Microfluidic Chip toward Cellular ATP and ATP-Conjugated Metabolic

Anal. Chem. 2005, 77, 573-578

Microfluidic Chip toward Cellular ATP and ATP-Conjugated Metabolic Analysis with Bioluminescence Detection Bi-Feng Liu,*,†,‡ Motoaki Ozaki,‡ Hideaki Hisamoto,‡ Qingming Luo,† Yuichi Utsumi,§ Tadashi Hattori,§ and Shigeru Terabe‡

The Key Laboratory of Biomedical Photonics of Ministry of Education, Huazhong University of Science and Technology, Wuhan 430074, P.R. China, and Graduate School of Science and Laboratory of Advanced Science and Technology for Industry, University of Hyogo, Kamigori, Hyogo, 678-1297, Japan

In this article, a microfluidic platform integrating capillary electrophoresis and bioluminescence (BL) detection that was fabricated in poly(dimethylsiloxane) (PDMS) with labon-a-chip technology was demonstrated for cellular metabolic analyses. A microchannels network, “cross combining with Y”, was designed to perform on-chip sample preparation, separation, and BL detection of ATP and ATP-conjugated metabolites, using firefly luciferin-luciferase BL system. A dynamic modification of the channel wall of PDMS proved to be crucial to reverse the direction of electroosmotic flow (EOF), which was uniquely achieved by a prewash cycle with a cationic surfactant didodecyldimethylammonium bromide. The influences of surfactant on the EOF and BL reaction were also investigated. Quantitative analyses revealed a dynamic linear range over 2 orders of magnitude for ATP, with a detection limit down to submicromolar (midattomole). The method was validated by measuring cellular ATP of E. coli. with direct on-chip cell lysis. Further work was emphasized on ATPconjugated metabolite analysis, using galactose as an example. Assays of galactose in human urine samples confirmed the reliability of the protocol, which revealed good prospect of this platform for ATP-conjugated submetabolomic profiling. With advent of the postgenome era, probing the behaviors of genes in complex biochemical networks has become of great interest, which involves emerging “-omics” research areas such as transcriptomics, proteomics, and metabolomics. It is currently a major challenge to comprehensively profile molecular components (e.g., proteins, metabolites, and signal molecules) at the system level, particularly their dynamic changes inside the cell in various physiological or pathological conditions, which strongly fosters and promotes new advancement in analytical chemistry. Many innovations in technique and device have been achieved. For example, miniaturized total analysis system * Corresponding author. E-mail: [email protected]. Tel: +86-2787792033. Fax: +81-27-87792034. † Huazhong University of Science and Technology. ‡ Graduate School of Science, University of Hyogo. § Laboratory of Advanced Science and Technology for Industry, University of Hyogo. 10.1021/ac0490447 CCC: $30.25 Published on Web 11/26/2004

© 2005 American Chemical Society

(µTAS), referred to as lab-on-a-chip,1-4 is undergoing rapid development. Based on µTAS strategy, microchip capillary electrophoresis (µCE), which promises the advantages of high-performance separation ability and low sample/reagent requirement, as well as integration and high-throughput capability, has made significant progress since its first introduction by Manz et al.1 Developments in many aspects of µCE, e.g., fabrication, separation, detection and signal process, integrated sample preparation, and reaction, have been recently accomplished4-6 that reveal great potentials for a wide range of chemical and biochemical applications,7-9 in which genomic10,11 and proteomic12-14 analyses are gaining precedence. For metabolomic analysis,15 however, things become complicated. Though the total number of metabolic species in cells (usually ∼1000) is far lower than that of proteins, which calls for less requirements of separation, the divergent characteristics of these metabolites make it a big problem for detection. There is no likely universal way to label all these metabolites as can be done for DNA fragments or proteins for highly sensitive laserinduced fluorescence (LIF) detection. Mass spectrometry (MS) seems suitable, considering the nondiscriminating detectability. (1) Manz, A.; Graber, N.; Widmer, H. M. Sens. Actuators, B 1990, 1, 244-248. (2) Harrison, D, J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895-897. (3) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 1114-1118. (4) Vilkner, T.; Janasek, D.; Manz, A. Anal. Chem. 2004, 76, 3373-3386. (5) Reyes, D. R.; Iossifidis, D.; Auroux, P. A.; Manz, A. Anal. Chem. 2002, 74, 2623-2636. (6) Auroux, P. A.; Iossifidis, D.; Reyes, D. R.; Manz, A. Anal. Chem. 2002, 74, 2637-2652. (7) Effenhauser, C. S. In Microsystem Technology in Chemistry and Life Science; Manz, A., Becker, H., Eds.; Springer: Berlin, 1999; pp 51-82. (8) Dolı´k, V.; Liu, S.; Jovanivich, S. Electrophoresis 2000, 21, 41-54. (9) Khandurina, J.; Guttman, A. J. Chromatogr., A 2002, 943, 159-183. (10) Emrich, C. A.; Tian, H.; Medintz, I. L.; Mathies, R. A. Anal. Chem. 2002, 74, 5076-5083. (11) Tabuchi, M.; Ueda, M.; Kaji, N.; Yamasaki, Y.; Nagasaki, Y.; Yoshikawa, K.; Kataoka, K.; Baba, Y. Nat. Biotechnol. 2004, 22, 337-340. (12) Ramsey, J. D.; Jacobson, S. C.; Culbertson, C. T.; Ramsey, J. M. Anal. Chem. 2003, 75, 3758-3764. (13) Jin, L. J.; Giordano, B. C.; Landers, J. P. Anal. Chem. 2001, 73, 55075512. (14) Li, J.; Kelly, J. F.; Chernushevich, I.; Harrison, D. J.; Thibault, P. Anal. Chem. 2001, 73, 599-609. (15) Terabe, S.; Markuszewski, M. J.; Inoue, N.; Otsuka, K.; Nishioka, T. Pure Appl. Chem. 2001, 73, 1563-1572.

Analytical Chemistry, Vol. 77, No. 2, January 15, 2005 573

But the high cost and large size are quite incompatible with the concept of µTAS. Consequently, the development of an alternative detector of high sensitivity, simplicity, and low cost and that can be directly integrated onto the microchip has been highly focused, e.g., electrochemical16-21 or chemiluminescent (CL)22-27 detection. In our previous report,23 different configurations of on-chip CL detection coupled to µCE were demonstrated for metal ion analysis and chiral recognition, with comparative sensitivity to LIF and MS. It is interesting to construct a µCE-CL platform for cellular metabolomic analysis. However, the same situation as for LIF detection is encountered in that few metabolites are applicable for CL detection without labeling. ATP, well known as energy currency in cells,28 plays a fundamental role in cell metabolism. ATP is involved in many metabolic processes. Because ATP can be easily detected by CL or, more biologically, by bioluminescence (BL), these ATP-conjugated metabolites can be analyzed indirectly by monitoring the changes (formation or loss) of ATP. Such a strategy was previously employed to analyze creatine kinase with CE.2929 In this work, a novel microfluidic platform that integrated CE and BL detection was proposed for ATP-conjugated submetabolome analyses. A microchannel network, “cross combining with Y”, was designed and fabricated in poly(dimethylsiloxane) (PDMS) to perform on-chip sample preparation, separation, and BL detection of ATP and ATP-conjugated metabolites, using the firefly luciferin-luciferase BL system. Dynamic modification of the channel wall surface with a cationic surfactant, didodecyldimethylammonium bromide (DDAB), was conducted to reverse the electroosmotic flow (EOF) for simultaneously efficient ATP and BL reagent transport, uniquely using a preseparation rinse cycle. Under the optimized conditions, ATP analysis could be accomplished within 30 s with this platform that was also applied to monitor intracellular ATP of Escherichia coli., coupled with on-chip cell lysis. Further work was emphasized on ATP-conjugated metabolite analysis, using galactose as an example. Assays of galactose in human urine samples validated the reliability of the protocol that paves a new avenue to metabolomic profiling, that is, ATP-conjugated submetabolome analyses. (16) Ertl, P.; Emrich, C. A.; Singhal, P.; Mathies, R. A. Anal. Chem. 2004, 76, in press. (17) Liu, Y.; Vickers, J. A.; Henry, C. S. Anal. Chem. 2004, 76, 1513-1517. (18) Wang, J.; Chatrathi, M. P.; Musameh, S. Anal. Chem. 2004, 76, 298-302. (19) Manica, D. P.; Mitsumori, Y.; Ewing, A. G. Anal. Chem. 2003, 75, 45724577. (20) Osbourn, D. M.; Lunte, C. E. Anal. Chem. 2003, 75, 2710-2714. (21) Zeng, Y.,; Chen, H.; Pang, D. W.; Wang, Z.-L.; Cheng, J.-K. Anal. Chem. 2002, 74, 2441-2445. (22) Qiu, H.; Yan, J.; Sun, X.; Liu, J.; Cao, W.; Yang, X.; Wang, E. Anal. Chem. 2003, 75, 5435-5440. (23) Liu, B.-F.; Ozaki, M.; Ustumi, Y.; Hattori, T.; Terabe, S. Anal. Chem. 2003, 75, 36-41. (24) Arora, A.; Eijkel, J. C. T.; Morf, W. E.; Manz, A. Anal. Chem. 2001, 73, 3282-3288. (25) Mangru, S. D.; Harrison, D. J. Electrophoresis 1998, 68, 2301-2307. (26) Hashimoto, M. H.; Tsukagoshi, K.; Nakajima, R.; Kondo, K.; Arai, A. J. Chromatogr., A 2000, 867, 271-279. (27) Su, R.; Lin, J.-M.; Qu, F.; Chen, Z.; Gao, Y.; Yamada, M. Anal. Chim. Acta 2004, 508, 11-15. (28) Berg, J. M.; Tymoczko, J. L.; Stryer, L. In Biochemistry; WH Freeman: New York, 2002. (29) Fujima, J. M.; Danielson, N. D, Anal. Chim. Acta 1998, 375, 233-241.

574 Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

Figure 1. Schematic layouts of µCE-BL system. Channel dimension, 40 µm in width and 50 µm in depth, except that the channel width of the stem in Y shape of chip was 80 µm. The length for each channel segment is given in millimeters.

EXPERIMENTAL SECTION Materials. Sylgard 184 silicone elastomer and curing agent were obtained from Dow Corning (Midland, MI). SU8 100 negative photoresist was purchased from Microchem (Newton, MA). AMP, ADP, ATP, ATP assay mix, galactose, galactokinase (from galactose-adapted yeast), and the powder of 2×YT culture media were bought from Sigma (St. Louis, MO). Cationic surfactants cetyltrimethylammonium chloride (CTAC) and DDAB were purchased from TCI chemicals (Tokyo, Japan). All other reagents were of analytical grade. Water purified by the Milli-Q Labo system was used for the preparation of all solutions that were further autoclaved (121 °C for 5 min). Fabrication of PDMS Microchip. Figure 1 showed the pattern of the channel network that was microfabricated in PDMS following the method described in a previous paper.30 In brief, the chip layout made by AutoCAD was written onto a glass mask for UV photolithography, through which a mask was further prepared in gold by electroplating for synchrotron radiation lithography. The glass wafer was used as the substrate of a synchrotron radiation-cured polymer mold for replication of the channel pattern to PDMS. A pretreated wafer was spin-coated with negative photoresist (SU8 100). After a preexposure bake, the wafer was lithographed with synchrotron radiation (1.5 GeV in storage ring) scanning using the beam line BL11 (6 keV) at NewSUBARU31 (Himeji Institute of Technology). Following a postexposure bake to bridge epoxy polymer, the wafer was developed, subsequently rinsed, and hard baked to remove organic solution. Thus, a patterned SU8 microstructure was formed, onto which a thermal cured Si elastomer-PDMS that was prepared by mixing Sylgard 184 silicone elastomer and curing agent at a ratio of 10:1 (w/w) was poured. After a soft baking procedure with sequentially cooling to room temperature, the PDMS layer that transcribed channel pattern was peeled off from the SU8 mold. Holes were directly punched on the PDMS layer to access the channels as sample and buffer reservoirs. Another flat PDMS layer was bonded to the patterned PDMS layer to form a closed microchannel structure. Before bonding, both two PDMS layers were washed to remove the contaminants. The two layers were then exposed to air plasma for 10 min in an air plasma chamber (model PC-101A, Yamato, Japan) for irreversible adhesion. Cell Culture of E. coli. E. coli (strain BL21) was used for cell culture in 2×YT culture media at 37 °C overnight. The culture was then diluted 1:200 with fresh 2×YT media and was grown to (30) Liu, B.-F.; Ozaki, M.; Hisamoto, H.; Ustumi, Y.; Hattori, T.; Terabe, S. Lab Chip 2004, 4, 368-371. (31) Mekaru, H.; Utsumi, Y.; Hattori, T. Nucl. Instrum. Methods Phys. Res., A 2001, 467-468, 741-744.

midlog phase when the optical density at 600 nm (OD600) was measured to be 0.424 using a Shimadzu spectrophotometer with a 1 × 1 cm cell. The bacterial cells were collected at 15 000 rpm for 1 min at 4 °C. The bacteria pellet was immediately frozen at -80 °C for further use. Preparation of Human Urine Sample. The human urine samples were prepared as described in ref 32. Human urine was collected from three voluneer adult donors (2 males and 1 female) immediately after rising in the morning and stored at 4 °C. Before analysis, urine samples were diluted 5-fold in 20 mM phosphate-buffered saline (PBS) at pH 7.8 and centrifuged for 2 min at 12 000 rpm. Experimental Procedures. For new microchips, channels were washed sequentially using methanol, water, 1 M NaOH, 0.1 M HCl, 0.1 M NaOH, and water. For normal EOF conditions, they were then rinsed and conditioned with buffer solution, 20 mM PBS (pH 7.8). Between runs, channels were rinsed in order with 0.1 M NaOH, water, and buffer to ensure the reproducibility. In the case of reverse-EOF conditions, dynamic coating of the wall by a cationic surfactant CTAC or DDAB was performed. In detail, the channels were rinsed consequtively with 0.1 M NaOH, water, and 0.5 mM CTAC or 0.1 mM DDAB dissolved in buffer solution for 5 min. Then a flush with running buffer for 1 min was conducted for further electrophoresis. Between runs, the abovementioned coating procedures were repeated to ensure a stable reversed EOF. Gated injection mode was employed for injections in all experiments. To analyze ATP, all channels and reservoirs were first filled with buffer. Buffer solution was then replaced by BL solution in the BL reagent reservoir (BRR). Afterward, a vacuum was applied at the buffer waste reservoir (BWR) to drive the BL solution into the channel connecting to the BRR. Thus, the solution in the channel connecting to the BWR would be the mixture of buffer and BL reagent. Then the buffer solution in the sample reservoir (SR) was changed to sample. And the solution in the BWR was replaced by half-diluted BL reagent with buffer. For initialization, voltages for all reservoirs were -1567 V at SR, -867 V at sample waste reservoir (SWR), -1800 V at buffer reservoir (BR), grounding at BWR, and -834 V at BRR, respectively. At injection, the voltages were changed to -1567 V, -1345 V, -1322 V, grounding, and -834 V, respectively. After injection, the voltages at all reservoirs were back to the values at the initialization phase for separation. Instruments. All µCE-BL analyses were performed on a home-built µCE system for BL detection. Briefly, a PMT (Hamamatsu Photonics) powered with a voltage of -840 V was mounted to an inverted microscope (Carl Zeiss) that was placed in a plastic box as a dark room. BL light was collected by a 20× objective. No filter was used in the optical path. The PMT current was converted and amplified into a voltage by a picoamperometer (Takeda, Tokyo, Japan) and then acquired by a data acquisition board (National Instrument) at a sampling rate of 50 Hz. A moving average (35 points) was used for all data processes. Programs locally written in Labview 6.0 (National Instrument) were used to control the high-voltage supply group (Matsusada Precision Devices, Kusatsu, Japan) and data acquisition. (32) Britz-McKibbin, P.; Markuszewski, M. J.; Iyanagi, T.; Matsuda, K.; Nishioka, T.; Terabe, S. Anal. Biochem. 2003, 313, 89-96.

RESULTS AND DISCUSSION The purpose of this work was to develop a microfluidic chip for metabolic analysis. The BL detection method was chosen because of its benefits such as high sensitivity, low cost, and simple instrument configuration, which has proved to be quite compatible with µCE from our previous investigations. ATP has been well-characterized using the firefly luciferin-luciferase BL reaction,33 which could provide a very low detection limit down to one or two yeast cells. Considering many metabolic processes were ATP-dependent, they could be indirectly analyzed by monitoring the change (formation or loss) of ATP. Consequently, we attempted to construct a µCE-BL platform for ATP analysis, as the first step for ATP-conjugated submetabolome profiling. ATP Analysis. To achieve µCE-BL analysis, a straightforward idea was a µCE separation hyphenated to a postseparation BL reaction. In our previous paper,23 a comprehensive comparison and evaluation of different strategies for on-chip CL detection coupling to µCE was demonstrated. It was suggested that a microchip pattern of “cross combining with Y” as illustrated in Figure 1 would be excellent for those BL systems where BL reagent could be effectively delivered by EOF. The sample was injected at the “cross”, delivered to the joint of “Y”, where it mixed and reacted with the BL reagent electroosmotically driven through one arm of “Y” from BRR, and then emitted luminescence in the stem of “Y”. While such a strategy was applied to this work, however, a big problem occurred. The EOF in the PDMS channel was not strong enough to deliver ATP, whose mobility was larger than that of EOF in the reversed direction. Thus, there are two possible choices in practice to simultaneously transport ATP and BL reagent, enhancing or reversing EOF. Both methods involved a modification of the channel surface wall, unless a radial electric field control to the chip channel was performed.34 Because PDMS was an emerging material for µCE, relevant knowledge on its surface chemistry and physics was poor.35 Liu et al.36 and Dou et al.37 recently studied the dynamic modification of the PDMS channel to stabilize EOF. Although the reproducibility of separation could be remarkably improved, no enhancement of EOF was found in the neutral pH range.37,38 To the best of our knowledge, no report concerning EOF reversal was found for the PDMS microfluidic chip. In this experiment, EOF enhancement was initially attempted by an addition of SDS into the running buffer that was a useful approach in capillary electrochromatography with a neutral stationary phase. However, no obvious enhancement was accomplished. Thus, reversing EOF was further investigated by dynamic modification using the cationic surfactant CTAC as a running buffer additive. Sulforhodamine B dissolved in 20 mM PBS (pH 7.8) running buffer at a concentration of 10 µM was chosen as a test sample, with LIF detection. That running buffer was chosen mainly for matching the best buffer condition for BL (33) Campbell, A. K. Chemiluminescence: Principles and Applications in Biology and Medicine; Ellis Horwood and VCH: Weinheim, 1988. (34) Buch, J. S.; Wang, P.-C.; DeVoe, D. L.; Lee, C. S. Electrophoresis 2001, 22, 3902-3907. (35) Makamba, H.; Kim, J. H.; Lim, K.; Park, N.; Hahn, J. H. Electrophoresis 2003, 24, 3607-3619. (36) Liu, Y.; Fanguy, J. C.; Bledsoe, J. M.; Henry, C. S. Anal. Chem. 2000, 72, 5939-5944. (37) Dou, Y. H.; Bao, N.; Xu, J. J.; Chen, H. Y. Electrophoresis 2002, 23, 35583566. (38) Yassine, M. M.; Lucy, C. A. Anal. Chem. 2004, 76, 2983-2990.

Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

575

Figure 2. Influence of CTAC concentration on EOF mobility and BL reaction sensitivity. (a) EOF; (b) normalized BL reaction sensitivity.

Figure 3. Comparison of EOF reversal by dynamic coating with rinse cycle using DDAB and CTAC. Buffer, 20 mM PBS (pH 7.8). Sample, 10 µM sulforhodamine B in buffer. Injection time, 0.2 s. Strength of electric field for injection and separation, 333 V/cm.

reaction of ATP. In such a buffer, sulforhodamine B was negatively charged very slightly. As a result, it could be approximately considered as an EOF marker, with no respect to its charge and possible retention by channel surface. Figure 2a showed the migration behavior of the analyte with different CTAC concentrations in the buffer. With the increase of CTAC concentration, EOF mobility increased rapidly from 0.1 to 0.3 mM and then increased slowly until it reached a plateau at 0.5 mM, where the surface modification proved to be saturated. Although the reversal of EOF could be achieved with 0.1 mM CTAC in running buffer, the stability of EOF was unreliable until the concentration of CTAC reached 0.4 mM. However, the BL sensitivity for ATP analysis declined ∼50% at 0.4 mM CTAC as shown in Figure 2b. 576 Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

Figure 4. Electropherogram of ATP assay by µCE-BL microfluidic platform. Buffer, 20 mM PBS (pH 7.8). Sample, 5.0 µM ATP in buffer. BL reagent, ATP assay mix. Injection time, 0.2 s. Strength of electric field for injection and separation, 333 V/cm. 1, 2, and 3 are three injections in a run and their corresponding peaks.

It was evident that the addition of surfactant would significantly deteriorate the activity of luciferase. Moreover, CTAC in buffer could compete with Mg2+ to form an ion pair, or further an ATP-Mg2+-CTAC complex, each of which would lead to an inhibition of the formation of ATP-Mg2+, which was the right species participating in the BL reaction. Recently, there were several reports38,39 published concerning dynamic coating of cationic surfactant DDAB on the inner surface of the glass capillary to reverse EOF direction in conventional CE by simply using rinsing procedures. The presence of DDAB in the running buffer was not needed. This methodology was tested in our work to reverse the EOF on the PDMS surface. The coating procedure is described in the Experimental Section. Figure 3 depicts the result. The reversed EOF was very stable over 300 s, which was already sufficient for routine µCE in which the analysis is often achieved in less than 1 or 2 min. And it could be ensured with a repeated preanalysis rinse cycle. The instability after 350 s was believed to be the limited capacity of the low volume of running buffer. The reversed EOF velocity on PDMS was lower than that on glass. A possible reason was that the adsorption of DDAB onto the PDMS surface was not as effective as onto a glass surface, because of the lower charge density on the PDMS surface. A comparison with using CTAC was also included in Figure 3. Obviously, CTAC was less effective in terms of both the EOF velocity and stability. Thus, the µCE-BL platform shown in Figure 1 with effective EOF reversal using a prewash of DDAB could be employed for further ATP analysis. The sample ATP dissolved in buffer solution of 20 mM PBS (pH 7.8) was injected at the cross with gated injection mode. It was then transported through the separation channel to the Y joint where ATP mixed and reacted with BL reagent (firefly ATP assay mix dissolved in the same buffer solution as the sample), continuously driven by EOF through the (39) Melanson, J. E.; Baryla, N. E.; Lucy, C. A. TrAC, Trends Anal. Chem. 2001, 20, 365-374.

Figure 5. Schematic description of µCE-BL procedures coupling with on-chip cell solubilization. (1) Picking up log-phase E. coli cells from cell culture. (2) On-chip cell lysis and sample extraction. (3) Injection with gated mode. (4) Separation from cell background. (5) Mixing and reacting with BL reagent. (6) Result read-out.

second arm of Y from BRR, and eventually emitted BL light in the stem of Y. Half-diluted BL reagent with buffer solution was also presented in the BWR. The voltage manipulation at all reservoirs was described in the Experimental Section. Figure 4 showed a typical electropherogram involving three consecutive injections at 0, 10, and 20 s. As depicted in Figure 4, the analyte ATP could be detected within 30 s, with high reproducibility. The detection limit for ATP was found to be 0.20 µM (∼100 amol) calculated from a peak height of three times of the noise, with a standard deviation of 5.6% for peak height and 4.2% for migration time. A calibration curve was established that exhibited a dynamic linear range of over 2 orders of magnitude between peak height and concentration of ATP ranged from 0.2 to 50 µM, with a correlation coefficient of 0.9956. An interference test of ADP and AMP, which could inhibit the BL reaction, was conducted by adding 50 µM ADP or AMP into a 0.25 µM ATP sample. No inhibition was found within the standard deviation, which guaranteed the analytical accuracy of ATP in the presence of ADP and AMP that was definitely important to indirectly analyze those metabolites by reactions converting ADP to ATP. The interference effect of Triton X-100 that was widely used for cell lysis was also evaluated with an addition of 0.5% concentration in the ATP sample. Fortunately, the same result was revealed as found for ADP and AMP. Hence, no dilution of the sample from cell extract is needed, which is essential in the usual case of ATP analysis without a cell background separation procedure. Cellular ATP Monitoring. To test the validation of the platform for real-world samples, analysis of cellular ATP of E. coli was implemented. For a µTAS chip, one of important advantages is its integration capability that could combine sample preparation onto microchip. Here, the cell lysis and sample extraction was directly conducted in the sample reservoir. In detail as described in Figure 5, cultured E. coli cells at logarithm phase with an OD600 value of 0.424 were resuspended using 42.4 µL of 20 mM PBS (pH 7.8) in a tube. An aliquot of 10 µL was added in the sample reservoir. Then, cell extraction solution of an equal

volume containing 20 mM PBS (pH7.8) and 1% Triton X-100 was added and mixed with the cell sample. After a 3-min extraction, the program of ATP analysis was started. A concentration of 1.62 µM ATP corresponding to ∼1.62 amol/E. coli cell (strain BL21, OD600 value of 1.0 representing 100 million cells of E. coli) on average (n ) 4; RSD ) 4.6%) was determined. ATP-Conjugated Metabolite Analysis. It is well known that many metabolic processes inside cells are ATP dependent, e.g., biosynthesis, active mass transport thru cellular membranes, etc. Such dependences provide very efficient approaches to monitor these so-called ATP-conjugated metabolites through measuring ATP change. In this work, galactose, as an example, was introduced to show the potential capability of the above platform for ATP-conjugated submetabolome analysis. Galactose can react with ATP to form galactose 1-phosphate and ADP, in the presence of galactokinase and Mg2+. Because the formed ADP does not interfere with the measurement of ATP as shown above, galactose can be analyzed by monitoring the ATP consumption in the sample. In practice, 10 µL of 0.1 mM ATP prepared in 20 mM PBS (pH7.8) buffer was added into the sample reservoir that was employed as a biochemical reactor. A 5-µL enzyme solution containing 0.1 unit/mL galactokinase, 10 mM Mg2+, and 1 mg/mL BSA prepared in 20 mM PBS (pH7.8) buffer and a 5-µL galactose sample dissolved in 20 mM PBS (pH7.8) were then successively added and mixed with ATP solution. After a reaction time of 5 min, ATP concentration in the sample reservoir was measured following the protocol as described in the above sections. A series of galactose standards were utilized for calibration that showed a dynamic linear range from 10 µM to 1 mM between concentration and BL intensity, with a correlation coefficient of 0.9904. To test the applicability of this method, galactose in urine from three adult donors were analyzed. The results suggested the galactose amount in these urine samples (0.24 ( 0.01, 0.17 ( 0.01, and 0.45 ( 0.02 mM) were within normal range (for adult, 0-0.57 mM)40 based on triplicate measurements. Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

577

CONCLUSION A novel platform of a microfluidic chip integrating CE separation and BL detection was demonstrated for ATP-conjugated metabolic analyses. Dynamic modification of the channel surface with the cationic surfactant DDAB was performed to effectively reverse EOF, which proved to be necessary to deliver ATP whose intrinsic mobility was faster than that of the original EOF in the reversed direction. A microchip pattern of cross combining with Y was designed to implement µCE separation with the cross and BL detection with the Y. Under optimized conditions, ATP analysis could be realized within 30 s with a detection limit down to 0.20 µM, representing a potential platform for ATP-conjugated metabolic analyses. No interferences from ADP, AMP, and cell lysis detergent Triton X-100 were found within analysis standard deviation, which was crucial to perform ATP-conjugated metabolites analyses and minimize cell background. The method was successfully employed to monitor cellular ATP of E. coli coupled with on-chip cell lysis and sample extraction. Further work was (40) Hicks, J. M.; Young, D. S. Directory of Rare Analyses; AACC Press: Washington, 1997.

578

Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

emphasized on ATP-conjugated metabolite analysis, using galactose as an example. Tests of galactose in human urine samples validated the reliability of the protocol. It revealed a promising prospect for this platform or method in ATP-conjugated submetabolomic profiling. ACKNOWLEDGMENT The authors gratefully acknowledge the fellowship and the Grant-in-Aid for Scientific Research (No. P01082) supported by Japan Society for the Promotion of Science (to B.-F.L.) and financial support in part from National Natural Science Foundation of China (No. 20105004, No. 20405006). Special thanks are given to Dr. Z.-W. Guan and Prof. T. Iyanagi in the Department of Life Science for their kind help on cell culture and discussion.

Received for review June 30, 2004. Accepted October 14, 2004. AC0490447