Anal. Chem. 1999, 71, 1815-1819
Microfabricated Porous Membrane Structure for Sample Concentration and Electrophoretic Analysis Julia Khandurina, Stephen C. Jacobson, Larry C. Waters, Robert S. Foote, and J. Michael Ramsey*
Chemical and Analytical Sciences Division, Oak Ridge National Laboratory, P.O. Box 2008, Oak Ridge, Tennessee 37831-6142
A microfabricated injection valve incorporating a porous membrane structure is reported that enables electrokinetic concentration of DNA samples using homogeneous buffer conditions followed by injection into a channel for electrophoretic analysis. The porous membrane was incorporated in the microchannel manifold by having two channels separated from each other by 3-12 µm and connected by a thin porous silicate layer. This design allows the passage of current to establish an electrical connection between the separated channels but prevents large molecules, e.g., DNA, from traversing the membrane. Concentrated DNA can be injected into the separation channel and electrophoretically analyzed. Experiments exhibit a nonlinear increase in concentration with time, and DNA fragments can be concentrated up to 2 orders of magnitude as shown by comparison of peak intensities for analysis performed with and without concentration. Developments in microfluidics and microinstrumentation during the last several years have opened the possibility of fabricating devices with increased functionality and complexity for chemical and biochemical applications. Various electrically driven separation techniques for liquid-phase analysis such as capillary electrophoresis (CE),1-6 electrochromatography,7 micellar electrokinetic chromatography,8 capillary gel electrophoresis for DNA restriction fragment sizing,9,10 and DNA sequencing11 have been successfully adapted to a microchip format. A number of integrated microstructures that combine chemical and biochemical reactions for (1) Manz, A.; Harrison, D. J.; Verpoorte, E. M. J.; Fettinger, J. C.; Paulus, A.; Ludi, H.; Widmer, H. M. J. Chromatogr. 1992, 593, 253. (2) Harrison, D. J.; Manz, A.; Fan, Z.; Ludi, H.; Widmer, H. M. Anal. Chem. 1992, 64, 1926. (3) Seiler, K.; Harrison, D. J.; Manz, A. Anal. Chem. 1993, 65, 1481. (4) Harrison, D. J.; Fluri, K.; Seiler, K.; Fan, Z.; Effenhauser, C. S.; Manz, A. Science 1993, 261, 895. (5) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Warmack, R. J.; Ramsey, J. M. Anal. Chem. 1994, 66, 1107. (6) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 1114. (7) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 2369. (8) Moore, A. W.; Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1995, 67, 4184. (9) Effenhauser, C. S.; Paulus, A.; Manz, A.; Widmer, H. M. Anal. Chem. 1994, 66, 2949. (10) Woolley, A. T.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1994, 91, 11348. (11) Woolley, A. T.; Mathies, R. A. Anal. Chem. 1995, 67, 3676. 10.1021/ac981161c CCC: $18.00 Published on Web 03/19/1999
© 1999 American Chemical Society
sample pretreatment,12 post-13 and precolumn14 derivatization, and PCR amplification15 and analysis on a single device have been demonstrated. In addition to integration and multifunctionality, an intrinsic advantage of microfabricated devices is the potential to produce arrays of separation channels for high-throughput applications with negligible additional cost.16 As with many analytical techniques, sensitive detection methods are also required for microinstrumentation. Recently, the detection of single chromophore molecules has been demonstrated on-chip using confocal fluorescence detection.17 Detection capabilities can also be enhanced by sample concentration methods. For electrokinetically driven separation techniques, sample stacking18-20 and field-amplified sample injection21,22 have been demonstrated. Also, sample stacking has been implemented on-chip to concentrate the sample at the inlet of the separation column.23 More elaborate preconcentration techniques utilize twoand three-buffer systems, i.e., transient isotachophoretic concentration,24-26 and have been demonstrated in capillary electrophoresis. Several approaches27,28 have been described for the on-tube concentration of ionic solutes in capillary electrophoresis. These techniques are based on decreasing the electrophoretic velocities by moving the sample toward a pH gradient, into a small-pore gel, by dialysis tubing attached to one end of the capillary, or by (12) Jacobson, S. C.; Hergenro¨der, R.; Koutny, L. B.; Ramsey, J. M. Anal. Chem. 1994, 66, 4127. (13) Jacobson, S. C.; Hergenro¨der, R.; Moore, A. W.; Ramsey, J. M. Anal. Chem. 1994, 66, 3472. (14) Fluri, K.; Harrison, D. J. Anal. Chem. 1996, 68, 4285. (15) Woolley, A. T.; Hadley, D.; Landre, P.; deMello, A. J.; Mathies, R. A.; Northrup, M. A. Anal. Chem. 1996, 68, 4081. (16) Simpson, P. C.; Roach, D.; Woolley, A. T.; Thorsen, T.; Johnston, R.; Sensabaugh, G. F.; Mathies, R. A. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 2256. (17) Fister, J. C., III; Jacobson, S. C.; Davis, L. M.; Ramsey, J. M. Anal. Chem. 1998, 70, 431. (18) Moring, S. E.; Colburn, J. C.; Grossman, P. D.; Lauer, H. H. LC-GC 1990, 8, 34. (19) Burgi, D. S.; Chien, R.-L. Anal. Chem. 1991, 63, 2042. (20) Aebersold, R.; Morrison, H. D. J. Chromatogr. 1990, 516, 79. (21) Jandik, P.; Jones, W. R. J. Chromatogr. 1990, 546, 431. (22) Chien, R.-L.; Burgi, D. S. J. Chromatogr. 1991, 559, 141. (23) Jacobson, S. C.; Ramsey, J. M. Electrophoresis 1995, 16, 481. (24) Hjerten, S.; Elenbring, K.; Kilar, F.; Liao, J.-L.; Chen, A.; Siebert, C. J.; Zhu, M.-D. J. Chromatogr. 1987, 403, 47. (25) Foret, F.; Szoko, E.; Karger, B. L. J. Chromatogr. 1992, 608, 3. (26) Schwer, C.; Lottspeich, F. J. Chromatogr. 1992, 623, 345. (27) Hjerten, S.; Liao, J.-L.; Zhang, R. J. Chromatogr., A 1994, 676, 409. (28) Liao, J.-L.; Zhang, R.; Hjerten, S. J. Chromatogr., A 1994, 676, 421.
Analytical Chemistry, Vol. 71, No. 9, May 1, 1999 1815
Figure 1. Schematic of a porous membrane microchip for DNA concentration and electrophoretic analysis.
displacement electrophoresis combined with hydrodynamic counterflow. In this paper, an alternative approach to concentrating DNA fragments on a microchip has been developed using a porous membrane structure. The microchip fabrication incorporated a low-temperature bonding method using a silicate solution as an adhesive to attach the cover plate to the substrate.29 Two adjacent channels were then connected electrically by a narrow layer of porous material formed as a result of polycondensation of silicate film. The thin polysilicate layer serves as a semipermeable membrane allowing ionic current to pass but preventing large DNA molecules from crossing the membrane. This arrangement allowed substantial enhancement in DNA concentration in the channel adjacent to the membrane using homogeneous buffer conditions. Concentrated in this manner, the sample plug can then be injected into the separation channel and detected. The fabrication and performance of this device are reported. EXPERIMENTAL SECTION Microfabrication. The fabrication of microchips involves standard photolithographic procedures followed by wet chemical etching as previously described.5,30 The microchannel design, shown in Figure 1, was transferred onto the glass substrate using a positive photoresist, photomask, and UV exposure. The channels are etched in a dilute, stirred HF/NH4F bath. Bonding of a cover plate to the glass substrate to form the closed network of channels was accomplished by a low-temperature bonding process using a spin-on silicate layer as the adhesive.29 Briefly, this procedure includes the following steps: cleaning the glass substrate and cover plate, hydrolyzing the glass surfaces and rinsing with water, spinning diluted silicate solution onto the cover plate surface and bringing the treated cover plate into contact with the glass substrate, and annealing the microchip at 200 °C. Two silicate solutions were used: potassium silicate solution (KASIL 2130, The PQ Corp., Valley Forge, PA) and sodium silicate solution (E, The PQ Corp.) both diluted to a final concentration of 0.1 M. It was found that use of a potassium silicate solution results in somewhat better bonding performance than sodium silicate. To help control (29) Wang, H. Y.; Foote, R. S.; Jacobson, S. C.; Schneibel, J. H.; Ramsey, J. M. Sens. Actuators 1997, B45, 199. (30) Ko, W. H.; Suminto, J. T. in: Gopel, W.; Hasse, J.; Zemmel, J. N. (Eds.) Sensors Weinheim, 1989, 1, VCH: pp 107-168.
1816 Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
the thickness of the porous layer and improve chip-to-chip reproducibility, the same concentration of silicate solution (0.1 M) and the same spin rate and time (4000 rpm for 10 s) were used. Annealing at 200 °C was used to reinforce the bonding through dehydration and siloxane bond formation. The channels were typically about 8-10 µm deep and 60-65 µm wide at half-depth. The width of the porous membrane (the distance between the separated channels) varied from 3 to 12 µm, depending on the photomask used and the etch time. These dimensions were measured using a stylus-based surface profiler. The analyte, buffer, separation, and side channel lengths were approximately 1, 1, 3, and 1 cm, respectively. Cylindrical glass reservoirs were affixed with epoxy at the points where the channels extend beyond the cover plate. The electroosmotic mobility was minimized, when necessary, by the covalent immobilization of linear polyacrylamide on the channel walls.31 For microchip gel electrophoresis, all the channels were filled with 3% linear polyacrylamide or 3.5% Super Non-Adulterated Polymer (SNAP; ABI Applied Biosystems, Foster City, CA) in 1× TBE buffer (89 mM Tris, 89 mM boric acid, and 2 mM EDTA at pH 8.3). The acrylamide, used for polymerization, is toxic and should be handled appropriately. Instrumentation. Microchip performance was monitored by laser-induced fluorescence (LIF). For charge-coupled device (CCD) imaging, the argon ion laser beam (514.5 nm, 100 mW, 543-AP, OmNichrome, Chino, CA) was expanded to a diameter of approximately 5 mm at the microchip surface. The CCD (SSCC374, Sony) camera was mounted on a stereomicroscope (SMZU, Nikon). The fluorescence signal was filtered spectrally (550 nm cut-on) and detected by the CCD. For single-point detection,32 an argon ion laser (514.5 nm, 5 mW) was focused to a spot in the separation channel using a 100-mm focal length lens. The signal was collected by a 20× objective (0.42 NA; Newport Corp., Irvine, CA), followed by spatial filtering with a 1.0-mm pinhole (Melles Griot, Irvine, CA) and spectral filtering using a 514-nm holographic notch filter (10-nm bandwidth, Kaiser Optical Systems, Inc., Ann Arbor, MI) and a 540-nm band-pass filter (30-nm bandwidth 540DF30, Omega Optical, Brattleboro, VT), and measured with a photomultiplier tube (77348, Oriel Instruments, Inc., Stratford, CT). The data acquisition and voltage switching apparatus were computer-controlled using programs written in-house in Labview 3.0 (National Instruments, Austin, TX). Platinum electrodes provided electrical contacts from the power supply to the solutions in the reservoirs. Electrical currents in the microchannels were measured with an autoranging picoammeter (Keithley 485, Keithly Instruments Inc., Cleveland, OH). Analytes. The following DNA products were used for DNA concentration and electrophoretic separation performance studies: a ΦX174-HaeIII digest (Sigma Chemicals, St. Louis, MO), a PCR marker (Promega Corp., Madison, WI) containing six fragments of 50, 150, 300, 500, 750, and 1000 base pairs (bp), each present at approximately equal weight/volume concentration, and a 500-bp PCR product amplified from bacteriophage λ DNA. All DNA solutions were diluted in 1× TBE buffer. An intercalating dye, TO PRO (Molecular Probes, Inc., Eugene, OR), was added to the DNA solutions at the ratio 1:10 (dye/base pairs). (31) Hjerten, S. J. Chromatogr. 1985, 347, 191. (32) Jacobson, S. C.; Ramsey, J. M. Anal. Chem. 1996, 68, 720.
Figure 2. Schematic of the porous membrane structure with top and side views.
RESULTS AND DISCUSSION A schematic of the microfabricated channels and porous membrane structure is shown in Figure 2. The injection tee and side channel are separated from each other by a distance of 3-12 µm, depending on the mask used and the etch time. A silicate bonding process29 was employed to join the substrate and cover plate glass surfaces, and the resulting thin polysilicate layer also provides the porous membrane between the analyte and side channels. Due to a high impedance at the membrane region as compared to the channel, the electric field strength changes abruptly and stepwise across this area when a high potential is applied between the analyte and side reservoirs. Electric current measurements monitored by applying potentials to different pairs of reservoirs allowed the resistances in the microchannel manifold and across the porous membrane to be determined. The electric field strength across the membrane connection is approximately 3-4 orders of magnitude higher than that along the channel filled with a sieving matrix solution, e.g., 105-106 and 102 V/cm, respectively, with 500 V applied between the analyte and side channel. To evaluate the performance of the porous membrane for double-stranded (ds) DNA concentration, the microchip design in Figure 1 was utilized. The concentration takes place in the tee region of the channel adjacent to the membrane followed by injection of the concentrated plug into the separation channel for electrophoretic analysis. The CCD images in parts b and c of Figure 3 show the microchip porous membrane region and DNA concentration steps. For reference, an overlay of the injection tee and side channel is shown in Figure 3a. During sample loading, the potentials at the analyte and side reservoirs were 0 and 1 kV, respectively, and the buffer and waste reservoirs had no potentials applied. DNA molecules accumulate in the tee region due to their hindered transport through the membrane, and the concentration,
Figure 3. (a) Overlay of injection tee and porous membrane. CCD images of analyte concentrated for (b) 2 and (c) 3 min. (d) Injection of concentrated analyte plug. Porous membrane region width is 7 µm. All channels are filled with 3% linear polyacrylamide in 1× TBE buffer. DNA sample: 25 µg/mL ΦX174-HaeIII digest with 6.0 µM TO PRO added. Preconcentration mode (b, c): 1 kV applied between analyte and side reservoirs with no potential applied to buffer and waste reservoirs. Injection of the 3-min concentrated plug (d) with relative potentials applied to buffer, analyte, and waste reservoirs 0, 0.4, and 1.0 kV, respectively (no potential applied to side reservoir).
observed by monitoring the fluorescence intensity, increased with time. Figure 3b shows the DNA concentration after 2 min and Figure 3c after 3 min. Reconfiguration of the voltage distribution for the separation mode results in an injection of the DNA plug concentrated for 3 min into the separation channel (Figure 3d). The potentials at the waste, buffer, and analyte reservoirs in this mode were 1, 0, and 0.4 kV, respectively, and no potential was applied to the side reservoir. The potential at the analyte reservoir in the separation mode prevents bleeding of the excess sample into the separation column. Also, a bright spot remains in the membrane (Figure 3d) where DNA has embedded itself in the porous silicate layer. There was no observed release of the embedded material in the form of increased background of spurious peaks during subsequent injection and separation cycles. Marked changes in the membrane performance were not noticed for up to 10 concentration, injection, and separation cycles. Figure 4 shows three electropherograms of the Promega DNA marker concentrated for 150, 200, and 250 s. In the top panel (150 s) there is a noticeable bias against the larger base pair fragments (see the inset) due to their lower mobility. As expected, this bias is reduced with longer concentration times. All preconcentrated DNA fragments are well separated, although the resolution of the larger fragments decreases slightly for longer concentration times. Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
1817
Figure 4. Successive electropherograms of a DNA PCR marker (50, 150, 300, 500, 750, and 1000 bp) after (a) 150, (b) 200, and (c) 250 s of preconcentration. Porous membrane region width is 3.4 µm. Sieving medium is 3.5% SNAP in 1× TBE buffer. Initial DNA concentration is 10.7 µg/mL (0.9 µM TO PRO). Separation distance is 2.5 cm. Separation electric field strength is 170 V/cm. Potential applied in the concentration step is 1.2 kV.
Further extension of concentration time leads to gradual deterioration of separation efficiency. This is related to marked changes in the conductivity of the highly concentrated analyte plug. This loss of resolution is also observed with high concentrations of DNA separated on-chip without the concentration step. The plots in Figure 5 depict increases in fluorescence response with concentration time for different DNA fragments. The data in Figure 5a include those presented in Figure 4. Panels b and c of Figure 5 depict the relative concentration enhancement for a ΦX174-HaeIII digest sample and a 500-bp PCR product sample, respectively. Their electropherograms are not shown. To estimate the concentration enhancement of DNA fragments by the porous membrane structure, the peak heights are normalized to those obtained with a simple cross microchip with similar channel dimensions operated with the pinched injection scheme.33 Relative baseline signals for the two types of chips were first obtained by continuously flowing identical DNA samples through their separation channels and measuring the signal intensity. Discrete injections were then made with the simple cross microchip, and the peak heights for the separated fragments were normalized by multiplying by the ratio of baseline signal intensities in the preconcentrator versus simple cross chips. These normalized values were used to determine the relative enhancement of peak heights due to the concentration step as shown in Figure 5. The preconcentration data were generated from three sets of sample loading and separation sequences on three separate microchips for the three different DNA samples. All experiments demonstrated a similar nonlinear dependence of the concentration enhancement with time (Figure 5). An exponential fit for these (33) Waters, L. C.; Jacobson, S. C.; Kroutchinina, N.; Khandurina, J.; Foote, R. S.; Ramsey, J. M. Anal. Chem. 1998, 70, 158.
1818 Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
Figure 5. DNA concentration in porous membrane microchips relative to conventional microchip pinched injection. Sieving medium is 3.5% SNAP in 1× TBE buffer. Porous membrane connection region width is (a) 3.3, (b) 7.4, (c) 12 µm. Separation distance is (a) 2.5, (b) 1.5, and (c) 1.0 cm. Separation electric field strength is (a) 170, (b) 185, and (c) 180 V/cm. Potential applied for concentration is (a) 1.2, (b) 0.8, and (c) 0.7 kV. DNA samples are (a) 50 bp PCR marker, 10.7 µg/mL (0.9 µM TO PRO); (b) ΦX174-HaeIII digest, 3.9 µg/mL (1 µM TO PRO); (c) 500-bp PCR fragment (0.8 µM TO PRO).
plots is shown for the 50-bp fragment in Figure 5a, the 194-bp fragment in Figure 5b, and the 500-bp fragment in Figure 5c. The exponential increase in concentration is not fully understood. However, these results demonstrate the capability of concentrating a DNA sample on-chip by up to 2 orders of magnitude prior to
the electrophoretic separation, without a significant loss of the separation performance. For example, the peak height of the 194bp fragment of ΦX-HaeIII digest obtained after concentration for 250 s was 100 times that obtained with a conventional pinched injection (Figure 5b). In conclusion, a novel porous membrane structure was designed and integrated onto a microchip device. This feature allows DNA to be concentrated up to 2 orders of magnitude prior to an electrophoretic analysis on the same microchip. Refinements of the microfabricated structures and performance of the concentration-separation cycle, as well as elucidation of the underlying physicochemical principles, are under continued investigation. ACKNOWLEDGMENT This research is sponsored by Oak Ridge National Laboratory (ORNL) Laboratory Directed Research and Development Program
and by National Institutes of Health Grant R01HG01398. ORNL is managed by Lockheed Martin Energy Research Corp. for the U.S. Department of Energy under Contract DE-AC05-96OR22464. This research was supported in part by appointment for J.K. to the Oak Ridge National Laboratory Postdoctoral Research Associates Program administered jointly by ORNL and Oak Ridge Institute for Science and Education. The authors thank Justin E. Daler, Judith Eggers, and John Cockfield for assistance in microchip fabrication.
Received for review October 22, 1998. Accepted February 9, 1999. AC981161C
Analytical Chemistry, Vol. 71, No. 9, May 1, 1999
1819
