MS with Chromatographic Peak

Anal. Chem. 2005, 77, 3090-3100

Automated 20 kpsi RPLC-MS and MS/MS with Chromatographic Peak Capacities of 1000-1500 and Capabilities in Proteomics and Metabolomics Yufeng Shen, Rui Zhang, Ronald J. Moore, Jeongkwon Kim,† Thomas O. Metz, Kim K. Hixson, Rui Zhao, Eric A. Livesay, Harold R. Udseth, and Richard D. Smith*

Biological Science Division and Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, Washington 99352

Proteomics analysis based-on reversed-phase liquid chromatography (RPLC) is widely practiced; however, variations providing cutting-edge RPLC performance have generally not been adopted even though their benefits are well established. Here, we describe an automated format 20 kpsi RPLC system for proteomics and metabolomics that includes on-line coupling of micro-solid phase extraction for sample loading and allows electrospray ionization emitters to be readily replaced. The system uses 50 µm i.d. × 40-200 cm fused-silica capillaries packed with 1.4-3-µm porous C18-bonded silica particles to obtain chromatographic peak capacities of 1000-1500 for complex peptide and metabolite mixtures. This separation quality provided high-confidence identifications of >12 000 different tryptic peptides from >2000 distinct Shewanella oneidensis proteins (∼40% of the proteins predicted for the S. oneidensis proteome) in a single 12-h ion trap tandem mass spectrometry (MS/MS) analysis. The protein identification reproducibility approached 90% between replicate experiments. The average protein MS/MS identification rate exceeded 10 proteins/min, and 1207 proteins were identified in 120 min through assignment of 5944 different peptides. The proteomic analysis dynamic range of the 20 kpsi RPLC-ion trap MS/MS was approximately 106 based on analyses of a human blood plasma sample, for which 835 distinct proteins were identified with high confidence in a single 12-h run. A single run of the 20 kpsi RPLC-accurate mass MS detected >5000 different compounds from a metabolomics sample. Reversed-phase liquid chromatography (RPLC)-mass spectrometry (MS) using an electrospray ionization (ESI) interface has become an increasingly routine tool for complex proteomics. Separations prior to ESI play an important role in the overall sensitivity, dynamic range, and general effectiveness of proteomic analysis. Separation efficiency (e.g., how many components can be isolated from a mixture) is often quantitated by separation peak capacity.1 The separation peak capacity of isocratic chromato* Corresponding author. † Current address: Barnett Institute, Northeastern University, Boston, MA 02115.

3090 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

graphy is directly associated with the column plate number that is determined by factors, including the column length and particle size in packed column chromatography.2 Use of either longer columns or smaller particles to achieve high-efficiency separations requires higher pressures, and Jorgenson and co-workers have modified gas amplifiers for RPLC operation pressure up to 100 kspi.3 Similarly, 50 kspi RPLC instrumentation has also been developed by Lee and co-workers to speed separations.4 Although high-efficiency separations are important for addressing highcomplexity samples, proteomics and metabolomics applications to date have been limited to 10 kspi5-11 or less.5-17 Several issues must be addressed to make ultra-high-pressure RPLC broadly accepted for proteomic and metabolomic analyses. The first is whether extending pressure actually improves separation peak capacity of gradient RPLC that is used in all proteomic (1) Giddings, J. C. United Separation Science; John Wiley & Sons Inc.: New York, 1991; p 105. (2) Shen, Y.; Lee, M. L. Anal. Chem. 1998, 70, 3853-3856. (3) Patel, K. D.; Jerkovich, A. D.; Link, J. C.; Jorgenson, J. W. Anal. Chem. 2004, 76, 5777-5786. (4) Lippert, J. A.; Xin, B.; Wu. N.; Lee, M. L. J. Microcolumn Sep. 1999, 11, 631-643. (5) 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. (6) Shen, Y.; Tolic, N.; Zhao, R.; Pasa-Tolic, L.; Li, L.; Berger, S. J.; Belov, M. E.; Smith, R. D. Anal. Chem. 2001, 73, 3011-3021. (7) Shen, Y.; Zhao, R.; Berger, S. J.; Anderson, G. A.; Rodriguez, N.; Smith, R. D. Anal. Chem. 2002, 74, 4235-4249. (8) Shen, Y.; Moore, R. J.; Zhao, R.; Blonder, J.; Berger, S. J.; Auberry, D. L.; Auberry, K. J.; Masselon, C.; Pasa-Tolic, L.; Hixson, K. K.; Anderson, G. A.; Smith, R. D. Anal. Chem. 2003, 75, 3596-3605. (9) Shen, Y.; Tolic, N.; Masselon, C. D.; Pasa-Tolic; Camp, D. G.; Hixson, K. K.; Zhao, R.; Anderson, G. A.; Smith, R. D. Anal. Chem. 2004, 76, 144154. (10) Shen, Y.; Tolic, N.; Masselon, C. D.; Pasa-Tolic; Camp, D. G.; Lipton, M. S.; Anderson, G. A.; Smith, R. D. Anal. Bioanal. Chem. 2004, 378, 1037-1045. (11) Shen, Y.; Jacobs, J. M.; Camp, D. G.; Fang, R.; Moore, R. J.; Smith, R. D. Anal. Chem. 2004, 76, 1134-1144. (12) Washburn, M. P.; Wolters, D.; Yates, J. R. Nat. Biotechnol. 2001, 19, 242247. (13) Chen, J.; Balgleg, B. M.; DeVoe, D. L.; Lee, C. S. Anal. Chem. 2003, 75, 3145-3152. (14) Peng, J.; Elias, J. E.; Thoreen, C. C.; Licklider, L. J.; Gygi, S. P. J. Proteome Res. 2003, 2, 43-50. (15) Janiszewski, J. S.; Rogers, K. J.; Whalen, K. M.; Cole, M. J.; Liston, T. E.; Duchoslav, E.; Fouda, H. G. Anal. Chem. 2001, 73, 1495-1501. (16) De Baere, S.; Cherlet, M.; Baert, K.; De Backer, P. Anal. Chem. 2002, 74, 1393-1401 (17) Miao, S.-X.; Metcalfe, C. D. Anal. Chem. 2003, 75, 3731-3738. 10.1021/ac0483062 CCC: $30.25

© 2005 American Chemical Society Published on Web 04/06/2005

Figure 1. Schematic diagram of the automated 20 kpsi LC system. All operationssmicroSPE sample introduction and transfer to 50-µm-i.d. LC column, gradient LC separation, mixture purging for next run, and the pump filling/purgingswere performed using positive feedback switching valves. A replaceable ESI emitter was used to ensure effective and rugged utilization of the efficiency quality provided by the LC separations. Operational details are described in the Experimental Section.

analyses. This is debatable even in chromatography. For example, Haskins et al.18 have reported that short 2-cm capillary columns did not result in compromised RPLC resolution relative to longer columns (e.g., 10 cm) and ascribed this to the large role of the solvent strength parameter19 for the relatively low molecular weight peptide analytes. As proteolytic digests of proteins behaved more similar to intact proteins due to the large size of resultant polypeptides, the separation peak capacity in gradient RPLC would be essentially unrelated to the isocratic RPLC efficiency or the pressure used. A second related issue is if the gradient RPLC peak capacity can be improved with the use of longer columns or smaller particles, how much benefit is actually obtained in the coverage of complex peptide or metabolite mixtures using MS or MS/MS; a prerequisite for acceptance of the technology for practical proteomics and metabolomics. A third issue involves the convenience, robustness, and stability of automated instrumentation. A final issue is whether extending LC pressure can enable one-dimensional separations to approach those obtained using conventional separations in a two-dimensional (2D) format.12-14 Here we reported the development of an automated 20 kspi RPLC system and issues associated with proteomic and metabolomic applications. (18) Haskins, W. E.; Wang, Z.; Watson, C. J.; Rostand, R. R.; Witowski, S. R.; Powell, D. H.; Kennedy, R. T. Anal. Chem. 2001, 73, 5005-5014. (19) Snyder, L. R.; Stadalius. M. A.; Quarry, M. A. Anal. Chem. 1988, 55, 1412A1430A.

EXPERIMENTAL SECTION Implementation of an Automated 20 kpsi RPLC System. Figure 1 shows a schematic of the automated 20 kspi (alternatively, 1379 bar, 138 MPa, or 1361 atm) RPLC system developed in this work. Two ISCO cylinder pumps (65-mL volume each, Lincoln, NE) were used to deliver LC mobile phases A and B, respectively. The pump pressure was set at 20 kspi and all operationsspump filling, mobile-phase gradient generation, mobilephase mixer purge, on-line micro-solid phase extraction (microSPE) for loading samples, on-line sample transfer from microSPE to RPLC, and gradient RPLC separationswere completed using 20 kspi positive feedback switching valves (Valco Instruments, Houston, TX). Specifically, two four-port valves were used to fill/ purge the two pumps; the mobile-phase gradient was generated through a homemade static stainless steel mixer (2.5 mL, resistant to >40 kspi pressure) by switching the mobile phase from A to B with a four-port valve; the microSPE was coupled on-line prior to the 50-µm-i.d. RPLC column as described elsewhere8 with two sixport and one four-port valves; gradient split and mixer purge after each gradient RPLC run was accomplished by using four-port valves; the mobile phase A was used as the back pressure fluid for sealing all positive feedback valves. The RPLC solvents used in this study included a mixture of H2O/trifluroacetic acid (TFA)/ acetic acid (100:0.05:0.2, v/v/v, Aldrich, Milwaukee, WI) for mobile phase A and acetonitrile/H2O/TFA (90:10:01, v/v/v, Aldrich) for mobile phase B. After loading the sample, the gradient was immedAnalytical Chemistry, Vol. 77, No. 10, May 15, 2005

3091

Table 1. Pressure Limits of Typical Chromatography Components Used for the 20 kpsi LC System limits (kpsi) SS tubing connections 1/4-in. o.d., 1/16-in. i.d. 1/8-in. o.d., 1/16-in. i.d. 1/16-in. i.d., 50a >50 ∼20 >50

collar connections collar connections sleeve connections sleeve connections

∼50 ∼30 ∼40 ∼20 ∼20 ∼13 ∼20 ∼ 8.5

pore i.d. of 0.015 in. (∼380 µm, homemade) pore i.d. of 0.016 in. (∼400 µm) pore i.d. of 0.015 in. (homemade) pore i.d. of 0.016 in. pore i.d. pore i.d. of 0.016 in. tubing i.d. of 0.015 in. tubing i.d. of 0.016 in.

>50 >50

1/4-in.-o.d. tubing and fittings 1/16-in.-o.d. tubing and fittings

>50K: no leaking observed at 50 kpsi.

iately started; complete elution of the peptide mixtures required ∼65% solvent B using the previously described gradient profile.5 The pressure resistances of chromatography components (e.g., unions, tees, and column/tubing connections) were evaluated using a Haskel pump (Haskel, Burbank, CA) at pressures of up to 50 kspi with mobile phase B to evaluate the operational safety of the system. The pressure limits for typical chromatographic accessories, as used in this study, are listed in Table 1. Preparation of Packed Capillary Columns. All RPLC columns in this study used 50-µm-i.d. fused-silica capillaries (Polymicro Technologies, Phoenix, AZ) packed with either porous 3- (300-Å pore size, Phenomenex, Torrance, CA), 2- (100-Å pore size, Phenomenex), or 1.4-µm (120-Å pore size, Waters, Milford, MA) C18-bonded silica particles. The capillaries were packed using procedures as previously described7 at a pressure of 20 kspi. For microSPE columns, 150 µm i.d. × 4 cm fused-silica capillaries (Polymicro Technologies) were packed with 5- or 3-µm C18bonded silica particles (300-Å pore size, Phenomenex). It took ∼2 min to load a 20-µL sample solution on to the microSPE; the RPLC gradient split was controlled by adjusting the length of the 150-µm-i.d. capillaries that contained 5-µm particles (i.e., the gradient slope is proportional to the capillary splitter length). MS, MS/MS, and Data Analysis. The RPLC-MS was performed using a 11.4-T FTICR MS.6,20 The LC-MS/MS used either a conventional ion trap (LCQ XP; ThermoElectron, San Jose, CA) or a linear ion trap (LTQ; ThermoElectron) mass spectrometer. A replaceable ESI emitter coupled the RPLC column outlet to the mass spectrometers.7 The ESI voltage and the MS heated capillary temperature were held at 1.5 kV and 150 °C, respectively. The LC eluent was scanned over a mass (m/z) range of 400-2000 (1 microscan, maximum ion accumulation time 10 ms), followed by MS/MS scans (1 microscan, 100 ms maximum ion accumulation time) of the three most abundant peaks from the MS scan, using a collision energy setting of 35%. Dynamic exclusion software settings allowed for data-dependent discrimination against previ-

ously analyzed ions by exclusion of ions falling within -0.5 to +0.5 m/z unit of the analyzed ion for 1 min. The FTICR MS data were processed using ICR-2LS,21 and ion trap MS/MS peptide identifications were obtained using the SEQUEST (ThermoQuest Corp., San Jose, CA) according to a set of criteria reported by Washburn et al.12 Sample Preparation. A “global” human plasma tryptic digest was prepared as described elsewhere.11 The Shewanella oneidensis strain MR-1 protein global tryptic digest was prepared as follows. S. oneidensis was grown in a batch culture at 30 °C with shaking under aerobic conditions to either mid-logarithm phase (phase of bacterial growth in batch culture where generation time is constant) or late-logarithmic growth phase (i.e., where limitation of growth in batch culture applies due to nutrient depletion byproduct buildup, etc.). Cells were harvested with centrifugation and washed with cold PBS (pH 6.8). Equal cell volumes were combined and were resuspended in twice the volume of 50 mM ammonium bicarbonate (pH 7.8). Lysis was achieved by bead beating the cell mixture with 0.1 mm zirconia/silica beads in a minibead beater (Biospec, Bartlesville, OK) for 180 s at 4500 rpm. The lysate was collected and placed immediately on ice to inhibit proteolysis and then denatured (7 M urea and 2 M thiourea) and reduced by the addition of 0.5 M neutralized TCEP (Bond breaker, Pierce, Rockford, IL) to a final concentration of 0.5 mM. The solution was incubated for 30 min at 60 °C and then diluted 10fold with 50 mM ammonium bicarbonate (pH 7.8). Calcium chloride (1 M) was added until a final concentration of 1 mM. Digestions (3 h at 37 °C) were achieved by adding trypsin (Promega, Madison WI) in an approximate ratio of 1:50 (w/w) protease to sample protein. The product was desalted using a Supelco Supelclean C-18 tube (St. Louis, MO). The C-18 resin was conditioned with one column volume of acetonitrile followed by one column volume of 5% acetonitrile in deionized water. After the peptide mixtures were loaded onto the resin, the peptides were washed with four column volumes of 5% acetonitrile in deionized

(20) Bruce, J. E.; Anderson, G. A.; Wen, J.; Harkewicz, R.; Smith, R. D. Anal. Chem. 1999, 71, 2595-2599.

(21) Anderson, G. A., Bruce, J. E., Eds. ICR-2LS, Pacific Northwest National Laboratory, Richland, WA, 1995.

3092

Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

water. Peptide elution was accomplished with one-third column volume of 80% acetonitrile deionized water, followed by one-third column volume of 100% acetonitrile. All eluted peptides were concentrated via SpeedVac (ThermoSavant, San Jose CA) until protein concentrations were 1.0 mg/mL. Peptide concentrations were determined by BCA assay (Pierce). For metabolomic samples, S. oneidensis from culture was suspended in 1 mL of NH4HCO3 (100 mM, pH 8) and aliquotted to siliconized 0.6-mL centrifuge tubes (200 µL/tube). The samples were then covered with 0.1 mm zirconia/silica beads (BioSpec Products, Inc., Bartlesville, OK) and homogenized in a MiniBeadbeater-8 (BioSpec Products, Inc.). Cell lysates transferred to 1.5-mL siliconized centrifuge tubes by puncturing the bottoms of the 0.6-mL tubes with a 26-gauge needle and centrifuging at 14 000 rpm for 5 min. The remaining beads were washed with 100 µL of NH4HCO3 followed by an additional 5-min spin at 14 000 rpm. The supernatants were transferred by pipet to polycarbonate ultracentrifuge tubes and spun at 100 000 rpm for 10 min (Optima TL Ultracentrifuge; Beckman, Fullerton, CA) to remove membrane material. The supernatants were removed by pipet, and proteins were precipitated using acetone (80:20 v/v) and stored overnight at -20 °C. Proteins were then sedimented by centrifugation at 14 000 rpm for 5 min, and the supernatants were removed and dried in vacuo. The residues were reconstituted in 1 mL of Nanopure water and filtered through a 0.2-µm syringe filter. RESULTS AND DISCUSSION Fundamental Relationship between Gradient RPLC Peak Capacity and Operation Pressure. For simple linear gradient, the RPLC peak capacity (Cp) can be expressed as22

Cp ) 1 +

xN B∆c 4 B∆c(t0/tg) + 1

xN B∆c ) a 4

x

L dp

dp (µm)

L (cm)

Cp

T (min)

proteins/peptides (in 24 h)

3 2 1.4 1 0.5

200 90 40 20 5

1500 1200 1000 800b 550b

2000 900 400 200b 50b

1019/4623 993/4676 1012/4542 c c

a The column length (L) usable at 20 kpsi for different sizes of particles (dp) is determined by the pressure drop dependence on L and dp (see ref 23). The peak capacity (Cp) values for different column lengths and particle sizes are experimentally measured using 50-µmi.d. columns at a gradient slope of 100 min/10 cm within the separation time (T). The numbers of proteins/peptides are obtained within 24 h on various columns for 15 µg of a S. oneidensis global tryptic digest using ion trap MS/MS analysis and reported criteria (ref 12) as filters for identification of peptides/proteins. b Such sizes of porous particles are not yet available, and the peak capacities and required analysis times are predicted according to eq 5, with a gradient slope of 100 min/10 cm. c Data not available.

as follows from the column permeability expression:23

Cp ∝ xL ∝ xP

(2)

where a is a constant, L is the column length, and dp is the particle size. This expression shows that Cp for slow gradient RPLC separations is proportional not only to the square root of the plate number but also to the change in organic composition over the gradient. The ∆c term can be combined into the constant a term, since similar values (e.g., ∼65% of acetonitrile) will be required to elute peptides from the same or different porous particle-packed RPLC columns. For a fixed dp, the Cp expression can be simplified (22) Neue, U. D.; Carmody, J. L.; Cheng, Y.-F.; Lu, Z.; Phoebe, C. H.; Wheat, T. E. Design of Rapid Gradient Methods for the Analysis of Combinatorial Chemistry Libraries and the Preparation of Pure Compounds. In Advances in Chromatography; Brown, P. R., Grushka, E., Eds., Marcel Dekker: New York, 2001; Vol. 41, pp 93-136.

(3)

where P is the RPLC operation pressure. This expression highlights the significance of using longer columns and extended pressure to obtain high peak capacity separations. When column L is fixed, Cp can be simplified as 4

(1)

where N is the column plate number for isocratic elution, tg is the retention time in gradient elution, t0 is the column dead time, B is the slope of the relationship of ln k (where k is the retention factor) to the solvent composition, and ∆c is the change in organic composition over the gradient. If the gradient is sufficiently slow (i.e., tg . t0), then Cp can be simplified as follows:

Cp )

Table 2. Chromatographic Separation Peak Capacities and the Resultant Capabilities for Gradient 20 kpsi RPLC-Ion Trap (LCQ) MS/MS Proteomic Analysis Using Various Small Particle Packed Columnsa

Cp ∝ x1/dp ∝ xP

(4)

This expression illustrates the role of using smaller packing particles and extended pressures for improving the gradient RPLC peak capacity. Note that eqs 2-4 are expected to be suitable for other nonlinear gradient separations, provided the gradient is sufficiently slow so that the variance of the mobile-phase gradient becomes negligible. However, if the gradient is not sufficiently slow, then Cp will depend on both N and the gradient speed; thus, a comparison of RPLC peak capacities among different lengths of columns should be performed by using the same mobile-phase gradient along the column length (i.e., retaining a constant ratio of gradient duration to column length), not simply by using the same gradient duration or time scale.22 Experimental Generation of Chromatographic Peak Capacities of 1000-1500 at 20 kpsi Using Porous Particles. Figure 2 shows the gradient 20 kpsi RPLC-MS total ion current chromatogram for a S. oneidensis tryptic digest. Narrowing the m/z range simplifies the chromatogram to reveal peak baselines, as illustrated in Figure 2A-C for three time windows. According to the definition of Giddings,1 a peak capacity of 1500 was measured over the effective separation window of 200-1900 min (the values of 2σ range from 0.4 to 0.8 min when the narrowest and broadest peaks were discarded). Shortening the separation (23) Knox, J. H. Kinetic Factors Influencing Column Design and Operation. In Techniques in Liquid Chromatography; Simpson, C. F., Ed., John Wiley & Sons: New York, 1983; p 31.


3093

Figure 2. Achieving a chromatographic separation peak capacity of 1500 using a 200 cm × 50 µm i.d. capillary containing 3-µm porous C18 particles operated at 20 kpsi for a cellular global tryptic digest of S. oneidensis. The total ion current chromatogram (upper) was obtained with m/z range of 400-2000; the separation was simplified by narrowing the m/z ranges to (A) 700-800, (B) 750-900, and (C) 800-1000 to calculate the peak capacity between the effective peak elution time period of 200-1900 min. Conditions: 2 µg of the S. oneidensis global tryptic digest was first loaded to the microSPE column and then transferred to the RPLC column; a 11.4-T FTICR MS was used for detection (data were collected after 80-min gradient, scan speed of 6 s/scan); a linear velocity of 0.12 cm/s (measured with the RP mobile phase A) at 20 kpsi was obtained for this 200-cm-length column and the gradient was selected with reference to a conventional 10-cm column for a 100-min gradient RPLC separation (simply referred to as “100 min/10 cm”). Other conditions are given in the text.

time to 600 min by speeding the gradient reduced the peak capacity to ∼1400. According to the column pressure drop dependence on the column length and particle size,23 90- and 40-cm lengths were selected for use with 2- and 1.4-µm porous particle packed columns, respectively, at 20 kpsi to generate the peak capacities given in Table 2. These data support the use of eq 2 to describe the dependence of gradient RPLC peak capacity on column length and particle size for resolving peptides and, furthermore, lead to the following simple expression for estimating the gradient RPLC peak capacities for a given column length and particle size:

Cp ≈ 180xL/dp

(5)

where L and dp are in units of centimeter and micrometer, respectively. From this expression, an 8-cm column packed with 5-µm porous particles would be expected to achieve a Cp of ∼230, which is in agreement with the Cp of 220 (in a 140-min separation) 3094


reported by Wolters et al.;24 similarly, an 85-cm-length column packed with 3-µm porous particles would generate a peak capacity of ∼1000, which is in agreement with our previous results.5,7 Thus, the present data support the use of long RPLC columns for obtaining high peak capacity separations of tryptic peptides. Figure 3 shows the gradient 20 kpsi RPLC of S. oneidensis metabolomes using the conditions of Figure 2. The metabolomic components had lower average retention than peptides (see time scales of Figures 2 and 3), and a peak capacity of ∼1500 was measured across the effective separation window of 80-1800 min. This peak capacity value is much higher than obtained using shorter columns (e.g., 10-20 cm),15-17 and exhibits column length and LC pressure dependences similar to those for peptides. Packing Materials of Smaller Porous Particles for Gradient 20 kpsi RPLC. Smaller packing particles favor the LC (24) Wolters, D. A.; Washburn, M. P.; Yates, J. R. Anal. Chem. 2001, 73, 56835690.

Figure 3. Global metabolite analysis by 20 kpsi RPLC-MS. In this single LC-MS experiment, >5000 metabolites were detected. Conditions: 20 µL of the S. oneidensis metabolomic sample (unknown concentrations) was loaded to the RPLC system; other conditions are the same as for Figure 2, except that an m/z range of 100-1500 was used for detection.

kinetics of fast separations,25-29 and we attempted completing the separation within 1 h (as shown in Figure 4) by use of a gradient slope of 25 min/10 cm and a 20-cm column packed with prototype 1.4-µm porous particles (proprietary to Waters Corp.). A peak capacity of ∼350 was measured for both peptides and metabolites. Further improvement in separation speed requires use of even smaller particles, and Table 2 lists the predicted column lengths usable at 20 kpsi and peak capacities obtainable for 1- and 0.5-µm porous particles according to eq 5. It is noted that the use of smaller particles does not provide higher peak capacity (e.g., >1000) within pressure limit of 20 kpsi, but improves the peak capacity generation rates (e.g., to 550/50 min). Packing Materials of Nonporous Particles for Gradient 20 kpsi RPLC. Smaller nonporous particles (1-1.5 µm dp) are of interest primarily for ultra-high-pressure fast RPLC.3,4,29-32 However, their lower solute retention could significantly affect the gradient RPLC peak capacity since it is determined by both chromatographic peak width (dictated by N1/2) and the separation (25) Giddings, J. C. Dynamics of Chromatography, Part I, Principle and Theory; Marcel Dekker: New York, 1965; p 62. (26) Giddings, J. C. Anal. Chem. 1964, 36, 1890-1892. (27) Giddings, J. C. Anal. Chem. 1964, 37, 60-63. (28) Giddings, J. C. Anal. Chem. 1967, 39, 1027-1028. (29) MacNair, J. E.; Lewis, K. C.; Jorgenson, J. W. Anal. Chem. 1997, 69, 983989. (30) Wu. N.; Lippert, J. A.; Lee, M. L. J. Chromatogr., A 2001, 911, 1-12. (31) MacNair, J. E.; Patel, K. D.; Jorgenson, J. W. Anal. Chem. 1999, 71, 700708.

window (determined by ∆c) (see eq 2). We compared the separation windows supplied by porous (1.4 µm) and nonporous (1.5 µm, Micra Scientific, Northbrook, IL) C18-bonded silica particles for very slow gradient separation of a S. oneidensis tryptic digest, as shown in Figure 5. Elution of most peptides could be correlated between the two columns, and on average, the porous particle packed column resulted in ∼1.6-fold longer retention than the nonporous particle packed column. If porous and nonporous particle packed columns have the same reduced plate heights,33 eq 4 indicates that an ∼7-fold increase in pressure would be needed to compensate for the 40% reduction in peptide elution window using nonporous particle packed columns. Thus, the peak capacity for nonporous particle packed columns can be roughly estimated as

Cp ≈ 110xL/dp

(6)

According to eq 6, a 38-cm-length column packed with 1-µm nonporous particles should provide a peak capacity of ∼650, roughly consistent with that reported by Jerkovich et al (i.e., a Cp of 500).34 Using nonporous particles to obtain the peak capacities (32) Tolley, L.; Jorgenson, J. W.; Moseley, M. A. Anal. Chem. 2001, 73, 29852991. (33) Mellors, J. S.; Jorgenson, J. W. Anal. Chem. 2004, 76, 5441-5450. (34) Jerkovich, A. D.; Mellors, J. S.; Jorgenson, J. W. LC-GC 2003, 21, 600610.


3095

Figure 4. The 50-min separations of peptides and metabolites by 20 kpsi RPLC-MS. Conditions: 20 cm × 50 µm i.d. capillary packed with 1.4-µm porous particles as the LC column. (A) 0.1 µg of the S. oneidensis global tryptic digest (the data were collected after 10-min gradient) and (B) the S. oneidensis metabolomic sample (the sample was diluted by 4-fold prior to analysis; the data were immediately collected after gradient). In both experiments, the sample was directly loaded to the RPLC column without use; microSPE to eliminate the influence of the microSPE on short column efficiency (see ref 8). FTICR MS scan speed: 1.5 s/scan. Other conditions are described in the text.

of 1000-1500 requires much higher pressures. For example, a peak capacity of 1000 can be generated by 1-µm nonporous particle packed 90-cm columns, but would require >80 kpsi pressure; while obtaining peak capacities of 1500 with 0.5-µm nonporous particles would require pressures of >300 kpsi. Protein Identification and Metabolite Detection Power Supplied by Chromatographic Peak Capacities of >1000. Table 2 lists the numbers of peptides and proteins identified by ion trap (LCQ) MS/MS from use of various peak capacities of columns. The same analysis time of 24 h was used for these experiments to exclude influence of the number of acquired MS/ MS spectra on the identification. These columns enabled ∼1000 different S. oneidensis proteins to be identified through assignment of >4500 different peptides in each single 20 kpsi RPLC-MS/MS experiment, even though the physical-chemical properties may be somewhat different for each of the three sizes of packing particles and their associated gradient LC peak capacities vary from 1000 to 1500. The observation of no significant difference in the number of peptide and protein identifications among these columns may be due to the minor influences of the properties of different size particles on peptide detection under conditions of long (24 h) MS/MS analysis times and high peak capacity separations, as these conditions enabled most of the detectable peptides to be observed based on the sensitivity provided by the 3096 Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

MS/MS and the loaded sample amount (i.e., 15 µg). Increasing the sample amount to 30 µg was found having little influence on the number of peptides/proteins detected. Approximately 20% of the S. oneidensis proteome,35 which contains 4898 predicted open reading frames (ORFs), identified in a single 20 kpsi RPLC-LCQ MS/MS run makes this technology super to the conventional 2DLC/LC-MS/MS approach with the same type mass spectrometer for the detection,12 where three-cycle 2D-LC/LC-MS/MS runs for the enriched three subproteomes resulted in a total of 5540 peptide and 1484 protein identifications, covering 24% of the possible proteome proteins. Shortening the 3-µm porous particle packed column to 15 cm (operated at 2 kpsi pressure) detected 2360 different peptides from 691 proteins within the same analysis time, evidencing that the use of the long column doubled the number of identified peptides, increased the number of detected proteins by 47%, and improved the ratio of peptides/proteins from 3.3 to 4.6. These improvements become more significant for samples containing many low-abundance proteins of interest. For example, the 1.4-µm porous particle packed column analysis identified 78 different plasma proteins through assignment of 285 different peptides in 50 min, compared to 35 proteins and 105 peptides in 100 min using a 10-cm column packed with the 3-µm (35) Heidelberg, J. F.; et al. Nat. Biotechnol. 2002, 20, 1118-1123.

Figure 5. Correlation of peptide elution on the C18-bonded porous silica (PS) and nonporous silica (NPS) particle packed capillary columns. Conditions: for the porous particle packed column experiment, 40 cm × 50 µm i.d. capillary packed with the 1.4-µm porous C18 particles as the LC column and 4 cm × 150 µm i.d. capillary packed with 3-µm porous C18 particles as the microSPE column. The separation was completed in 12 h for 10 µg of the S. oneidensis tryptic digest sample. For the nonporous particle packed column experiment, the same gradient and column dimensions as for the porous column experiment were used, except 1.5-µm nonporous C18 particles were used in the LC and the microSPE columns. To eliminate the influence of sample overloading (sample overloading slightly reduces retention times), 1 µg of the S. oneidensis tryptic digest sample was loaded on the nonporous particle packed column compared to 10 µg of the sample on the porous particle packed column. A LCQ ion trap MS/MS was used for the analyses in which 561 peptides were commonly identified in the two experiments (assigned according to the criteria reported in ref 12). The earliest peptide to elute from the porous particle packed column experiment was adjusted to the origin to normalize the peptide elution windows, and the mass spectrum number was directly used to represent the elution time (the same MS and MS/MS scanning rates were used for the two experiments). Other conditions are described in the text.

porous particles under the same LCQ MS/MS detection conditions. That is, the use of high peak capacity RPLC columns for human plasma proteomic analysis at least doubled the number of identified proteins, tripled the number of identified peptides, and shortened the analysis time by 50% compared to the traditional columns that are commonly used for most current proteomic analysis. Table 3 gives the numbers of proteins and peptides identified for the proteomic sample using the 20 kpsi RPLC coupled with a linear (LTQ) ion trap for MS/MS (for comparison, data obtained on the LCQ are also given). In a single 12-h, 20 kpsi RPLC-MS/ MS experiment, >2000 distinct proteins were identified (covering >40% of all predicted proteins) based on assignment of >12 000 different tryptic peptides (of the assigned peptides, ∼99.5% are tryptic), which doubled the number of proteins obtained using an LCQ ion trap for MS/MS and increased the number of peptide identifications by 2.8-fold. This coverage improvement is attributed to the higher LTQ sensitivity, since doubling the analysis time of the LCQ ion trap MS/MS analysis time from 12 to 24 h (i.e., doubling the number of spectra) did not significantly increase the numbers of identified peptides and proteins (see Tables 2 and 3). The analysis reproducibility was 89% for proteins and 78% for peptides, higher than those obtained from use of LCQ MS/MS

Figure 6. Global proteomic analysis time and coverage supported by the 20 kpsi RPLC-ion trap MS/MS within 12 h. Conditions: the same conditions as for the porous particle packed column experiment in Figure 5 were used, except both conventional and linear ion trap instruments (i.e., LCQ and LTQ) were used for peptide identification. The gradient was adjusted as described in the text. Table 3. Capability for the Gradient 20 kpsi RPLC with Conventional and Linear Ion Trap MS/MS Analyses in 10-12 h, and Corresponding Protein and Peptide Detection Reproducibilitiesa peptides and proteins identified from a S. oneidensis global tryptic digest conventional ion trap (LCQ) MS run 12 h 12 h duplicate overlaps reproducibilityc 10 h 10 h duplicate overlaps reproducibilityc

linear ion trap (LTQ) MS

proteinsb

peptidesb

proteinsb

peptidesb

966 964 784 81% 951 956 775 81%

4467 4265 2965 68% 4207 4421 2862 66%

2036 2016 1796 89% 1961 1983 1721 87%

12487 12366 9731 78% 11995 12300 9398 77%

a A 40 cm × 50 µm i.d. × 1.4 µm d column was used for these p investigations; the data listed are those obtained in individual experiments for 10 µg of the S. oneidensis global tryptic digest. b The numbers listed are for distinct proteins and different peptides identified with criteria reported in ref 12. c The method reproducibility values were calculated according to the overlapped proteins and peptides, not simply the numbers of proteins and peptides detected; the variance of numbers of peptides and proteins was as small as 1-5%.

with the same identification criteria (see Table 3). Decreasing the analysis time to 10 h slightly decreased the number of proteins/ peptides identified, and Figure 6 shows the further reduction in the peptide and protein identifications with shortening the analysis times. A 4-h 20 kpsi RPLC-(LTQ) MS/MS analysis identified ∼1750 proteins, i.e., ∼87% of the proteome proteins detectable (i.e., 2000 proteins), through assignment of 8709 different peptides; Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

3097

Figure 7. The 20 kpsi RPLC-linear ion trap LTQ MS/MS base peak chromatograms for global proteomic analyses with analysis times ranging from 12 (upper) to 2 h (lower). Experimental conditions are the same as for Figure 6.

the 2-h experiment identified 1200 proteins, i.e., 60% of the proteome proteins detectable, through assignment of 5944 different peptides. The high separation quality shown in Figure 7 suggests that the reduced identifications in the 2-h analysis were primarily due to the MS/MS analysis speed. Comparing with results given in Table 2, the coverage obtained in 2-h, 20 kpsi RPLC-(LTQ) MS/MS was still not matched by a 24-h LCQ MS/ MS analysis. Additionally, the ratio of peptides/proteins identified was ∼6 for the LTQ compared to ∼4.5 for the LCQ ion trap MS/ MS. For the metabolomic sample, a single 20 kpsi RPLC-FTICR MS run (shown in Figure 3) detected 5036 putative metabolites. Proteomic Analysis Dynamic Range Supplied by Chromatographic Peak Capacities of >1000. The proteomic analysis dynamic range was estimated using a human blood plasma proteome sample in which abundances for some proteins are relatively well established.36 Figure 8 shows the base peak chromatogram and two MS/MS spectra for low-abundance full and partial tryptic peptides from the human plasma. This experiment resulted in identification of 572 different human plasma proteins (after excluding 263 sequences from antibodies) from (36) Anderson, N. L.; Anderson, N. G. Mol. Cell. Proteomics 2002, 1, 845-867.

3098


the nondepleted human plasma sample11 by assignment of 2244 peptides through searching against the human database (ftp:// ftp.ncifcrf.gov) and filtering according to conventional MS/MS criteria.12 The number of identified proteins is ∼7-fold greater than that obtained using 10 kpsi RPLC-LCQ MS/MS and is equivalent to the ultra-high efficiency SCXLC-RPLC-LCQ MS/MS.11 Proteins having concentrations of 103-fold lower than the most abundance human serum albumin (HSA, ∼50 mg/mL) were identified by detection of >2 peptides/protein, while 104-fold lower concentration proteins were identified by assignment of single peptides, but with high SEQUEST scores, e.g., XCorr 3.017 and ∆Cn 0.1266 for the M2+ peptide R.EMEDTLNHLK.F identified from 2 to 4 µg/mL of insulin-like growth factor binding protein 3. The detected lowest concentration human blood protein thyroglobulin (3-42 ng/mL), which is 106-fold lower in abundance than HSA, was detected by assignment of a single M2+ partial tryptic peptide. The three MS/MS spectra for this partial tryptic peptide yielded SEQUEST scores of XCorr 2.2058 (∆Cn 0.1137), XCorr 1.9898 (∆Cn 0.107), and XCorr 1.8916 (∆Cn 0.055) with only one score higher than the threshold. If this situation is considered as defining the protein detection sensitivity limit, the 20 kpsi RPLC-(LTQ) MS/ MS is estimated to provide ∼6 orders of magnitude protein

Figure 8. Human plasma tryptic digest analysis by the 20 kpsi RPLC-linear ion trap (LTQ) MS/MS that resulted in identification of 853 human plasma proteins and antibody sequences (peptide assignment was completed according to the criteria reported in ref 12) in 12 h. The MS/MS spectra show the detection of low-abundance human plasma proteins. Conditions: the same conditions as for the porous particle packed column experiment in Figure 5 were used, except 5 µg of the human plasma global tryptic digest was loaded to the system. Other conditions are described in the text.

detection dynamic range (i.e., detection of human blood plasma proteins having concentrations from 3-42 ng/mL to ∼50 mg/ mL). The 20 kpsi RPLC System Convenience and Robustness of Operation. In the present 20 kpsi RPLC system, requirements for the LC pumps are simplified since the mobile gradients are generated in a static mixer and two ISCO 20 kpsi pumps have been used for >12 months without failure (e.g., leaking). The 20 kpsi switching valves (Valco) have also been successfully used over the same time period; however, they have occasionally leaked from the back pressure chamber when operated at lower pressures, such as 2000 distinct proteins with a reproducibility of 89% through assignment of >12 000 different tryptic peptides from a proteomics sample and >5000 different metabolite candidates from a global metabolomic sample. The separation quality also enabled ion trap MS/ MS analyses to simultaneously identify proteins at concentration Analytical Chemistry, Vol. 77, No. 10, May 15, 2005

3099

differences of 6 orders of magnitude lower and allow identification of up to 1200 proteins in 2 h. The utilization of even higher LC pressure (i.e., >20 kpsi) would be most effective in combination with smaller (e.g., e1 µm) porous particles, as 20 kpsi is already sufficient to deliver the mobile phase through a 40-cm column packed with 1.4-µm porous particles. Nonporous particles are less suitable for achieving high chromatographic peak capacities for peptides because of the reduced separation window arising from their smaller retention with additional disadvantage of the low sample capacity. The faster and higher peak capacity separations that result from the use of smaller porous particles and higher pressures are expected to significantly improve and extend proteomics and metabolomics measurements. ACKNOWLEDGMENT We thank the NIH National Center for Research Resources (RR 018522) and the Environmental Molecular Sciences Labora-

3100


tory at the Pacific Northwest National Laboratory for their support of portions of this research. We also thank Phenomenex and Waters for donation of the 2- and 1.4-µm porous particles, respectively, and also ISCO and Valco for timely providing the 20 kpsi LC pumps and switching valves, respectively. Pacific Northwest National Laboratory is operated by Battelle for the U.S. Department of Energy through Contract DE-ACO6-76RLO 1830. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review November 16, 2004. Accepted March 3, 2005. AC0483062

MS with Chromatographic Peak

Recommend Documents