Real-Time Fluorescence Imaging of Isotachophoretic

Anal. Chem. 1996, 68, 2766-2770

Real-Time Fluorescence Imaging of Isotachophoretic Preconcentration for Capillary Electrophoresis Jonas Johansson,† Dirk T. Witte,‡ Marita Larsson,‡ and Staffan Nilsson*,§

Division of Atomic Physics, Lund Institute of Technology, P.O. Box 118, S-221 00 Lund, Sweden, Astra Ha¨ ssle AB, S-431 83 Mo¨ lndal, Sweden, and Department of Technical Analytical Chemistry, Chemical Center, P.O. Box 124, University of Lund, S-221 00 Lund, Sweden

Real-time studies of the dynamic processes that take place during isotachophoresis (ITP) were performed. The experimental arrangement utilized was a real-time fluorescence imaging system based on a dye laser at 488 nm, pumped by a XeCl excimer laser. Fluorescence emitted from the migrating sample molecules was recorded by an image-intensified, thermoelectrically cooled CCD camera. The camera signals were processed by a computer and displayed on a screen in real time, allowing the ITP concentration to be monitored continuously. Real-time analysis provided additional information concerning ITP hardly obtainable using conventional detection systems or by theoretical calculations. Such experimental data can be evaluated and be compared with theoretical calculations. Information obtained by this detection technique is useful if ITP is to be used, for example as an on-line preconcentration technique in combination with capillary zone electrophoresis. Although capillary zone electrophoresis (CZE) is a separation technique that is developing rapidly, it still requires high sample concentrations. The CZE systems that are currently available commercially involve absorbance and use of fluorescence detectors. The concentration detection limit can be improved by 4-5 orders of magnitude, however, by use of more sensitive methods such as laser-induced fluorescence (LIF). A basic feature of CZE is the small sample volume that can be analyzed. Combining the injection of larger volumes with on-line, preconcentration techniques is an effective way of improving the sensitivity of CZE. One approach of this sort utilizes a small amount of packing made of reversed phase1 or antibody2 material placed on-line with the capillary. Another approach3 is the injection, by either hydrodynamic flow or electromigration, of sample components dissolved in a low-conductivity medium. In the case of hydrodynamic injection, the low conductivity of the injected medium causes field amplification at the start of electromigration. In the case of electromigration injection, the solutes will effectively migrate into the capillary due to the low conductivity and high field strength that is experienced by the sample. Sample focusing occurs in †

Lund Institute of Technology. ‡ Astra Ha ¨ssle AB. § University of Lund. (1) Swartz, M. E.; Merion, M. J. Chromatogr. 1993, 632, 209. (2) Guzman, N. A.; Trebilcock, M. A.; Advis, J. P. J. Liq. Chromatogr. 1991, 14, 997. (3) Chien, R.-L.; Burgi, D. S. Anal. Chem. 1992, 764, 1046.

2766 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

the conductivity interface because of the higher conductivity in the electrolyte inside the capillary. Isotachophoresis (ITP) is a stacking technique that can be used on-line with CZE. It is carried out in a discontinuous buffer system.4,5 The analytes migrate between a leading and a terminating electrolyte, i.e., between an electrolyte with higher and one with lower mobility than the analytes. When the mobilities of the analyte ions are intermediate to those of the leading and the terminating ions, Kohlrausch’s law6 stipulating that an isotachophoretic train is established applies. In such a train, the sample concentration adapts to the concentration of the leading ion,6 which is often up to 5 orders of magnitude higher. In practice, a more than a 500-fold larger injection volume than normal can be analyzed without resolution loss by CZE after ITP focusing. Theoretical models for different ITP-CZE systems have been presented, showing that both stacking (focusing and concentration of the injected sample plug through formation of an ITP train) and destacking (opening of the ITP train through exchange of the terminating electrolyte for the background electrolyte and initiation of the CZE process) can be controlled.7,8 The conditions for stacking and destacking (at the start of CZE) should be selected carefully. If destacking is initiated too early, only a part of the sample plug will be collected. If the destacking starts too late, on the other hand, the separation of the linked zones that takes place during CZE is not complete when the end of the capillary is reached. The combination in CZE of ITP focusing and laser-induced fluorescence provides a considerable gain in sensitivity, as Reinhoud et al.9 have shown. In the present study, a novel realtime detection technique10 for the visualization of ITP focusing processes was evaluated. Real-time detection was achieved by illuminating the central part of the capillary by a laser beam and detecting fluorescence using a CCD camera, a major part of the separation capillary serving as a detection window. The camera collects a series of images which are processed by a computer and displayed in real-time. Use of this experimental approach in CZE and isoelectric focusing applications10-13 has been reported earlier. (4) Foret, F.; Szo ¨ko ¨, E.; Karger, B. L. J. Chromatogr. 1992, 608, 3. (5) Witte, D. T.; Någård, S.; Larsson, M. J. Chromatogr. 1994, 687, 155. (6) Bocek, P.; Deml, M.; Gebauer, P.; Dolnik, V. In Analytical Isotachophoresis; Radola, B. J., Ed.; VCH: Weinheim, Germany 1988; pp 40-57. (7) Gebauer, P.; Thormann, W.; Bocek, P. J. Chromatogr. 1992, 608, 47. (8) Foret, F.; Bocek, P. Electrophoresis 1990, 11, 661. (9) Reinhoud, N. J.; Tjaden, U. R.; van der Greef, J. J. Chromatogr., A 1994, 673, 239. S0003-2700(95)01071-7 CCC: $12.00

© 1996 American Chemical Society

Real-time detection makes it possible to follow the stacking profile of a sample as it migrates through the capillary. Through observing the formation of this profile, one can easily establish the time needed for collecting the whole of the injected sample. This allows the start of destacking to be controlled in a much more effective way than is possible using conventional single-point detection. This technique is valuable too for studies of focusing processes that utilize a hydrodynamic pressure14,15 that is superimposed. In addition, it provides unique possibilities for the evaluation and/or the application of theoretical models of peak broadening in CE. EXPERIMENTAL SECTION Chemicals. Ultra-high-purity grade water (18 MΩ‚cm resistivity) was obtained from an ELGA purification system (High Wycombe, U.K.). 6-Aminocaproic acid (EACA) (puriss) was obtained from Fluka (Buchs, Switzerland). Acrylamide, ammonium persulfate, and N,N,N′,N′,-tetramethylethylenediamine (TEMED), used for the preparation of the coating, were from BioRad (Richmond, CA). 3-(Methacryloxypropyl)trimethoxysilane, used to silylate the capillary before coating, was obtained from ABCR (Karlsruhe, Germany). Glacial acetic acid (HOAc) was received from E. Merck (Darmstadt, Germany). Rhodamine B (base) was from Janssen Chimica (Beerse, Belgium). The following electrolytes were studied: 20 mmol/L EACA, brought to pH 3.6 with acetic acid, was used as the leading (LE) and the background electrolyte (BGE); 50 mmol/L acetic acid, pH 3.1, served as the terminating electrolyte (TE); rhodamine B, dissolved in ethanol and diluted in EACA to a concentration of 10-7-10-10 mol/L, was chosen as the test compound because of its charge properties and good fluorescence characteristics. LIF Detection System. The experimental arrangement employed an excimer laser-pumped dye laser as the excitation source and an image-intensified CCD camera for detecting the fluorescence from the capillary, as shown in Figure 1. An XeCl excimer laser (Estonian Academy of Science, ELI-76E, Tallin, Estonia) with an output energy of up to 200 mJ/pulse at 308 nm served as a pump source for the dye laser. The laser system was operated at a 2 Hz repetition rate during the experiments. The dye laser (ELTO Ltd., VL 2200, Tartu, Estonia) was tuned to 488 nm (laser dye, coumarin 500, Exciton, Dayton, OH). The output from the dye laser was coupled into a 1 mm diameter optical quartz fiber and was imaged to a streak of light along the capillary. Use of the optical fiber provided flexibility and a smooth beam profile. An interference filter was employed to remove any residual light of longer wavelengths. The excitation energy at the capillary was ∼0.25 mJ/pulse, equal to 0.5 mW of excitation power. A typical ITP run was followed for 60 s, resulting in a total exposure of the samples to the laser of less than 30 mJ. Considering this low light power and the fact that the line focus of the excitation beam was broad to assure that the capillary was (10) Nilsson, S.; Johansson, J.; Mecklenburg, M.; Birnbaum, S.; Svanberg, S.; Wahlund, K.-G.; Mosbach, K.; Miyabayashi, A.; Larsson, P.-O. J. Capillary Electrophor. 1995, 2, 46. (11) Sweedler, J. V.; Shear, J. B.; Fishman, H. A.; Zare, R. N.; Scheller, R. H. Anal. Chem. 1991, 63, 496. (12) Wu, J., Pawliszyn, J. Am. Lab. 1994, 26, 48. (13) Beal, S., Sudmeier, J. Anal. Chem. 1995, 67, 3367. (14) Reinhoud, N. J.; Tjaden, U. R.; van der Greef, J. J. Chromatogr., A 1994, 641, 155. (15) Reinhoud, N. J.; Tjaden, U. R.; van der Greef, J. J. Chromatogr., A 1993, 653, 303.

Figure 1. System setup for real-time fluorescence imaging of capillary zone electrophoresis. The excitation light from the excimer laser-pumped dye laser was focused onto a decoated separation capillary by means of an optical fiber and cylindrical quartz lenses. The main section of the capillary (between the holder arms) was decoated. Light emitted from fluorescent bands in the capillary was collected by a thermoelectrically cooled, image-intensified CCD camera. The signal was processed and stored in a computer. During the run the signal could be displayed on screen in real time.

in focus over its whole length, photobleaching of the sample can be considered to have been negligible. The fluorescence emitted from the capillary was imaged onto an 578 × 384 pixel image-intensified CCD camera (Spectroscopy Instruments, CCD-576R, Gilching, Germany) equipped with a standard 50 mm objective (Nikkor 50 mm f 1.8, Nikon, Tokyo, Japan). The CCD was operated at a temperature of -20 °C to keep the detector noise at a low level. The distance from the capillary to the CCD was ∼30 cm. A cutoff filter (Schott, GG530, Mainz, Germany) was placed in front of the CCD to remove scattered laser light. This allowed all fluorescence emission from 530 nm and above to be sampled. The CCD signal was transferred to a PC 486 computer for data processing and display. Full resolution using 578 channels was employed to image a capillary length of ∼7 cm, corresponding to 0.12 mm/pixel. To collect the total signal, the CCD was binned over 50 pixels in a direction perpendicular to the capillary. The resulting profile was a line spectrum displaying fluorescence intensity versus position along the capillary. Since the signal-to-noise ratio was found to be sufficient for each laser shot, no signal integration was performed. Electrophoresis Equipment. The CZE equipment used for the ITP experiments was built in-house. A stabilized power supply (0-30 kV, Zeta-elektronik, Ho¨o¨r, Sweden) was employed. The electrode vessels consisted of Eppendorf microfuge tubes. The capillary was mounted on the Plexiglas holder described below. The capillary was fixed in a horizontal position, allowing the laser light to access the capillary, except for short stretches at each end. The voltage was kept constant, but the current decreased during the run due to the amount of terminating electrolyte entering the capillary. Injections of the analytes were made by hand using a 5 µL Hamilton microsyringe. The internal diameter of the fused-silica capillary was 100 µm (Polymicro Technologies, Phoenix, AZ). Its total length was 18 cm, a 7 cm segment of it Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

2767

Figure 2. Theoretical view of the ITP system with HOAc as the TE and EACA as the BGE and LE. (A) Start of ITP. The entire capillary was filled with sample dissolved in EACA. (B) Stacking complete.

being covered by the camera. To prevent electroosmotic flow, the fused-silica capillary walls were coated with linear polyacrylamide.16 Alignment of the Capillary. The capillary was cut to a 18 cm length and was mounted by being glued at each end within a U-shaped Plexiglas holder, as shown in Figure 1. The capillary was fixed in a stretched position in the holder in order that it would remain within the linear streak of the laser excitation light. This arrangement also reduced the risk of breakage. The outer protective polyimide coating on the capillary had to be removed since it produced a high background fluorescence. The central portion (from 4 to 14 cm) of the capillary was treated for 30 min with bichromic acid in an oven at 50 °C. The outside of the capillary was then washed with water and ethanol and wiped with tissue paper. The decoating of the outside was checked under the microscope, any remaining patches of the coating being removed by additional acid treatment. A more gentle decoating method was employed when transparent coated capillaries were used. The part of the capillary to be decoated was simply immersed in Fluorinert electronic liquid FC-77 (3M, St. Paul, MN) for 10 min, allowing the clear coating to be peeled off with Kleenex. RESULTS AND DISCUSSION To take full advantage of this detection system, a video recording was employed. Some of the experiments presented here are available on World Wide Web http://www.kc.lu.se./ teknlk/Forskn.htm. The present paper indicates to a considerable extent the results that can be obtained using the real-time detection system. Figure 2 shows an outline of the different stages involved in the ITP system that was employed. First, the capillary is filled with the EACA buffer, which is used as the LE. Next, the sample is dissolved in EACA and is then injected into the capillary displacing ∼90% of the buffer in the capillary. Finally, a vial of acetic acid, used as the TE, is placed at the capillary inlet end. When the voltage is applied across the capillary, a rapid analyte focusing occurs due to the differences in conductivity between the different electrolytes present in the capillary. After focusing, destacking can be initiated by replacing the vial of TE at the inlet by a vial of LE. Thereafter, the CZE separation can be started. The destacking and CZE were not performed in our case, however, since our main goal was to monitor and evaluate the preconcentration process. Figure 3 presents an example of an ITP experiment. Our intention was to visualize the stacking effect of the ITP on the (16) Hjerten, S. J. Chromatogr. 1985, 347, 191.

2768 Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

Figure 3. 3-D representation of the rhodamine B profiles obtained by use of the fluorescence imaging system after injection of a long sample plug (90% of the capillary length) during ITP focusing. The fluorescence intensity is presented as a function of capillary position (in CCD pixels) and time after injection. About 7 cm of the capillary is monitored, the ITP system as in Figure 2. The analytes move from the left to the right.

sample which results in a narrowing of the sample zone (∼1 µL ( 1 nL). The fluorescence intensity is displayed as a function of CCD pixels, which corresponds to the position in the capillary, of a selected number of CCD frames. For the sake of clarity, only one-third of the recorded frames are included in this figure. The peak observed, moving from left to right, is the zone of concentrated analyte that follows the leading ion. The leading ion controls the speed of the boundary. In an ITP system run at constant voltage, the electric field strength and the linear speed of the LE gradually decrease as the TE migrates into the capillary. Accordingly, there is a gradual reduction in the linear zone velocity over time from 0.97 mm/s at the start of the window to 0.80 mm/s at the end of the window as shown in Figure 3. Theoretical relationships between ITP zone migration and current have been discussed in detail by Reinhoud et al.9 Although beyond the scope of the present study, the real-time imaging detector device provides an excellent means of verifying such relationships, also in more complex systems involving the superimposing of electroosmotic and hydrodynamic flow. As is evident in Figure 3, the peak width in this example already equals the optical resolution of the system. Before the voltage was turned on, the entire capillary was filled with the rhodamine/EACA sample. Due to the presence of the TE, the sample zone became concentrated at the rear end, the analyte collected moving through the capillary from the left to the right in a narrow zone. The stacking process was already in progress before the zones entered the 7 cm detection window in the capillary, and it continued as the plug moved through the window. Such stacking results in a gradual increase in peak height. With the present setup, however, it is not possible to monitor the process taking place at the capillary inlet. We plan to construct a quartz cuvette that will make this possible. A normal CZE run using only one electrolyte in the whole system was made in order to determine whether the focusing effect in the experiment above was indeed the result of ITP. Again a long sample plug was injected. One can clearly see in Figure 4, that the analyte moves through the capillary as a broad zone and that no focusing occurs. The limited injection volume used in CZE with regard to efficiency can readily be understood in

Figure 4. 3-D representation of rhodamine B profiles obtained with the imaging system after injection of a long sample plug, using a continuous electrolyte system (no ITP stacking initiated). All other parameters are the same as in Figure 3.

Figure 5. Fluorescence imaging profiles during focusing. Solute, rhodamine B. The sample plug was forced into the detection window by injection of TE into the capillary before applying the voltage.

terms of the differences in shape of the sample zones, as is evident in Figures 3 and 4. Two different approaches to facilitating the recording of the stacking process from the start were examined. In the first of these, the injected sample plug (rhodamine/EACA) was pushed by pressure from a syringe containing the TE away from the inlet to the part of the capillary that was covered by the camera. With such a procedure it should be possible to observe a sharp sample peak developing in the central part of the capillary. The result of an experiment of this sort, in which basically the same system as in Figure 3 is employed, is shown in Figure 5. The narrow peak that results from ITP can be seen developing, in the middle of the screen. It can be observed that, after only a brief time (less than 1 s), collection is already under way. During collection, the peak moves toward the negative electrode at the outlet of the capillary, being followed by a mixed zone of the TE and the as yet uncollected sample. This initial mixing of the sample with the TE results from the pressure-driven introduction of TE behind the sample. In the other approach to facilitating observation of the stacking process, the system was forced to progress at a lower rate through ethanol being added to the injected rhodamine sample to a total concentration of 80% (v/v, ethanol/LE). All the other parameters were the same as in previous runs. Due to the low conductivity of ethanol and the resulting low current in the capillary, the whole stacking process is slowed down and can thus be more readily observed, as illustrated in Figure 6. Seven equidistant frames

Figure 6. Fluorescence imaging profiles during focusing. Solute rhodamine B dissolved in 80% (v/v) ethanol/TE. The inset presents selected frames from the main figure that show the concentration process. This recording is available on the World Wide Web http:// www.kc.lu.se./teknlk/Forskn.htm.

Figure 7. Peak width at half-maximum as a function of time based on the data in Figure 6. The observed peak, moving from left to right, is the zone of concentrated analyte, following the leading ion. The leading ion controls the speed of the boundary. In an ITP system run at constant voltage, the electric field strength and the linear speed of the leading electrolyte decreases gradually as the terminating electrolyte migrates into the capillary. A gradual reduction in linear zone velocity is therefore observed over time in Figure 3, from 0.97 mm/s in the beginning of the window to 0.80 mm/s at the end of the window, respectively.

taken from the main figure are shown in the inset of Figure 6 so as to better show the peak focusing. The peak widths calculated as full width at half-maximum (fwhm) found in the individual frames in Figure 6 are shown in Figure 7. The peak width appears initially to decrease linearly with time until it reaches a steady state level at a width of ∼3.5 pixels. This lower limit is probably set by the optical resolution of the system. It is constrained by the resolving power of the optics used and also by the quantification error due to the limited number of CCD pixels, which makes it difficult to determine the exact peak height and thus the fwhm. The total current in the ITP process follows Ohm’s law as the ratio of the applied voltage and the electrical resistance of the liquid in the capillary. The sum of the leading, the terminating, and the sample zone resistances determines the total resistance in the ITP capillary and thus too the possibility of determining the current and the position of the concentrating zone. This is possible as long as the ion composition from sample to sample is well defined through pure samples being dissolved in appropriate buffer systems. Good reproducibility can then be achieved. When the sample matrix involves other ions or differing concentrations, Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

2769

as in “real” samples such as plasma, urine, and lacrimal or interstitial liquid, the applicability of the equations that have been derived is limited to the steady state of ITP. Although it is sometimes possible to transfer the analytes to a well-defined matrix by removing the excess of the sample matrix components, trace analyses in the nanomolar range are very susceptible to any additional manipulation steps. Attempts were made to compare the results of our experiments, one of which is illustrated in Figure 3, with theoretical calculations made for isotachophoretic preconcentration. The velocity ν at which the stacking front migrates through the capillary was calculated as the effective mobility of the leading ion mL times the electric field strength in the leading zone EL, the latter calculated according to ref 9. However, calculating in this way the transition times for the passing of the stacking peak through the window yields migration times that are a factor of 1.5-2 too large. We have not found any satisfactory explanation for this deviation, although working as we did at constant voltage makes the ITP process rather difficult to describe theoretically in a simple way. It should be emphasized that the ITP process is affected by variations in the composition of the sample matrix. High salt content, for example, will change the rate and position of the zone migration in the capillary. Real-time imaging can monitor the momentary position, either of the analyte itself or of a fluorescent marker of suitable mobility. Thus, the proper focusing time can be determined before the CZE mode is switched, also for samples in which the matrix has not been fully controlled. With this study we have shown the advantages for ITP of realtime fluorescence imaging and indicated its considerable potential for providing an understanding of the isotachophoretic preconcentration. Employing the option observing the peak and continuously monitoring peak width, the stacking can be stopped at an optimal point. When the stacking has been fully achieved, the subsequent steps, including destacking and CZE, can be initiated. Too long an ITP time results in a shorter capillary part that is left for CZE separation and thus limits resolution power. It should be pointed out that a very strong dynamic impression was experienced during the ITP experiments and that some of this is lost by the way the figures have to be presented on paper. In some cases, as in Figure 5, fluctuations in intensity between different electropherograms were very pronounced. These fluctuations originate in the pulse-to-pulse fluctuations of the excimer/ dye laser. However, the fluctuations were less disturbing when the line profiles were presented on the screen as a moving peak. Pulse-to-pulse fluctuations can be circumvented by utilizing a continuous laser, which also provides a much higher output power and thus greater system sensitivity. On the other hand, our pulsed laser allows gated detection to be employed, enabeling the system to be operated in full daylight without being shielded against ambient light. In the present setup, the system sensitivity was

2770

Analytical Chemistry, Vol. 68, No. 17, September 1, 1996

estimated to be 5 × 10-10 M for fluorescein,10 which was adequate for these experiments. Employing ITP results in a 2-3 order of magnitude higher concentration sensitivity makes this combination attractive for analysis in the subnanomolar range. CONCLUSION The development of sample focusing techniques is crucial for the successful use of CZE in trace analysis. The processes that take place during ITP-induced stacking were studied in real-time by means of laser-induced fluorescence imaging. An optimal way to perform ITP-CZE on “real” samples, such as plasma, urine, and lacrimal or interstitial liquid, without pretreatment or normalization of the sample matrix is the use of real-time LIF/CCD detection combined with adding to the sample a fluorophore with mobility very similar to that of the analyte ion. An even better system would be to use one fluorophore with a somewhat lower mobility than the analyte ion and another with a somewhat higher mobility than the analyte. Real-time surveillance of this ITP process facilitates choice of the proper time for changing the TE into a leading electrolyte for CZE separation, which can be combined with conventional UV detection of the analyte. With such a detection system, specific information on the position and shape of the sample zones in the capillary at any particular moment during the ITP process can be obtained. A conventional point-monitoring detection system does not provide this kind of data. A practical outcome of this is that access is gained to information on the time needed to focus the plug that is injected when a particular electrolyte system is used or when to stop the stacking process and to initiate the subsequent steps, such as separation by CZE. Furthermore, postrun processing enables detailed studies of the peak profile and position to be carried out. Future modification of the detection setup should facilitate measurements being made at the capillary inlet end as well. ACKNOWLEDGMENT The help of Petr Bocek for the theoretical interpretation of part of the data is gratefully acknowledged. Martin Andersson is thanked for his computer support. Saul Perry’s help with the theoretical calculations is gratefully acknowledged. The work was supported by the Swedish Natural Science Research Council, the Carl Trygger Foundation, the Crafoord Foundation, the Royal Physiografic Society of Lund, and the Magnus Bergwall Foundation. The results described were presented in part at the Fifth Annual Frederick Conference on Capillary Electrophoresis October 25-26, 1994, Frederick, MD, and at HPCE 95, January 29February 2, 1995, WA° rzburg, Germany. Received for review October 25, 1995. Accepted March 27, 1996.X AC951071X X

Abstract published in Advance ACS Abstracts, July 1, 1996.