Anal. Chem. 2005, 77, 7763-7773
Making Broad Proteome Protein Measurements in 1-5 min Using High-Speed RPLC Separations and High-Accuracy Mass Measurements Yufeng Shen, Eric F. Strittmatter, Rui Zhang, Thomas O. Metz, Ronald J. Moore, Fumin Li, Harold R. Udseth, and Richard D. Smith*
Biological Science Division, Pacific Northwest National Laboratory, Richland, Washington 99352 Klaus K. Unger and Dipika Kumar
Institut fuer Anaorganische Chemie und Analytische Chemie, Johannes Gutenberg-Universitaet, Saarstrasse 30, D-55099 Mainz, Germany Dieter Lubda
Life Sciences Analytics, Merck KGaA, Frankfurter Strasse 250, D-64271 Darmstadt, Germany
The throughput of proteomics measurements that provide broad protein coverage is limited by the quality and speed of both the separations as well as the subsequent mass spectrometric analysis; at present, analysis times can range anywhere from hours (high throughput) to days or longer (low throughput). We have explored the basis for proteomics analyses conducted on the order of minutes using high-speed capillary RPLC combined through online electrospray ionization interface with high-accuracy mass spectrometry (MS) measurements. Short 0.8-µm porous C18 particle-packed 50-µm-i.d. capillaries were used to speed the RPLC separations while still providing high-quality separations. Both time-of-flight (TOF) and Fourier transform ion cyclotron resonance (FTICR) MS were applied for identifying peptides using the accurate mass and time (AMT) tag approach. Peptide RPLC relative retention (elution) times that were generated by solvent gradients that differed by at least 25-fold were found to provide relative elution times that agreed to within 5%, which provides the basis for using peptide AMT tags for higher throughput proteomics measurements. For fast MS acquisition speeds (e.g., 0.2 s for TOF and either ∼0.3 or ∼0.6 s for FTICR), peptide mass measurement accuracies of better than (15 ppm were obtained with the highspeed RPLC separations. The ability to identify peptides and the overall proteome coverage was determined by factors that include the separation peak capacity, the sensitivity of the MS (with fast scanning), and the accuracy of both the mass measurements and the relative RPLC peptide elution times. The experimental RPLC relative elution time accuracies of 5% (using high-speed capillary RPLC) and mass measurement accuracies of better than (15 ppm allowed for the confident identification of >2800 peptides and >760 proteins from >13 000 dif10.1021/ac051257o CCC: $30.25 Published on Web 11/02/2005
© 2005 American Chemical Society
ferent putative peptides detected from a Shewanella oneidensis tryptic digest. Initial results for both RPLCESI-TOF and RPLC-ESI-FTICR MS were similar, with ∼2000 different peptides from ∼600 different proteins identified within 2-3 min. For 150-s analysis due to the improved mass accuracies attained using longer spectrum acquisition times. While peptide-level (bottom-up) measurements of proteomes are now broadly applied, many potential proteomics applications are limited by analysis sensitivity and throughput. Rapid proteome analysis can be achieved by making mass spectrometry (MS) measurements without the use of separations,1 but measurements made in conjunction with separations provide increased dynamic range and much broader protein coverage. We recently have demonstrated the potential of capillary reversed-phase liquid chromatography (RPLC)-electrospray ionization (ESI)-linear ion trap tandem MS (MS/MS) for much higher throughput proteomic analysis (occurring on the order of minutes).1 Further gains in analysis throughput and coverage are limited by the MS/MS data acquisition speed (e.g., 25 MS/MS spectra obtainable in 6.5 s using the linear ion trap mass spectrometers) with ∼40% of the acquired spectra useful for identifying different peptides.1 Analysis coverage and throughput can be significantly enhanced by replacing one-at-a-time MS/MS measurements with accurate mass measurements2 that allow multiple peptides, including those of low relative abundance, to be identified from each spectrum. The mass measurement accuracy (MMA) needed for distinction of peptides has been typically achievable using Fourier (1) Shen, Y.; Smith, R. D.; Unger, K. K.; Kumar, D.; Lubda, D. Anal. Chem. 2005, 77, 6692-6701. (2) Bruce, J. E.; Anderson, G. A.; Wen, J.; Harkewicz, R.; Smith, R. D. Anal. Chem. 1999, 71, 2595-2599.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005 7763
Table 1. Standard Proteins and Peptides Used To Evaluate Mass Measurement Accuracya proteins
content (µg)
Sigma No.
SPro No.
MW
65 66 65 65 66
A7638 C3934 L3908 T1408 G2267
PO2769 Q865Y7 P02754 Q29443 P46406
69293 28982 19883 77753 35688
68 65 67 66 66 66
G5635 L6010 M0630 A2512 C2037 P6635
P00722 P00711 P02188 P01012 P00006 P00489
116351 16246 16951 42750 11572 97158
8 1 16 4 6 6 21 2 2 1 3 5
B1901 B4791 B3259 B4764 F3254 C9781 O3632 S2525 L7288 P7967 P2613 V0131
904 921 1240 1223 1536 3780 1407 1507 1527 2123 1533 1425
5 2
G9898 E9520
2150 2318
71 1 3 10 5
F4799 P2490 D7017 N6383 A9650
2861 2089 1603 1672 1296
bovine serum albumin bovine carbonic anhydrase bovine β-lactoglobulin bovine serotransferrin rabbit glyceraldehyde-3-phosphate dehydrogenase E. coli β-galactosidase bovine R-lactalbumin equine skeletal muschle myoglobin chickin ovalbumin bovine cytochrome c rabbit phosphorylase b peptides bradykinin fragment 2-9 Des Pro Ala bradykinin bradykinin Try bradykinin acetate salt fibrinopeptide A Tyr c peptide osteocalcin fragment 7-19 human syntide 2 leptin fragment 93-105 human [Ala92]-peptide 6 ProteoMassa¨P14R MALDI-MS standard vasoactive intestinal peptide fragment1-12 human, porcine, rat diazepam binding inhibitor epidermal growth factor receptor fragment 661-681 3X FLAGPeptide presenilin-1 N-terminal peptide dynorphin A porcine fragment 1-13 neurotensin angiotensin
a Both peptides and protein tryptic digests were dissolved in H O to a final concentration of 0.22 µg/µL. Prior to analysis, the concentration was 2 further diluted to the desired sample loading concentration as described in the text. Sigma No., Sigma product number; SPro No.. Swiss-Prot accession number. The accurate molecular masses were established by calibration using known peptides.
transform ion cyclotron resonance (FTICR) MS.3 Achieving high MMA with FTICR requires some sacrifice in the dynamic range of measurements to restrain ion populations to a level that minimizes deleterious effects due to space charge;4 while obtaining a greater dynamic range (and better detection of low relative abundance species) with larger ion populations generally lowers the MMA and thus the effectiveness (or confidence) of peptide identifications. The accurate mass and time (AMT) tag approach5 partially overcomes this limitation by combining peptide RPLC elution times and accurate mass measurements to provide a 2-D specificity that improves the identification.6-8 The approach not only provides increased throughput and sensitivity but can also (3) Goodlett, D. R.; Bruce, J. E.; Anderson, G. A.; Rist, B.; Pasa-Tolic, L.; Fiehn, O.; Smith, R. D.; Aebersold, R. Anal. Chem. 2000, 72, 1112-1118. (4) Belov, M. E.; Zhang, R.; Strittmatter, E. F.; Prior, D. C.; Tang, K.; Smith, R. D. Anal. Chem. 2003, 75, 4195-4205. (5) Smith, R. D.; Anderson, G. A.; Lipton, M. S.; Pasˇa-Tolic´, L.; Shen, Y.; Conrads, T. P.; Veenstra, T. D.; Udseth, H. Proteomic 2002, 2, 513-523. (6) Shen, Y.; Zhao, R.; Belov, M. E.; Conrades, T. P.; Anderson, G. A.; Tang, K.; Pasa-Tolic, L.; Veenstra, T. D.; Lipton, M. S.; Udseth, H.; Smith, R. D. Anal. Chem. 2001, 73, 1766-1775. (7) Shen, Y.; Tolic´, N.; Zhao, R.; Pasˇa-Tolic´, L.; Li, L.; Berger, S. J.; Harkewicz, R.; Anderson, G. A.; Belov, M. E.; Smith, R. D. Anal. Chem. 2001, 73, 30113021. (8) Shen, Y.; Tolic, N.; Masselon, C. D.; Pasa-Tolic, L.; Camp, D. G.; Hixson, K. K.; Zhao, R.; Anderson, G. A.; Smith, R. D. Anal. Chem. 2004, 76, 144154.
7764 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
enable the use of somewhat lower MMA, for example, from the use of time-of-flight (TOF) MS for proteomics analyses.9 The use of much faster separations with the AMT tag approach to realize ultrahigh-throughput proteomic analyses gives rise to issues associated with both MMA and the time measurement accuracy (TMA) of peptide RPLC relative retention time (or normalized elution time, NET) information. In particular, NET values from high-speed separations may not always be well correlated to those developed from slower separations needed to provide sufficient time for MS/MS identification of detectable peptides to develop an extensive AMT tag database. Additionally, MMA can also be somewhat degraded by very fast separations, e.g., due to the more rapidly varying composition of the mixture delivered to the MS or to the use of shorter MS spectrum acquisition times. Furthermore, speeding the RPLC separation simply by speeding the solvent gradient inevitably degrades the separation peak capacity,1 raising questions as to whether sufficient separation power can be achieved to characterize complex peptide mixtures. In this work, we report on the development of much higher throughput proteomics using high-speed capillary RPLC separations with ESI-accurate TOF and FTICR MS. Issues associated (9) Strittmatter, E. F.; Ferguson, P. L.; Tang, K.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2003, 14, 980-981.
with the analytical technology and the AMT tag methodology used for ultrahigh-throughput analysis are also addressed. EXPERIMENTAL SECTION Capillary RPLC Experiments. An RPLC system similar to one recently described1 was used for this study, with the exception that a shorter (10 cm) column was used, allowing LC separations at moderate pressures. Briefly, 0.8-µm porous C18-bonded silica particles were packed into a 50-µm-i.d. fused-silica capillary (Polymicro Technologies, Phoenix, AZ) at 8.5 kpsi. An integrated ESI emitter was made after packing the capillary to minimize the dead volume.1 The sample was loaded directly on the head of the capillary RPLC column, and separations were performed at a constant pressure of 8.5 kpsi. The RPLC gradient was controlled by adjusting the gradient split capillary (10-µm i.d., Polymicro Technologies) length. Solvent mixtures of H2O/trifluroacetic acid/ acetic acid (100:0.05:0.2, v/v/v, Aldrich, Milwaukee, WI) and acetonitrile/H2O/trifluroacetic acid (90:10:01, v/v/v, Aldrich) were used as mobile phases A and B, respectively. It is important to note that the same type of packing material and gradient composition were used for all separation speeds. RPLC-ESI-MS Studies. Three mass spectrometers capable of accurate mass measurements were used in this study: (1) a linear ion trap (LTQ)-FTICR MS (Thermo Electron, Bremen, Germany); (2) an in-house-developed 11.4-T FTICR MS;10 and (3) an Agilent TOF MS (Agilent Technologies, Palo Alto, CA). The RPLC column effluent was directly delivered to the LTQ-FTICR MS without the use of a sheath gas or makeup liquid. The heated capillary temperature and ESI voltage were 200 °C and 1.85 kV, respectively. The LTQ-FTICR MS was used in normal MS mode, and data were collected with an automatic gain control (AGC) setting of 106 and an m/z range of 400-2000, which yielded an acquisition speed of 0.3 s/spectrum for the denser data zones (near the midpoint of the separation). The ESI conditions used for the LTQ-FTICR MS (i.e., no use of a sheath gas or makeup liquid, 150 °C for the heated capillary, and 1.8 kV for ESI) were also used for the 11.4-T FTICR MS. Except where specified, this instrument was operated at a spectrum acquisition speed of 0.6 s/spectrum (with a constant ion accumulation time of 0.3 s) and an m/z range of 400-2000. The Agilent TOF MS was interfaced with an ion funnel11 and used at 120 °C heated capillary and ESI voltage of 1.6 kV. TOF spectra were obtained by summing spectra for 0.2-s periods (i.e., summing 2000 TOF spectra, 100 µs/TOF spectrum) for all analyses; the use of shorter 0.1-s periods (i.e., summing 1000 TOF spectra) resulted in fewer identifications, which is attributed to mimimum cutoffs applied in the data analysis to provide acceptable S/N. Data Analysis. The ICR-2LS software package8,12 was used to convert the raw data from RPLC-MS analyses into m/z spectra that were subsequently transformed to generate a list of neutral masses. The output data files from these analyses were then plotted as 2-D displays, and peptides were identified by searching (10) Harkewicz, R.; Belov, M. E.; Anderson, G. A.; Pasa-Tolic, L.; Masselon, C. D.; Prior, D. C.; Udseth, H. R.; Smith, R. D. J. Am. Soc. Mass Spectrom. 2002, 13, 144-154. (11) Shaffer, S. A.; Anderson, G. A.; Udseth, H. R.; Smith, R. D. Anal. Chem. 1998, 70, 4111-4119.
Figure 1. Examination of MMA obtained with a fast LC-MS analysis. Conditions: a 10 cm × 50 µm i.d. capillary packed with 0.8-µm porous C18-bonded silica particles was used as the RPLC column. A linear ion trap-FTICR MS with AGC was used for detection as described in the Experimental Section. The base peak chromatogram (m/z of 4002000 with a spectrum acquisition speed of ∼0.3 s/spectrum for the peak zone) (A); mass accuracy histogram for analysis of 70 ng of the standard test mixture (B); mass accuracy histogram for analysis of 700 ng of the standard test mixture (C). The histogram bin size (B, C) was 0.5 ppm for MMA.
against a mass versus time peptide tag database generated from separate RPLC-MS/MS analyses.1 The RPLC-MS/MS data were filtered according to previously reported criteria13 to increase the confidence of peptide identifications, and the rate of false positive identifications was evaluated using a previously reported method.14 Sample Preparation. A set of standard peptides and tryptic digest proteins (Table 1) was separated prior to mass analysis using the conditions described above to evaluate the performance of the instrument (separation, sensitivity, and MMA distribution for each mass spectrometer). A Shewanella oneidensis strain MR-1 global protein tryptic digest was used to examine the proteomics (12) Tolic´, N.; Monroe, M. E.; Anderson, G. A.; Smith, R. D. ICR-2LS software package; Pacific National Laboratories, Richland, WA. (13) Florens, L.; Washburn, M. P.; Raine, J. D.; Anthony, R. M.; Grainger, M.; Haynes, J. D.; Moch, K. J.; Muster, N.; Sacci, J. B.; Tabb, D. L.; Witney, A. A.; Wolters, D.; Wu, Y.; Gardner, M. J.; Holder, A. A.; Sinden, R. E.; Yates, J. R.; Carucci, D. J. Nature 2002, 419, 520-526. (14) Peng, J.; Elias, J. E.; Thoreen, C. C.; Licklider, L. J.; Gygi, S. P. J. Proteome Res. 2003, 2, 43-50.
Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
7765
Figure 2. Correlation of proteome peptides eluted from various RPLC separations ranging from 10 to 600 min. Conditions: a 20 cm × 50 µm i.d. capillary packed with 0.8-µm porous C18-bonded silica particles was used as the RPLC column. Peptides were identified using a linear ion trap for MS/MS using previously described conditions1,15 and reported criteria.13 A 10-µg S. oneidensis global tryptic digest was used for the 600-min RPLC-linear ion trap MS/MS experiment, and 2.5 µg was used for the 10- and 50-min experiments. (A1), (B1), and (C1) are the correlation plots and (A2), (B2), and (C2) are their corresponding error distributions.
analysis performance; the sample preparation procedure is described elsewhere.15 RESULTS AND DISCUSSION Accurate Mass Measurements with Subsecond Spectrum Acquisition Speed Using FTICR and TOF MS with HighSpeed LC Separations. Figure 1 shows the results obtained using the RPLC-LTQ-FTICR MS instrumentation at an acquisition speed of 0.3 s/spectrum for a standard test mixture (containing the peptides and protein tryptic digests listed in Table 1). Under the experimental conditions for this particular analysis, chromatographic peaks that differed by ∼0.6 s were partially resolved (see the elution times labeled on the peaks in Figure 1A), potentially providing 98% identified peptides) accurate to within (5 ppm (Figure 1B). Increasing the test mixture sample size from 70 (Figure 1B) to 700 ng resulted in slight degradation of the MMA (Figure 1C), presumably dominated by the somewhat greater variations in the trapped ion populations for the FTICR measurements with the larger sample size (there were no significant differences in peak widths between the two separations). However, we note that the two most (15) Shen, Y.; Zhang, R.; Moore, R. J.; Kim, J. K.; Metz, T. O.; Hixson, K. K.; Zhao, R.; Livesay, E. A.; Udseth, H. R.; Smith, R. D. Anal. Chem. 2005, 77, 3090-3100.
7766 Analytical Chemistry, Vol. 77, No. 23, December 1, 2005
abundant peptides K.HGTVVLTALGGILK.K (from myoglobin) and A.LIVTQTMK.G (from β-lactoglobulin) were identified with
