Anal. Chem. 1991, 63, 1233-1236
1239
Separation of DNA Restriction Fragments by Capillary Electrophoresis Using Coated Fused Silica Capillaries Mark Strege and Avinash Lagu* Eli Lilly and Company, Lilly Research Laboratories, Indianapolis, Indiana 46285 The aMIities of several different capillary electrophoresis techdqum to separate DNA resbktbn fragments up to 23 000 bp were Invmtlgated. Methods -yhg eb&w"tk f b w In an untreated dika capillary were found to provide, at best, only partial resolution of the 23 fragments In a 1-kbp DNA ladder. By coating the inner walls of a slllca capillary with poly(acrylam1de) and filling these caplllarles with buffers contdnhg ~hylceiluloseas a &vhg medium, aH fragments In the l-kbp DNA ladder were separated. I n addition, this technique facilitated the separation of the very large fragments in a X D N A - H M I I I digest. Optimum resolution was obtalnd at iow separation potentials using buffers containing at loart 0.5% methylcellulose. The performance of this technique, Le., resolution and quantitatlon, make capillary electrophoresis a powerful complement to slab gel electrophoreds and may make It a preferred alternative to both agarose gel electrophoresis and HPLC for appilcatlons such as the confirmation of plasmid integrity.
INTRODUCTION The manipulation of DNA has become the foundation of contemporary molecular biology, and the discovery of restriction endonucleases has prompted the development of techniques for the separation and characterization of large DNA molecules. Since individual restriction enzymes cleave double-stranded DNA at specific base sequences, the separation and identification of individual fragments can provide important information regarding the structure of the DNA. Restriction mapping of double-stranded DNA fragments is currently a major application of slab gel electrophoresis, where separations are based upon size differences as the linear DNA molecules are sieved during their migration through the pores of the gel. The separation of DNA fragments within the size range 500-15000 bp has proven to be an application well-suited to agarose gel in particular (I, 2). This method is currently employed in the confirmation of the integrity of plasmids utilized in the fermentation of recombinant DNA products. While successful as a qualitative technique, agarose gel electrophoresis is a time-consuming (>2 h), tedious, nonquantitative process prone to error, particularly in regard to the unreliable detection of small fragments and the incomplete resolution of fragments larger than 15000 bp. The analytical separation of DNA restriction fragments by liquid chromatography is a technique offering several advantages compared to agarose gel electrophoresis. Potentially, high-performance liquid chromatography (HPLC) offers rapid separation and quantitation when combined with reliable area integration. However, although contemporary anion-exchange HPLC packings appear very suitable for the analysis of mixtures of DNA fragments less than 1000 bp in size, resolution of larger fragments was found to decrease significantly with increasing fragment size (3). Capillary electrophoresis (CE) is another technique offering the benefits of rapid separation and integration (4-6). Since its mechanism of separation is electrophoretic, it is not restricted by limitations associated with analyte-packing interactions, as is HPLC. CE has displayed the ability to sep0003-2700/91/0363-1233$02.50/0
arate a wide variety of molecules, including nucleic acids and their subunits, using very high potential electrical fields, with extreme efficiency in regard to theoretical plate height, sample volume, and analysis time (7,8). Various techniques associated with CE, such as micellar electrokinetic capillary chromatography (MECC) or poly(acrylamide) gel capillary electrophoresis, have been successfully employed to separate nucleosides, nucleotides, and oligonucleotides (9-15). Large double-stranded DNA molecules such as restriction fragments, however, possess similar charge/mass ratios and have proven much more difficult to separate. This problem was predicted by theoretical analysis, which proposed that the electrophoretic mobility of these molecules in free solution was independent of molecular weight (16). Early attempts by Kaspar et al. at the separation of large linear DNA fragments in capillaries filled with viscous buffers were unsuccessful (17). Derivatization of the capillary wall with [3-(methacryloxy)propyl]trimethoxysilane,in an effort to reduce electroosmotic flow, did not enhance separation, although partial resolution of fragments less than 1500 bp was achieved in a poly(acrylamide) gel filled capillary. Yamamoto et al. reported attempts at the separation of nucleic acids with a wide range of molecular size, including uncut X phage DNA (50000 bp), employing techniques previously used for the isotachophoretic analysis of proteins (18). Poor separation efficiency was obtained in these separations. Cohen et al. reported a high-resolution separation of a DNA mixture containing fragments from 72 to 23 130 bp in size, using a Tris-borate buffer containing 7 M urea and 0.1% SDS (19). Although borate, urea, and SDS were presumed to have bound to the DNA and altered its conformation, the mechanism of the separation was not well understood. Employing a polymer-modified capillary, Zhu et al. demonstratedthe separation of restriction fragments up to 4182 bp (fragments larger than 1000 bp were only partially resolved) by electrophoretically sieving these fragments through a buffer containing methylcellulose (20). Chin and Colburn reported the development of a buffer that can facilitate the separation of DNA fragments between 100 and 7000 bp in an uncoated capillary (21). Recently, Heiger et al. achieved high-efficiency separations of restriction fragments up to 12000 bp in size by using a poly(acry1amide) gel filled capillary (22). If one of the techniques noted above could be optimized to separate fragments up to 15000 bp, it could efficiently replace agarose gel electrophoresis as an analytical procedure for plasmid integrity. To obtain both an optimum separation and an understanding of the separation mechanism, the effects of variables such as buffer components and separation potential were also investigated. EXPERIMENTAL SECTION Capillary electrophoresis was performed with an automated PACE 2000 instrument (Beckman Instruments, Inc., Palo Alto, CA) controlled by an IBM PS/2 Model 80 386 computer fitted with PACE software (Beckman) running in a WINDOWS (Micrmft, Redmond, WA) environment The capillary cassette used was fitted with a 75-pm-i.d. fused silica capillary, 50 cm in length (43 cm to the detector). Injection of the sample was by low pressure (0.5 psi) for 5 s, avoiding any bias problems that may occur with electrokineticinjection (23). On-column detection was 0 1991 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 63,NO. 13, JULY 1, 1991
1234
L
0.0501
s
I I
P , - I
II
w
n 4
0.ooo-l 0.0
0.~~~1
I\
I
‘
I
i
4
I
Tlme (min)
30.0
-0 1. Electropherogram of the l k b p DNA ladder separated at 20 kV in 0.1 M Tris-borate, pH 8.1, 2.5 mM EDTA, 7 M urea, and 0.1 %
SDS.
performed by UV absorption at 254 nm, and the temperature was controlled at 30 f 0.1 O C . A 1-kbp DNA ladder (0.9 pg/pL, Lot ANC702; BRL, Gaithersburg, MD) and a X DNA-Hindm digest (0.25 pg/pL, Lot D105; Promega, Madison, WI) were utilized as protein-freesources of DNA restriction fragments. The 1-kbp DNA ladder contained 23 fragments of 75,134,154,201,220,298,344,396,506,517,1018, 1636,2036,3054,4072,5090,6108,7126,8144,9162,10180,11198,
and 12 216 bp and was solubilized in 10 mM Tris-HC1 (pH 7.5), 50 mM NaCl, and 0.1 mM EDTA. The X DNA-Hind111 digest contained eight fragments of 125,564,2028,2322,4371,6557,9419, and 23 130 bp and was solubilized in 10 mM Tris-HC1 (pH 7.5), 10 mM NaCl, and 1 mM EDTA. Acrylamide, ammonium persulfate, and N,N,N’,”-tetramethylethylenediamine (TEMED) were purchased from Bio-Rad (Richmond, CA). Methylcellulose (high viscosity), urea, and sodium dodecyl sulfate (SDS) were obtained from Fluka (Ronkonkoma, NJ). Tris(hydroxymethy1)aminomethane(Tris)was purchased from Fisher Scientific (Fair Law, NJ), and boric acid, sodium hydroxide, and (ethy1enedinitrilo)tetraacetic acid (EDTA) were obtained from EM Science (Gibbstown, NJ). Sepragene 5000, a commercially available buffer, was a gift from Applied Biosystems, Inc. (Foster City, CAI. The untreated capillaries used in this investigation were conditioned prior to use with a sequential bmin rinse of 0.1 M NaOH, a 3-min rinse with water, and a 5min rinse with the starting buffer. Poly(acry1amide)-coatedcapillaries were conditioned for at least 15 min in 1 M NaOH and then prepared in the manner described by Hjerten (24). All buffers were fiitered through a 0.25-pm pore size filter (Millipore) or centrifuged in a model TJ-6 centrifuge (Beckman) at 3000 rpm for 2 min to remove particulates. All solutions were vacuum degassed.
RESULTS AND DISCUSSION The 1-kbp DNA ladder was subjected to CE a t 20 kV (cathode at the detector end) in a buffer composed of 0.1 M Tris-borate, pH 8.1,2.5 mM EDTA, 7 M urea, and 0.1% SDS, as suggested by Cohen et al. (19). The resulting electropherogram, displayed in Figure 1, revealed the DNA fragments to elute as one unresolved peak. Pretreatment of the sample by heating at 60 OC for 15-20 min, followed by injection of the hot sample, did not improve separation. Dilution of the sample in the running buffer and adjustment of the separation potential were also found to have no effect upon resolution. It may be important to note that the silica tubing used in this study was obtained from a different manufacturer than that employed by Cohen et al. (19). Since analyte-wall interactions can play an important role in separations inside uncoated capillaries, difference in capillary composition may have had a significant effect upon these separations. Separation of the 1-kbp DNA ladder was next attempted with the Sepragene 5000 buffer, a Tris-borate-based media containing hydrophilic cellulose derivatives, developed for the separation of DNA fragments in an uncoated open capillary (21). This separation system employs a mechanism based upon the attraction and interaction of DNA fragments to the
-0.010
0.0
Time (min)
90.0
Figwe 2. Electropherogram of the l k b p DNA ladder separated at 10 kV in Sepragene 5000 buffer. Resdved fragments are labeled by thek sizes in base pairs.
cellulose derivatives in the buffer: the larger the fragment, the stronger the interaction. Thus, in principle, the largest fragment should elute first, as the smaller fragments more easily migrate upstream through the buffer toward the anodic inlet, resisting the electroosmotic flow. The elution profile of the 1-kpb DNA ladder, obtained at 10 kV (cathode at the detector end), is displayed in Figure 2, with the resolved fragmenta labeled by their sizes in base pairs (21). Reaolution of fragments 3054-bp and smaller was evident, but the separation appeared to diminish rapidly in regard to fragments larger than 2036 bp. Extreme peak broadening was found to occur as fragment size decreased, and the five smallest fragments present in the sample eluted as two broad bands at approximately 70 min. This effect was a function of fragment migration velocity. A separation at 20 kV required only 25 min and sharpened these late-eluting peaks, but resolution of the larger fragments decreased. This evidence suggested that this separation mechanism may perform best when used in conjunction with a voltage gradient, an option currently not available on the PACE instrument. Another strategy that may reduce peak broadening and improve resolution would be the employment of capillaries of narrower bore. The third restriction fragment separation technique attempted in this investigation was molecular sieving in the absence of electroosmotic flow, as had been reported by Zhu et al. (20). Capillaries were coated with poly(acrylamide) to “izethe effects of electroosmotic flow (24)and then filled with 0.5% methylcellulose in 50 mM Tris-borate, pH 8.0, and 2.5 mM EDTA. The acrylamide coating may also serve to reduce adsorption effects, since phosphate, which comprises the backbone of the DNA molecule, has been shown to interact strongly with bare silica (25,26). The separation of the 1-kbp DNA ladder presented in Figure 3, obtained at 10 kV (anode a t the detector end), revealed that all fragments were separated within 17 min. The fragments were identified by their sizes in base pairs, the assignments agreeing with the reported separation of the 1-kbp DNA ladder obtained in a poly(acrylamide) gel filled capillary, where peaks were assigned by spiking with slab gel isolated fragments (22). In addition to the relatively rapid separation provided by this technique, extremely high resolution was evident. Two fragments differing by only 11bp which are usually not separable on a slab gel (27),the 506- and 517-bp fragments, were baseline resolved by sieving through methylcellulose. An investigation of the effect of methylcellulose concentration upon resolution of the 1-kbp DNA ladder revealed that separation improved as methylcellulose concentration increased up to 0.5%. At concentrations above 0.5%, no improvement in resolution was evident. Inside a poly(acrylamide)-coated capillary filled with 0.5% methylcellulose, 50 mM Tris-borate, pH 8.0, and 2.5 mM
ANALYTICAL CHEMISTRY, VOL. 63, NO. 13, JULY 1, 1991
o*olol
4072 3054 I
I
1235
6557
- 0.001 25.0
- 0.001 4
I
18.0
10.0
Time (min) Flgu. 3. Eiectropherogamof the l k b p DNA ladder separated at 10 kV In a poMacrylaWWated capMary filled with 50 mM Tris-borate, pH 8.0,2.5 mM EDTA, and 0.5% methyicellukse. Resohred fragments are labeled by their sizes in base pairs.
I
o,Mo] Figure 4b
5 m
I3 1 O.OIO 1 Figure 4c 0.0oO 0.0
-0.001
6.0
I
J
0.0
Time (mln)
12.0
Flguro 4. Electropherogramsof the l k b p DNA ladder separated at 20 kV In (a)a poly(acrylamkle)coatedcaplllary (no condltlonlng with NaOH prior to coating) fHled with 50 mM Trls-borate, pH 8.0, 2.5 mM EDTA, and 0.5% methylcellulose; (b) a poly(actytamlde~oatedcapillary filled with 50 mM Trls-borate, pH 8.0, and 2.5 mM EDTA: (c) a noncoated caplllary filled with 50 mM Trls-borate, pH 8.0, 2.5 mM EDTA, and 0.5% methylcellulose.
EDTA, electroosmotic flow is virtually eliminated, as determined by the fact that a neutral marker (1% acetone) does not appear at the detector after 60 min at 20 kV (anode at the injection end) following injection of the sample into the short end of the capillary 7 cm from the detector window. Activation of the silanols on the capillary wall by conditioning the capillary with NaOH prior to the coating step was found to be crucial for obtaining optimum resolution, as evidenced by the comparativelypoorer separation obtained in a coated capillary that had not been preconditioned with NaOH (see Figure 4a). The separation of the 1-kbp DNA ladder in 50 mM Tris-borate, pH 8.0, and 2.5 mM EDTA at 20 kV, displayed in Figure 4b, suggested that the contribution of the wall-anchored poly(acrylamide1 chains to sample separation was minimal and that the presence of methylcellulose in the
Time (mln)
41.O
Figwe 5. Electropherogram of the A DNA-HMIII digest separated at 10 kV In a poIy(acrylamkle)-cmted capillary filled with 50 mM Trls-borate, pH 8.0, 2.5 mM EDTA, and 0.5% methylcellulose (high viscosity). Resolved fragments are labeled by theh size In base paks.
buffer was crucial to achieving the separation displayed in Figure 3. A separation of the sample in 0.5% methylcellulose, 50 mM Tris-borate, pH 8.0, and 2.5 mM EDTA at 20 kV inside a noncoated capillary (cathodeat the detector end) (see Figure 4c) revealed resolution similar to that achieved in the Sepragene 5000 buffer, displayed in Figure 2. By using a coated capillary filled with the methylcellulose buffer, the complete separation of the eight fragments in a A DNA-Hind111 digest (see Figure 5) was achieved at 10 kV (the 125-bp fragment was not detected). The fragments in this sample were labeled, their assignments agreeing with size-based separations of the A DNA-HindIII digest achieved via both electrophoresis on a 1%agarose gel and anion-exchange HPLC (3). It was evident that although fragment resolution decreased with size, perhaps in response to the reported tendency of large DNA molecules to align themselves parallel to the applied field, resulting in a snakelike, size-independent mode of migration known as reptation (28,29),this method remains applicable to the separation of very large fragments. The poly(acry1amide)-coated capillaries described here exhibited a run-to-runrelative standard deviation of migration times (1636-bp fragment) of 0.4% (n = 91, a day-to-day reproducibility of 0.8% (n = 5), and a capillary-to-capillary reproducibility of 0.9% (n = 5). For each coated capillary, these separations were reproducible for a period of approximately 2 weeks of use or at least 50 injections before the effecta of coating degradation (peakbroadening, decreased resolution) became evident. In addition to storage of the capillaries in dilute acetic acid when not in use, operation with acidic buffens may prolong the effective lifetime of the coating, as silica is known to possess a pronounced solubility in water at alkaline pH values, and the siloxane bond linking the coating to the capillary wall is known to be prone to nucleophilic cleavage under basic conditions (30,31). We are currently studying methods to regenerate the coated columns to their initial performance.
CONCLUSIONS We have found the use of poly(acry1amide)-coatedcapillaries in conjunction with buffers containing 0.5% methylcellulose to be well suited for capillary electrophoretic separations of a wide size range of DNA restriction fragments. Poly(acry1amide)gel filled capillaries have been shown to provide comparable resolution (261,but we feel that the use of coated open tubular capillary electrophoresis allows one to easily avoid difficulties associated with electrophoresis following polymerization inside a closed tube, such as poor gel-to-gel reproducibility, bubble generation, and gel matrix collapse (32).Also, in comparison, techniques developed for
1236
ANALYTICAL CHEMISTRY, VOL.
63,NO. 13, JULY 1, 1991
use with noncoated capillaries demonstrated either no resolution or partial separations that were limited by fragment size and the effects of electroosmotic flow. Another benefit provided by coated capillaries may be increased elution time reproducibility, since the electroosmotic flow present in uncoated capillaries is highly dependent upon numerous environmental parameters, such as pH, temperature, and buffer viscosity. Recent studies have suggested that complete equilibration of a silica capillary surface is very slow and may take weeks (33). The development of highly stable capillary coatings, such as the poly(acry1amide) coating recently reported by Cobb et al. (34), may serve to further enhance the performance of DNA sieving techniques employing solutions of hydrophilic polymers such as methylcellulose. The advantages offered by capillary electrophoresis in poly(acry1amidel-coated capillaries, in regard to resolution and quantitation suggest that this technique may be a preferred alternative to agarose gel electrophoresis or HPLC for applications such as the confirmation of plasmid integrity and is at the very least a powerful complement to slab gel electrophoresis. Registry No. Silica, vitreous, 60676-86-0; (acry1amide)(TEMEDMammonium persulfate) polymer, 133522-77-7; methylcellulose, 9004-67-5.
LITERATURE CITED (1) Serwer. P. Ekctrophoresls 1963, 4 . 375. (2) Andrews, A. 1.E k m & : Tschnlques,end&bchemlcal end CWnlcal Apphtbns; Oxford Unlversky Press: New York, 1986;Chapter 6. (3) Strege, M. A.; Lagu. A. L. J. Chrmtogr.. in press. (4) Eby, M. J. BloITechnokgy 1969,7 , 903.
m,
( 5 ) Hjerten, S. J . Chromatog. 1963, 270. 1. (6) Ewing. A. J.; WaHingford, R. A.; Olefirowicz, T. M. Ana/. Chem. 1969, 61, 292.
(7) Gordon. M. J.; Huang, X.; Pentoney, S. L.; Zare, R. N. SckMcs 1966, 242, 224. (8) Kuhr, W. 0.Anal. Chem. 1990, 62, 403 R. (9) Burton. D.; Sepaniak, M.; Maskarinec, M. CIvometOgrepMe 1966, 21, 583. (10) Row, K. H.; &lest. W. H.; Maskarinec, M. P. J. Chrometogr. 1967, 409, 193. (11) Liu, J.; Banks, J. F. J.; Novotny, M. J. Microcd. Sep. 1969, 1 , 136. (12) m e n . A.; Terabe, S.; Smith, J.; Karger, B. Anal. Chem. 1967, 59,
1021. 330. T.; Nakagawa, G.; Sato, M.; Yagi, K. Appl. Blochem. 1963, 5 , (13) Tsuda, (14) Ooinwc, v.; LIU, J.; Banks, J. F.; Novotny, M.; Bocek, p. J . -tog. 1969. 480. 321. (15) S u s t k k , V.; Foret. F.; Bocek, P. J. Chrometogr. 1969, 480, 271. (16)Hermans, J. J. J. Wm.Scl. 1953. 78. 257.
(17) Kaspar, T. J.; Melera; M.; Gozei, P.; Brownlee. R. G. J. Chrometogr. 1966. 458, 303. (18) Yamamoto, H.; Manabe, T.; Okuyama, T. J. J. Chrmtogr. 1969, 480. 331.
(20) Zhu, M.; &nsen, D. L.; Burd, S.; Gannon, F. J. Chromatog. 1969, 480,311. (21) Chin, A. M.; Coiburn, J. C. A m . Blotech. Lab. 1969, 7 (1OA).16. (22) Heiger, D. K.; Cohen, A. S.; Karger, B. L. J . Chrmtogr. 1990. 33. (23) Huang, X.; Gordon, M. J.; Zare, R. N. Anal. Chem. 1966, 60. 375. (24) Hjerten, S.J. Chromatogr. 1965, 347, 191. (25) Mitsyuk, B. M. R w s . J. " r g . a m . 1972, 17, 471. (26) McCormick, R. M. Anal. Chem. 1966,60, 2322. (27) 1 Kb DNA Ladder Product Information. Bethesda Research Laboratories, Gaithersburg. MD, 1990. (28) Lumpkin, 0.J.; Dejardin, P.; Zimm, 8. H. SlOpdLmers 1985, 24, 1573. (29) Slater, G. W.; Noolandi. J. Bbpo&mefs 1969, 28, 1781. (30) Iler, R. K. The chemrsby of Silica; Wlley: New York, 1979; p 42. (31) Unger, K. K. Porous Sllica; Elsevier Scientific Publishing Co.: New York, 1979;Chapter 3. (32) Rickwood, D.; Hams, B. D. &I Electrophoresis of Nwkk A&: A &3Ct/C8/ Approach; IRL Press: Oxford, England. 1982; Chapter 1. (33) Lambert. W. J.; Middieton. D. L. Anal. Chem. 1990. 62, 1585. (34) Cobb, K. A.; Doinik, V.; Novotny, M. Anal. Chem. 1990, 62, 2478.
RECEIVED for review October 16,1990. Accepted March 20, 1991.
