Anal. Chem. 2003, 75, 1765-1768
Technical Notes
Hadamard Transform Microchip Electrophoresis Combined with Diode Laser Fluorometry Kazuki Hata, Yasuhiko Kichise, Takashi Kaneta, and Totaro Imasaka*
Department of Applied Chemistry, Graduate School of Engineering, Kyushu University, Hakozaki, Fukuoka, 812-8581 Japan
The application of a Hadamard transform technique to microchip electrophoresis is described. The sample is electrokinetically injected into a separation channel and is then detected by diode laser-induced fluorometry. The sample and buffer solutions are introduced into the channel by controlling the high voltages applied to the solutions, according to a code determined by a Hadamard matrix. The S/N ratio of the signal in the electropherogram can be improved by a factor of 5 in comparison with that obtained by a conventional single-injection method, although an 8-fold improvement is theoretically predicted when a 255-order matrix is used. Several techniques involving the use of mathematical methods are available for the improvement of analytical instrumentation. Included among such methods is the application of a Fourier transform technique to infrared absorption spectrometry and nuclear magnetic resonance spectrometry. In our previous report, the employment of a Hadamard transform technique was described for use in capillary electrophoresis, where the S/N ratio was substantially improved, as would be predicted from theory.1-3 In this approach, we injected the sample solution into a capillary, according to a pseudorandom code determined by Hadamard matrix. This was accomplished by optically gated sample injection, in which the fluorescent analyte molecule was photodegraded by intense laser radiation. However, this approach has several disadvantages. For example, it can only be applied to a sample that can be converted into nonfluorescent compounds by photodegradation. In addition, a high-power laser, which is cumbersome and expensive, is required to achieve efficient photodegradation. Therefore, this sample injection technique is not suitable for applications to practical analysis. Several reports have appeared on the application of mathematical methods to microchip electrophoresis, involving the Shah convolution4,5 or the Shah convolution differentiation Fourier transform,6 the Shah and sine convolution Fourier transform,7 and * Corresponding author. Phone: 81-92-642-3563. Fax: 81-92-632-5209. Email:
[email protected]. (1) Kaneta, T.; Yamaguchi, Y.; Imasaka, T. Anal. Chem. 1999, 71, 5444-5446. (2) Kaneta, T. Anal. Chem. 2001, 73, 540A-547A. (3) Kaneta, T.; Kosai, K.; Imasaka, T. Anal. Chem. 2002, 74, 2257-2260. (4) Crabtree, H. J.; Kopp, M. U.; Manz, A. Anal. Chem. 1999, 71, 2130-2138. (5) Kwok, Y. C.; Manz, A. Electrophoresis 2001, 22, 222-229. (6) Kwok, Y. C.; Manz, A. J. Chromatogr., A 2001, 924, 177-186. 10.1021/ac026330e CCC: $25.00 Published on Web 02/22/2003
© 2003 American Chemical Society
cross-correlation techniques.8,9 Microchip technology has several advantages, e.g., in compactness, ruggedness, and integration capability. In this approach, electrokinetic sample injection is successfully used because of its excellent reproducibility.8,10-12 This approach using electrokinetic injection would be potentially useful as a tool for multiple sample injection in Hadamard transform electrophoresis. In this study, we report on Hadamard transform microchip electrophoresis. In this approach, the sample and buffer solutions are injected electrokinetically. Therefore, this technique is applicable even to an analyte that is resistant to photodegradation. Furthermore, this approach greatly simplifies the analytical instrumentation required, since a large high-power laser is not required. We demonstrate herein an improvement in the S/N ratio by the use of the Hadamard transform technique in microchip electrophoresis. We used a compact diode laser as an excitation source in fluorescence spectrometry for practical trace analysis. EXPERIMENTAL SECTION Apparatus. The experimental apparatus used in this study is illustrated in Figure 1A. A confocal fluorescence detection system was constructed by modifying a commercial microscope (Eclipse E600, Nikon, Tokyo, Japan). A diode laser (LDA1035-3, ILEE AG) emitting at 635 nm was used as an excitation source. The diode laser was mounted onto a three-dimensional (XYZ) translation stage (Σ-707-60PS, Sigma Koki, Saitama, Japan). The laser beam passing through a band-pass filter (630AF50, Omega Optical, Brattleboro, VT) is reflected by a dichroic mirror (650DRLP, Omega Optical) and is then focused onto the channel of a microchip at the position 10 mm away from the injection cross by a microscope objective (Nikon, 50× magnification). The fluorescence from the sample is collected by the same microscope objective. After passing through the dichroic mirror and a band(7) McReynolds, J. A.; Edirisinghe, P.; Shippy, S. A. Anal. Chem. 2002, 74, 5063-5070. (8) van der Moolen, J. N.; Poppe, H.; Smit, H. C. Anal. Chem. 1997, 69, 42204225. (9) Fister, J. C.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1999, 71, 44604464. (10) Effenhauser, C. S.; Manz, A.; Widmer, H. M. Anal. Chem. 1993, 65, 26373642. (11) Jacobson, S. C.; Hergenro ¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 1114-1118. (12) Jacobson, S. C.; Koutny, L. B.; Hergenro ¨der, R.; Moore, A. W.; Ramsey, J. M. Anal. Chem. 1994, 66, 3472-3476.
Analytical Chemistry, Vol. 75, No. 7, April 1, 2003 1765
and simplest injection methods,10,14 typically used in microchip electrophoresis, would be difficult to apply to the present study, since the sample injection volume cannot be changed. For this reason, the gated injection method12 was employed since the length of the sample plug can be controlled. The injection time for the sample and buffer solutions was changed by turning on or off the reed relay that is interfaced with the personal computer, according to the Hadamard sequence code. Initially, reservoirs 1 and 2, and the channel between reservoirs 1 and 2 were filled with the sample solution. The other parts of the microchip, i.e., reservoirs 3 and 4 and the channel between reservoirs 3 and 4, were filled with buffer solution. Before starting the sample injection, a high voltage of 0.9 kV was applied to reservoir 1, containing the sample solution, and 1.1 kV to reservoir 3, containing the buffer solution, simultaneously. Reservoir 2 and reservoir 4 were then grounded. To inject the sample, the voltage applied to reservoir 3 was turned off. A high voltage of 1.1 kV was applied to buffer reservoir 3 by turning on (closing) the reed relay for injection of the buffer solution. Data Processing. A fundamental equation of Hadamard transformation is given by15
[η] ) [S] × [E]
Figure 1. (A) Schematic diagram of Hadamard transform microchip electrophoresis. (B) Schematic diagram for application of high voltages. Two power supplies are employed; one is applied to reservoir 1 (0.9 kV) and the other to reservoir 3 (1.1 kV).
pass filter (695AF55, Omega Optical), the fluorescence is focused by a lens (SLB-30-50PM, Sigma Koki) onto a spatial filter (homemade, 1.0-mm i.d.) and is detected by a photomultiplier (R3896, Hamamatsu Photonics, Shizuoka, Japan). The fluorescence signal is digitized by an A/D converter (AD12-16(98), Contec, Tokyo, Japan) and recorded on a personal computer (PC286VE, Epson, Tokyo, Japan). Microchip. Figure 1B shows a schematic diagram that outlines the application of high voltages in microchip electrophoresis. Two power supplies, model HCZE-30PN0.25 (Matsusada Precision Devices, Shiga, Japan), are used for this purpose. A microchip device was purchased from Shimadzu (Microchip Type U, Kyoto, Japan). The device has a cross-shaped channel network consisting of a sample injection channel and a separation channel, at the end of which four outlet holes are manufactured. Four plastic tubes were glued to connect them to the microchip using epoxy resin to form four reservoirs. All the channels are designed to be 20 µm deep and 50 µm wide. A homemade electronic circuit consisting of a photocoupler (TLP521-1Y, Toshiba, Tokyo, Japan) and an amplifier using a transistor (2SC372, Toshiba) was constructed for driving a reed relay (DP1A05-10, Mersault, Tokyo, Japan), which is controlled by the same personal computer as is used to record the fluorescence signal. The reed relay functions as a switch for turning on or off the high voltage applied to buffer reservoir 3. Several electrokinetic injection methods have been reported thus far for use in microchip electrophoresis, e.g., pinched, simplest, and gated injection methods.10,12-14 However, the pinched13 1766
Analytical Chemistry, Vol. 75, No. 7, April 1, 2003
(1)
where η is a series of data, i.e., the observed fluorescence intensities, encoded by a Hadamard matrix, S, which is the n × n matrix consisting of “0” and “1” elements, and E is a series of data representing an electropherogram. The encoded electropherogram, η, is decoded to the electropherogram, E, by multiplying an inverse matrix of S, S-1, as follows.
[E] ) [S]-1 × [η]
(2)
The electropherogram was calculated using the same personal computer described above. Details of the procedure for a Hadamard transformation have been presented in previous papers, as well as in a review article.1-3 Chemicals. Boric acid and tris(hydroxymethyl)aminomethane (Tris) were purchased from Wako Pure Chemical Industries (Osaka, Japan). Monofunctional Cy5 was obtained from Amersham Life Science (Tokyo, Japan). Deionized water was supplied by a water purification system (Elix, Millipore Co. Ltd., Molsheim, France). A borate buffer solution (4 mM) was prepared by dissolving boric acid in deionized water and adjusting the pH to 9.0 using Tris. A Cy5 solution was prepared by dissolving 0.2 mg of Cy5 in 500 µL of anhydrous dimethylformamide followed by dilution with the Tris-Borate buffer. RESULTS AND DISCUSSION Sample Injection Period. In Hadamard transform electrophoresis, a sample is introduced into a separation channel, according to a series of Hadamard codes consisting of 0 and 1, i.e., 10110.... To optimize the sample injection period, i.e., the time (13) Jacobson, S. C.; Hergenro ¨der, R.; Koutny, L. B.; Warmack, R. J.; Ramsey, J. M. Anal. Chem. 1994, 66, 1107-1113. (14) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895-897. (15) Harwit, M.; Sloane, N. J. A. Hadamard Transform Optics; Academic Press: London, 1979.
Figure 2. Electropherogram obtained by a single-injection technique. Conditions: analyte, Cy5; concentration, 1.6 nM; Tris-Borate buffer, pH 9.0; injection time, 1.0 s; effective separation length, 10 mm. The inset is an electropherogram obtained by increasing the analyte concentration 100 times (160 nM) and by extending the effective separation length 2.5 times (25 mm), to identify the components in the sample.
for introducing the sample or buffer solution into a single segment, the solutions were alternately injected into the channel at intervals of 0.5, 1.0, and 1.5 s. When the injection period exceeded 1.0 s, the fluorescence signal was sufficiently modulated. The sample solution was, then, injected into the channel according to the Hadamard code, to investigate peak broadening arising from a lengthy injection time. The signal peak obtained by the inverse Hadamard transform became slightly broader when the injection period was increased from 1.0 to 1.5 s. Thus, a shorter injection period, i.e., 0.5 s, provides a poor S/N ratio, and a longer injection period, i.e., 1.5 s, provides a poor separation resolution under the conditions used here. These results suggest that the dispersion required to cover the distance between plugs was separated by 1.0 s. For optimum sample injection, the time duration was adjusted to 1.0 s throughout this work. Electropherogram. Figure 2 shows the electropherogram obtained by a single injection of a sample solution containing 1.6 nM Cy5. As shown in the inset of Figure 2, two components are present in the sample and are assigned to the hydrolysis product of Cy5 and an impurity in the labeling reagent, as reported in our previous paper.16 Due to the poor S/N ratio, these peaks are not clearly identified in an electropherogram obtained by the singleinjection method. Figure 3A shows the raw data obtained by a pseudorandom 255-step sample injection, and Figure 3B shows the result obtained by the inverse Hadamard transform of the signal shown in (A). The S/N ratio is substantially improved, and two peaks are more clearly visible in Figure 3B. The resolution in Figure 3B seems to be slightly poorer than that in (A). It was, however, found to arise not from the difference in the separation resolution but from irreproducibility in the electroosmotic flow. Improvement of Sensitivity. A Hadamard transform technique improves the S/N ratio by a factor of (n + 1)/2n1/2,15 where n is the order of the matrix, i.e., the number of sample injections used to obtain an electropherogram. In the present experiment, the value of n is 255, and therefore, the S/N ratio should be improved, in theory, by a factor of 8.0. However, only a 5-fold improvement was observed in Figure 3B. This discrepancy may arise from several factors, e.g., errors induced in the process of numerical calculation by a computer.3 To investigate the source (16) Jing, P.; Kaneta, T.; Imasaka, T. Electrophoresis 2002, 23, 550-555.
Figure 3. (A) Raw data obtained by multiple sample injection, according to the Hadamard sequence code. The 255-order matrix was used in the Hadamard transformation; i.e., the sample was injected 255 times according to the Hadamard sequence code. Other conditions are the same as those shown in Figure 2. (B) The result obtained by transformation of the data shown in (A). A sequence of the data between 255 and 509 in (A) is used for calculation of the result shown in (B).
Figure 4. Expanded view of the data shown in Figure 3A (solid line) and the pseudorandom sequence code of Hadamard matrix (dashed line).
of noise in the electropherogram, the observed raw data (Figure 3A) were expanded and are shown in Figure 4, in which the Hadamard code used for sample injection is also presented. The signal apparently increases for “1” and decreases for “0”, as would be expected from the theory. However, the signal level varies in a random manner, which can be attributed to a shot noise arising from weak fluorescence or to light scattering arising from the excitation source. In addition, such fluctuation in the signal level might be caused by imperfect switching of the power supplies between “0” and “1” states and also interdiffusion of the plugs of sample as they migrate along the channel. The concentration detection limit has been reported to 10-13 M in Hadamard transform capillary electrophoresis.3 However, the signal intensity in the present study is much smaller, since a near-infrared weakly fluorescent dye was used as a sample for the diode laser Analytical Chemistry, Vol. 75, No. 7, April 1, 2003
1767
fluorometry. Moreover, a notch filter, which is effective in reducing the scattered light, was not available for use in the present study. To improve the separation resolution, increasing the applied voltage and the length of the separation channel is a possibility, although they also improve the resolution in conventional singleinjection microchip electrophoresis.
On the other hand, the present method based on electrokinetic injection may be applied not only to fluorescence spectrometry but to other analytical methods such as absorption spectrometry or electrochemical detection. Thus, the present method may have a variety of applications and would be of potential use in practical trace analysis.
CONCLUSION The use of electrokinetic sample injection in Hadamard transform microchip electrophoresis is demonstrated. The sensitivity was improved by a factor of 5, although an 8-fold enhancement would be expected from theory under present conditions (n ) 255). The photodegradation technique previously used for sample injection in Hadamard transform electrophoresis can be applied only when the analyte molecule can be decomposed to form a nonfluorescent compound by intense laser radiation.3
ACKNOWLEDGMENT This work was supported by a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture of Japan.
1768
Analytical Chemistry, Vol. 75, No. 7, April 1, 2003
Received for review November 20, 2002. Accepted January 30, 2003. AC026330E
