Anal. Chem. 1994,66, 2446-2450
Separation of Large DNA Fragments by Capillary Electrophoresis under Pulsed-Field Conditions Jan Sudor and Milos V. Novotny' Department of Chemistry, Indiana University, Bloomington, Indiana 4 7405
Capillary zone electrophoresis (CZE), using an entangled polyacrylamide solution, was applied to large DNA samples under pulsed-field conditions. Highly efficient separationswere achieved under biased sinusoidalfield and field-inversion pulsing regimes. The separations that were obtained with 8.3-48.5 kb XDNA standards and 48.5 kb-1 Mb XDNA concatamers (modified with ethidium bromide) clearly demonstrate a dramatic improvement in the separation time (roughly, 10-50 times) over the conventionally used slab-gel techniques. Moreover, the CZE method appears much more sensitive and amenable to component quantificationand method automation.
Approximately ten years ago, Schwartz and Cantor' developed pulsed-field gel electrophoresis (PFGE) to cope with difficulties in separating large DNA strands in an electric field. Since then, PFGE has evolved into a set of separation techniques of great utility to geneticists and molecular biologists.2 McDonnel et al. had previously shown3 that the electrophoretic separation of DNA fragments, under constant field conditions, ceases to be effective at approximately 2030 kilobase (kb) pairs in size. The difficulties of a size-dependent separation of "giant" DNA molecules are now largely attributed to the molecular stretching and reptation behavior of these moleculesk6 in electric fields. This reptation concept was originally introduced by de Gennes7 and Doi and Edwards8 for polymer melts. The reptation of biopolymer chains can be countered by induction of polymer motions through various voltage-pulsing techniques. The electrophoretic movement of charged polymers in entangled polymer matrices, or "gels", has attracted considerable interest from theoreticians as ell.^-'^ One of the most successful techniques of PFGE is field-inversion gel electrophoresis (FIGE), introduced by Carle et al.15 in 1986,where pulsing takes place at a 180' angle. Since the introduction (1) Schwartz, D. C.; Cantor, C. R. Cell 1984, 37, 67-75.
(2) Burmeister, M., Ulanovsky, L., Eds. Pulsed Field Gel Electrophoresis: Methods in Molecular Biology; Humana Press: Totowa, NJ, 1992. (3) McDonnel, M. W.; Simon, M. N.; Studier, F. W. J . Mol. B i d . 1977. 110, 114-146. (4) Lumpkin, 0. J.; Dejardin, P.; Zimm, B. H. Biopolymers 1985, 24, 15731593. ( 5 ) Slater, G. W.; Noolandi, J. Phys. Reu. Left. 1985, 55, 1579-1583. (6) Noolandi, J.; Slater, G. W.; Lim, H. A,; Viovy, J. L. Science 1989, 243, 1456-1458. (7) de Gennes, P.-G. J. Chem. Phys. 1971, 55, 572-579. (8) Doi, M.; Edwards, S. F. J. Chem. Soc., Faraday Trans. 2 1978, 74, 17891801. (9) Grossman, P. D.; Soane, D. S . Biopolymers 1991, 31, 1221-1228. (IO) Viovy, J. L.; Duke, T. A. J. Elecfrophoresis 1993, 14, 322-329. (11) Deutsch, J. M. J. Chem. Phys. 1989, 90, 7436-7441. (12) Zimm, B. H. J. Chem. Phys. 1991, 94, 2187-2206. (13) Noolandi, J. Annu. Reu. Phys. Chem. 1992, 43, 237-256. (14) Duke, T. A. J.; Viovy, J. L. J . Chem. Phys. 1992, 96, 8552-8563. (15) Carle, C. F.; Frank, M.; Olson, M. Science 1986, 232, 65-68.
2448
Analytical Chemisty, Vol. 66, No. 15, August 1, 1994
of this simple technique, the theory of FIGE has been studied extensively, while the applications to increasingly large DNA molecules have been gradually pushed to new limits.I6J7 Kotaka et al. 18-20 have recently developed a technique called biased sinusoidal field gel electrophoresis (BSFGE). In BSFGE, a sinusoidal field strength, Es, of frequencyf, is superimposed on a steady, biased field, Eb (Et, = (El + E2)/2, where E1 and E2 are the intensities of forward and backward pulses, respectively). The chief advantage of this technique over conventional FIGE is its capability to overcome dynamic self-trapping and, consequently, band inversion21.22 of sample polymers in a separation medium. The use of entangled polymer solutions instead of a fixed gel network should also result in less frequent occurrence of band inversion because of the reduced effect of the solid friction on mobility.23 Capillary zone electrophoresis (CZE)24.25has evolved during the last decade as an extraordinarily efficient and fast separation method. The recent interest in extending CZE to larger DNAs (up to 48.5 kb) is evident from the studies of Chrambach et a1.,26 but similarly to the electrophoresis in slab gels, the disadvantages of working with constant fields become evident. Combining the FIGE approach with CZE occurred to Heiger et al.,27who briefly studied the effect of pulsed field on the resolution of small restriction fragments. Similarly, velocity modulation of small DNA fragments was studied by a different group.28 In both studies, the effects of alternating electric fields were marginal because of the small size (0.5-1.5 kb) of the DNA fragments used. In our l a b ~ r a t o r ya, ~pulsed-field ~ CZE technique was successfully applied to the separation of dextran polysaccharides ( lo51 06-Da molecular mass); because of very different polymer topologies and chain flexibilities, these results cannot be readily translated to the situation with large DNA molecules. We report here our initial results on separating large XDNA (from 8.3 to 48.5kb; and from 48.5 kb to 1 Mb, concatamers of 48.5 kb) by CZE under various conditions of pulsed-field electrophoresis. There are very distinct advantages of the (16) Turmel, C.; Brassard, E.; Slater, G. W.; Noolandi, J. Nucleic Acids Res. 1990, 18, 569-575. (17) Schwartz, D. C.; Koval, M. Nature 1989, 338, 520-522. (18) Shikata, T.; Kotaka, T. Biopolymers 1991, 31, 253-254. (19) Shikata, T.; Kotaka, T. Macromolecules 1991, 24, 48684813. (20) Kotaka, T.; Adachi, S.; Shikata, T. Electrophoresis 1993, 14, 313-321. (21) Slater, G. W. J . Phys. IIFr. 1992, 2, 1149-1158. (22) Ulanovsky, L.; Drouin, G.; Gilbert, W. Nature 1990, 343, 190-192. (23) Burlatsky, S.;Deutch, J. Science 1993, 260, 1782-1784. (24) Jorgenson, J. W.; Lukacs, K. D. Anal. Chem. 1981, 53, 1298-1302. (25) Jorgenson, J. W.; Lukacs, K. D. Science 1983, 222, 266-272. (26) Guszczynski, T.; Pulyaeva, H.; Tietz, D.; Garner, M. M.; Chrambach, A. Electrophoresis 1993, 14, 523-530. (27) Heiger, D. N.;Cohen, A.S.; Karger, B. L. J . Chromatogr. 1990,516,3348. (28) Demana, T.;Lanan, M.; Morris, M. D. Anal. Chem. 1991,63,2795-2797. (29) Sudor, J.; Novotny, M. V. Proc. Nafl. Acad. Sci. U.S.A. 1993, 90, 94519455.
0003-2700/94/036&2446~04.50/0
0 1994 American Chemlcai Society
n a
I
I 160 min
100
13 I
25 30 35 10 45 50 55 Time (min) Flgure 1. Separation of the 8.3-48.5kb DNA size standards under (a) constant-field and (b) pulsed-field ceplllary electrophoresis. Peak assignments: (1) 8.3, (2) 8.0, (3) 10.1, (4) 12.2, (5) 15.0, (0) 17.1 (7) 19.4, (8) 22.6, (9) 24.8, (10) 29.9, (11) 33.5, (12) 38.4, and (13) 48.5 kb: the asterisk indicates two bands (1.1 kb, and 1.5 kb; not identified). Conditlons: buffer, 0.5X TBE (44.5 mM Trla, 44.5 mM boric add, and 1.25 mM EDTA), pH 8.1,0.0 % linear polyacrylamide (MW (5-0) X lo8, Polysclences, Inc., Wanington, PA): electromigration injectbn, 30 8 (25 V/Cm): applied voltage, 25 V/cm for (a); Eb = 50 Vlcm and E, = 100 V/cm for (b); frequency of the applied sine-wave input signel, 12 Hz.
5
0
15
10
20
22 ;L 21 8.3 kb 8.6 kb 10.1 kb 12.2 kb 15.0 kb 17.1 kb 19.4 kb 22.6 kb 24. 8kb 29.9 kb 33.5 kb 38.4 kb
20
19 18 17 16
48.5 kb I3
1
I
I
I
I
0.0
0.2
0.4
0.6
0.8
,
1.o
1.2
Log (freq) Flgure 2. Dependence of the electrophoretic mobilities of 13 DNA fragments on log (frequency) of the input slne-wave signal. Mobiilty is shown in (cm*/V.s) X lo6 units. Range of tested frequencies, 1-15 Hz. Other conditbns are the same In Figure lb.
methodology reported here over other state-of-the-art techniques? (a) CZE separations are considerably faster than those in slab gels; (b) separations in the capillaries permit more quantitative and more sensitive detection; (c) the CZE method is potentially more reproducible and amenable to automation.
EXPERIMENTAL SECTION The DNA size standards (8.3-48.5 kb; a mixed digest of XDNA) and XDNA concatamers (48.5 kb-1 Mb; X cI857Sam7) were from Bio-RadLaboratories(Hercules, CA). Boric acid was purchased from Mallinckrodt, Inc. (Paris, KY), and T r i m a base (tris(hydroxymethyl)aminomethane), EDTA, Analytical Chemistry, Vol. 66, No. 15, August 1, 1994
2447
and ethidium bromide were received from Sigma (St. Louis, MO.). A home-built CE system, described p r e v i o ~ s l ywas , ~ ~used for all experiments. The separation capillaries, 10-20-cm effective length (20-40-cm total length; 75-pm id.; 194-pm o.d.), were purchased from Polymicro Technologies (Phoenix, AZ). The inner surfaceof the separation column was modified by the attachment of linear p~lyacrylamide.~~ The column was enclosed in a Plexiglas box with an interlock safety system. A variable-wavelength UV detector (UVIDEC-100-V; Jasco, Tokyo, Japan) operating at 260 nm, was modified for capillary on-column detection. On-column fluorescence measurements were carried out with an argon ion laser (Model 543, Omnichrome, Chino, CA) used as a light source (2-mW output power at 488 nm). The incident laser beam was aligned to its optimum position by adjusting the position of collecting optics between the optical cell and the detector. Fluorescence emission from 620 to 680 nm was collected through a 600-pm fiber-optic placed at a right angle to the incident laser beam. Signals isolated by broad-band interference filter with the peak wavelength at 650 nm (Oriel, Stradford, CT) were monitored with a R928 photomultiplier tube (Hamamatsu Photonics K.K., Shizuoka Pref., Japan), and amplified with a Model 128A lock-in amplifier (EG&G Princeton Applied Research, Princeton, NJ). The high-voltage dc power supply used in this work (Spellman High Voltage Electronics, Plainview, NY) was capable of delivering 0-30 kV. For the pulsed-field experiments, a 20-kV operational amplifier (Model 20/20; Trek, Inc., Medina, NY) controlled with a function generator FG2 (Beckman Industrial, Emerson Electric, Brea, CA) was used. Wave forms were monitored at the ground end of the column with a 5103N Tektronix oscilloscope (Tektronix, Inc., Beaverton, OR). The resistance of the column was calculated from the current at the biased electric field strength. The system’s RC time constant was -2.4 ms.
RESULTS AND DISCUSSION We have initially studied the migration mechanisms for the 8.3-48.5-kb DNA sample in a linear polyacrylamide solution by CZE under constant-field conditions. As expected, the electrophoretic mobilities, p, of all DNA fragments under such conditions were independent of the DNA chain lengths, L ( p a Lo, where L is a contour length of DNA molecule). Decreases in gel concentrations and field strength (to avoid stretching of the end-to-end vector of DNA molecules), as well as increases in gel concentration (to decrease a constraintrelease effect of a polymer solution), respectively, did not improve their separation. However, migration times increased substantially when either higher gel concentrations or low electric field strengths were applied. To force DNA chains into conformational changes during the separation process (stretched vs coiled conformation), for an improvement in component resolution, we applied alternating electric fields (sine- or square-wave input signals) along the capillary. When the sine-wave input signal was applied, resolution of DNA chains (8.3-48.5 kb) improved dramatically in comparison to (30) Liu, J.; Hsieh, Y . - 2 . ;Wiesler, D.; Novotny, M. V. Anal. Chem. 1991, 63, 408412. (31) Cobb,K. A,: Dolnik, V.; Novotny, M. V. Anal. Chem. 1990,62,2478-2483.
2448
AnalyticaIChemistry, Vol. 66, No. 15, August 1, 1994
a
C
0
5
10
15
20
25
30
35
40
45
50
55
Time (min) Flgure 3. Resolution of the mixed digest XDNA (8.3-48.5 kb) In dependence on frequency of the applied electric field. (a) f = 1 Hz, (b) f = 5 Hz, and (c) f = 15 Hz. Other conditions are the same as in Figure l b .
the constant-voltage regime (Figure 1, (a) vs (b)). This separation is extremely sensitive to the applied frequency of the sine wave. In fact, the separation shown in Figure 1b was obtained at 12-Hz optimum frequency (duty cycle), at the field intensities of forward and backward pulses, 150 and 50 V/cm, respectively (Eb = 50 V/cm, E, = 100 V/cm; 200 V/cm peak to peak). During a change in a frequency setting to either 10 or 15 Hz, the first two or the last four mixture components, respectively, did not separate from each other. For a symmetrical square-wave input signal, the optimum frequency of the same amplitudes of forward and backward
I
I
I
I
I
110
120
130
140
150
I
I
I
160
170
180
I
Time (min) Flgurs 4. Separationof tte XDNA concatamers (48.5 kb-1 Mb) by field-Inversion capillary electrophoresis. Condltions: buffer, 0.5X TBE, ethidium bromide (concentration,0.075 pglmL), pH 8.1,0.4% linear polyacrylamide (MW (5-6) X loe);vacuum Injection, 5 s; the orlglnal sample (agarose insert) was diluted 40 tlmes wlth 0.2X TBE buffer and prestainedwlth ethidium bromide (total Concentrationof ethidium bromlde was 0.075 pglmL); applied voltage, forward dlrectlon, 100 Vlcm, and reverse direction, 50 Vlcm; & = 2; cyde durations of forward and backward pulses were the same, R, = 1; frequency of the applied square-wave Input signal, 30 Hr.
pulses, as originally used in BSFGE, increased to 15 Hz, but still the first (8.3 kb) and the second (8.6 kb) fragment did not separate from each other. In addition, the peaks were relatively asymmetrical in comparison to BSFGE, and their migration times were longer (results not shown). The dependence of electrophoretic mobility on applied frequency is shown in Figure 2 for 13 DNA fragments. It is obvious that no separation between the fragments occurs at 1-Hz frequency (Figure 3a). Under such conditions, the stretching times of all DNA fragments (8.3-48.5 kb) were shorter than the duty cycle of the sine-wave input signal, and therefore, the DNA molecules could be stretched by the electric field and caused to “reptatenin the gel solution in both forward and reverse directions. In other words, the DNA chains spent most of their migration time in stretched conformations. When the frequency setting was increased to 5 Hz, we observed improvement in resolution for the largest fragments (Figure 3b). Consequently, through increasing frequency to 15 Hz, the largest fragments started to comigrate, while resolution between small fragments (8.3, and 8.6 kb) improved (Figure 3c). These measurements indicate that further refinements in component resolution can potentially be achieved through frequency or amplitude programming. In the following sets of experiments, separation of larger DNAchains (XDNA; 48.5 kb-1 Mb, concatamersof 48.5 kb) was attempted. This XDNA sample is sold as agarose inserts (0.75% Bio-Rad’s Low Melt Preparative Grade Agarose; 9 pg of XDNA per insert). When the sample was simply melted (at 65 OC for 5 min) and injected into the capillary, the DNA chains migrated together, entangled in the agarose matrix, resisting good resolution. In order to decrease entanglement of DNA chains in the agarose inserts, we diluted the sample with a TBE buffer, so that the final concentration of XDNA (after dilution) was 1 pg/mL (the concentration of agarose was 0.019%). For such a dilute solution, we had to use laser-
-
induced fluorescent detection of the ethidium bromide-stained DNA molecules. It was previously shown32s33that ethidium bromide (an intercalating fluorescent dye) increases the contour and persistence length of the DNA chains. It also decreases their charge-to-mass ratio and, consequently, their electrophoretic mobility. At first, we tried to separate this mixture under constant-field conditions ( E = 12.5-200 V/cm; 0 . 2 4 8 % linear polyacrylamide; molecular mass 5-6 MDa), but no size-dependent separation was achieved. BSFGE mode under low-bias conditions was tried next (Eb = 25 V/cm; E, = 75 V/cm), but we observed only a hint of separation at a frequency of 30 Hz. By use of the square-wave input signal (the amplitudes and the duty cycle of forward and backward pulses remained the same as in BSFGE), the resolution between fragments increased dramatically. Figure 4 shows the separation of XDNA concatamers at an optimized frequency of the alternating field (30 Hz). The separation time was -3 h. With increasing (decreasing) frequency (f5 Hz), the resolution of longer (shorter) chains decreased (f a l/L). Besides the main fragments, there are additional (smaller) peaks which probably correspond to DNA fragments. We suggest that at the field intensities used in this work (100 and 50 V/cm for forward and reverse pulses, respectively), longer DNA chains might partially break up. It is well-known from the slab-gel pulsed-field electrophoresis that the separation frequency is proportional to the intensities of the applied electric field and inversely proportional to the concentration of polymer solution cf a E a 1/C, where E is an electric field strength and C is a concentration of polymer solution). In order to decrease the dynamic self-trapping of DNA chains in a gel network, the electric field strength has to be decreased (32) Smith, S. B.; Aldridgc, P. K.; Callis, J. B. Science 1989, 243, 203-206. (33) Yoshikawa, K.; Matsuzawa, Y.; Minagawa, K.; Doi, M.; Matsumoto, M. Biochem. Biophys. Res. Commun. 1992, 188, 1274-1279.
Ana&ticalChemistry, Vol. 68, No. 15, August 1, 1994
2449
with increasing size of separated DNA molecules,34 and therefore, the frequency of the applied alternating field has to be decreased as well (if all other parameters are kept constant). This was not the case in our experiments. While the mixed digest of XDNA (8.3-48.5 kb) was separated at 12 Hz in 0.6% linear polyacrylamide (molecular mass 5-6 MDa; [b 32 nm; where [b is a blob size which represents a dynamic pore sizelo), the XDNA (48.5-1000 kb) was separated at 30 Hz in 0.4% linear polyacrylamide (molecular mass 5-6 MDa; [b = 43 nm). In addition to the influence of different concentrations of polymer solutions, the shape of an input signal also seems to play an important role. As mentioned before, the optimized frequency for the separation of DNA (8.3-48.5 kb) increased when the sine-wave input signal was changed to a square wave. The most distinct advantage of using pulsed-field CZE for large DNAs appears to be the time of separation. As seen in Figure lb, optimum separation of 13 fragments was accomplished in less than 50 min, and even large DNA chains (48.5 kb-1 Mb) were separated in -3 h. In contrast, similar separation in a slab gel takes 24 h or more.35 Furthermore, it should be feasible to quantify the individual sample ~~
~
(34) Vollrath, D.;Davis, R. W. Nucleic Acids Res. 1987, IS, 7865-7871
2450
AnalyticalChemistry, Vol. 66, No. 15, August 1, 1994
components reliably in various applications to comparative biochemistry and genetics. Sensitivity of the CZE technique with UV detection appears to be at least comparable to that observed with fluorescently stained slab gels. In fact, components marked with the asterisk (1.1, and 1.5 kb; not identified; Figure 1b), whose presence was a n t i ~ i p a t e dcould ,~~ not be readily discerned in the slab gels. Significant improvement in sensitivity can be achieved by employing laser-induced fluorescence detection (Figure 4), while optimized fluorescentlabeling procedures can still be developed.
ACKNOWLEDGMENT This research was supported by Grant 24349-1 3 from the National Institute of General Medical Sciences, the U.S. Department of Health and Human Services. We thank Chris Siebert of the Bio-Rad Laboratories for supplying us with the samples of DNA fragments. Received for review M a y 26, 1994. Accepted M a y 26, 1994." ( 3 5 ) LifeScience Research Products; Bio-Rad Laboratories: Hercules, CA, 1993; p 197. Abstract published in Aduance ACS Abstracts, June IS, 1994.
