimensional gel electrophoresis followed by MS analysis for use in proteomics. The present method involves the off-line coupling of a narrow-bore, polymer-based, reversed-phase column using an acetonitrile gradient in an alkaline mobile phase in the first dimension with octadecylsilanized silica (ODS)-based nano-LC/MS in the second dimension. After the first separation, successive fractions were acidified and dried off-line, then loaded on the second dimension column. Both columns separate peptides according to hydrophobicity under different pH conditions, but more peptides were identified than with the conventional technique for shotgun proteomics, that is, the combination of a strong cation exchange column with an ODS column, and the system was robust because no salts were included in the mobile phases. The suitability of the method for proteomics measurements was evaluated. Keywords: Shotgun proteomics • LC/MS • Peptide identification • Polymeric column
Introduction Over the past decade, high-performance liquid chromatography coupled with mass spectrometry (LC/MS) has become an indispensable tool for proteomics and much work has been done to improve LC/MS technology.1–3 Most peptide separations are performed with columns packed with an octadecylsilanized silica (ODS) operated with an acidic acetonitrile gradient, and MS can detect different m/z peaks from coeluting peptides. However, tens to hundreds of peptides may be coeluted, and it is generally not possible to obtain MS/MS spectra for all peptides since the scan speed of the mass spectrometry is limited. In other words, single-dimension separations using one ODS column lack sufficient resolution to resolve complex biological matrices or cellular extracts, and ionization of some components might be suppressed due to overlapping peaks. Complex samples require analytical methods with extremely high resolving power in order to provide thorough analysis of the sample components. Multidimensional analytical tools having orthogonal separation modes are required to overcome the problem of insufficient resolution in analysis of complex biological matrices to reduce the sample complexity4 and, thus, to support the identification of especially low-abundance peptides.5–7 Multidimensional liquid chromatography can be generally carried out in two different modes: heart-cutting and comprehensive chromatography. Heartcutting chromatography involves the transport of one or more fractions of the eluate from the first separation column to the second separation column.8,9 In comprehensive multidimen* Towhomcorrespondenceshouldbeaddressed.E-mail:
[email protected]. Tel: +81-29-847-7084. Fax: +81-29-847-7614. 10.1021/pr7005878 CCC: $40.75
2008 American Chemical Society
sional chromatography, the entire eluate from the first dimension is transferred to the second dimension. For proteomics, comprehensive analysis is generally required. The most efficient high-resolution separation technique currently available for protein separation is two-dimensional gel electrophoresis, which was introduced by O’Farrell.10,11 However, separation of peptides by comprehensive two-dimensional LC followed by MS analysis is also a powerful technique in proteomics. Yates et al. introduced this method for peptide analysis, naming it multidimensional protein identification technology (MudPit).12,13 They developed an automated MudPit method to separate complex mixtures of proteins after tryptic cleavage, making it possible to detect a protein at a level of 100 copies per cell on a background of 1 000 000 protein molecules per cell. Most of the reported methods employ a strong cation exchange (SCX) column as the first dimensional separation. In an SCX column, separation depends mainly on the nature of the basic moiety and is only slightly affected by the hydrophobicity of peptides. In an ODS column, the analytes are separated mainly on the basis of hydrophobicity, though basic moieties such as amines sometimes interact with silica beads. Therefore, the combination of these two separation modes should maximize the number of chromatographically resolvable components. The nature of the mobile phase used in the first-dimension SCX column is also generally compatible with the second-dimension ODS column. A low concentration of organic solvent in an acidic environment in the first step of separation allows trapping with minimal band broadening of analytes on the ODS column. However, the main drawback of this approach is the need for complete, labor-intensive desalting of the SCX-fractionated samples, though a volatile aqueous buffer is often needed to Journal of Proteome Research 2008, 7, 1007–1011 1007 Published on Web 02/02/2008
research articles permit robust LC/MS analysis for a long period. The SCX column is preferred for the first-dimension separation because of its higher sample capacity. In this paper, we report the new combination of a polymer-based, reversed-phase (PolyRP) column with an ODS column for proteomics. The first dimension again consists of RP separation, but with an acetonitrile gradient in an alkaline environment (0.1% ammonium hydroxide). Such an approach is rare in the literature14,15 but has the potential for high performance in proteomics applications. We evaluated a range of analytical conditions in the first dimension in order to optimize the identification rate of tryptic peptides.
Materials and Methods Acetonitrile, methanol, trifluoroacetic acid (TFA), acetic acid, formic acid, urea, dithiothreitol (DTT), iodoacetamide, and Lys-C were obtained from Wako Pure Chemical Co. (Osaka, Japan). Sequence-grade modified trypsin was obtained from Promega (Madison, WI). The water used for preparing the mobile phases was purified using a Milli-Q SP TOC (Millipore, Marlborough, MA). Other reagents of analytical grade were used without further treatment. Sample Preparation for the Two-Dimensional LC-MS/ MS Analysis. C57BL mouse forebrain was extracted in the presence of a protease inhibitor cocktail (Roche Diagnostics, Basel, Switzerland), and the soluble fraction was collected from the supernatant after centrifugation (100 000g) with a Hitachi Himac CS150GXL (Tokyo, Japan). Protein amounts were measured using a BCA protein assay kit (Pierce, Woburn, MA). The pH of cell lysates was adjusted by adding 500 mM Tris-HCl buffer (pH 9.0) and the solutions were made up to 8 M urea (final concentration). These mixtures were reduced with DTT, alkylated with iodoacetamide, and digested with Lys-C (Wako, Osaka, Japan). After 4-fold dilution with 50 mM ammonium bicarbonate, trypsin was added and the mixtures were kept at 37 °C overnight. The digested solutions were desalted and concentrated using an Empore C18-HD disk cartridge (3M, MN). First-Dimensional LC Separation. The desalted samples were subjected to the first dimensional SCX, or SAX or PolyRP separation. For SCX separation using a Mini S 3.2 × 30 mm (GE Healthcare, Uppsala, Sweden), the mobile phase A was 0.1% formic acid in water/acetonitrile (9:1), and the mobile phase B was 500 mM sodium chloride in mobile phase A. For SAX separation using a Mini Q 3.2 × 30 mm (GE Healthcar), the mobile phase A was 25 mM Tris-HCl (pH 8.0) in water/ acetonitrile (9:1), and the mobile phase B was 500 mM sodium chloride in mobile phase A. For PolyRP separation using a PLRP-S 1.0 × 150 mm (stylene/divinilbenze copolymer, 3 µm spherical particle, pore size: 100 A°, Polymer Laboratories, Shropshire, U.K.), the mobile phase A was 50 mM ammounium hydroxide in water, and the mobile phase B was 50 mM ammounium hydroxide in water/acetonitrile (2:8). The gradient in all first dimensional SCX, SAX, and PolyRP separations was 0–60% B (0–45 min), 60–100% B (45–50 min), 100% B (50–60 min), and 0% B (60–75 min). Fractions were collected every 2 min. Nano-LC-MS/MS and Data Analysis. An Empore C18-HD disk cartridge (3M) packed into a 200 µL tip was used for desalting/concentration before LC-MS/MS analysis. LC/MS was performed using an Ultimate 3000 nano LC pump (Dionex Co., Sunnyvale, CA), a HTC-PAL autosampler (CTC Analytics, Switzerland), and a LTQ (Thermo Electron, San Jose, CA) with an in-house-built nanosprayer (100 µm inner diameter, 6 µm 1008
Journal of Proteome Research • Vol. 7, No. 3, 2008
Nakamura et al. opening, 150 mm length). ReproSil-Pur C18 materials (3 µm, 120 A°, Dr. Maish, Ammmerbuch, Germmany) were packed into the nanosprayer. The mass spectrometer was operated in the data-dependent mode to automatically switch between MS full scan and MS2 at a spray voltage of 2400 V. The mobile phases A and B for ODS separation at LC/MS consisted of 0.2% acetic acid in water and in water/acetonitrile (1:4), respectively. The gradient was 5% B (0–5 min), 5–40% B (5–80 min), 40–100% B (80–85 min), 100% B (85–95 min), and 5% B (95–115 min) at a flow-rate of 500 nL/min. Peptides and proteins were identified via automated database searching using Mascot v2.1 (Matrix Science, London, U.K.) against NCBInr database with a mass tolerance of 2 Da for precursor ions, 0.8 Da for product ions, and strict trypsin specificity. Peptides were considered identified if the Mascot score was over the 95% confidence limit.
Results and Discussion Silica-based columns sometimes show low recovery and poor peak shapes to basic compounds. To improve the efficiency of chromatography, the ion-pairing effect of acids such as trifluoroacetic acid is used for peptide analysis on ODS columns. Compared with the silica based columns, polymer-based PolyRP columns are suitable for the separation of such molecules.16 Besides providing higher yield, PolyRP columns allow the use of alkaline mobile phases, which have more dissolving power than acidic solutions for proteins in general. In a twodimensional separation system, orthogonal separations are preferred, because the different retention mechanisms of the two columns permit the resolution of some components that are coeluted in the first-dimension separation. Therefore, it is important to examine whether PolyRP and ODS show similar separation characteristics.17 We compared the retention times on an ODS column of an early eluted fraction 6, a middle-eluted fraction 16, and a late-eluted fraction 24 from PolyRP, as shown in figure 1A. If the two columns had similar retention behavior, components in fraction 6 from PolyRP would be eluted earlier on ODS, while components in fraction 24 would be slowly eluted from ODS, and the components of the two fractions would be well-separated on ODS. In fact, the retention window on the ODS column of fraction 6 from the PolyRP column was 90% overlapped with that of fraction 24, which means that the two columns, PolyRP and ODS, actually show almost orthogonal separation. The reason different separation mechanisms appeared to operate in the two reversed-phase columns is probably as follows. Under the conditions used for PolyRP chromatography, the physicochemical properties of tryptic peptides are different from those under acidic conditions in ODS column separation, because peptides have negative charges under alkaline conditions, but positive charges under acidic conditions. In fact, it has been reported that changing the pH of the mobile phase is a useful tool for achieving orthogonal 2D-LC separation of peptides prepared from digested protein standards.18 It has also been reported that separation on the PolyRP column involves both hydrophobic and cation exchange mechanisms at basic pH values of the mobile phase.19 Also, silanol residues on ODS generally produce weak retention, especially under acidic conditions, but sometimes interact strongly with basic amino acid residues of tryptic peptides. Therefore, the retention behaviors of tryptic peptides on the first PolyRP column and the second ODS column are quite different. The SCX column did not show hydrophobic behavior, because the early fraction 7 and the late fraction 25 from SCX had almost 100% overlapping retention windows on
Evaluation of Comprehensive Multidimensional Separation
research articles
Figure 2. Comparison of numbers of identified peptides between three different fractionations in the first dimension. The peptide sample was prepared from the soluble protein fraction of mouse forebrain after reduction, carbamidomethylation of cysteine residues and enzymatic digestion. The sample was subjected to first-dimensional LC separation, and the separated fractions were analyzed by ODS column LC-MS/MS. Brown bars indicate overlapped peptides among the three columns, pink bars indicate overlapped peptides between PolyRP and SAX, yellow bars indicate overlapped between PolyRP and SCX, light blue bars indicate overlapped between SCX and SAX, and dark blue bars indicate unique peptides for each column.
Figure 1. Elution profiles of identified peptides from the ODS column. The peptide sample was prepared from soluble protein fraction of mouse forebrain after reduction, carbamidomethylation of cysteine residues, and enzymatic digestion. The sample was subjected to first-dimensional LC separation, and the separated fractions were analyzed by ODS column LC-MS/MS. The normalized number of identified peptides is plotted against retention time on the ODS column. Fraction 7 (black line) was early eluted and fraction 25 (blue line) was late eluted. (A) PolyRP fractionation as the first dimension. (B) SCX fractionation.
the ODS column, as shown in Figure 1B. Interestingly, more hydrophobic peptides, which have stronger retention on ODS, were identified from the PolyRP column (Figure 1). The separation conditions used with PolyRP might be more suitable for hydrophobic peptides than those used with SCX in terms of recovery from the column. We next compared three different candidates for a comprehensive 2D-LC system. The numbers of identified peptides were 4035 from PolyRP-ODS, 1579 from SCX-ODS, and 3469 from SAX-ODS, as shown in Figure 2. We then explored the reasons PolyRP gave the best result and SCX the worst as the first-dimensional separation. Figure 3 shows the overlap of peaks before and after fraction 16 to evaluate peak broadening in the first dimension. In the case of PolyRP, 47.8% of peaks were unique to fraction 16, whereas 46.5% peaks overlapped among fractions 15-17 in the case of SCX. Peak shapes of all identified peptide eluted from SCX were broadened, and SAX was intermediate between PolyRP and SCX in this respect. Because of the relatively low plate numbers, PolyRP columns are not widely used, but SCX columns, which
Figure 3. Evaluation of peak broadening in the first-dimensional separation. The peptide sample was prepared from the soluble protein fraction of mouse forebrain after reduction, carbamidomethylation of cysteine residues, and enzymatic digestion. The sample was subjected to first-dimensional LC separation, and the separated fractions were analyzed by ODS column LC-MS/ MS. The relative number of unique peptides identified in only fraction 16 is showed by yellow bars, and the relative number of peptides observed in three sequential fractions (fractions 15, 16, and 17) is shown by light blue bars. Purple bars indicate the relative number of peptides overlapped across two fractions (fractions 15 and 16 or fractions 16 and 17).
are popular in multidimensional LC/MS for shotgun proteomics, also gave broadened peaks on chromatograms. However, PolyRP separation with gradient elution in the first dimension allowed a narrower fraction to be directed into the second dimension than the SCX/SAX separation, because equilibration between the stationary phase and mobile phase is faster in the RP mode than in the ion exchange mode. This peak broadening in the first dimension was directly related to the number of identified peptides. Figure 4 shows the elution profile of identified peptides obtained by SCX and PolyRP fractionations. PolyRP eluted peptides almost equally over all the retention windows, whereas SCX eluted more than twice the number of peptides in the earlier fraction 7. Ideally, the number of peaks in each fraction from the first dimension should be equally Journal of Proteome Research • Vol. 7, No. 3, 2008 1009
research articles
Figure 4. Elution profile of identified peptides from SCX and PolyRP column. The peptide sample was prepared from the soluble protein fraction of mouse forebrain after reduction, carbamidomethylation of cysteine residues, and enzymatic digestion. The sample was subjected to first-dimensional SCX or PolyRP column separation, and the separated fractions were analyzed by ODS column LC-MS/MS. The plot shows the fraction number 6–25 versus the number of identified peptides in each fraction.
Figure 5. Number of ionic amino acids in tryptic peptides identified in three different first-dimensional separations. Charge values were calculated by subtraction of the total number of acidic amino acids (Asp, Glu) from the total number of basic amino acids (Arg, Lys, His) in a tryptic peptide. The blue line indicates PolyRP, the red one indicates SCX, and the green one indicates SAX as the first dimension.
distributed to increase the number of identifiable peptides. Therefore, PolyRP seems to be better than SCX for comprehensive analysis. Next, the physicochemical properties of the identified peptides were compared among the three separations in first dimension. Charge number of peptides was calculated as the number of basic amino acids (Arg, Lys, His) minus the number of acidic amino acids (Asp, Glu), as shown in Figure 5. The charge numbers of peptides identified from SCX were mainly -1, 0, +1, and +2, while those in the case of PolyRP showed a similar pattern, but shifted by one charge (-2, -1, 0, and +1). On the other hand, SAX mainly identified negatively charged peptides and the distribution was wider (-1 to -5). Completely cleaved tryptic peptides contain one Arg or Lys, and sometimes His, and SCX separation seems to be based on the variety of these basic amino acids, mostly identifying peptides of charge number +1. Therefore, improvement of separation in SCX might be difficult even if the further optimization of the chromatographic condition is performed. 1010
Journal of Proteome Research • Vol. 7, No. 3, 2008
Nakamura et al.
Figure 6. UV chromatograms of human serum albumin digest in polyRP (A), SAX (B), SCX (C), and SCX by shallow gradient (D). Human serum albumin (1∼3 mg) were subjected to reduction, carbamidomethylation of cysteine residues, and enzymatic digestion. Then digested samples were injected to SCX, SAX, or PolyRP column. Each LC condition of A-C is described in the Materials and Methods, and LC condition of D was 0–60% (B conc.) in 90 min.
On the other hand, tryptic peptides contain various numbers of acidic residues, so the charge number of identified peptides is biased toward negative charges. SAX separation seems to be more suitable than SCX separation for tryptic peptides, and indeed, the number of identified peptides from SAX was much greater than that from SCX. Peptides with negative charge number were quite frequent in the SCX fraction, being eluted in fraction 6–7, as shown in Figure 4. We compared peptides separation profiles among three columns by using tryptic albumin in Figure 6, because real biological samples are too complicated to observe peak shapes. The separation capacity (peak width and retention window) of the SCX seemed to be the worst (Figure 6C,D). On the other hand, more peaks were able to be found in the PolyRP (Figure 6A) and SAX (Figure 6B) separations. We plotted the molecular weights of peptides identified from the three different columns, as shown in Figure 7. SAX identified larger peptides than those from the other two columns. This difference may contribute to the better performance of SAX than SCX, and the characteristics of SAX separation seem to be quite different from those of the other two columns. Thus, we repeated 2D-LC/MS based on PolyRPODS twice. Overlap of peptides between the two analyses was 62.4%, and there were no marked differences in terms of peptides character and peak broadening (data not shown). In Figure 2, overlap of peptides between PolyRP and SAX was less than 40%, and peptide characters and peak broadening were different between the two methods, as mentioned above. Therefore, the PolyRP-ODS and SAX-ODS 2D-LC/MS methods would be better for comprehensive proteome analysis. However, PolyRP showed narrower peak shapes and does not require any salt in the mobile phases, which is important for maintaining a robust LC/MS system. Automated online multidimensional chromatography is convenient and rapid in use, but online systems are technically more complex and often need specific interfaces. The main limitation of such multidi-
research articles
Evaluation of Comprehensive Multidimensional Separation
Figure 7. Distribution of molecular weight of identified peptides in three different first-dimensional separations. The blue line indicates PolyRP, the red one indicates SCX, and the green one indicates SAX as the first dimension.
mensional separation systems, in which the two dimensions are operated in different modes, is mobile-phase incompatibility. The introduction of large volumes of an incompatible solvent into a separation column yields broadened and distorted peaks. When PolyRP is used as the first dimension, the mobile phase transferred to the secondary column, which is also operated in RP-LC mode, has very similar elution power to the mobile phase at the head of the secondary column. In this case, effective concentration of the sample in the secondary column is difficult to achieve in the online mode. On the other hand, the use of an off-line system permits transfer from the first to the second dimension after appropriate sample processing to remove incompatible solvents. The off-line approach is also very easy and robust. Furthermore, there is no limitation in terms of the number of fractions from the first dimension in the off-line mode, which can increase separation power. Disadvantages when compared with the online mode include labor-intensive nature, slowness, sensitivity to sample loss, vial contamination, sample dilution, and poorer reproducibility. Nevertheless, the PolyRP column in the off-line mode does appear to be useful for first-dimensional separation for comprehensive shotgun proteomics. The combination of our system with quantitative methods to measure protein expression levels20,21 should lead to a better understanding of gene networks.
Acknowledgment. The authors are grateful to Takashi Seiki for animal care and Tsuyoshi Tabata for data management in Eisai. This work was supported by funds from the New Energy and Industrial Technology Development Organization (NEDO), Japan. References (1) Peng, J.; Gygi, S. P. Proteomics: the move to mixtures. J. Mass Spectrom. 2001, 36, 1083–1091.
(2) Wu, C. C.; Yates, J. R., III. The application of mass spectrometry to membrane proteomics. Nat. Biotechnol. 2003, 21, 262–267. (3) Qian, W. J.; Jacobs, J. M.; Liu, T.; Camp, D. G., II; Smith, R. D. Advances and challenges in liquid chromatography-mass spectrometry-based proteomics profiling for clinical applications. Mol. Cell. Proteomics 2006, 5, 1727–1744. (4) Giddings, J. C. Two-dimensional separations: concept and promise. Anal. Chem. 1984, 56, 1258A–1264A. (5) Wu, C. C.; MacCoss, M. J. Shotgun proteomics: tools for the analysis of complex biological systems. Curr. Opin. Mol. Ther. 2002, 4, 242–250. (6) Wang, H.; Hanash, S. Multi-dimensional liquid phase based separations in proteomics. J Chromatogr., B: Anal. Technol. Biomed. Life Sci. 2003, 787, 11–18. (7) Frahm, J. L.; Howard, B. E.; Heber, S.; Muddiman, D. C. Accessible proteomics space and its implications for peak capacity for zero-, one- and two-dimensional separations coupled with FT-ICR and TOF mass spectrometry. J. Mass Spectrom. 2006, 41, 281–288. (8) Wong, V.; Shalliker, R. A. Isolation of the active constituents in natural materials by ‘heart-cutting’ isocratic reversed-phase twodimensional liquid chromatography. J. Chromatogr., A 2004, 1036, 15–24. (9) Dobrev, P. I.; Havlicek, L.; Vagner, M.; Malbeck, J.; Kaminek, M. Purification and determination of plant hormones auxin and abscisic acid using solid phase extraction and two-dimensional high performance liquid chromatography. J. Chromatogr., A 2005, 1075, 159–166. (10) Patton, W. F. A thousand points of light: the application of fluorescence detection technologies to two-dimensional gel electrophoresis and proteomics. Electrophoresis 2000, 21, 1123–1144. (11) Wittmann-Liebold, B.; Graack, H. R.; Pohl, T. Two-dimensional gel electrophoresis as tool for proteomics studies in combination with protein identification by mass spectrometry. Proteomics 2006, 6, 4688–4703. (12) Link, A. J.; Eng, J.; Schieltz, D. M.; Carmack, E.; Mize, G. J.; Morris, D. R.; Garvik, B. M.; Yates, J. R., III. Direct analysis of protein complexes using mass spectrometry. Nat. Biotechnol. 1999, 17, 676–682. (13) Washburn, M. P.; Wolters, D.; Yates, J. R., III. Large-scale analysis of the yeast proteome by multidimensional protein identification technology. Nat. Biotechnol. 2001, 19, 242–247. (14) Sasagawa, T.; Ericsson, L. H.; Teller, D. C.; Titani, K.; Walsh, K. A. Separation of peptides on a polystyrene resin column. J. Chromatogr. 1984, 307, 29–38. (15) Mohammad, J.; Jaderlund, B.; Lindblom, H. New polymer-based prepacked column for the reversed-phase liquid chromatographic separation of peptides over the pH range 2–12. J. Chromatogr., A 1999, 852, 255–259. (16) Linde, S; Welinder, B. S. Silica versus polymer-based stationary phases for reversed-phase high-performance liquid chromatographic analyses of rat insulin biosynthesis. A comparison of resolution and recovery. J. Chromatogr. 1991, 548, 195–206. (17) McCalley, D. V. Comparison of an organic polymeric column and a silica-based reversed-phase for the analysis of basic peptides by high-performance liquid chromatography. J. Chromatogr., A 2005, 1073, 137–145. (18) Gilar, M.; Olivova, P.; Daly, A. E.; Gebler, J. C. Orthogonality of separation in two-dimensional liquid chromatography. Anal. Chem. 2005, 77, 6426–6434. (19) Ruiz, R; Ruiz-Angel, M. J.; Garcia-Alvarez-Coque, M. C.; Rafols, C; Roses, M; Bosch, E. Hydrophobic and cation exchange mechanisms in the retention of basic compounds in a polymeric column. J. Chromatogr., A 2004, 1028, 139–148. (20) Oda, Y.; Huang, K.; Cross, F. R.; Cowburn, D.; Chait, B. T. Accurate quantitation of protein expression and site-specific phosphorylation. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 6591–6596. (21) Ishihama, Y.; Sato, T.; Tabata, T.; Miyamoto, N.; Sagane, K.; Nagasu, T.; Oda, Y. Quantitative mouse brain proteomics using culturederived isotope tags as internal standards. Nat. Biotechnol. 2005, 23, 617–621.
PR7005878
Journal of Proteome Research • Vol. 7, No. 3, 2008 1011
