A Microfabricated Fluidic Device for Performing Two-Dimensional

View: PDF | PDF w/ Links | Full Text HTML. Citing Articles ... J. Michael Ramsey. Analytical Chemistry 2013 85 (8), 4100-4106 ..... Analytical Chemist...
0 downloads 0 Views 107KB Size
Download PDF
Anal. Chem. 2000, 72, 5244-5249

A Microfabricated Fluidic Device for Performing Two-Dimensional Liquid-Phase Separations Roy D. Rocklin,†,‡ Roswitha S. Ramsey,§ and J. Michael Ramsey*,§

Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831-6142, and Caliper Technologies Corporation, Mountain View, California 94043-2234

A microfabricated fluidic device that combines micellar electrokinetic chromatography and high-speed openchannel electrophoresis on a single structure for the rapid automated two-dimensional analysis of peptides has been devised and demonstrated. The microchip operates by rapidly sampling and analyzing effluent in the second dimension from the first dimension. Second-dimension analyses are performed and completed every few seconds, with total analysis times of less than 10 min for tryptic peptides. The peak capacity of the two-dimensional separations has been estimated to be in the 500-1000 range. The orthogonality of the separation techniques, an important factor for maximizing peak capacity or resolution elements, was verified by examining each technique independently for peptide separations. The two-dimensional separation strategy was found to greatly increase the resolving power over that obtained for either dimension alone. The information obtainable from any separation scheme is largely dependent upon the resolution provided by the system. Giddings has shown using Poisson statistics that as the number of components in a complex mixture increases, the probability of resolving all the individual elements decrease.1,2 This implies that the resolving power of the separation method must also be increased or the complexity of the original sample reduced (by prefractionation, for example) to accurately determine individual constituents. A chromatographic (or electrophoretic) system can be defined as having a certain peak capacity (nc) or maximum number of separated elements that may be placed in the space provided by the separation method. Over a given separation length (L), assuming unit resolution of adjacent peaks,

nc ) L/w where w is the mean value of the zone width.3 Peak capacities for conventional capillary electrophoresis (CE) range from about 50 * To whom correspondence should be addressed: (phone) (865) 574-5662; (fax) (865) 574-8363. † Caliper Technologies Corp. ‡ Present address: Analytical Solutions, Inc., Sunnyvale, California 94086. § Oak Ridge National Laboratory. (1) Giddings, J. C. Anal. Chem., 1984, 56, 1258. (2) Giddings, J. C. J. High Resolut. Chromatogr. Chromatogr. Commun. 1987, 10, 319-323. (3) Giddings, J. C. Unified Separation Science; John Wiley and Sons: New York, 1991; pp 112-141.

5244 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

to 150 while values for high-performance liquid chromatography (HPLC) range from about 20 to 100.4 One-dimensional separation methods are, therefore, frequently unable to provide the resolution required for the analysis of complex biological samples. Twodimensional (2D) separations, however, greatly increase resolving power. Peak capacities in multidimensional systems are approximately equivalent to the product of the capacities of each individual system (i.e., nc ) ninj ...), provided that the separation mechanisms of each are uncorrelated (i.e., orthogonal).2,3 In addition to providing far greater resolving power, 2D approaches also offer versatility. Techniques with different separation mechanisms may be paired in various combinations to provide the selectivity and resolving power necessary to address a certain problem. Jorgenson et al. have described a number of coupled column separation schemes for performing comprehensive multidimensional analysis.5-9 The systems are configured so that the effluent from one column passes through an interface that samples the material onto another column, subjecting analytes to displacement mechanisms in both dimensions. Comprehensive 2D column analyses have been achieved using automated switching valves where a given fraction is “parked” while another is separated by the second column,5,6 using parallel columns in the second dimension7 or using optically gated techniques to rapidly sample the first dimension.6,8 Size exclusion chromatography (SEC) has been coupled with CE,6 reversed-phase liquid chromatography (RPLC) with CE,5,6 and ion-exchange chromatography with RPLC9 for the generation of 2D systems. CE has also been coupled to ultrathin channel gel electrophoresis by moving the outlet end of the CE capillary across the channel entrance.10 Three-dimensional systems have been configured by coupling SEC with RPLC and fast CE.8 In this system, a 6-h analysis on the SEC column was followed by a 6-min analysis by RPLC and a final 2-s analysis by CE. The speed of the separations is an important parameter, with each successive technique requiring a faster analysis than the preceding to adequately sample the effluent. Various 2D systems (4) Meyer, V. R.; Welsch, T. LC-GC 1996, 10, 670-681. (5) Bushey, M. M.; Jorgenson, J. W. Anal. Chem. 1990, 62, 978. (6) Moore, A. W., Jr.; Larmann, J. P., Jr. ; Lemmo, A. V.; Jorgenson, J. W. In Methods in Enzymology; Karger, B. L., Hancock, W. S., Eds.; Academic Press: New York, 1996; Vol. 207, pp 401-419. (7) Opitek, G. P.; Jorgenson, J. W.; Anderegg, R. J. Anal. Chem. 1997, 69, 2283-2291. (8) Moore, A. W., Jr.; Jorgenson, J. W. Anal. Chem. 1995, 67, 3465-3468. (9) Opitek, G. J.; Lewis, K. C.; Jorgenson, J. W. Anal. Chem. 1997, 69, 15181524. (10) Liu, Y.-M.; Sweedler, J. V. Anal. Chem. 1996, 68, 3928. 10.1021/ac000578r CCC: $19.00

© 2000 American Chemical Society Published on Web 09/27/2000

have also been combined with mass spectrometry, which provides in effect a third dimension given the mass to charge discrimination of the latter.7,9 Microfabricated, microfluidic devices constructed on planar substrates have been shown to be highly advantageous for manipulating small sample volumes, rapidly processing materials, and integrating sample pretreatment and separation strategies.11-14 The dexterity with which materials can be manipulated and the ability to machine structures with interconnecting channels with essentially zero dead volume contribute to the high performance of microchips. These features coupled with the ease of automating all fluidic manipulations make these devices good candidates for configuring multidimensional separations. A variety of electrically driven separation techniques have been demonstrated on microchips, including free zone electrophoresis,15-17 open-channel electrochromatography (OCEC),18,19 gel electrophoresis,20-22 and micellar electrokinetic chromatography (MEKC).23-25 These miniature separation devices exhibit speed advantages due to the short axial extent of the injection plugs that can be generated with the chips. In this paper, we demonstrate a simple microfabricated microfluidic device for 2D analysis. Peptide mixtures including tryptic digests are separated by MEKC in the first dimension and by free zone electrophoresis in the second. EXPERIMENTAL SECTION Sample Preparation. Peptides and proteins were obtained from Sigma (St. Louis, MO) and were used without further purification. Cytochrome c (bovine heart) was digested with trypsin by pumping 400 µL of a 12.5 mg/mL solution of the protein in N-ethylmorpholine acetate buffer (N-EMA; 400 mM N-ethylmorpholine, 100 mM acetic acid, 10 mM CaCl2, 5% methanol, pH 8.2) through a Poroszyme immobilized trypsin cartridge (Perceptive Biosystems, Inc., Framingham, MA) at 10 µL/min at room temperature. An additional 450-µL volume of buffer was used to rinse the cartridge and the effluent combined with the initial eluent. Ribonuclease A and R-lactalbumin were reductively alkylated prior to digestion with dithiothreitol and iodoacetic acid using (11) Jacobson, S. C.; Hergenro ¨der, R.; Moore, A. W., Jr.; Ramsey, J. M. Anal. Chem. 1994, 66, 4127-4132. (12) Woolley, A. T.; Hadley, D.; Landre, P.; deMello, A. J.; Mathies, R. A.; Northrup, M. A. Anal. Chem. 1996, 68, 4081-4086. (13) Waters, L. C.; Jacobson, S. C.; Kroutchinina, N.; Khandurina, J.; Foote, R. S.; Ramsey, J. M. Anal. Chem. 1998, 70, 158-162. (14) Chiem, N. H.; Harrison, D. J. Clin. Chem. 1998, 44, 591-598. (15) Effenhauser, C. S.; Manz, A.; Widmer, H. M. Anal. Chem. 1993, 65, 26372642. (16) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895-897. (17) Jacobson, S. C.; Hergenro ¨der, R.; Koutny, L. B.; Warmack, R. J.; Ramsey, J. M. Anal. Chem. 1994, 66, 1107-1113. (18) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 2369-2373. (19) Kutter, J. P.; Jacobson, S. C.; Matsubara, N.; Ramsey, J. M. Anal. Chem. 1998, 70, 3291-3297. (20) Effenhauser, C. S.; Paulus, A.; Manz, A.; Widmer, H. M. Anal. Chem. 1994, 66, 2949-2953. (21) Woolley, A. T.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 1134811352. (22) Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1996, 68, 720-723. (23) Moore, A. W., Jr.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1995, 67, 4184-4189. (24) Kutter, J. P.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1997, 69, 51655171. (25) von Heeren, F.; Verpoorte, E.; Manz, A.; Thormann, W. Anal. Chem. 1996, 68, 2044-2053.

Figure 1. Schematic diagram of the microchip used for 2D separations. The first-dimension separation channel extends 69 mm from the first cross intersection to the second. The second separation channel extends 10 mm from the second cross to the point of detection. Reservoirs contain sample, S, buffer 1, B1, waste 1, W1, buffer 2, B2, waste 2, W2, and waste, W.

standard procedures.26 Excess thiol and iodoacetic acid were removed by dialysis in N-EMA using Spectra/Por Dispo-Dialyzers (500-Da cutoff) (Fischer Scientific Co., Pittsburgh, PA) for 24 h. The reductively alkylated proteins were then digested with trypsin (50:1, w/w) for 24 h at 37 °C. The tryptic peptides from cytochrome c were fluorescently labeled with the mixed isomer, 5(6)-carboxytetramethylrhodamine succinimidyl ester (5(6)-TAMRA, SE; Molecular Probes Inc., Eugene, OR), while R-lactalbumin and ribonuclease A were derivatized with the single isomer, 6-TAMRA, SE (Molecular Probes Inc.). A 0.25-mL aliquot of the tryptic digest solutions (∼6 mg/mL) was diluted with 0.25 mL of digestion buffer and then reacted with 40 µL of a 10 mg/mL solution of the dye dissolved in DMSO for 2 h at 37 °C. Undigested protein was removed by size exclusion chromatography using HiTrap desalting cartridges (5 kDa cutoff) (Amersham Pharmacia Biotech Inc., Piscataway, NJ). Prior to analysis, the samples were concentrated 10-50-fold by evaporation. Bioactive peptides were derivatized for 2 h at 37 °C by addition of 100 µL of 10 mM peptide in water and 100 µL of 25 mM fluorescein isothiocyanate (FITC), isomer I (Sigma) in DMSO to 800 µL of 100 mM sodium bicarbonate buffer adjusted to pH 9. Microchip Device. Soda-lime glass chips were fabricated using standard wet chemical etching, photolithography, and bonding techniques.17 The bonded cover plate of the chips contained ∼25-µL volume holes that served as sample, buffer, and waste reservoirs. A schematic diagram of the device used for the 2D separations is shown in Figure 1. Fluid reservoirs contained sample, S, buffer 1, B1, sample waste 1, W1, buffer 2, B2, sample waste 2, W2, and waste, W. The channels were 35 µm wide at half-depth, 10 µm deep, and 69 and 10 mm long for the first and second dimension separations, respectively. Following loading with sample and buffers, the microchips were placed in a holder with a lid containing platinum electrodes connected to a computer(26) Allen, G. Laboratory techniques in Biochemistry and Molecular Biology, 2nd ed.; Elsevier: New York, 1989.

Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5245

controlled, multichannel high-voltage power supply. Separations were monitored 10 mm from the second cross, in the second separation channel, using a fluorescence microscope with a xenon lamp. The signal was collected with a 20× microscope objective and measured with a PMT. An in-house-written data acquisition program controlled the timing and voltage script for the reservoir power supply and also collected the PMT data. Two-dimensional plots were constructed from the serial time-dependent data collected by the detector monitoring the repetative seconddimension separations and were prepared using the program Transform from Fortner software (Sterling, VA). The gated injection scheme27 was used to introduce controlled volumes of sample at the uppermost four-way intersection of the microfluidic device (Figure 1) onto the first separation channel. Effluent from this channel generally flows to the waste reservoir, W2, with aliquots periodically introduced into the lower channel every few seconds for further separation using another gated injection at the lower four-way intersection. The power supply used to apply the electrical potentials to the microchip was operated in current control mode. In this mode of operation, the electrical potential is fixed at one reservoir and the currents flowing into the remaining reservoirs are specified rather than the voltages. The separation buffers were 50 mM triethylamine (TEA), 25 mM acetic acid, pH 10.7, with 10 mM sodium dodecyl sulfate (SDS), B1, and without SDS, B2. RESULTS AND DISCUSSION To maximize the peak capacity of a multidimensional separation system, the individual separation techniques must be orthogonal. Although there are any number of liquid-phase separation techniques that may be paired, we have chosen initially to couple MEKC and CE, the simplest, nearly orthogonal techniques that may be linked. Surface treatment or modification of the channel walls is not required to affect separations with these methods, and liquid phases having similar primary composition and pH may be used to reduce problems that could be encountered with grossly dissimilar buffers when aliquots from one dimension are transferred to the next. Sudden large changes in conductivity resulting in increased Joule heating in the transferred volume could affect overall resolution, as could large changes in viscosity or transfer of immiscible fractions, for example. The orthogonality of the MEKC and CE techniques for peptide separations was investigated by monitoring the separation of a five-component mixture of fluorescently labeled compounds. Electropherograms obtained on a simple cross chip27 using a gated injection, detecting 27 mm from the injection cross on the separation channel, and using TEA buffer with and without SDS for peptides prelabeled with FITC are shown in Figure 2a and b, respectively. Changes in migration order and time are clearly evident between the runs, indicating different separation mechanisms are responsible for migration behavior. The effect of SDS was greatest on positively charged, hydrophobic molecules where electrostatic interaction with the negatively charged micelles would be expected to delay their elution. Bradykinin, which contains two arginines, migrated first in the CE run and was found to elute last when SDS was added to the buffer, for example. To (27) Jacobson, S. C.; Koutny, L. B.; Hergenro¨der, R.; Moore, A. W., Jr.; Ramsey, J. M. Anal. Chem. 1994, 66, 3472-3476.

5246 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

Figure 2. MEKC and CE separations for FITC-labeled peptides. Separations performed on a 27-mm-long channel, at a field strength of 500 V/cm, using 0.5-s sample injection. Buffers were 50 mM TEA, 25 mM acetic acid, pH 10.7, with 10 mM SDS (MEKC, top) and without SDS (CE, bottom).

Figure 3. Plot of the relative retention of FITC-labeled peptides as a function of SDS concentration in the buffer. Conditions as in Figure 2.

examine the effects of SDS concentration on migration behavior, several runs were performed with increasing surfactant concentration. The relative migration times of the test compounds plotted as a function of SDS concentration are shown in Figure 3. Although changes were apparent from 5 to 50 mM surfactant, MEKC runs for 2D separations were performed with 10 mM SDS where changes in migration order were evident but buffer conductivity was also relatively low. The operation of the 2D device shown in Figure 1 is described with the use of the schematic diagram shown in Figure 4. The

Figure 4. Schematic diagram illustrating the direction of fluid flow in the operation of the 2D microchip shown in Figure 1 during sample loading (solid arrows) and in the separation mode (hatched arrows).

basic architecture of the experimental device used (Figure 1) and the schematic (Figure 4) are identical, i.e., an upper injection cross followed by a first-dimension separation channel and a lower injection cross connected to a second-dimension separation channel. The device shown in Figure 1 uses folded channels to connect to reservoirs that have fixed locations on the chip and to allow a longer first-dimension separation channel. The arrows shown in the schematic of Figure 4 depict the direction of fluid transport for the valves under the two operating modes, i.e., separation and sample loading. Voltages are applied to the reservoirs so that the buffers flow primarily into the respective separation channels with some spillover into the waste channels to shunt the sample to waste in the separation mode. This flow condition is easily achieved using current control, given that the fluid flow is linearly proportional to the current under homogeneous buffer conditions. For example, this condition is met when the current supplied to the sample reservoir is less than the current arriving at the waste reservoir, W1. An alternative, but equivalent, condition is that the current into buffer reservoir, B1, exceeds the current through the separation channel. (Note that the sample for the second dimension is the effluent of the first dimension.) Under load mode, the currents are adjusted to electrokinetically transport material from the sample (or firstdimension separation) channels to the respective separation channels. This can be done, for example, by reducing the current through reservoirs B1 and W1 to zero for the duration of the loading or injection process. The actuation of the two gate valves shown in Figure 4 is asynchronous with the second-dimension valve operating at a fixed frequency that is more than 100 times greater than for the first-dimension valve. The experiment is started by switching the first-dimension valve for 2 s to inject a sample. This valve is held in the separation mode for the remainder of the 2D separation (500-600 s) while the lower valve is switched from the separation mode to the load condition every 3-4 s for 0.3 s to inject samples of effluent from the first dimension into the second dimension. Thus, ∼10% of the sample injected into the first dimension is sampled by the second dimension. A 2D separation of the five bioactive peptides using MEKC in the first dimension and fast CE in the second is shown in Figure 5. Effluent from the first dimension was injected onto the CE channel every 4 s. Corresponding one-dimensional separations

Figure 5. 1D and 2D separations of FITC-labeled peptide standards. Effluent from the first dimension injected every 4 s into the second. Peaks: (1) leucine enkephalin, (2) angiotensin I, (3) angiotensin III, (4) neurotensin, (5) bradykinin. Buffers as in Figure 2.

are also shown adjacent to their respective axes on the 2D plot. These electropherograms were obtained by moving the detection point to just prior to the second injection cross for the MEKC run and by filling the entire first-dimension channel with sample for the CE analysis to allow a mixture to be introduced onto the short CE channel. Peak identities were confirmed by injecting the analytes individually. Bradykinin was not present in the mixture separated by the single-dimension CE analysis, but its position was determined by an individual, independent injection. In general, peaks in the one-dimensional plots line up with “spots” shown in the 2D plot. However, there were some slight changes in elution time between the various runs that may be caused by fluctuations in temperature in the laboratory as well as changes in electroosmotic flow as a result of solute adsorption on the channel walls. Although only five peptides were present in the sample mixture, additional “spots” are shown on the 2D plot. Those at (340, 4.5 s) and (440, 2.5 s) were present in the FITC blank while others probably arose from low concentrations of multiply labeled peptides for those that contain lysine. Our first attempt at a 2D separation of a tryptic digest of cytochrome c, using FITC as a the derivatization reagent to fluorescently label the peptides, resulted in a 2D plot where the fragments appear along a diagonal line, suggesting correlation between the free zone and micellar electrophoresis used in these experiments for these peptides. The most likely cause is a reduction in the charge spread for the tryptic fragments resulting from attachment of the FITC label. This reagent binds to the N-terminus and to lysine -amino groups replacing positive amino charges with a doubly negative charged residue. Switching to TAMRA for derivatization, which also conjugates at amino groups, alleviated this problem because the label is zwitterionic at the electrolyte pH. A 2D separation of the TAMRA-labeled tryptic fragments of cytochrome c is shown in Figure 6. The spots are generally distributed in the separation space. The single-dimension separations are also shown along their respective axes. A few Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5247

Figure 7. Linear recording of the 2D separation shown in Figure 5.

Figure 6. 1D and 2D separations of peptides from a tryptic digest of cytochrome c, fluorescently labeled with 5,6-TAMRA. Two spots labeled F are from unreacted derivatizing reagent, L from labeled lysine. Injections into the second dimension were made every 3 s. Buffers as in Figure 2.

peaks were identified by injecting standards: the large peaks labeled with an F at (265, 2.3 s) and (290, 2.3 s) are unreacted TAMRA, while the peaks at (280, 1.9 s) and (360, 1.9 s) are from lysine. Although many components appear along the first-dimension axis, few of them are baseline resolved. The seconddimension separation greatly improves the overall resolution. Comparison of the first-dimensional separation with the 2D plot also shows that components eluting after 500 s do not appear in the 2D display. Electrophoretic bias during the second-dimension injection may have prevented the introduction of these substituents on the CE channel. The total number of spots is also greater than the number of fragments (i.e., 17 peptides including lysine) predicted from a complete tryptic digestion of cytochrome c. Multiply labeled products and incomplete digestion resulting in a larger total number of tryptic fragments may account for the increase. The additional complexity of the sample, in this case, however, serves to further demonstrate the resolving power of the 2D separation. The entire analysis was completed in less than 10 min with aliquots from the first dimension being gated onto the CE channel every 3 s. The linear output of the detector from the 2D analysis of the sample is shown in Figure 7. The major peaks shown in the figure are 3 s apart corresponding to the period between actuation of the second (CE) gate valve. An expanded segment from 300 to 320 s is shown in Figure 8. Average peak widths (fwhm) are ∼150 ms. The total peak capacity of the 2D separation was estimated to be between 500 and 1000 (20-40 for the MEKC run and ∼25 for the CE dimension, as estimated from the sample analysis in these dimensions). Tryptic digests of R-lactalbumin and ribonuclease A are shown in Figures 9 and 10, respectively. These proteins were digested using trypsin in solution rather than immobilized on a support, and the single isomer, 6-TAMRA, was used for derivatization to reduce multiple peaks from the same peptide. Spots identified as coming from the unreacted fluorescent label are marked with an 5248 Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

Figure 8. Expanded time scale (300-320 s) of the linear recording of the 2D separation of cytochrome c.

Figure 9. Plot of the 2D microchip separation of tryptic peptides of R-lactalbumin, labeled with 6-TAMRA. Spots at F resulting from the derivatizing reagent and at L from labeled lysine. Injections into the second dimension were made every 4 s. Buffers as in Figure 2.

F on the figure, while those from labeled lysine are marked with an L. The resolution obtained for the R-lactalbumin sample was greater than for the peptides from ribonuclease A. The latter were found primarily along a diagonal line of the 2D plot, indicating correlation between the two separation mechanisms. The tryptic fragments from ribonuclease A also have a narrower m/z distribution compared to that of lactalbumin, as predicted from the PeptideMass software in the ExPASy proteomics server available on the Internet (http://www.expasy.ch/tools/peptide-mass.html)

Figure 10. Plot of the 2D microchip separation of tryptic peptides of ribonuclease A, labeled with 6-TAMRA. Spots at F resulting from the derivatizing reagent and at L from labeled lysine. Injections into the second dimension were made every 4 s. Buffers as in Figure 2.

and may therefore be more difficult to separate by size in the CE dimension. CONCLUSION A microchip device integrating two different separation modes, MEKC and high-speed CE, has been developed and demonstrated. It operates by rapidly sampling and analyzing effluent in the second dimension from the first, with second-dimension analyses performed and completed every few seconds. The ability to microfabricate a fluidic structure that includes valves and connections with low dead volume allows multiple separation dimen-

sions to be serially connected with little loss in separation efficiency and high volumetric sampling efficiency. The electrokinetic control features allow the multidimensional separation process to be easily automated once reagents are loaded onto the microchip. The 2D separation strategy was found to significantly increase the resolving power over that obtained for either dimension alone for a peptide test mixture as well as for tryptic digests of various proteins. The resolution may be further improved by pairing different separation techniques and increasing the channel lengths. Given the overall increased speed of analysis for separations performed on microchip platforms relative to conventional benchscale chromatographic or electropheric systems, the device and approach described here hold considerable promise for rapidly processing complex samples and enabling high-throughput peptide mapping. ACKNOWLEDGMENT This research was sponsored by the Laboratory Directed Research and Development Program of Oak Ridge National Laboratory and by the National Cancer Institute under Grant CA 83238-02. Oak Ridge National Laboratory is managed by UTBattelle, LLC for the U.S. Department of Energy under Contract DE-AC05-00OR22725. The authors acknowledge Mac McReynolds and Knute Stevenson at Caliper Technologies, Inc. for microchip fabrication. Received for review May 19, 2000. Accepted August 10, 2000. AC000578R

Analytical Chemistry, Vol. 72, No. 21, November 1, 2000

5249