Anal. Chem. 1994,66, 3081-3085
Pulsed Field Capillary Electrophoresis of Multikilobase Length Nucleic Acids in Dilute Methyl Cellulose Solutions Yongseong Kim and Michael D. Morris' Department of Chemistry, University of Michigan, Ann Arbor, Michigan 48 109- 1055
Pulsed field capillary gel electrophoresis in dilute methyl cellulose solutions is used to separate nucleic acid fragments in the size range 75-23 000 base pairs. Field inversion is shown to increase resolution for fragments longer than about 500 base pairs. Methyl cellulose solutions as dilute as 0.01%can be used. Intermolecular hydrogen bonding is suggested as the cause of apparent cellulose fiber entanglementat concentrations below the calculated entanglementlimit. The 1-kbDNA ladder and the XDNAIHidIII restriction fragment mixtures are each baseline resolved in a 28-cm capillary in less than 9 min at 180 V/cm (dc component). Capillary electrophoresis (CE) in linear polymer solutions, typically derivatized cellulose, or linear polyacrylamide is finding increasing acceptance for the rapid and efficient separation of many kinds of biopolymers. Applications to nucleic acid separations include (single-stranded) dideoxy chain termination products, polymerase chain reaction products, and DNA restriction The first two applications are to short fragments, generally less than 1000 bases or base pairs (bp) long. Restriction endonucleases yield much larger fragments, and separations of DNA fragments as long as 48 000 base pairs (kbp) have been reported.6 However, above about 10 kbp long, resolution has been poor and separation times have been long. Pulsed field capillary gel electrophoresis (PFCGE) is now emerging as a promising technique to improve nucleic acid separations, as well as separations of other biopolymers. Our own group has shown that sinusoidal modulation can be used7 and that separations can often be improved by random variation of the pulse protocol during the course of a run.* However, previously reported separations have been to shortchain nucleic acids, usually less than 1.5 kbp. Little improvement is expected or found below about 500 bp, because the mechanism of separation is sieving rather than reptation for short fragments. The sieving medium is important for the separation of DNA fragments. Migration behavior depends on both the chemical composition and the concentration of the polymer solution. Cellulose derivatives have been widely investigated, because ( I ) Cohen, A. S.; Najarian, D. R.; Karger, B. L. J. Chromatogr. 1990,516,4940. (2) Schwartz, H. E.; Ulfedler, K.; Sunjeri, F.J.; Busch, M. P.; Brownlee, R. G. J . Chromatogr. 1991, 559, 267-283. (3) Strege, M.; Lagu, A. Anal. Chem. 1991.63, 1223-1236. (4) McGregor, D.; Yeung, E. D. J . Chromatogr. A 1993, 652, 67-73. ( 5 ) Chiari, M.; Nesi, M.; Righetti, R. G. J. Chromatogr. A 1993, 652, 31-39. (6) Guszczynski, T.; Pulyaeva, H.; Tietz, D.; Garner, M. M.; Chrambach, A. Electrophoresis 1993, 14, 523-530. (7) Demana, T.; Lanan, M. Morris, M. D. Anal. Chem. 1991, 63, 2795-2797. (8) Navin, M.; Rapp, T.; Morris, M. D. Anal. Chem. 1994, 66, 1179-1182.
0003-2700/94/0366-308 1$04.50/0 @ 1994 Amerlcan Chemlcal Soclety
they are water soluble, have low viscosities, and are stable under CE c o n d i t i o n ~ . ~ J ~ Cellulose physical and chemical properties have been well characterized.' Although their solution behavior is conventionally described by the formalisms derived for random coil polymers and athermal solvents,12 it is clear that these descriptions are inadequate. While celluloses do form semidilute solutions in which the polymers are entangled, the scaling law is different from the classical expression known to be valid for polyethylene and other random coils. For cellulose solutions, a different scaling law, eq 1, has been empirically found by Barron and co-workers.13
In eq 1, c* is the entanglement threshold concentration and N is the number of monomer units. Using the viscosity dependence on chain length, Barron et al. reported an entanglement threshold which scales with the -1.2 power, rather than the classically predicted -0.8 power.12 That classical polymer solution theory should not apply exactly might be expected if the polymers do not consist of random coils and are subject to intra- or intermolecular chemical effects such as hydrogen bonding. Hydrogen bonding is well-known in carbohydrates, of course, and recent Raman spectroscopic work has shown that celluloses strongly hydrogen bond to water and disrupt the water structure.14 Similar behavior is observed for linear polya~rylamide,~~ the other major class of CE sieving polymers. Clearly, these polymers are not random coils and water is not a structureless athermal solvent. Barron and co-workers have shown that it is possible to obtain CE separation of nucleic acid fragments as large as 1353 bp at concentrations lower than the nominal entanglement threshold of hydroxyethyl c e l l ~ l o s e . Because ~~ the viscosity of such solutions is low, the separations are relatively fast. The separation is surprising and warrants further investigation because it is contrary to the conventional descriptions of capillary gel electrophoresis, which require entangled polymers. In this paper we describe the rapid and (9) Guttman, A.; Horvath, J.; Cook, N. Anal. Chem. 1993, 65,
199-203.
(IO) Grossman, P.; Sonan, D. S.; J . Chromazogr. 1991,559,257-266. (1 1) Nevell, T. P.; Zeronian, S. H. Cellulose Chemistry and its Application; Ellis Horwood Limited: West Sussex, England, 1985. (12) de Gennes, P. G. Scaling Concepts in Polymer Physics; Cornell University Press: Ithaca, NY, 1979. (13) Barron, A. E.; Soane, D. S.; Blanch, H. W. J . Chromatogr. A 1993,652,3-16. (14) Meada, Y.; Tsukida, N.; Kitano, H.; Terada, T.; Yamanaka, J. J . Phys. Chem. 1993, 97, 13903-13906. (15) Takana, N.; Ito, K.; Kitano, H. Macromolecules 1994, 27, 54C-544.
AnalYticalChemlstry, Vol. 66,No. 19, October 1, 1994
3081
I
12
1;
16
10
20
i2
24
time, minutes
Figure 1. Electrophoretic separation of 1-kb DNA ladder in 1X TBE containing 0.4% MC: (a) 180 V/cm dc only; (b) pulsed field, 180 V/cm dc 500-Hz sine wave, 130% modulation; (c) same conditions as in (b), but modulation 16 min after injection. Conditlons: DNA sample concentration, 17 ng/pL; electrokinetic InJection, 2 kV for 4 s. Peak legend: (1) 75, (2) 134, (3) 154, (4) 201, (5) 220, (6)298. (7) 344, (8) 396, (9) 506, (10) 517, (11) 1018, (12) 1636, (13) 2016, (14) 3054, (15)4072, (16) 5090, (17) 6108, (18) 7128, (19) 8144, (20) 9162, (21) 10 180, (22) 11 198, and (23) 12 216 base pairs.
+
high-resolution separation by pulsed field capillary gel electrophoresis of DNA fragments as largeas 23 kbpin methyl cellulose solutions at concentrations well below the nominal entanglement threshold.
EXPER IMENTAL SECT1ON The capillary electrophoresis apparatus used in this work has been previously described in detail.16 Briefly, a highvoltage amplifier (Trek 20/20) driven by a dc voltage source was used for dc electrophoresis. For pulsed field experiments the amplifier was driven by the summed voltages of a dcvoltage source and signal generator, Fixed frequency sinusoidal modulation at frequencies in the 100-2000-Hz range was employed. Modulation depths of 130-140% were used. Typically the dc voltage was 180 V/cm in dc and pulsed field separations. Detection was by laser-induced fluorescence, using a 543-nm He-Ne laser and filter/photomultiplier detection. The 75-pm-i.d. silica capillaries were coated with polyacrylamide, as previously described.I6 The capillaries had 28-cm inlet-detector length and were 33 cm long overall. Two different buffer solutions, 1X TBE (89 mM tris, 89 mM boric acid, 2 mM EDTA), and 2.5X TAE (100 mM tris, 50 mM acetic acid, 5.5 mM EDTA) were used. Methyl cellulose (MC) (Aldrich Chemical, 4000 CPat 2% solution) was added to the buffer solutions to yield 0.4%, 0.3%, 0.02%, 0.015%, and 0.01% (w/w) methyl cellulose. Ethidium bromide was added to the running buffer prior to CE at 3 pg/mL for laserinduced fluorescence detection. The test fragment mixtures were a 1-kb DNA ladder and XDNA/HindIII digest (Bethesda Research Labs) at nominal concentrations of 17 and 12 ng/ pL, respectively. (16) Kim,Y.;Morris, M. D.Anal. Chem. 1994, 66, 1168-1174.
3082
Analytical Chemistry, Vol. 68, No. 19, October 1, 1994
11
ia
15
17
19
21
I
Time, minuter
Figure 2. Eiectrophoretlc separation of XDNA/HindIII digest in 1X TBE Containing 0.3% MC: (a) 180 V/cm dc only; (b) pulsed field, 180 V/cm dc i-500-Hz sine wave, 140% modulation depth. Conditions: DNA sample concentration, 12 ng/pL; electrokinetic injection, 0.7 kV for 4 8. Peak legend: (1) 554, (2) 2028, (3) 2322, (4) 4371, (5) 6657, (6) 9419, and (7) 23 130 base palrs.
RESULTS AND DISCUSSION Figure 1 shows the separation of a 1-kb DNA ladder in 0.4% methyl cellulose at 180 V/cm dc field and under pulsed field conditions. The peaks are identified by their increasing peak area and are assumed to be eluting in order of increasing chain length. As shown in Figure 1a, under dc-onlyconditions, all 23 fragments ranging in size from 75 bp to 12 kbp in 1-kb DNA ladder are separated. For relatively small fragments (-300 bp), resolution is excellent. However, resolution becomes worse for larger fragments (>500 bp). The 509/ 517- and the 1636/2106-bp fragments are not baseline resolved, and resolution degrades rapidly above 10 kbp. Similar behavior has been observed by others3 Generally, the poor performance at long chain length is attributed to the onset of reptation or biased reptation in the entangled polymer network. In the pulsed field mode (Figure lb), much improved resolution is observed, particularly for the larger fragments. All large (> 1 kbp) fragment bands are sharpened, the 1636/ 2 106-bp fragments are well resolved, and near-baseline resolution is observed for the 11 198/12216-bp fragments. Mobilities are increased, an effect we have previously attributed to the temperature increase in the The improvement due to pulsed field operation is much clearer here than in previous reports, because about 12 fragments migrate in the reptation regime, instead of 2 or 3. Because of the increased temperature, resolution suffers in the 500-bp region, although two incompletely resolved bands are still observed. There are many ways to improve the short chain length separation. In this case, we simply delay the start of field pulsation until almost all of the smaller fragments have passed through the capillary, about 16 min. There is, of course, some loss of resolution at longer chain lengths, but high resolution is obtained for all fragments. Figure 2 shows separation of the XDNA/HindIII digest fragments in 0.3% methyl cellulose solution. The digest
Table 1. Entanglement Threshold (c') and Contour Length (L.) between Entanglement Points at c = c' for Methyl Cellulose M" c* (%, w/w) r, (nm)
h7
24 000 45 000 86 000 90 000 180 000 320 000 450 000
1.04 0.493 0.25 1 0.221 0.103 0.0575 0.0422
56.66 106.3 202.5 212.5 425.0 755.6 1063
b I 4
6
8
10
12
I
14
Time, minuter
Figure 3. Electrophoretic separation of XDNA/HlndIII digest in 1X TBE containing 0.02% MC: (a) 180 V/cm dc only: (b) pulsed field, 180 Vlcm dc 1-kHr sine wave, 140% modulation; (c)field isorheic with (b), 220 V / c m dc. Other conditions are as in Figure 2.
+
contains seven fragments ranging in size from 554 bp to 23 kbp. They are widely spaced and all are baseline resolved in a dc separation, as shown in Figure 2a. However, the 23-kbp peak is broad and badly tailed. Under pulsed field conditions, all fragments larger than 6 kbp yield narrower bands. The effect is especially pronounced for the 23-kbp fragment, although it remains broader than is desirable. If the broad band of 23 kbp arises from the relatively small pore size of the polymer network, it should be sharper when more dilute methyl cellulose solutions are employed. When the methyl cellulose concentration is reduced to 0.1%, 0.05%, and 0.02%, the migration times become short as the solution viscosity decreases. We show the 0.02% behavior in Figure 3. Under dc conditions, the peaks for relatively small fragments (
