Anal. Chem. 2000, 72, 2154-2159
High-Efficiency Capillary Isoelectric Focusing of Peptides Yufeng Shen, Scott J. Berger, Gordon A. Anderson, and Richard D. Smith*
Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, P.O. Box 999, Richland, Washington 99352
Separation and identification of peptides derived from an individual protein or from protein mixtures is becoming increasingly important in the burgeoning field of proteomics.1 Both capillary zone electrophoresis (CZE)2-7 and liquid chromatography (LC)8-10 have been widely developed for separations of peptides. A significant advantage of CZE for separations of peptides is its high separation efficiency.2-7 A theoretical plate number of ∼105 can be generally achieved using high electric field strengths.
A major disadvantage of CZE, however, is the limited sample injection volume. For a typical capillary column with a dimension of 50 µm i.d. × 30-100 cm length, the capillary column has a volume of 0.6-2 µL. If the allowable injected sample volume is ∼1% of the total capillary column volume, i.e., for obtaining highefficiency (∼105 plates) separations,11 the injected sample volume should be 1000 atm) to deliver the mobile phase through a packed chromatographic bed of very small particles (∼1 µm) can provide column efficiencies of up to ∼105 plates.16 When a mobile-phase composition gradient is used, the allowable injected sample volume can be increased because the sample can be initially concentrated at the column inlet head in the low solvating power mobile phase and then mobilized by the solvent gradient. As base-acid amphoters, the isoelectric point (pI) of peptides is determined by the composition (and sequence) of amino acid residues.17 Separations based upon the pI values, i.e., using isoelectric focusing (IEF), have been performed in both planar gel18 and capillary (CIEF) formats.19 In IEF, the separation is accompanied by sample concentration; that is, the sample separation and concentration occur simultaneously during the focusing
(1) Yates, J. R., III J. Mass Spectrom. 1998, 33, 1-19. (2) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (3) Moseley, M. A.; Deterding, L. J.; Tomer, K. B.; Jorgenson, J. W. Anal. Chem. 1991, 63, 109-114. (4) Huang, M.; Yi. G.; Bradshaw, J. S.; Lee, M. L. J. Microcolumn Sep. 1993, 5, 199-205. (5) Schmalzing, D.; Piggee, C. A.; Foret, F.; Carrilho, E.; Karger, B. L. J. Chromatogr., A 1993, 652, 149-159. (6) Righetti, P. G.; Bossi, A. Anal. Chim. Acta 1998, 372, 1-19. (7) Hutteerer, K. M.; Jorgenson, J. W. Anal. Chem. 1999, 71, 1293-1297. (8) High-Performance Liquid Chromatography, Advances and Perspectives; Horva´th, C., Ed.; Academic Press: New York, 1983; Vol. 3. (9) HPLC of Proteins, Peptides, and Polynucleotides; Hearn, W. T. W., Ed.; VCH Publishers: New York, 1991. (10) High-Performance Liquid Chromatography of Peptides and Proteins: Separation, Analysis, and Conformation; Mant, C. T., Hodges, R. S., Eds.; CRC Press: Boca Raton, FL, 1991.
(11) Chien, R.-L. Sample Introduction and Stacking. In High Performance Capillary Electrophoresis, Theory, Techniques, and Applications; Khaledi, M. G., Ed.; John Wiley & Sons: New York, 1998. (12) Stegehuis, D. S.; Tjaden, U. R.; van der Greef, J. J. Chromatogr. 1992, 591, 341-349. (13) Foret, F.; Szoko, E.; Karger, B. L. J. Chromatogr. 1992, 608, 3-12. (14) Strausbauch, M. A.; Landers, J. P.; Wettstein, P. Anal. Chem. 1996, 68, 306-314. (15) Tong, W.; Link, A.; Eng, J. K.; Yates, J. R., III. Anal. Chem. 1999, 71, 22702278. (16) MacNair, J. E.; Patel, K. D.; Jorgenson, J. W. Anal. Chem. 1999, 71, 700708. (17) Cohn, E. J.; Edsall, J. Proteins, Amino Acids, and Peptides as Ions and Dipolar Ions; Hafner Publishing Co.: New York, 1965. (18) Righetti, P. G. Isoelectric Focusing: Theory, Methodology and Applications; Elsevier Biomedical Press: Amsterdam, 1983. (19) Hjerten, S.; Zhu, M.-D. J. Chromatogr. 1985, 346, 265-270.
Several approaches are presently being developed for global proteome characterization that are based upon the analysis of polypeptide mixtures resulting from digestion of (often complex) mixtures of proteins. Improved methods for peptide analysis are needed that provide for sample concentration, higher resolution separations, and direct compatibility with mass spectrometry. In this work, methods for the high-efficiency capillary isoelectric focusing (CIEF) separation of peptides have been developed that provide for simultaneous sample concentration and separation according to peptide isoelectric point. Under typical nondenaturing CIEF conditions, peptides are concentrated ∼500-fold, and peptides present at 100 successive runs. The stable zones observed after ∼80 min focusing suggest that limited electroosmotic flow exists under typical CIEF conditions with these capillaries.24 The entire coated capillary was filled with the sample premixed with 1% Pharmalyte (pH range of 3-10, Pharmacia Biotech, Uppsala, Sweden). Ammonium hydroxide (1%, w/w, pH ∼10.7) and acetic acid (1%, w/w, pH ∼2.5) were used as the catholyte and anolyte, respectively, since these are desirable solutions for future CIEF-MS studies. The CIEF experiments were carried out at a voltage of 20 kV for focusing, and hydrodynamic mobilization was obtained by elevating the anode reservoir relative to the cathode reservoir. The focusing and mobilization processes were monitored by UV absorbance at 280 nm. Standard protein pI markers (Sigma) were used for calibrating of CIEF separations and mobilization linear velocity. CZE Experiments. Poly(vinyl alcohol) (PVA, average MW 85 000-146 000, Aldrich) coated fused-silica capillaries (65 cm × 50 µm i.d. × 190 µm o.d.) were prepared for CZE experiments according to a previously described procedure.25 The sample was hydrodynamically introduced into the coated capillary by inserting the capillary into the sample solution vial and elevating the vial by 30 cm for 1 min, yielding an injection volume of ∼10 nL. The CZE was carried out on a home-built instrument using a phosphate buffer solution (0.1 M, pH 2.5) by applying a voltage of 20 kV (Glassman High Voltage supply). The separation was monitored at 215 nm (Spectra 100 UV/visible detector, Spectra-Physics). (24) Shen, Y.; Smith, R. D. J. Microcolumn Sep. 2000, 12, 135-141. (25) Gilges, M.; Kleemiss, M. H.; Schomburg, G. Anal. Chem. 1994, 66, 20382046.
Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
2155
Figure 1. UV detection sensitivity for angiotensin I under nondenaturing CIEF conditions. Conditions: 65 cm × 50 µm i.d. HPCcoated fused-silica capillary columns; sample solutions contain 1% (v/v) Pharmalyte (pH 3-10) and angiotensin I with a final concentration of 1 µg/mL; 20 kV; NH4OH (1%, v/v, pH ∼10.7) as the catholyte; acetic acid (1%, v/v, pH ∼2.5) as the anolyte; UV (280 nm) detection; 40 min focusing; gravity mobilization from anode to cathode (7 cm height difference). Time is given from the start of mobilization.
Packed Capillary RPLC/UV and RPLC/ESI-MS Experiments. For separations of protein tryptic digests, a fused-silica capillary column (16 cm × 150 µm i.d. × 360 µm o.d.) was packed with 3-µm-diameter nonporous octadecyl-bonded silica particles (ODS, Micra Scientific, Northbrook, IL). For separation of yeast cytosol tryptic digest, a long capillary column (100 cm × 150 µm i.d. × 360 µm o.d.) was packed with 5-µm porous ODS particles (300-Å pores, Alltech, Deerfield, IL) to achieve a high total column efficiency. The capillaries were packed using a liquid slurry packing method26 with H2O/ACN (75:25, v/v) as the solvent at 60 MPa using an Isco micropump (µL-500, Isco, Lincoln, NE). A stainless steel screen (2-µm pores, Valco) was used to hold the particles in the packed capillary column, and the packed capillary column connections were made using PEEK tubing and zero dead volume unions (Valco, Houston, TX). A valve injector with a 200nL sample loop (Valco) was used for sample introduction. An Isco pump was used to deliver the mobile phase through the separation column. A 16-cm packed capillary column containing 3 µm nonporous ODS demonstrated a total column efficiency of 19 000 plates. A 100-cm column containing 5-µm porous ODS yielded a total column efficiency of 86 000 plates at a mobile-phase linear velocity of ∼0.11 cm/s (using benzene as the test solute). For packed capillary RPLC/UV of peptides, the separation was monitored at 215 nm (Spectra 100 UV/visible detector, SpectraPhysics). For packed capillary RPLC/ESI-MS, an LCQ ion trap mass spectrometer (Finnigan, San Jose, CA) was used. ESI (2000 V) was performed in the positive ion detection mode with a m/z scan range of 200-2000 using a heated capillary inlet temperature of 200 °C. Spectra were acquired with 2 microscans/scan. RESULTS AND DISCUSSION Concentration Coefficient and UV Detection for Peptides under CIEF Conditions. A wavelength of 215 nm is commonly used for UV detection of most peptides in HPLC and CZE. However, the distribution of carrier ampholytes after isoelectric focusing produces nonuniform absorption at this wavelength. We (26) Andreolini, F.; Borra, C.; Novotny, M. Anal. Chem. 1987, 59, 2428-2432.
2156 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
Figure 2. CIEF measurement of peptide pI values under nondenaturing conditions. (A) CIEF of peptide mixture (each peptide with a concentration of ∼3 µg/mL); (B) CIEF of standard protein pI markers (each at a concentration of ∼5 µg/mL); (C) CIEF of mixed mixtures used in (A) and (B) (1:1, v/v); 6 cm height difference for gravity mobilization. Other conditions are the same as in Figure 1. Peak identifications in (B) and (C): (1) myoglobin (from horse heart, pI ∼7.2), (2) myoglobin (from horse heart, pI ∼6.8), (3) carbonic anhydrase I (from human erythrocytes, pI ∼6.6), and (4) carbonic anhydrase II (from bovine erythrocytes, pI ∼5.9). The pI values of standard protein pI markers are those given by Sigma. Calc, calculated pI value; Exp, experimental pI value calibrated using standard protein pI markers.
measured the UV absorption at 215, 254, and 280 nm under CIEF conditions using a carrier ampholyte from Phamacia (pH 3-10) without sample, and only 280 nm proved practical for detection. This wavelength is obviously not optimal for the detection of most peptides. Due to the focusing process, CIEF sample zones have concentrations much higher than for the original solution. Figure 1 shows the detection of 1 ng/µL angiotensin I at 280 nm under CIEF conditions. For this oligopeptide, a very narrow peak was obtained (peak width of ∼8 s at the base). On the basis of standard protein pI markers, and the mobilization linear velocity of 0.9 cm/ min, the ∼8 s peak width at base corresponds to a zone length of ∼0.12 cm for the 65-cm-length capillary column. This corresponds to a sample concentration coefficient of ∼540 (65-cm column
Table 1. Major Peptides for a BSA Tryptic Digest Determined Using Packed Capillary RPLC/MSa fragment
RTb
MWc
sequence
fragment
RTb
MWc
sequence
t29 t64 t25 t76 t35 t74 t29,2 t32.2 t13 t27.2 t34 t18 t24,2 t4 t12 t2,2 t25.2 t64.2 t42 t66 t77 t61 t18,2 t28.2 t35.2 t6 t41 t1,2
43.2 46.6 29.1 39.2 1.2 27.9 27.8 37.5 56.1 42.0 1.3 26.1 48.4 15.0 54.8 33.7 40.0 46.2 14.5 53.7 55.3 49.1 59.5 58.4 80.6 38.4 71.1 43.2
507.3 659.3 702.4 724.2 788.5 817.4 819.5 846.5 885.4 917.5 921.5 926.5 959.6 973.5 976.4 985.5 986.6 987.6 1014.5 1023.4 1049.3 1051.4 1082.6 1137.6 1152.7 1162.6 1176.6 1192.6
FGER TPVSEK VLASSAR CCAADDK LVTDLTK ATEEQLK FGERALK LSQKFPD DDSPDLPK LRCASIQK AEFVEVTK YLYEIAR EKVLASSAR DLGEEHFK NECFLSHK SEIAHRFK VLASSARQR TPVSEKVTK SHCIAEVEK CCTESLVNR EACFAVEGPK CCTKPESER YLYEIARR CASIQKFGER LVTDLTKVHK LVNELTEFAK ECCDKPLLEK DTHKSEIAHR
t44.2 t26,3 t3,2 t33,2 t7 t65,2 t10 t75 t27,3 t30,3 t55 t51 t31,3 t14 t46 t37 t22,3 t48,2 t17,3 t11,2 t39,2 t71,3 t49,2 t47,3 t63,3 t23.4 t64,3
70.0 26.2 55.4 59.2 10.4 76.5 19.1 60.0 71.5 59.6 80.6 66.2 42.7 58.8 49.5 50.1 53.7 61.1 81.7 76.5 81.7 53.3 62.9 54.8 66.7 70.8 53.9
1196.6 1201.7 1248.6 1293.7 1348.5 1351.7 1363.5 1398.7 1406.7 1456.9 1478.8 1496.6 1516.9 1518.7 1566.7 1577.6 1587.8 1594.9 1600.9 1615.7 1626.8 1632.0 1699.8 1751.0 1810.0 1874.0 1992.9
DVCKNYQEAK QRLRCASIQK FKDLGEEHFK FPKAEFVEVTK TCVADESHAGCEK VTKCCTESLVNR ETYGDMADCCEK TVMENFVAFVDK LRCASIQKFGER ALKAWSVARLSQK LGEYGFQNALIVR DDPHACYSTVFDK AWSVARLSQKFPF LKPDPNTLCDEFK DAFLGSFLYEYSR ECCHGDLLECADDR GACLLPKIETMREK HPEYAVSVLLRLAK FWGKYLYEIARR QEPERNECFLSHF YICDNQDTISSKLK KQTALVELLKHKPK LAKEYEATLEECCAK RHPEYAVSVLLRLAK LCVLHEKTPVSEKVTK IETMREKVLASSARQR TPVSEKVTKCCTESLVNR
a Mobile-phase gradient from A (ACN/H O, 10:90, v/v, 0.1% TFA) to B (ACN/H O, 75:25, v/v, 0.1% TFA) in 80 min. Other conditions are 2 2 described in the text. b Retention time (min). c Calculated molecular weight (and in agreement with MS measurements).
length/0.12-cm zone length). Clearly, even though the UV detection wavelength is not optimal, sensitive detection is still obtainable (0.5 ng/µL sample detected with a signal/noise of >3). Our previous studies of CIEF for proteins found that ∼20 min focusing was sufficient for obtaining high-resolution separtations.27 In this work, we found that long focusing times (∼40 min) were beneficial for obtaining high-resolution CIEF separations of test peptides, and all experiments used 40 min for focusing. Total analysis times were less than 100 min. CIEF Measurement of pI Values for Peptides. The isoelectric point of amphoters is determined by the pKa values of carboxylic acid and amine groups.17 In most theoretical models for calculating pI values of peptides,28 only the contributions of the carboxylic acid and amine groups to the pI are considered. Improved models that properly account for the peptide’s threedimensional structure influence on its net pI values are not yet available. The pI values for peptides can be experimentally determined (as for proteins) by either titration or electrophoretic methods,17,29 but CIEF using standard pI markers is much more convenient since many peptides can be studied in a single separation. Figure 2 shows the experimental CIEF measurement of pI values for angiotensin I and γ-endorphin. The measurement was carried out by mixing a two-component peptide mixture with a mixture of standard protein pI markers. Theoretical (calculated) (27) Shen, Y.; Xiang, F.; Veenstra, T. D.; Fung, E. N,; Smith, R. D. Anal. Chem. 1999, 71, 5348-5353. (28) http://trex.musc.edu/manuals/Unix/isoelectric.html. (29) Gianazza, E.; Miller, I.; Eberini, I.; Castiglioni, S. Electrophoresis 1999, 20, 1325-1338.
pI values are 7.7 and 6.4 for angiotensin I and γ-endorphin, respectively.28 However, the pI values were measured to be 7.0 and 5.7, respectively, under the present nondenaturing conditions. The pI values of standard protein pI markers were experimentally determined by the manufacturer (Sigma), and these values are used in the present work. The deviation of ∼0.7 of the calculated pI value of the peptides from the experimentally measured values may result from various interactions between carboxylic acid and amine groups, which are neglected in the calculations. Under the denaturing conditions (9.8 M urea), the calculated pI values agree with those measured by IEF.6 Clearly the three-dimensional structures of peptides under CIEF conditions lead to pI values that are not well treated by present theoretical approaches. Comparisons of calculated and experimental pI measurements for peptides were further investigated using a BSA tryptic digest that was also analyzed using LC/MS (Table 1) to identify the major polypeptides (some peptides are lost in sample handling and cleavage at certain sites can be inefficient). The calculated pI values (from pI 3 to 9) for the detected peptides is shown in Figure 3 (bottom). The calculated pI distributions of the listed digestion product tend to cluster in the pI ranges of 8.1-8.3, 6.0-6.1, and 4.4-4.7. The CIEF experimental results (Figure 3, top) yield pI distributions for major components that cluster primarily in the pI ranges of 7.5-8.3, 6.1-6.6, and 4.6-5.1. Thus, we obtain qualitative agreement but also note that calculated basic peptide pI values shift to higher pH (i.e., lower pI value), and acidic peptides appear shifter to lower pH. Again, this observation is consistent with some retention of higher-order structure due to the nondenaturing conditions of the separation. Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
2157
Figure 3. CIEF electrophoregram for a BSA tryptic digest. Bottom: the calculated pI distribution of major peptides detected from a BSA tryptic digest using LC/MS; the vertical axial represents the peptide number detected at the specific pI point. Top: the pI distribution of the BSA (3.8 mg/mL) tryptic products measured by CIEF; the pI scale was calibrated using standard pI markers. Other experimental conditions are the same as in Figure 2.
Separation Power of CIEF for Peptides. A complex peptide mixture from a yeast cytosol tryptic digest was used to demonstrate high-efficiency CIEF of peptides. Figure 4 shows the typical reproducibility obtained for CIEF separations of the cytosol digest sample. (The sample concentration was 0.31 µg/µL and the pI scale was calibrated by adding standard pI markers to the sample solution and performing CIEF under the same experimental conditions.) Very well resolved peaks having widths at the halfheight of ∼3 s for the less abundant components were reproducibly achieved (see Figure 4). The mobilization linear velocity used for detection after focusing was ∼0.9 cm/min, as determined using standard pI markers. A peak width at half-height of ∼3 s corresponds to a zone length of ∼0.045 cm in the 65-cm-length capillary and ∼0.005 pI unit for the pH range of 3-10 used for these separations. Thus, an estimated capacity of 720 resolved peaks (65 cm/2 × 0.045 cm) can be projected for baseline separations (Rs ∼1.5). This corresponds to a peak capacity of ∼1000 (Rs ∼1.0).30 The minimal pI difference that can be separated is ∼0.01 [2 × 0.045 cm × (10 - 3) pH/65 cm]. These results indicate CIEF is potentially the most powerful microcolumn technique yet reported for separation of complex peptide mixtures. Comparison of HPLC, CZE, and CIEF for Separations of a Complex Yeast Cytosol Digest. To compare various micro(30) Giddings, J. C. Unified Separation Sciences; John Wiley & Sons: New York, 1991.
2158 Analytical Chemistry, Vol. 72, No. 9, May 1, 2000
Figure 4. Example showing the reproducibility of CIEF separations obtained for a yeast cytosol tryptic digest. Conditions: the yeast cytosol (5 mg/mL) tryptic digest was diluted 16-fold with 1% Pharmalyte (pH 3-10); the pI scale was calibrated using standard pI markers. Other conditions are the same as in Figure 2. (A) run 1 and (B) run 3.
column techniques for complex peptide mixture separations, a yeast cytosol tryptic digest sample was also separated under typical CZE and RPLC conditions. For CZE, a neutral PVA-coated capillary column was used since previous reports indicated the ability of this type of coated capillary to yield high-efficiency separations due to the suppression of electroosmotic flow and reduced interactions between sample and capillary inner wall (even for acidic and basic proteins).25 Figure 5A shows a CZE separation for a sample concentration 16 times higher than that used for CIEF. A relatively high-efficiency separation was obtained in ∼1 h. However, a major limitation was the low detection sensitivity (the detection baseline shifted at 215 nm under the experimental conditions of high-sensitivity range of the UV detector). In our experiments, ∼10 nL (∼0.8% of the total capillary column volume) of the sample (5 µg/µL) was injected. This is a widely recognized limitation of CZE that has led to extensive efforts to develop improved sample preconcentration methods (e.g., using solidphase extraction15). Figure 5B shows a capillary RPLC separation of the yeast cytosol tryptic digest. The sample concentration used was again 16 times greater than that for CIEF. We used a 100-cm-long capillary column packed with 5-µm porous ODS particles in order to obtain high column efficiency (at the cost of speed). This packed capillary column provided a total column efficiency of
concentration of 5 µg/µL, the detection sensitivity was similar to that obtained in CZE but lower than that for CIEF. Obviously, increasing injected sample volume and enhancing the pump pressure limit (e.g., the use of ultrahigh pressure pump) would improve the detection sensitivity and column efficiency and shorten the analysis time, respectively. However, compared with these conventional CZE and RPLC separations, CIEF provides higher efficiency separations with improved detection sensitivity and simultaneously yields useful pI information for the peptides. Similar to LC and CZE, CIEF can be coupled with on-line MS.20,21 Through manipulation of the sheath liquid interface, both basic or acidic spray conditions can be generated to facilitate the analysis of both positive and negative ions, although the positive ion mode is preferred for reasons of sensitivity and generally affords broad coverage for peptides. Combination of MS with each of these separation methods can provide complementary information to assist peptide identification from extremely complex mixtures, such as those relevant to proteomics.
Figure 5. CZE and packed capillary RPLC separations for a yeast cytosol tryptic digest. (A) CZE conditions: 65 cm × 50 µm i.d. PVAcoated fused-silica capillary columns; injecting ∼10 nL of the yeast cytosol (5 mg/mL) tryptic digest using a hydrodynamic method; 20 kV; 0.1 M phosphate buffer (pH ∼2.5); 215-nm detection. (B) Packed capillary RPLC conditions: 100 cm × 150 µm i.d. fused-silica capillary column packed with 5-µm porous ODS; sample injection volume of 200 nL of the yeast cytosol (5 mg/mL) tryptic digest; mobile-phase gradient from A (H2O, 0.1% TFA) to B (H2O/ACH, 10:90, v/v, 0.1% TFA) in 180 min; 215-nm detection.
∼86 000 plates. The column back pressure was 60 MPa (8600 psi) at a mobile-phase linear velocity of ∼0.1 cm/s. This is the longest packed capillary column (containing 5-µm particles) for which commercially available pumps can deliver the mobile phase at optimal linear velocities. A high-efficiency separation was obtained in ∼170 min. Compared with CZE and CIEF, capillary RPLC took significantly longer (∼170 min) and extra time (∼3 h) was also required to wash and recondition the column with mobile phase A after the gradient. Using a 200 nL sample loop and a sample
CONCLUSIONS Low-concentration (