Anal. Chem. 2004, 76, 2298-2305
Continuous Electrophoretic Purification of Individual Analytes from Multicomponent Mixtures David G. McLaren and David D. Y. Chen*
Department of Chemistry, University of British Columbia, Vancouver, BC, V6T 1Z1 Canada
Individual analytes can be isolated from multicomponent mixtures and collected in the outlet vial by carrying out electrophoretic purification through a capillary column. Desired analytes are allowed to migrate continuously through the column under the electric field while undesired analytes are confined to the inlet vial by application of a hydrodynamic counter pressure. Using pressure ramping and buffer replenishment techniques, 18% of the total amount present in a bulk sample can be purified when the resolution to the adjacent peak is ∼3. With a higher resolution, the yield could be further improved. Additionally, by periodically introducing fresh buffer into the sample, changes in pH and conductivity can be mediated, allowing higher purity (g99.5%) to be preserved in the collected fractions. With an additional reversed cycle of flow counterbalanced capillary electrophoresis, any individual component in a sample mixture can be purified providing it can be separated in an electrophoresis system. Since its introduction in 1981,1 capillary electrophoresis (CE) has become one of the most powerful analytical separation techniques. Because it utilizes both electric field and chemical equilibrium as driving forces for the separation, CE often succeeds in separating species that cannot be resolved by other methods. Furthermore, because the separation takes place in free solution and the mechanisms of separation are gentle, the integrity of delicate or labile samples can often be preserved.2-5 For these reasons, there exists a significant demand for methodologies that allow for recovery of components separated by CE. A number of groups have developed methods for performing fraction collection in CE,4,6-12 and optimization of the relevant parameters necessary * To whom correspondence should be addressed. Tel: +604-822-0878. Fax: +604-822-2847. E-mail:
[email protected]. (1) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (2) Lerner, E. A.; Nelson, R. J. LC-GC 1994, 12, 36, 38. (3) Landers, J. P.; Oda, R. P.; Spelsberg, T. C.; Nolan, J. A.; Ulfelder, K. J. Biotechniques 1993, 14, 98-111. (4) Rose, D. J.; Jorgenson, J. W. J. Chromatogr., A 1988, 438, 23-34. (5) Pacakova, V.; Suchankova, J.; Stulik, K. J. Chromatogr., B: Biomed. Appl. 1996, 681, 69-76. (6) Zhang, H. Y.; Caprioli, R. M. J. Mass Spectrom. 1996, 31, 1039-1046. (7) Minarik, M.; Foret, F.; Karger, B. L. Electrophoresis 2000, 21, 247-254. (8) Minarik, M.; Kleparnik, K.; Gilar, M.; Foret, F.; Miller, A. W.; Sosic, Z.; Karger, B. L. Electrophoresis 2002, 23, 35-42. (9) Magnusdottir, S.; Heller, C.; Sergot, P.; Viovy, J. L. Electrophoresis 1997, 18, 1990-1993. (10) Muller, O.; Foret, F.; Karger, B. Anal. Chem. 1995, 67, 2974-2980. (11) Huang, X.; Zare, R. N. Anal. Chem. 1990, 62, 443-446.
2298 Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
to maximize the amount of analyte recovered in a single run has also been investigated.13-15 Though effective, fraction collection produces only small amounts of purified sample components due to limitations associated with the small dimensions of the capillary columns. As such, beyond fraction collection, CE has not been widely investigated as a preparative technique in its own right. Alternatively, many researchers have attempted to utilize CE indirectly as a tool for sample purification. In large part, these efforts have been focused on developing CE methods that can then be transferred to other, larger-scale techniques such as continuous free flow electrophoresis (CFFE). In CFFE, the sample is introduced as a constant stream into a chamber formed by two closely spaced plates. A continuous flow of buffer moves in one direction through the chamber, and this serves to sweep the sample from its introduction point toward a series of collection vials at the chamber exit. Sample components separate according to their electrophoretic mobilities under an electric field that is applied perpendicularly to the direction of buffer flow. Several instrumental designs have been reported, and a number of commercial systems are also available for the practice of CFFE.16,17 As in CE, a variety of operating modes are available in CFFE simply by changing the nature of the background electrolyte (BGE); free zone electrophoresis,18 chiral separations,19-22 and isoelectric focusing23,24 have all been demonstrated. Because method development in CFFE is generally time-consuming and often costly (in terms of both financial expense and sample consumption), several attempts have been made to predict optimal conditions for operation of CFFE from analogous CE separations. A certain number of these attempts were successful,24-26 while (12) Cheng, Y. F.; Fuchs, M.; Andrews, D.; Carson, W. J. Chromatogr. 1992, 608, 109-116. (13) Bergman, A. C.; Bergman, T. FEBS Lett. 1996, 397, 45-49. (14) Cifuentes, A.; Xu, X.; Kok, W. T.; Poppe, H. J. Chromatogr. 1995, 716, 141156. (15) Yin, H. F.; Keelytemplin, C.; McManigill, D. J. Chromatogr. 1996, 744, 4554. (16) Krivankova, L.; Bocek, P. Electrophoresis 1998, 19, 1064-1074. (17) Thome, B.; Ivory, C. F. J. Chromatogr., A 2002, 953, 263-277. (18) Bondy, B.; Bauer, J.; Seuffert, I.; Weber, G. Electrophoresis 1995, 16, 9297. (19) Glukhovskiy, P.; Vigh, G. Electrophoresis 2001, 22, 2639-2645. (20) Glukhovskiy, P.; Vigh, G. Electrophoresis 2000, 21, 2010-2015. (21) Stalcup, A. M.; Sutton, R. M. C.; Painuly, P.; Rodrigo, J. V.; Gratz, S. R.; Yanes, E. G. Analyst 2000, 125, 1719-1724. (22) Rizzi, A. Electrophoresis 2001, 22, 3079-3106. (23) Spanik, I.; Vigh, G. J. Chromatogr., A 2002, 979, 123-129. (24) Spanik, I.; Lim, P.; Vigh, G. J. Chromatogr., A 2002, 960, 241-246. (25) Kasicka, V.; Prusı´k, A.; Sa´zelova´, P.; Jira´cek, J.; Barth, T. J. Chromatogr. 1998, 796, 211-220. 10.1021/ac0350460 CCC: $27.50
© 2004 American Chemical Society Published on Web 03/09/2004
others demonstrated that CE runs failed to accurately predict analyte behavior in CFFE,17,27,28 suggesting that CE is only of limited use as a predictive tool in this regard. Another drawback of CFFE is that most of these techniques are not well suited to the handling of complex mixtures. Purification of multicomponent samples is often complicated by significant overlap of adjacent sample bands, leading to decreased purity in the collected fractions.29,30 Operating CE directly in a preparative fashion holds the potential for resolving a number of deficiencies associated with other continuous processes. In particular, CE could be preferentially applied to the purification of complex mixtures or when working with limited amounts of unique samples. The principle and practice of flow counterbalanced CE (FCCE) was first demonstrated by Dovichi et al. with a continuous back pressure applied through a pressurized sheath flow cuvette, and plate numbers of up to 2.5 million were demonstrated for peaks of separated fluorescently labeled amino acids.31 Culbertson and Jorgenson also used FCCE for the analytical-scale separation of closely related species, including isotopomers of fluorescently derivatized amino acids.32,33 By applying a hydrodynamic counterpressure in opposition to the direction of migration, the analytes could be kept on-column until the desired resolution was achieved. In order for this technique to be successful, very narrow internal diameter capillaries had to be used to prevent excessive broadening of the peaks arising from application of the counterpressure. Chankvetadze et al demonstrated that this technique could also be used in a continuous fashion in normal-scale capillaries for preparative purposes by recovering pure fractions of the R-isomer of the dipeptide aspartame from a binary mixture.34 Our work with FCCE has been focused on developing ways to improve the purity and recovery of the separated components.35 By constructing an instrument with precision pressure control (0.01 psi), we were able to purify closely migrating species. Additionally, we have demonstrated that judicious choice of a low-conductivity background electrolyte and a large-bore capillary are of primary importance for practical application of this technique. We have also developed a new technique called pressure-ramped FCCE that is capable of nearly doubling yields without significantly increasing processing time. In our earlier studies, buffer depletion was found to limit the amount of pure fractions that could be recovered, and isolation of only the fastest migrating species in a sample mixture (under forward or reversed polarity) was demonstrated. If continuous (26) Wind, M.; Hoffmann, P.; Wagner, H.; Thormann, W. J. Chromatogr., A 2000, 895, 51-65. (27) Gratz, S. R.; Schneiderman, E.; Mertens, T. R.; Stalcup, A. M. Anal. Chem. 2001, 73, 3999-4005. (28) Schneiderman, E.; Gratz, S. R.; Stalcup, A. M. J. Pharm. Biomed. Anal. 2002, 27, 639-650. (29) Loseva, O. I.; Gavryushkin, A. V.; Osipov, V. V.; Vanyakin, E. N. Electrophoresis 1998, 19, 1127-1134. (30) Bahre, F.; Maier, H. G. Fresenius J. Anal. Chem. 1996, 355, 190-193. (31) Cheng, Y. F.; Wu, S.; Chen, D. Y.; Dovichi, N. J. Anal. Chem. 1990, 62, 496-503. (32) Culbertson, C. T.; Jorgenson, J. W. Anal. Chem. 1994, 66, 955-962. (33) Culbertson, C. T.; Jorgenson, J. W. J. Microcolumn Sep. 1999, 11, 175183. (34) Chankvetadze, B.; Burjanadze, N.; Bergenthal, D.; Blaschke, G. Electrophoresis 1999, 20, 2680-2685. (35) McLaren, D. G.; Chen, D. D. Y. Electrophoresis 2003, 24, 2887-2895.
FCCE is to become a viable alternative for preparative applications, it must demonstrate satisfactory recoveries as well as greater versatility in purification of any individual species in a complex mixture. The present study details our attempts to improve sample recovery in continuous FCCE through the use of buffer replenishment, and we also demonstrate the feasibility of isolating any individual species in a multicomponent mixture. EXPERIMENTAL SECTION Apparatus. Experiments were performed on a home-built electrophoretic purification system that has been previously described.35 Uncoated capillaries (Polymicro Technologies, Phoenix, AZ) 40 cm in length (31 cm to detector) with internal diameters of either 75 or 100 µm and 365-µm outer diameters were used. New capillaries were first rinsed with 0.1 M NaOH (20 min), followed by rinsing with the BGE (20 min), and finally electrolyzed at 20 kV (20 min) prior to use. All separations were preceded by a 3-min, high-pressure rinse with 0.1 M NaOH, followed by a 3-min, high-pressure rinse with the BGE. Appreciable decreases in the EOF were noticed over extended periods of operation, likely due to adsorption of sample components to the capillary wall during the extended run times. As such, capillaries were used for at most 1 week before being replaced. Separations were performed at 20 kV and ambient temperature with forced-air cooling using either normal or reversed polarity as indicated in the text. Injections were performed hydrodynamically at a pressure of 0.30 psi for 3 s. All data were collected and processed on a desktop PC operated by software written in-house using LabVIEW 6.0. Reagents. 2-(N-Cyclohexylamino)ethanesulfonic acid (CHES), sodium dodecyl sulfate (SDS), L-alanine, L-valine, and L-isoleucine were all purchased from Sigma (Sigma-Aldrich, Oakville, ON, Canada). Dimethyl sulfoxide, HPLC grade methanol, NaOH, and HCl were from Fisher (Fisher Scientific, Nepean, ON, Canada), and the fluorescent probe, tetramethylrhodamine isothiocyanate (TRITC, G-isomer), was from Molecular Probes (Eugene, OR). The derivatized amino acids were separated using micellar electrokinetic chromatography. The BGE used in all separations was composed of 50 mM CHES, 15 mM SDS, and 10% MeOH (v/v) and was prepared in distilled, deionized water. The pH of these solutions was adjusted to 9.1 using 1 M NaOH. The derivatization procedure for the amino acids has been previously described.35 Stock solutions of the labeled amino acids were prepared, aliquoted into 250 µL volumes, and stored at -20 °C. Aliquoting minimizes the number of freeze-thaw cycles each stock solution is subjected to and helps to prevent accelerated degradation of the labeled amino acids. Derivatized sample solutions were found to be stable for ∼2 months when stored in this manner. For preparative separation by continuous FCCE, samples consisting of a mixture of the 50 or 300 µM derivatized amino acids were prepared by diluting appropriate amounts of the individual amino acid stock solutions with BGE to a total volume of 350 µL. For quantitative analysis of the fractions recovered at the outlet following purification, an aliquot was injected at the capillary inlet and separated by conventional CE. Peak areas of the recovered sample (corrected for dilution) were compared to calibration curves to determine the final concentration of the purified components. Percentage yields are quoted Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
2299
Figure 1. Schematic representation of continuous FCCE. (a) The inlet is positioned in the sample vial while the outlet is positioned in a buffer vial. (b) A potential is applied to begin the separation. Simultaneously, a counterpressure is applied at the outlet end of the capillary that results in the slower migrating component having a net velocity toward the inlet and the faster migrating component having a net velocity toward the outlet. (c) At the end of the run, a portion of the faster migrating component has been isolated in the outlet vial.
Figure 2. Representative separation of the three labeled amino acids in a 100-µm-i.d. capillary. The concentration of each individual amino acid in the sample solution was 50 µM. Inset: general structure of the tetramethylrhodamine thiocarbamyl derivative of an amino acid. Charge state indicated is as it would appear at pH 9.1. R ) CH3 (Ala); CH(CH3)2 (Val); CH(CH3)CH2CH3 (Ile).
relative to the initial concentration of the analyte in the original sample. RESULTS AND DISCUSSION Buffer Replenishment. Figure 1 shows a schematic representation of the continuous FCCE process, and Figure 2 shows the migration pattern of the three labeled amino acids for a typical separation. The first component to elute is alanine, followed by valine, isoleucine, and finally unreacted TRITC. In our previous experiments, we found there to be a maximum amount of pure alanine that could be recovered before valine began to appear as a contaminant. The appearance of valine in the collection vial was 2300 Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
attributed to an increase in the forward linear velocity of all the analytes stemming from protonation of amine functionalities in the structure of the labeled amino acids (shown in the inset of Figure 1). At the initial BGE pH of 9.1, the net charge on the labeled amino acids is zero. However, as the separation is carried out, the pH in the inlet (sample) vial begins to drop due to the generation of protons through the electrolysis of water in the aqueous BGE. As the pH drops, full protonation of the alkylamine moiety occurs (pKa ≈ 10), followed by protonation of the arylamine moieties (pKa ≈ 5). Under normal polarity, this results in all analytes acquiring an increased electrophoretic mobility toward the outlet (cathodic) vial. Adding an aliquot of fresh buffer to the sample vial should help to mediate such pH changes, thereby preventing slower migrating species from interfering with the collection of the desired analyte. To investigate the effects of buffer replenishment on the purity of collected fractions, a series of purification cycles were carried out on a mixture containing alanine, valine, and unreacted TRITC using a 75-µm-i.d. capillary. A counterpressure of 0.38 psi was applied at the outlet, and sample was allowed to migrate onto the capillary for a period of 20 min. Under normal FCCE conditions, alanine requires 30 min to migrate from the inlet of the capillary to the detector window, and as such, no signal was observed at the detector during this process. After the 20 min had elapsed, the separation was discontinued and the sample that had migrated on-column was flushed into the outlet vial. Eight consecutive cycles were performed in this manner, and all collections were made into the same vial. Figure 3a shows the results of the conventional CE analyses of the samples collected following selected individual 20-min cycles. A distinct valine peak was observed in the sample following the eighth collection (total elapsed time of purification, 160 min), and this agrees well with our previous experience. Obviously, after 160 min of continuous operation, the forward linear velocity of valine has increased sufficiently to allow some of that component to migrate on-column. Following the eighth cycle, an attempt to restore the pH of the sample to its original value was made. A new BGE consisting of 150 mM CHES, 15 mM SDS, and 10% MeOH (v/v) was prepared at a pH of 9.3. A 10-µL aliquot of this concentrated CHES BGE was added to the sample, and a ninth purification cycle was performed. To properly assess whether any valine had entered the capillary as a contaminant during the ninth cycle, the sample loaded onto the capillary was collected into a fresh buffer vial. The profile of the sample collected in the ninth cycle is shown in Figure 3b. The height of the peak in Figure 3b cannot be directly compared to those in part a because different detection conditions were used in the conventional CE analyses. However, it is evident that no appreciable amount of valine appears in the sample collected for the ninth cycle, indicating that buffer replenishment is successful in preventing contamination of the purified fractions of the desired analyte. Optimization of Production Yield Using Buffer Replenishment. To get the greatest benefit possible from buffer replenishment, the composition and volume of the restoring BGE must be carefully selected. In the previous demonstration, arbitrary choices regarding concentration and volume were made to evaluate the potential of buffer replenishment for preventing contamination of
Figure 3. Purification of Ala from a mixture also containing Val with and without buffer replenishment. Traces have been offset for ease of comparison. (a) Recovery of alanine at the outlet for increasing cycle number. (b) Comparison of sample components purified in the eighth and ninth cycles. Buffer replenishment of the sample was carried out between the eighth and ninth cycles. Capillary, 75-µm i.d. The concentration of each amino acid in the sample prior to any purification was 50 µM. The counterpressure applied during the purification was 0.38 psi.
Figure 4. pH and volume drift plots for the BGE in the inlet and outlet vials after electrolysis at 20 kV (normal polarity); 9, change in inlet vial; 2, change in outlet vial. The solid line corresponds to pH changes, and the dashed line corresponds to volume changes: (a) 75-µm-i.d. capillary; (b) 100-µm-i.d. capillary.
the purified fractions. If an optimal composition is to be chosen, changes in the BGE during the continuous FCCE process must first be documented. Experiments were thus carried out to profile the pH and volume variations occurring in the inlet and outlet vials during normal FCCE operation. The 350-µL volumes of BGE were subjected to electrolysis at 20 kV for fixed time intervals using either a 75- or 100-µm-i.d. capillary. The pH’s of both the inlet and outlet solutions were measured using a microelectrode after 15 min for the 100-µm capillary and after 20 min for the 75µm capillary. Additionally, the weights of both vials were recorded before and after each time increment to document the increase/ decrease in volume of the outlet/inlet vials due to the EOF. Counterpressures were applied to accurately reflect purification conditions and were set at 0.22 and 0.38 psi for the 100- and 75µm capillaries, respectively, corresponding to the values normally used for the purification of alanine from valine. All runs were performed in triplicate. The changes in pH (left axis) and volume (right axis) of the inlet and outlet solutions were plotted as a function of electrolysis time, and the results are shown in Figure 4. The buffering range of CHES extends from pH 8.6 to 10.0 at 25 °C. As Figure 4 shows, the pH of the inlet vial approaches this lower limit after 120 min at 20 kV in the 75-µm capillary and after
just 75 min in the 100-µm capillary. Once the lower limit has been passed, the pH of the solution drops rapidly. This results in proportional increases in the electrophoretic mobilities of all analytes, leading to contamination of the collected fractions. In a typical FCCE run, the time required for alanine to migrate through the column to the outlet/collection vial is ∼40 min (regardless of the internal diameter of the capillary). After this time, alanine migrates continuously into the collection vial, leaving 35 min of collection time before the buffering capacity of the BGE is exhausted when using a 100-µm capillary. The amount of pure analyte collected during this time is directly related to the initial concentration of that component in the sample. When replenishing BGE is added to the inlet vial it will dilute the sample, thereby reducing the amount that can be recovered. Thus, to maximize the amount of alanine recovered, collection should be continued for as long as possible before potential contamination becomes an issue. To ensure high purity of the collected fraction, we opted to cease collection once the pH in the sample vial reached the lower limit of the buffering range of the CHES BGE. All further experiments were carried out using the 100-µm-i.d. capillary, and replenishment of the sample was thus performed after 75 min of purification. Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
2301
Table 1. Changes in pH and Volume of the Sample Vial during Three Consecutive 75-min Continuous FCCE Runs pH
current (µA) volume (µL) after at at starting ending replenishment starting ending start end run 1 run 2 run 3
Figure 5. Amounts of Ala purified from a mixture also containing Val and Ile after three consecutive cycles of FCCE. Capillary, 100µm i.d.; run time, 75 min; counterpressure, 0.22 psi. Buffer replenishment was carried out between each consecutive FCCE cycle. The concentration of each amino acid in the sample prior to any purification was 50 µM. The peaks before and after Ala are fluorescent degradation products of FITC-Ala.
The composition (pH and concentration) of the buffer required to restore the pH of the sample to its original value was calculated based on the starting and ending pH of the solution and the total volume transferred between the inlet and outlet vials. The volume of the “spike” of replenishing buffer was initially set equal to the amount of solution lost from the inlet vial during the 75-min continuous FCCE process. This was done to maintain a relatively constant sample volume of 350 µL. Appropriate values for the pH and concentration of the replenishing buffer were then calculated using the Henderson-Hasselbach relationship and finally optimized experimentally. Adding a 55-µL aliquot of 75 mM CHES, 15 mM SDS, 10% MeOH (v/v) at a pH of 9.9 to the depleted sample was found to provide the best performance in terms of restoring pH and conductivity values to their original levels. For the sake of convenience, we will refer to this replenishing BGE as 1.5× CHES buffer. Figure 5 shows the amounts of pure alanine recovered from a sample also containing valine, isoleucine, and unreacted probe in three consecutive continuous FCCE runs. Purification was carried out for 75 min at 20 kV (counter pressure, 0.22 psi) following which the sample was replenished with 55 µL of the 1.5× CHES buffer. At the end of each 75-min cycle, some additional alanine remains inside the capillary. However, some valine may also have migrated on-column, and to prevent this secondary analyte from appearing as a contaminant in the collected fractions, the capillary was rinsed with fresh BGE between continuous FCCE cycles. This practice also helps to maintain a consistent EOF through the capillary between different FCCE runs, allowing a uniform counterpressure to be used throughout the experiments. Collection was performed with a fresh outlet vial each time. During the initial run, 3.2% of the total amount of alanine present in the original sample was recovered at the outlet and this agrees well with our previous observations.35 The amounts purified in subsequent runs were lower, with yields of 1.7 and 0.8% relative obtained in runs 2 and 3. Valine contamination was observed following the fourth run (data not shown). The decrease in percent yield between runs 1 and 2 and between runs 2 and 3 can be partially attributed to dilution of the sample. As previously mentioned, the amount of alanine recovered is directly related to its starting concentration and we would thus expect a decrease between consecutive 2302 Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
9.100 8.981 8.864
8.590 8.146 3.173
8.981 8.864 8.692
350 350 353
295 298 304
83.7 94.3 103.1
57.0 58.6 63.9
purifications in line with the dilution effected by addition of the restoring buffer. The sample volume prior to the first run was 350 µL, finishing at 295 µL after 75 min of purification. The 55-µL aliquot of restoring buffer would result in a replenished sample volume of 350 µL, leading to a 19% reduction in sample concentration. However, the amount of alanine purified in the second 75min run is only 59% of what was recovered in the previous run. Similarly, the dilution due to buffer replenishment between runs 2 and 3 represents an 18% reduction in sample concentration, but the amount recovered in run 3 is only 42% of that in run 2. From these findings it is evident that sample dilution is not the only factor contributing to decreased recoveries between runs. Table 1 shows the changes in pH and volume of the sample vial during the 75-min runs depicted in Figure 5, as well as the average current during the first and final 5 min. The amount of solution transferred between the inlet and outlet vials can be seen to decrease between consecutive runs, suggesting a reduction in the EOF throughout the process. Additionally, the initial operating current increases by ∼10% after each buffer replenishment. Since continuous FCCE is analogous to a continuous electrokinetic injection, a higher conductivity environment would result in decreased sample introduction. Both of these factors will have contributed to the decreased recoveries of alanine observed in the second and third runs. Finally, the nature of the analytes themselves should also be considered in assessing losses during the purification process. Tetramethylrhodamine thiocarbamyl derivatives of amino acids are known to be unstable over the long term in aqueous solutions, and thus, the actual amount purified could be higher than what is reported in this paper. Pressure-Ramped FCCE with Buffer Replenishment. In our previous experiments, we were able to recover 9.0% of the total alanine content of a sample by ramping the counterpressure at the end of the run to prevent secondary analytes from migrating into the collection vial.35 Figure 6 shows the amounts of alanine recovered in three consecutive PR-FCCE runs employing buffer replenishment between each run. Purification was carried out for 120 min in each run with a fixed counterpressure of 0.22 psi maintained for the first 90 min. Though the pH in the inlet vial is known to drop below the useful buffering range after 75 min of continuous operation, extending the run by an additional 15 min allows some of the alanine already inside the capillary to elute into the outlet vial. At the 90-min mark, the counterpressure was increased by 0.01 psi every 5 min until the end of the run (final counterpressure, 0.28 psi). Between runs, the sample was replenished with 90 µL of the 1.5× CHES buffer. Additional peaks clustered around the central alanine peak are evident in the fractions recovered in the second and third PR-FCCE runs. These peaks are evidence of the sample degradation taking place
Figure 6. Amounts of Ala purified from a mixture also containing Val and Ile after three consecutive cycles of pressure-ramped FCCE. Capillary, 100-µm i.d.; run time, 120 min; counterpressure, 0.22 psi from 0 to 90 min, ramped at a continuous rate of 0.01 psi every 5 min to a final pressure of 0.28 psi. Buffer replenishment was carried out between each consecutive FCCE cycle. The concentration of each amino acid in the sample prior to purification was 50 µM.
throughout the purification process. Similar peaks, though much smaller, are also evident in Figure 5. The fact that these degradation product peaks are much more prominent in these runs suggests that the tetramethylrhodamine thiocarbamyl derivatives may be sensitive to large shifts in pH. Following the first 120-min run, the pH in the sample vial was found to have dropped to 8.02. After replenishment, this value was restored to 9.03. Following the second 120-min run, the pH in the sample vial had dropped to 3.04 and was subsequently restored to 8.91. At the end of the third cycle, the pH in the sample vial was found to be 2.40. These large pH variations are likely to have contributed to the degradation of the individual amino acids present in the sample. Identical peaks were observed in runs carried out on samples containing only alanine, confirming their identity as degradation products of that amino acid. Because these peaks can be linked definitively to alanine, they were included in the quantification of the total amount purified. Once again, all peak areas were corrected for the dilution occurring at the outlet due to the bulk flow of the EOF. It is interesting to note that the relative amounts recovered in the second and third PR-FCCE runs are greater than the corresponding amounts recovered using continuous FCCE. As previously noted, the amounts of alanine recovered in runs 2 and 3 using FCCE were 59 and 42% of the amounts collected in the prior run. In the PR-FCCE experiments, the yield in run 2 was 68% of that in run 1 while the yield in run 3 was 55% of that in run 2. This despite the fact that the dilution of the sample was much higher in the PR-FCCE runs. These results can be explained by the large increase in the forward mobility of Ala during the latter stages of the PR-FCCE runs. The total yield of FITC-alanine following the three PR-FCCE runs was calculated to be 17.5% of the amount present in the original sample, amounting to 3.1 nmol (1.3 µg) of purified material. This represents nearly a 2-fold increase over the recoverable amount under similar experimental conditions without buffer replenishment. Purifying Individual Analytes from Multicomponent Mixtures Using Continuous FCCE. Although free-flow electrophoresis methods perform well in terms of throughput and yield,
Figure 7. Purifying a nonfastest moving component in a mixture. In a group of analytes with electrophoretic mobilities of µep,1-µep,4, when a pressure is applied, both analytes 1 and 2 are allowed to migrate through the capillary. The net mobility of the analyte 2 is denoted as ∆µ1 in the first process. To obtain pure analyte 2, the polarity is reversd and pressure applied at the inlet vial. The net mobility of analyte 2 in this situation is illustrated by ∆µ2. It should be noted that the slowest migrating component in a mixture can be obtained directly by using step 2 alone.
their ability to separate unique species from complex mixtures is compromised by field strength limitations and the diffusional broadening that takes place in free solution. Such separations should prove easier to achieve using continuous FCCE and PRFCCE. The methodology necessary to isolate analytes other than the fastest migrating species from multicomponent samples is relatively straightforward. Initially, a group of peaks including the desired analyte may be purified under conditions of normal polarity. The counterpressure is selected to allow the desired analyte to migrate on-column while preventing any slower moving analytes from leaving the inlet (sample) vial. In this case, the sample collected at the outlet will contain the desired analyte and all other species that migrate ahead of it. Once a satisfactory amount has been collected, the desired analyte may be purified from all other sample components by separation using reversed polarity (under which the desired analyte is now the fastest migrating species). The principles of this operation are illustrated in Figure 7. Without the use of buffer replenishment, the recoverable amount of an analyte migrating behind the fastest moving species is relatively limited. This is due primarily to the mobility bias between the desired analyte and those that migrate more swiftly; greater amounts of faster species will enter the capillary, limiting the amount of the desired analyte that can be collectedsa Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
2303
phenomenon routinely observed in electrokinetic injection. However, the increase in yield attainable using buffer replenishment unlocks the possibility of purifying any individual component of a complex mixture. To illustrate how buffer replenishment can be useful for just such a purpose, several sequential continuous FCCE steps were carried out on a sample containing alanine, isoleucine, and unreacted TRITC with buffer replenishment being employed between cycles. The results of these trials are shown in Figure 8. Part a shows the separation obtained under normal polarity for the sample prior to purification. Part b shows the amounts of alanine and isoleucine recovered during three consecutive continuous FCCE cycles. Each sequential FCCE run was 75 min in duration with a counterpressure of 0.15 psi. The same collection vial was used for each run. Between runs, the sample vial was replenished with 55 µL of the 1.5× CHES buffer. The amount of alanine recovered in these steps is significantly more than the amount recovered for isoleucine, as expected. Nevertheless, the 4.8% of isoleucine recovered at the end of the third cycle is satisfactory for further purification. The final purification step under reversed polarity was carried out by loading a large plug of Ile onto the capillary using a counterpressure (applied at the inlet due to the reversal in direction of flow of the EOF) of 0.20 psi for 10 min. The plug was then mobilized through the column and collected at the outlet. The amount of isoleucine collected from this single, 10-min run is shown in Figure 8c. Further efforts to purify greater amounts of isoleucine were complicated by the appearance of a broad peak migrating at the same time as unreacted TRITC. This peak must arise due to degradation of the purified isoleucine, as no such peak is evident in the purified fractions shown in Figure 8b. Nevertheless, these experiments do show that isolation of components other than the fastest migrating species is possible, suggesting that continuous FCCE with buffer replenishment has potential for the purification of multicomponent mixtures. CONCLUSIONS Through continuous operation of flow-counterbalanced capillary electrophoresis, the purification of closely related species is possible. Furthermore, the use of buffer replenishment allows the useful operating time of the technique to be extended, resulting in improved recoveries. The pH changes that previously limited recovery of pure analyte fractions can be mediated by periodically introducing fresh buffer into the sample, allowing the purification process to be prolonged. Using buffer replenishment, a 3-fold increase in operating time was demonstrated for the continuous FCCE purification of alanine, resulting in a total recovery of 5.9%. Even greater recoveries are possible using the technique of pressure-ramped FCCE, with a total yield of 17.5% when buffer replenishment is employed as opposed to 9.0% without replenishment. The purity of the alanine recovered is also exceptional; g99.5% purity was attained in all cases, as determined by comparing the collected amounts of Ala to the detection limit for Val (LODVal ) 2 nM). Furthermore, the increased recoveries achieved using buffer replenishment also allowed the isolation of pure fractions of isoleucine, which previously could not be accomplished since it is not the fastest migrating analyte in our experimental system. Although the amount of Ile recovered in these studies was low, it is nonetheless significant as it demonstrates the potential of using CE as a purification technique for 2304 Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
Figure 8. Purification of Ile from a mixture also containing Ala and underivatized TRITC using both FCCE under normal polarity and CFCCE under reversed polarity. Capillary, 100-µm i.d. The concentration of each amino acid in the sample mixture prior to purification was 300 µM. (a) Representative separation of the sample components prior to any purification. (b) Amounts of Ala and Ile recovered after purification from TRITC in three consecutive FCCE runs with a counterpressure of 0.15 psi. (c) Amount of Ile recovered after a single 10-min CFCCE run using reversed polarity from the pooled samples collected in part b. The counterpressure during this step was 0.20 psi and was applied at the capillary inlet.
multicomponent samples. The results reported here should be taken as merely a starting point when assessing the potential of continuous FCCE as it can be applied to preparative endeavors. The yields reported in this investigation were limited by degradation of the fluorescently labeled analytes, which are known to suffer from instability in aqueous solutions. For samples with
greater solution stability, we estimate that much improved yields can be achieved under similar experimental conditions. Finally, by making minor instrumental modifications, the overall process time can be considerably shortened and the applicability of the method can be widened to include a broader class of analytes. Capillary lengths can be shortened to reduce process times, and multipoint UV detection can be implemented to monitor and dynamically control the separation.
ACKNOWLEDGMENT This work is supported by the Natural Sciences and Engineering Research Council of Canada. Received for review September 5, 2003. Accepted January 28, 2004. AC0350460
Analytical Chemistry, Vol. 76, No. 8, April 15, 2004
2305
