Anal. Chem. 2003, 75, 3675-3680
Mechanism for the Separation of Large Molecules Based on Radial Migration in Capillary Electrophoresis Jinjian Zheng and Edward S. Yeung*
Ames LaboratorysUSDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011
We demonstrate a novel separation mechanism for large molecules based on their radial migration in capillary electrophoresis with applied hydrodynamic flow (HDF). The direction of radial migration depends on the direction of the applied HDF relative to the electric field. The radial migration velocities are size-dependent, which could be attributed to the different degree of deformation under shear flow. Analytical separation was demonstrated on a sample plug containing λ DNA (48 502 bp) and OX174 RF DNA (5386 bp) with baseline separation. Alternatively, this separation mode can be performed continuously and is thus applicable to preparative separations. Without the need for gel/polymer or complex instrumentation, this separation technique is complementary to capillary gel electrophoresis and field-flow fractionation. Although large DNA molecules were used to demonstrate the separation mechanism here, these protocols could also be applied to the separation of proteins, cells, or particles based on size, shape, or deformability. Separation is usually required in chemical analysis for identification, quantification, purification, or fractionation. For the separation of large molecules, numerous techniques have been developed, such as electrophoresis, liquid chromatography (LC), field-flow fractionation (FFF), etc.1-3 Electrophoresis including gel electrophoresis and capillary electrophoresis (CE) has become the main tool for the separation of DNA. CE with polymer solutions is usually employed for the separation of DNA molecules less than 20 kb. Agarose gel electrophoresis with pulsed electric fields has been used to separate large DNA molecules ranging from 1 kb to 1 Mb, but the separation usually takes 10 h or longer.4-9 LC is generally not as efficient as CE for size separation, * Corresponding author. E-mail:
[email protected]. Phone: 515-294-8062. (1) Giddings, J. C. Science 1993, 260, 1456-1465. (2) Janca, J. Field-Flow Fractionation; Marcel Dekker: New York, 1988. (3) Schimpf, M. E.; Caldwell, K.; Giddings, J. C. Field Flow Fractionation Handbook; Wiley-Interscience: New York, 2000. (4) Braun, B.; Blanch, H. W.; Prausnitz, J. M. Electrophoresis 1997, 18, 19941997. (5) Birren, B. W.; Lai, E.; Hood, L.; Simon, M. Anal. Biochem. 1989, 177, 282-286. (6) Zhang, T. Y.; Smith, C. L.; Cantor, C. R. Nucleic Acids Res. 1991, 19, 12911296. (7) Isfort, R. J.; Robinson, D.; Kung, H. J. J. Virol. Methods 1990, 27, 311317. (8) Chu, G.; Gunderson, K. Anal. Biochem. 1991, 194, 439-446. (9) Denko, N.; Giaccia, A.; Peters, B.; Stamato, T. D. Anal. Biochem. 1989, 178, 172-176. 10.1021/ac034430u CCC: $25.00 Published on Web 07/02/2003
© 2003 American Chemical Society
and it cannot be used to separate DNA fragments larger than 2 kb. An alternative technique called “slalom chromatography” has been developed to separate DNA molecules in the range of 5-50 kb but the mechanism has not yet been elucidated.10-12 FFF has been utilized to separate DNA molecules larger than 5 kb, but the resolution is not satisfactory, and the complex instrumentation limits its practical application.13-15 Some other approaches using porous media16,17 and microfabrication18 have been reported, but resolution remains to be a problem. We reported the radial migration of DNA molecules during electrophoresis with applied Poiseuille flow.19 We demonstrated that this radial migration resulted from the deformation and orientation of DNA molecules under shear flow. Briefly, in the presence of a parabolic (Poiseuille) flow, deformable particles such as DNA are oriented with respect to the direction of bulk flow. When an electric field is applied to induce electrophoretic motion in the same direction as bulk flow, the drag force in the opposite direction creates a lift force on the DNA molecules and focuses them toward the center of the capillary. Similarly, when the electrophoretic motion is in the direction opposite to bulk flow, DNA molecules are defocused and move toward the capillary walls. Two applications have been developed based on this phenomenon.20,21 Recently, we noticed that the velocity of radial migration is sizedependent. Inspired by the concept of FFF, we propose herein a novel mechanism for the separation of large molecules based on radial migration. However, in contrast to FFF, the separation was performed in a fused-silica capillary with no perpendicular field applied. This greatly simplifies the instrumentation. Here, we employed λ DNA (48 502 bp) and φX174 RF DNA (5386 bp) as model molecules to demonstrate the feasibility of applying this mechanism to both analytical and preparative separations. (10) Hirabayashi, J.; Kasai, K. Nucleic Acids Symp. Ser. 1988, 20, 57-58. (11) Huber, C. G.; Oefner, P. J.; Bonn, G. K. Anal. Chem. 1995, 67, 578-585. (12) Boyes, B. E.; Walker, D. G.; McGeer, P. Anal. Biochem. 1988, 170, 127134. (13) Schallinger, L. E.; Yau, W. W.; Kirkland, J. J. Science 1984, 225, 434-437. (14) Schallinger, L. E.; Gray, J. E.; Wagner, L. W.; Knowlton, S.; Kirkland, J. J. J. Chromatogr. 1985, 342, 67-77. (15) Liu, M. K.; Giddings, J. C. Macromolecules 1993, 26, 3576-3588. (16) Cole, K. D. Biotechnol. Prog. 1997, 13, 289-295. (17) Cole, K. D.; Tellez, C. M.; Blakesley, R. W. Electrophoresis 2000, 21, 10101017. (18) Huang, L. R.; Tegenfeldt, J. O.; Kraeft, J. J.; Sturn, J. C.; Austin, R. H.; Cox, E. C. Nat. Biotechnol. 2002, 20, 1048-1051. (19) Zheng, J.; Yeung, E. S. Anal. Chem. 2002, 74, 4536-4547. (20) Zheng, J.; Yeung, E. S. Anal. Chem. 2003, 75, 818-824. (21) Zheng, J.; Yeung, E. S. Aust. J. Chem. 2003, 56, 149-153.
Analytical Chemistry, Vol. 75, No. 15, August 1, 2003 3675
EXPERIMENTAL SECTION Experimental Setup. The experimental setup for this study is similar to that reported previously.19 Briefly, a Pentamax intensified-CCD camera (Roper Scientific, Princeton, NJ) was mounted on the camera mount of a Zeiss Axioskop upright microscope and was used to record the migration of single DNA molecules in a capillary. A 488-nm argon ion laser (Uniphase, San Jose, CA) provided the excitation beam. By using a cylindrical lens, the laser beam was focused normal to the capillary as a thin sheet. Only molecules in the center plane of the capillary (∼10 µm in thickness) were excited and recorded. A Uniblitz mechanical shutter with shutter controller (Vincent Associates, Rochester, NY) was used to block the laser beam when the camera was off to reduce photobleaching. Fluorescence from single molecules was collected by a Zeiss 20×/0.75 NA plan Apochromat microscope objective lens. Two 488-nm holographic notch filters with optical density of >6 (Kaiser Optical, Ann Arbor, MI; HNFP) were placed between the objective and the ICCD to remove scattered light from the excitation beam. The camera was operated in the external synchronization mode with the intensifier disabled open for single-molecule imaging and in the free-run mode when used as a CE detector. Data were acquired with the WinView software provided by Roper Scientific. A more detailed description of the setup can be found in our previous publication.19 Capillary Electrophoresis. We used both square and round fused-silica capillaries (Polymicro, Phoenix, AZ, 75-µm i.d., 365µm o.d.) in this study. A clear window (1 cm long) was created by removing the polymer cladding on the capillary with a window maker (Microsolv, Long Branch, NJ). The window was then cleaned repeatedly with methanol-soaked lens cleaning paper before use. The capillary inner wall was coated with linear polyacrylamide (LPA) according to standard procedures. The purpose of the LPA coating was to suppress electroosmotic flow (EOF) to reduce variations in migration times caused by a fluctuating EOF. All capillaries were fixed in a sturdy aluminum block so that no realignment was necessary between runs. To rinse the capillary or to inject samples, a capillary adapter (InnovaQuartz, Inc., Phoenix, AZ) was used to couple the capillary to a syringe. Two 1.5-mL centrifuge tubes were used as buffer reservoirs. A positive power supply was used, and sample was injected from the ground end. Hydrodynamic flow (HDF) was applied by changing the height of the ground buffer reservoir. Buffer and DNA Samples. The buffer used in this study was 1 or 10 mM pH 7.9 Gly-Gly (Sigma, St. Louis, MO) buffer. Solutions were filtered with a 0.22-µm filter and photobleached overnight with a mercury lamp prior to use. The λ DNA (48 502 bp, Life Science, Grand Island, NY) was used without further purification and was labeled with YOYO-1 (Molecular Probes, Eugene, OR) at a ratio of 1 dye molecule/5 bp according to the manufacturer’s instructions. Typically, YOYO1-labeled DNA samples were first prepared as 200 pM stock solutions, incubated at room temperature for at least 1 h, and diluted with the corresponding buffer to the desired concentration just prior to the start of the experiment. Circular φX174 RF DNA (5386 bp, Life Science) was digested with the enzyme Stu1 (Life Science) to yield linear φX174 RF DNA and was also labeled with YOYO-1 at a ratio of 1 dye molecule/5 bp. The stock solution of linear φX174 RF DNA was 400 pM. 3676 Analytical Chemistry, Vol. 75, No. 15, August 1, 2003
Figure 1. Radial distribution of λ DNA and φX174 RF DNA in a square capillary: (a) before radial migration and (b) after radial migration. A 75-µm-i.d., 40-cm-long square capillary was filled with 10 fM λ DNA or 10 fM φX174 RF DNA. The low concentration of DNA was selected to reduce the molecule counting error resulting from the overlapping of molecules after radial focusing. The results shown were the sum of 100 frames. DNA molecules were driven through the detection window by lifting the cathodic reservoir 4 cm higher than the anodic reservoir. This generated a HDF with a maximum velocity at the center of the capillary of 350 µm/s. The HDF was applied throughout the experiment.
Data Acquisition and Analysis. The WinView data file recorded the fluorescence images of DNA molecules within the field of view. With an in-house computer program, the number of DNA molecules with respect to their radial positions was obtained as a function of time. Since the concentration of DNA is very low, the average molecule numbers at specific radial positions in 100 consecutive image frames was used. The ICCD was also used as a fluorescence detector when conventional CE was performed with this setup. In that case, the ICCD was operated in free-run mode with an exposure time of 0.25 or 0.5 s. Using the statistics function of Winview software, the total intensity in each frame was obtained. The electropherogram was then constructed by plotting the total intensity versus time. RESULTS AND DISCUSSION Size Dependence of the Radial Migration Velocity. Shown in Figure 1a is the radial distribution of λ DNA and φX174 RF DNA under HDF. The relative radial position 0 refers to the center of the capillary, while -1 and 1 are the walls. Both λ DNA and φX174 RF DNA were randomly distributed across the capillary.
Figure 2. Dependence of the deviation angle φ-θ on the particle eccentricity ratio e at different particle orientations of θ ) 5°, 15°, 30°, 45°, 60°, and 75°.
An electric field of 100 V/cm was then applied to the ends of the 30-cm-long capillary. As we have demonstrated in our previous work, when an electric field is applied together with a HDF from cathode to anode, DNA molecules migrate toward the center of the capillary in addition to their axial migration.19 Shown in Figure 1b is the radial distribution of λ DNA and φX174 RF DNA after applying the electric field. After 10 s, most of the λ DNA molecules were focused into a 15-µm region around the center of the capillary. However, even after applying the electric field for 60 s, φX174 RF DNA molecules were still distributed in a 50-µm region around the axis of capillary. This indicated that the radial migration velocity was size-dependent, such that the larger the molecules, the faster their radial migration. As discussed in our previous work,19 DNA molecules in a hydrodynamic (Poiseuille) flow are deformed by shear flow and orient themselves. Here, θ is the direction of electrophoretic motion of the DNA molecules relative to the reference axis of the oriented DNA molecules. The electrophoretic velocity vEP is decomposed into a tangential component, vEP cos θ and a normal component vEP sin θ. Since the Reynolds number is much smaller than unity, the flow over the DNA molecules can be considered as Stokes flow. Therefore, the tangential Ft and normal Fn drag forces associated with the two velocity components are calculated with the Stokes drag equation.19,22 For an ellipsoid with a maximum radius of a and a length of 2b, e ≡ b/a is the eccentricity ratio. Here, we assume that DNA molecules are deformed to have a shape similar to an ellipsoid. Thus, the ratio of the normal force to the tangential force is calculated
coefficients. At equilibrium, the net drag force is balanced by the electric force in opposite directions. This means that the inclination of the particle φ is
Fn cn′ sin θ (3 + 2e) sin θ 3 + 2e ) ) tan θ ) Ft ct′ cos θ (4 + e) cos θ 4+e
(22) Panton, R. L. Incompressible Flow, 2nd ed.; John Wiley & Sons: New York, 1996; Chapter 21. (23) Bloomfield, V. A.; Crothers, D. M.; Tinoco, I. Nucleic Acids: Structures, Properties, and Functions; University Science Books: Sausalito, CA, 2000; Chapter 9.
(1)
where cn′ and ct′ are the normal and tangential drag force
()
φ ) tan-1
(
Fn 3 + 2e ) tan-1 tan θ Ft 4+e
)
(2)
The angle of deviation of the trajectory of the particle is a line at the angle φ-θ.22 Using the above formula, we calculated the dependence of deviation angle φ-θ on the eccentricity ratio e at different θ. The results are shown in Figure 2. It is clear that particles with larger eccentricity ratios e have a larger deviation angle (φ-θ) from the electrophoresis direction for a given orientation angle (θ). That is, particles will migrate toward (or away from) the axis at different rates depending on e. Due to the limited optical resolution in our setup, we were not able to measure the value of θ directly. We note that the hydrodynamic radius RG of a large DNA molecule (assumed spherical) is given by23
RG2 )
2PN(3.4 × 10-8) cm 6
(3)
where P is the persistence length and N is the number of bases. That is, RG is proportional to base number. However, since λ DNA is larger than φX174 RF DNA, the difference in shear stress on the two sides of the λ DNA molecule is larger than that on the φX174 RF DNA. Therefore, a λ DNA molecule is more deformed.
Analytical Chemistry, Vol. 75, No. 15, August 1, 2003
3677
Figure 3. Schematic illustration of the novel separation mechanism utilizing the differences in radial migration velocity of particles under HDF and an electric field. Open circles: large or deformable particles. Solid circles: small or spherical particles.
This, in the extreme case, is similar to the concept of biased reptation in gel electrophoresis.24 The deformed λ DNA has a larger eccentricity ratio than that of φX174 RF DNA so that it exhibits a larger deviation angle (φ-θ) (Figure 2) and thus faster radial migration. Separation Mechanism. For a Poiseuille flow, the flow velocity is a function of the radial position. Therefore, if different molecules are distributed at different radial positions, they will migrate at different velocities and can be separated. This is the basic principle of field-flow fractionation (FFF).1-3 In Figure 1, we have shown that DNA molecules of different sizes have different radial migration velocities and thus show different radial distributions after application of an electric field. Therefore, it should be possible to separate the molecules in a manner akin to FFF. However, a simple combination of a constant electric field and a constant HDF does not provide satisfactory resolution for λ DNA and φX174 RF DNA. That is because the radial distribution of the smaller molecules will finally reach equilibrium like the large molecules, although after a longer time. We therefore propose a novel separation mode that is schematically shown in Figure 3. The empty circles represent large, deformable particles while the small dots represent small and nondeformable particles. For simplicity, we will refer to a LPA-coated capillary where the EOF can be assumed to be zero. The discussion to follow is applicable to bare fused-silica capillaries with EOF, except that the elution order will be reversed. At the start, both types of particles are injected and coexist at the same axial position. After an electric field is applied, both particles will migrate at the same velocity νEP from cathode to anode, assuming that these particles are negatively charged and have the same mobility. Then, HDF with a maximum velocity of νm is applied from anode to cathode. For the small and nondeformable particles, their migration velocity is the combination of bulk flow and electrophoresis and is radial position-dependent. (24) Lumpkin, O. J. Biopolymers 1982, 21, 2315-2316.
3678
Analytical Chemistry, Vol. 75, No. 15, August 1, 2003
Specifically, for small particles at the capillary wall, the HDF velocity is 0. Therefore, their migration velocity remains at νEP. For the same small particles at the center of the capillary, their migration velocity becomes νEP - νm. Conversely, if HDF is applied from cathode to anode, small particles at the wall still have a velocity of νEP while those at the center of the capillary have a velocity of νEP + νm. If HDF in each direction is applied for an identical period t, the final migration distances for small particles both at the wall and at the center are 2νEPt. If molecular diffusion is neglected, there should be no band broadening. For large and deformable particles, the situation is quite different. When HDF is applied, they will be deformed (shown in Figure 3 as empty ellipsoids) and orient at a certain angle given in Figure 2. When the HDF is applied from anode to cathode, they will be defocused onto the wall and all of them will have a migration velocity of νEP. When the HDF is applied from cathode to anode, all large particles will be focused into the center of the capillary and have a migration velocity of νEP + νm. If HDF in each direction is applied for an identical period t, the final migration distances for the large molecules will be 2νEPt + νmt. When compared with the final migration distance of the small particles, the large particles migrate farther (by a distance of νmt) after the complete HDF cycle. If the above process is repeated n times, the small particles and the large particles will be separated by a distance of nνmt. This is the basic separation scheme invoked in this study. A more thorough mechanistic discussion would have to include partially deformable particles as well as radial molecular diffusion. Analytical Separation of λ DNA and OX174 RF DNA. In actual experiments, we used λ DNA and φX174 RF DNA as model molecules to demonstrate the above separation scheme. Shown in Figure 4a are electropherograms for φX174 RF DNA. The buffer reservoirs were balanced to eliminate any HDF. In the first run, φX174 RF DNA was driven through the detector by electrophoresis without any HDF. A sharp peak at 1144 s was observed. In the second run, after applying electric field for 1 min, HDF was applied by changing the height of the cathode buffer reservoir (15 cm to generate HDF cycles with a maximum velocity of 1.08 mm/s. HDF was applied from cathode to anode for 1 min and then from anode to cathode for another minute. This process was repeated for four cycles. The peak position for φX174 RF DNA shifted to 1127 s, and peak broadening was observed. This can be attributed to the slow radial migration of φX174 RF DNA and radial and axial molecular diffusion. Figure 4b shows the corresponding electropherograms for λ DNA. Without HDF, the migration time of λ DNA was 1162 s, roughly corresponding to that of φX174 RF DNA. When four cycles of HDF with the same peak velocity and direction as in Figure 4a were applied, the peak position of λ DNA shifted to 715 s. Compared to the migration time of φX174 RF DNA in Figure 4a, a difference of 412 s was found. No additional band broadening was observed for λ DNA, which could be attributed to its fast radial migration and reduced molecular diffusion. Figure 4c shows the separation of a mixture of λ DNA and φX174 RF DNA with four cycles of HDF applied 1 min after application of the electric field. Baseline separation was achieved. The peak positions of φX174 RF DNA and λ DNA exactly corresponded to those in Figure 4a and b. These results confirm that DNA molecules can be separated on the basis of their size
Figure 4. (a) Electropherogram of φX174 RF DNA with and without HDF. Sample: 20 pM φX174 RF DNA in 1 mM pH 7.9 Gly-Gly buffer. (b) Electropherogram of λ DNA with and without HDF. Sample: 20 pM λ DNA in 1 mM pH 7.9 Gly-Gly buffer. (c) Separation of 10 pM φX174 RF DNA and 10 pM λ DNA in 1 mM pH 7.9 Gly-Gly buffer. HDF: four cycles, each cycle included 1 min from cathode to anode and 1 min from anode to cathode. The height of the cathodic buffer reservoir was changed (15 cm relative to the anodic buffer reservoir to generate a HDF with a maximum velocity of (1.08 mm/s. A 60cm-long round, LPA-coated capillary (75-µm i.d., 365-µm o.d.) with an effective length of 52 cm was used. DNA samples were injected at the cathode hydrodynamically. A positive power supply was used so that the anode was the high voltage end and the cathode was at ground. The high-voltage anode was confined in an interlock box. This reduced the risk of electrical shock when the cathode buffer reservoir was raised or lowered.
and deformability. It is worth noting that no gel/polymer or pulse electric field was used. The separation was also much faster than pulsed-field gel electrophoresis. In fact, the experimental migration time of 21 min (which caused part of the band broadening) was dictated by the length of the capillary even though four HDF cycles took only 8 min to implement. The inherent limit is that the cycle time must be fast compared to radial diffusion but slow compared to radial focusing.
Preparative Separation of Molecules. As we stated above, when HDF is applied from anode to cathode, a negatively charged, deformable molecule will be defocused toward the capillary wall. Since the flow velocity at the surface is close to zero, the net migration is always from cathode to anode. However, for nondeformable molecules, the migration direction is the difference between the electrophoresis and HDF, which is a function of their radial position. By choosing the appropriate buffer, electric field, EOF, and HDF velocity, it should be possible to make deformable molecules migrate from cathode to anode while forcing most nondeformable molecules to migrate from anode to cathode (HDF faster than electrophoresis). This type of motion can be generated continuously and therefore can be used for preparative separations. Shown in Figure 5 are six consecutive frames from such a separation. Again, 1 pM φX174 RF DNA and 0.2 pM λ DNA in 10 mM pH 7.9 Gly-Gly buffer were used as model analytes. HDF was applied from anode to cathode (downstream) with a maximum velocity of 720 µm/s. An electric field of 100 V/cm was applied, and the electrophoretic velocity was ∼452 µm/s from cathode to anode (upstream). Since both molecules were labeled with the same dye/bp ratio, we were able to distinguish φX174 RF DNA from λ DNA molecules based on the brightness of their images. The large, bright spots corresponded to λ DNA while the small, dim spots corresponded to φX174 RF DNA. From Figure 5, it is evident that λ DNA molecules were defocused toward the wall and migrate upstream, or from cathode to anode. The φX174 RF DNA molecules are harder to follow because their velocities are position-dependent. To aid in tracing these molecules, we highlighted the same group of φX174 RF DNA molecules with a circle in Figure 5. An AVI movie is included in Supporting Information showing all of the frames of this data file. It is clear that most φX174 RF DNA molecules migrate from anode to cathode. Calculations showed that molecules in a 45-µm region around the capillary axis migrated from anode to cathode. Since only φX174 RF DNA migrates toward the cathode, we should be able to collect pure φX174 RF DNA at the cathode. Eventually, all of the φX174 RF DNA molecules will migrate into the cathode buffer reservoir and the λ DNA molecules remain in the anode buffer reservoir. The throughput with one 75-µm-i.d. capillary is probably too low for practical applications. However, it would not be difficult to use thousands of capillaries to run the separation simultaneously. For example, a bundle of 1000 75-µm-i.d. and 150-µmo.d. capillaries measure only 5.4 mm in diameter. Large i.d. capillaries can also be used to further improve the throughput, but the time needed for focusing will be larger due to a larger radial distance for migration. CONCLUSIONS We have demonstrated the separation of large molecules based on their radial migration with hydrodynamic flow and an applied electric field. Both analytical separation and preparative separations were achieved. The latter scheme might be applicable in the preparation of total DNA from a cell. Also, the use of this method for the isolation of plasmid or chromosomal DNA and biological cells is especially intriguing because the shear gradient here is relatively low, thus reducing the risk of breaking fragile DNA chains or cell membranes. The actual preparative separation demonstrated here was limited to the separation of binary Analytical Chemistry, Vol. 75, No. 15, August 1, 2003
3679
Figure 5. Six consequential frames showing opposite migration directions for φX174 RF DNA and λ DNA. The φX174 RF DNA molecules migrated downstream, or from anode to cathode at the center, while λ DNA migrated in the opposite direction, from cathode to anode at the walls. The sample was 1 pM φX174 RF DNA and 0.2 pM λ DNA in 10 mM pH 7.9 Gly-Gly buffer. Maximum flow velocity, 720 µm/s. Electric field, 100 V/cm.
components. However, if desired, the experiment can be operated in a multistage mode to separate multiple components. Each stage would be designed to separate a pair of analytes in a series. At the end, different buffer vials will collect pure single components. The more similar the particles (size or deformability), the more difficult it will be to design a set of conditions to impose opposite migration behavior. Changing the ionic strength of the buffer affects the separation significantly to allow further control. Other factors may include buffer pH, temperature, viscosity, and the addition of surfactants. The deformability of DNA molecules can be roughly predicted based on the balance of electrostatic repulsion force, hydrophobic force, and hydrogen bonding in a manner similar to DLVO theory.25,26 For example, at high ionic strengths or low pH, DNA molecules tend to exist as coils and are less deformable due to the decreased electrostatic repulsion force.27 Modifying EOF alters the relative magnitudes of the two opposing forces on the particle and thus the selectivity of the separation. (25) Derjaguin, B. V.; Landau, L. Acta Physicochim. U.S.S.R. 1941, 14, 633642. (26) Verwey, E. J. W.; Overbeek, J. T. G. Theory of the Stability of Lyophobic Colloids; Elsevier: Amsterdam, 1948. (27) Kang, S. H.; Shortreed, M. R.; Yeung, E. S. Anal. Chem. 2001, 73, 10911099.
3680
Analytical Chemistry, Vol. 75, No. 15, August 1, 2003
Although we used DNA molecules as models to elucidate the separation mechanism, this separation scheme should be applicable to other macromolecules and particles with differences in size, shape, or deformability. Considering the simple instrumentation, versatility, and potentially high efficiency (preparative mode), this new methodology should be a meaningful addition to the family of separation tools. ACKNOWLEDGMENT The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract W-7405-Eng-82. This work was supported by the Director of Science, Office of Basic Energy Sciences, Division of Chemical Sciences, and the Office of Biological and Environmental Research. SUPPORTING INFORMATION AVAILABLE An AVI movie of the complete progression of events in Figure 5. This material is available free of charge via the Internet at http://pubs.acs.org.
Received for review April 25, 2003. Accepted June 13, 2003. AC034430U
