On-Line Concentration of Proteins and Peptides in Capillary Zone

Anal. Chem. 2002, 74, 3899-3905

On-Line Concentration of Proteins and Peptides in Capillary Zone Electrophoresis with an Etched Porous Joint Wei Wei and Edward S. Yeung*

Ames Laboratory, United States Department of Energy, and Department of Chemistry, Iowa State University, Ames, Iowa 50011

A novel approach for on-line concentration of proteins and peptides in capillary electrophoresis (CE) is presented. A short section (∼0.5-1 cm) along the capillary was etched with HF. The etched section became a porous membrane that allowed electrical conductivity but prevented passage of the analyte ions. The capillary was isolated into two parts by the etched section. Thus, we were able to use three buffer vials to perform CE experiments in the capillary by applying high voltages independently. Concentration and separation were performed at the two respective regions. When high voltage was applied to the concentration capillary (between the inlet end and the etched section), proteins and peptides were concentrated at the etched portion, because the porous capillary wall allowed only small buffer ions to pass through and there was no electric field gradient beyond that point. After focusing, the narrow sample zone was introduced into the separation capillary (between the etched section and the outlet end) by hydrodynamic flow or by electroosmotic flow. Finally, conventional CE was carried out for separation of the analytes. Several different concentration schemes for proteins and peptides were successfully demonstrated by using this new approach. Capillary electrophoresis (CE) has become a routine separation tool and is widely used in different areas of chemistry and biochemistry; however, one of the drawbacks of CE is the poor detection limit due to the short optical path length across the capillary. Although laser-induced fluorescence in CE can be employed, it is still difficult to detect many analytes at low concentrations because of problems in labeling. On-line concentration is an alternative in CE for improvement of concentration detection limits and separation efficiencies.1-14 A variety of sample(1) (2) (3) (4) (5) (6) (7) (8)

Huang, X.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1988, 60, 375-377. Burgi, D. S.; Chien, R. L. Anal. Chem. 1992, 64, 1046-1050. Zhang, C. X.; Thormann, W. Anal. Chem. 1996, 68, 2523-2532. Burgi, D. S. Anal. Chem. 1993, 65, 3726-3729. Chien, R.-L.; Burgi, D. S. Anal. Chem. 1992, 64, 489A-496A. Foret, F.; Szoko, E.; Karger, B. L. J. Chromatogr. 1992, 608, 3-12. Foret, F.; Sustacek, V.; Vocek, P. J. Microcolumn Sep. 1990, 2, 229-233. Reinhoud, N. J.; Tjaden, U. R.; Greef, J. V.-D. J. Chromatogr. 1993, 641, 155-162. (9) Krivankova, L.; Gebauer, P.; Bocek, P. J. Chromatogr. A 1995, 716, 3548. (10) Foret, F.; Szoko, E.; Karger, B. L. Electrophoresis 1993, 14, 417-428. 10.1021/ac025612b CCC: $22.00 Published on Web 06/01/2002

© 2002 American Chemical Society

stacking methods have been reported. Field-amplified sample stacking (FASS)1-5 is based on the conductivity difference between the sample and the running buffer. The low conductivity sample matrix experiences a higher electric field strength, and analyte ions move faster in the matrix than in the background electrolyte (BGE). Stacking takes place at the boundary between the sample zone and the BGE; however, when the conductivity of the sample matrix is higher than that of the running buffer, the stacking effect is lost.4 Isotachophoresis (ITP) is another method for sample concentration in CE.6-10 The sample zone is sandwiched between leading ions and terminating ions with the electric fields at the three different zones adjusted to maintain the same velocity. ITP is often used for concentrating samples in which the conductivity is similar to that of the running buffer, because the leading ions and terminating ions can be also used as the running buffer. One online ITP concentration method is transient ITP.6,7 The sample containing the leading ions or terminating ions is introduced into the capillary by pressure. The focusing process is carried out by introducing the terminating or leading ions, respectively. After focusing, the running buffer is used for separation. However, when the sample zone is too long, it takes a long time to complete the focusing process. In addition, the effective length of the separation capillary is shortened by the length of the sample zone, and the separation resolution decreases. Other stacking methods have been reported that utilize the properties of proteins or peptides, such as size, zwitterions, etc. Hjerte´n et al.11,12 used isoelectric focusing in a pH gradient for the concentration of proteins. Wu et al.13 reported an on-line protein concentration method using a semipermeable hollow fiber. This method is based on the fact that large proteins cannot pass the fiber membrane. The proteins were injected through the membrane and concentrated; however, the fiber membrane construction is difficult and a complex setup is needed to make the whole system functional. Schwer et al.14 reported a peptide stacking method using a discontinuous buffer system. The sample was introduced between zones of OH- and H+. As OH- and H+ migrate toward each other, a zone of low conductivity is formed. Concentration occurs by FASS. Furthermore, OH- and H+ change (11) (12) (13) (14)

Hjerte´n, S.; Liao, J. L.; Zhang, R. J. Chromatogr. A 1994, 676, 409-420. Liao, J. L.; Zhang, R.; Hjerte´n, S. J. Chromatogr. A 1994, 676, 421-430. Wu, X.-Z.; Hosaka, A.; Hobo, T. Anal. Chem. 1998, 70, 2081-2084. Schwer, C.; Lottspeich, F. J. Chromatogr. 1992, 623, 345-355.

Analytical Chemistry, Vol. 74, No. 15, August 1, 2002 3899

the charges of the peptides. When peptides move to the H+ region, they acquire positive charges and migrate cationically. Conversely, when peptides meet OH-, they migrate anionically. In this way, the sample zone is concentrated additionally. Other preconcentration schemes have been reported for related separation modes, such as micellar electrokinetic chromatography15-17 and electrochromatography.18,19 Still other on-column preconcentrators have been reported.20,21 Recently, we developed a sample stacking protocol for one-step concentration of analytes in CE.22 A sharp pH gradient along the capillary was created dynamically by the electrolysis of water when voltage was applied to a platinum wire inserted into the capillary. Concentration of a large volume of injected analyte was accomplished by the change in analyte charge due to the dynamic pH gradient. In this work, we demonstrate a new approach for sample concentration in CE in a single capillary by using a short etched porous section. The etched section allows electrical conductivity but blocks the analyte ions so that CE experiments can be performed with three buffer vials. Many sample stacking concepts in CE can be easily implemented in this system, because sample concentration and separation are achieved within different parts of the capillary isolated by the etched section. EXPERIMENTAL SECTION Chemicals. β-Lactoglobulin B (BLGB), L-1-tosylamide-2-phenylethyl chloromethyl ketone(TPCK)-treated trypsin, lysozyme (from chicken egg white, pI ) 10.5), ribonuclease A (from bovine pancreas, pI ) 9.45) and R-chymotrypsinogen A (from bovine pancrease, pI ) 9.2), ammonium acetate, HPMC (hydroxypropylmethylcellulose, viscosity of 2% aqueous solution ∼4000 centipoise), calcium chloride, Trizma-base (tri[hydroxymethyl]aminomethane), triethylamine (TEA), phosphoric acid, acetic acid (HAc), hydrochloric acid, sodium hydroxide, and ammonium hydroxide were of analytical reagent grade from Sigma (St. Louis, MO). Bovine serum albumin (BSA, 10 mg/mL) solution was obtained from Promega (Madison, WI). Hydrofluoric acid 49% was obtained from Fisher Scientific. Water used for all experiments was deionized with a Milli-Q water purification system (Millipore, Worcester, MA). All buffer solutions were filtered through a 0.22µm Corning Filter System (Corning, NY) prior to use. Protein digestion was performed as described in ref 23. A mixture of 1 mg/mL BLGB and trypsin was prepared with a 10 mM Trizma base and 50 mM ammonium acetate buffer (pH 8.2) containing 0.1 mM calcium chloride. Trypsin was added at a trypsin/protein ratio of 1:50 (w/w), and the digestion mixture was incubated at 37 °C for 5 h. Capillary Electrophoresis. A home-built CE setup with a UV absorption detector (Unimicro Technologies, Inc, CA, model HDVUV-20) equipped with an on-column capillary cell module was (15) (16) (17) (18) (19) (20) (21) (22) (23)

Quirino, J. P.; Terabe, S. Science 1998, 282, 465-468. Quirino, J. P.; Terabe, S. Anal. Chem. 1999, 71, 1638-1644. Palmer, J.; Munro, N. J.; Landers, J. P. Anal. Chem. 1999, 71, 1679-1687. Quirino, J. P.; Dulay, M. T.; Bennet, B. D.; Zare, R. N. Anal. Chem. 2001, 73, 5539-5543. Quirino, J. P.; Dulay, M. T.; Zare, R. N. Anal. Chem. 2001, 73, 55575563. Khandurina, J.; Jacobson, S. C.; Waters, L. C.; Foote, R. S.; Ramsey, J. M. Anal. Chem. 1999, 71, 1815-1819. Guzman, N. A.; Stubbs, R. J. Electrophoresis 2001, 22, 3602-3628. Wei, W.; Xue, G.; Yeung, E. S. Anal. Chem. 2002, 74, 934-940. Kang, S. H.; Gong, X.; Yeung, E. S. Anal. Chem. 2000, 72, 3014-3021.

3900

Analytical Chemistry, Vol. 74, No. 15, August 1, 2002

used in this work. The signal from the UV detector was fed into a 24-bit A/D converter (Lawson Labs, Kalispell, MT) and stored in a desktop computer at 2 Hz. A high-voltage power supply (Glassman High Voltage, Whitehorse Station, NJ, model PS/ MJ30p0400-11) was used to provide the potential across the capillary. Construction of a Porous Region. Bare fused-silica capillaries of 75-µm i.d. with 365-µm o.d. (Polymicro Technologies, Phoenix, AZ) were cut into the desired lengths. The polyimide coating of a short section of the capillary (∼0.5-1-cm) 10 cm from the inlet end and 30 cm from the detection window was removed by a Window Maker (MicroSolv Technology, NJ). To form a holder, two holes were punctured in the wall of a 5-mL LatexFree syringe (Becton Dickinson, NJ) by a 0.8 mm × 25 mm needle. This served as both a holder and a buffer reservoir. The exposed section of the capillary was placed inside the syringe through the two holes. Then DEVCON 5-minute epoxy (Riviera Beach, FL) was used for sealing the two holes. This section of the capillary was immersed in 49% HF for 60 min at room temperature in a well-ventilated hood. Latex gloves were used for protection against skin contact. HF etched only the polyimidefree region of the capillary. After etching, a detection window was created further down the capillary by removing another section of the polyimide coating. The capillary was then installed onto the UV detector. Once made, the joint is very rugged and can be used for weeks without failure. We found that an etching time of 50 min was insufficient (no conductivity) and 70 min was too long (breakage of column). RESULTS AND DISCUSSION Construction of the Concentration Device. The porous capillary joint has been used in capillary electrophoresis with electrochemical detection (CE-EC) for isolating the electrochemical detector from the CE electric field.24-28 Several methods have been employed to construct a porous section along the separation capillary, including a porous glass tube,24 bare fractures,25 covering the fracture with a porous graphite tube26 or polymer tubing,27 and HF etching.28 The HF etching method is a relatively simple and effective approach, because it is easy to implement and no dead volume is created. The capillary wall forms a porous glass membrane after exposure to HF. The pore size of this porous section (not measured) is presumably so small that only certain small buffer ions can pass through it to carry current. When proteins or peptides migrate to this region, they are blocked and, thus, concentrated; however, the process appears to be reversible, and clogging was not observed. Meanwhile, large buffer ions are also concentrated if the pore size is very small. If the ionic strength in the concentrated sample zone is lower than that of the running buffer, the sample zone can be additionally sharpened at the start of the separation via FASS. To handle samples with high ionic strengths, other techniques, such as ITP or discontinuous stacking, are used to keep the concentrated sample zone sharp during (24) Wallingford, R. A.; Ewing, A. G. Anal. Chem. 1987, 59, 1762-1766. (25) Zhou, J.; Lunte, S. M. Electrophoresis 1995, 16, 498-503. (26) Yik, Y. F.; Lee, H. K.; Li, S. F. Y.; Khoo, S. B. J. Chromatogr. 1991, 585, 139-144. (27) O’Shea, T. J.; Greenhagen, R. D.; Lunte, S. M.; Lunte, C. E.; Smyth, M. R.; Radzik, D. M.; Watanabe, N. J. Chromatogr. 1992, 593, 305-312. (28) Hu, S.; Wang, Z.-L.; Li, P.-B.; Cheng, J. K. Anal. Chem. 1997, 69, 264267.

Figure 1. Schematic diagram of integrated concentration and separation system.

the separation process. Electroosmotic flow (EOF) must also be considered. If a large EOF exists in the concentration capillary, the concentrated sample zone can migrate to the separation capillary before separation is initiated, thus degrading the separation efficiency, so the magnitude of EOF should be controlled. Here, a small amount (0.02%) of HPMC was included in all of the buffers. Figure 1 shows the schematic diagram of the experimental setup. The whole system functions as follows. The concentration step is carried out when switch K2 is turned on and switch K1 is turned off. Separation is performed when switch K1 is turned on and switch K2 is turned off. For the porous junction, the homemade holder serves as buffer vial 2. The buffer compositions of vials 1 and 2 can be changed according to the properties of the sample matrix and the mode of sample concentration; however, vial 3 always contains the running buffer. Concentration of Proteins in Low Ionic Strength Samples. When the ion strength in the sample matrix is lower than that in the running buffer, the FASS technique is often used for sample concentration.1-5 However, peptides or proteins are not always ionized in water, because ionization depends on their isoelectric points (pIs). Although all 3 test proteins here are basic, typically, some analyte molecules are positively charged, and others are negatively charged. Therefore, for some analyte species, FASS cannot be used for simultaneous concentration. To make all proteins fully positively charged, 5 mM H3PO4 was mixed with the protein samples. Unfortunately, this increased the ionic strength of the sample solution and lowered the concentration factor. Figure 2 shows the electropherograms of three proteins obtained at different conditions. These basic proteins were chosen, because they were commonly used for evaluating CE separations. In Figure 2a, the sample was dissolved in water and injected to the separation capillary by applying 300 V/cm for 10 s. The capillary functioned as a normal separation capillary. A concentration factor of ∼80 was observed, as compared to when the sample is dissolved in the running buffer (height for lysozyme at 47 units, as compared to 0.6 units). When the same sample was dissolved in 5 mM H3PO4 and injected under the same condition, the stacking factor was decreased to only 5 (3 units), as shown in Figure 2b. The reason is that the difference in electric field strength between the sample zone and the running buffer became small. Figure 2c shows the result obtained by using the new approach reported here. The whole capillary was first filled with running buffer. Vials 2 and 3 contained the running buffer. The same sample as in Figure 2b (containing 5 mM H3PO4) was introduced

hydrodynamically into the concentration capillary (10 cm length) before a high voltage was applied between the sample vial and the etched section. The sample was further stacked at 300 V/cm for another 17 min. During this process, the proteins were concentrated as soon as they migrated to the beginning of the etched section. After that, the sample was injected into the separation capillary by hydrodynamic flow. Separation was performed at 300 V/cm by applying voltage between vials 2 and 3. Because the concentration capillary was initially filled with the sample, FASS was not involved in the concentration step. The amount of sample introduced can be calculated from the migration time of analytes. In Figure 2c, ∼50- and 38-cm-long sample zones were introduced into the concentration capillary for the first and last peaks, respectively. These were much longer than the concentration capillary because of the additional stacking under applied voltage. We could not calculate the amount of sample injected when the sample was dissolved in water (Figure 2a) or 5 mM H3PO4 (Figure 2b), because the electric field strengths were very different during concentration and separation. From peak 3 in Figure 2c (35 units), ∼25 and 3-fold larger concentration factors were obtained than with normal FASS samples dissolved in 5 mM H3PO4 (Figure 2b, 1.5 units) and water (Figure 2a, 13 units), respectively. In addition, injection bias was more pronounced in Figure 2a, as compared to Figure 2b,c (comparing peak 3 in each case), whereby the highly charged analyte was preferentially injected. The peak width for lysozyme was much smaller than that predicted by the length of the porous joint, confirming that the actual dimensions of the joint were not important. The noise of 0.3 units peak-to-peak in all electropherograms was determined by the optical detector and was independent of the separation or concentration mode. Judging from Figure 2c, the detection limit for lysozyme is 0.15 µg/mL. In addition, there was some shift in migration times resulting from changes in the capillary surface (EOF) following sample introduction. It is important to note that the length of the etched section was irrelevant as long as one can pass sufficient current for the concentration step. This section was always at equal potential, so no broadening was introduced other than axial diffusion and residual EOF. Concentration of Proteins in Acidic Samples. When proteins are dissolved in H3PO4 solution at a similar concentration (50 mM) to that of the running buffer (50 mM H3PO4, pH 2.5 with Tris), if the proposed stacking mechanism is correct, sample stacking should still occur, because H+ can pass through the etched pores because of its small size, but proteins are retained. However, our initial results produced a stacking factor of only 3-5, as seen in Figure 3a,b (1.2 units vs 6.5 units for lysozyme). In addition, the peak height did not increase in proportion to an increase in injection time. We attributed the small stacking factor to the adsorption of these basic proteins in the etched section. The concentrated proteins are easily adsorbed on the capillary wall because of the large surface area after etching. To overcome this problem, a prerun step was added to treat the etched section. A portion of 0.5 mg/mL BSA in 50 mM H3PO4 was first filled into the concentration capillary. Then BSA was focused at the etched section by applying 300 V/cm voltage for 10 min. After that, the whole capillary was filled with running buffer. Following sample filling and focusing, as in Figure 3b, the electropherogram shown Analytical Chemistry, Vol. 74, No. 15, August 1, 2002

3901

Figure 2. Electropherograms of the protein test mixture. In the order of elution, these are lysozyme (16 µg/mL), ribonuclease (20 µg/mL), and R-chymotrypsinogen (15 µg/mL); (a) 300 V/cm for 10-s injection into the separation capillary, sample dissolved in water; (b) 300 V/cm for 10-s injection into the separation capillary, sample dissolved in 5 mM H3PO4; and (c) the same sample as in part b was filled hydrodynamically into the concentration capillary and was continuously stacked at 300 V/cm for 17 min, then the concentrated zone was transferred to the separation capillary hydrodynamically for electrophoresis. The BGE is 50 mM H3PO4 (pH 2.5 with Tris) in 0.02% HPMC buffer. Concentration capillary length, 10 cm; separation capillary length, 30 cm; 75-µm i.d.; voltage applied, 21 kV; UV detection at 214 nm. 3902 Analytical Chemistry, Vol. 74, No. 15, August 1, 2002

Figure 3. Electropherograms of the protein test mixture. In the order of elution, these are BSA (#), lysozyme (16 µg/mL), ribonuclease (20 µg/mL), and R-chymotrypsinogen (15 µg/mL). (a) Hydrodynamic injection for 10 s at 10-cm height into the separation capillary; (b) the same sample as in part a was filled into the concentration capillary hydrodynamically and was continuously stacked at 300 V/cm for 10 min, then the concentrated zone was transferred to the separation capillary hydrodynamically for electrophoresis; (c) the concentration capillary was pretreated with BSA; other conditions were the same as in part b. All samples were dissolved in 50 mM H3PO4. For other conditions, see Figure 2.

Analytical Chemistry, Vol. 74, No. 15, August 1, 2002

3903

Figure 4. Electropherograms of the protein test mixture. In the order of elution, these are lysozyme (32 µg/mL), ribonuclease (40 µg/mL), and R-chymotrypsinogen (30 µg/mL). (a) Hydrodynamic injection for 10 s at 10-cm height into the separation capillary; (b) the sample was filled into the concentration capillary hydrodynamically, was continuously stacked at 300 V/cm for 4 min, and was subjected to ITP focusing for 4 min at 300 V/cm, then the concentrated zone was transferred to the separation capillary hydrodynamically for electrophoresis. BGE is 20 mM TEAHAc, pH 4.4, with 0.02% HPMC. All samples were dissolved in BGE buffer. For other conditions, see Figure 2.

in Figure 3c was recorded. By comparing Figure 3b with 3c, an additional peak from BSA was observed. More importantly, the peak height for each protein increased substantially by using BSA pretreatment (39 units for lysozyme). These results show that even highly acidic samples can be concentrated and further confirm our proposed stacking mechanism. From these results, it can be seen that the experiments in Figure 2 could have benefited from BSA pretreatment as well; however, there stacking is mostly at the solution interface (low to high ionic strength buffers) and not entirely at the porous junction. There was, therefore, less contribution from adsorption losses. Concentration of Proteins Dissolved in Running Buffer. When the sample is dissolved in the running buffer, the standard FASS method does not work in the concentration capillary. Transient ITP is usually used instead, because a sample matrix containing leading or terminating ions can be used as the running buffer;6,7,10 however, the sample volume is then limited by the need to provide a reasonable effective length for the separation capillary. In our approach, we are able to introduce a much larger volume without a decrease in resolution, because the separation capillary is isolated from the concentration capillary. Figure 4a shows the results obtained under normal injection, whereas Figure 4b is an electropherogram obtained using our approach. The sample was first introduced into the concentration capillary (10 cm length) hydrodynamically. To further increase the amount of sample 3904

Analytical Chemistry, Vol. 74, No. 15, August 1, 2002

loaded onto the column, electrokinetic stacking was then performed at 300 V/cm for 4 min. Note that the analytes do not go past the porous joint, as explained above. Next, ITP focusing was performed for 4 min at 300 V/cm by replacing sample vial 1 with 10 mM HAc. The effective lengths of sample zones for the first and last peaks were ∼22 and 18 cm, respectively. Again, these were longer than the actual concentration capillary because of the additional electrokinetic loading. The poor peak shape in Figure 4a was due to the strong interaction of basic proteins with the uncoated capillary wall. Improved peak shape and peak height were observed after concentration. A 50-fold stacking factor was obtained for the last peak (0.8 vs 37 units). Even better concentration factors could have been obtained by using a coated capillary to further reduce the EOF and protein adsorption during concentration. Although there was no electric field applied to the separation capillary, EOF could drive the preconcentrated sample zone to the separation capillary during ITP focusing. If too long a sample zone is introduced into the separation capillary, the separation efficiency decreases. Concentration of Protein Digest Samples with High Ionic Strengths. Protein digestion is often used for peptide mapping. In previous work,23 we developed a high-throughput comprehensive peptide mapping method by using multiplexed capillary electrophoresis. In this case, the sample matrix had a higher conductivity than that of the separation buffer. To deal with the highly conductive sample matrix, a discontinuous buffer stacking

Figure 5. Electropherograms of peptides from tryptic digest of protein. (a) Hydrodynamic injection for 10 s at 10-cm height into the separation capillary, 1 mg/mL BLGB digestion sample in digestion buffer (50 mM NH4Ac-10 mM Tris-0.1 mM CaCl2, pH 8.2) and (b) discontinuous buffer stacking of 10-cm-length sample zone, 20 µg/mL BLGB digestion sample in digestion buffer. The BGE is 40 mM TEA (pH 2.5 with H3PO4); separation voltage, 14 kV. For other conditions, see Figure 2.

method was used. After the sample was introduced into the concentration capillary, 100 mM HCl was used at vial 1, and 1% NH4OH was used at vial 2. Focusing was achieved by applying potential between vials 1 and 2. During this process, the concentrated sample zone was delivered by the residual EOF to the separation capillary. Because the pH in the sample zone was different from that of the running buffer, the preconcentrated sample was stacked again during separation. Figure 5a is the electropherogram obtained by using normal injection. The peptide map from discontinuous buffer stacking is shown in Figure 5b. For the last several peptide fragments, the peak heights were about twice as large in Figure 5b as they were in Figure 5a, although the concentration was only 1/50 in the former case. So more than 100-fold stacking was achieved. More importantly, there were no major differences in resolution and peak heights for the peaks in the two electropherograms. The current method, thus,

provides a means to concentrate complicated mixtures prior to electrophoresis. In principle, any sample matrix can be used. In addition, even larger sample volumes can be effectively concentrated by reducing the EOF by using, , for example, a coated capillary. ACKNOWLEDGMENT The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract No. W-7405-Eng82. This work was supported by the Director of Science, Office of Basic Energy Sciences, Division of Chemical Sciences, and by the National Institutes of Health. Received for review March 4, 2002. Accepted May 2, 2002. AC025612B

Analytical Chemistry, Vol. 74, No. 15, August 1, 2002

3905