Anal. Chem. 1994,66, 1168-1174
Separation of Nucleic Acids by Capillary Electrophoresis in Cellulose Solutions with Mono- and Bis-Intercalating Dyes Yongseong Kim and Michael D. Morris' Department of Chemistty, University of Michigan, Ann Arbor, Michigan 48 109- 1055
The effects of bis- and mono-intercalatingdyes on the capillary electrophoretic separation of double-stranded DNA have been investigated in buffers containing (hydroxypropylmethy1)cellulose. Broad bands and incomplete separations of #X 174 DNA HaeIII digest were obtained with the bis-intercalators ethidium homodimer 1 (EthD-1) and ethidium homodimer 2 (EthD-2), under a wide range of dye and polymer concentrations. The above results were attributed to the presence of several intramolecular dye/DNA complexesfor each fragment. For EthD-2/DNA complexes, but not for EthD-1/DNA complexes, additional bands beyond the expected number of fragments were observed. The extra bands were attributed to formation of intermolecular dimers. Use of the monomeric intercalating dyes, ethidium and propidium, allowed complete separation of all fragments of #X 174 DNA HaeIII digest. A low-power green He-Ne laser (543.6 nm) was shown to be satisfactory for laser-inducedfluorescence detection with any of these dyes. Capillary gel electrophoresis (CGE) has been extensively used for the separation of nucleic acids.'-5 It has proven to be an efficient technique for such diverse purposes as the analysis of polymerase chain reaction (PCR) products,6 the separation of restriction digest fragment^,^ and the sequencing of DNA by the dideoxy chain termination procedure.8-'0 CE with linear polymer additives may have advantages over conventional cross-linked polyacrylamide gel capillary electrophoresis in many of these applications. The linear polymers provide homogeneous sieving structures and are less susceptible to damage via bubble formation at high electric fields. Cellulose derivatives are particularly attractive because the materials are commercially available and need only be dissolved and pumped into capillaries. DNA restriction fragments and PCR reaction products have been successfully analyzed with cellulose derivatives."-15
Except for sequencing by the dideoxy chain termination, UV absorption is the most common detection technique in nucleic acid CGE. However, laser-induced fluorescence (LIF) offers much lower detection limits and can be a surprisingly simple and inexpensive detector technology for nucleic acid CGE.I6J7The fluorophores which are covalently bonded to nucleotides and the intercalating dyes which are used with double-stranded DNA (dsDNA) can be excited with 1-2 mW of blue-green or green laser light. Illumination can be provided by compact air-cooled argon ion lasers (488 and 514.5 nm), green He-Ne lasers (543.6 nm), or diode-pumped Nd-YAG lasers (532 nm). The He-Nelaser may be especially attractive, because it is the least expensive of the currently available choices. There has been increasing interest recently in the use of homodimeric intercalating dyes for LIF detection of dsDNA because the fluorescence enhancement of these dyes upon intercalation is high and the binding constants to dsDNA are so large that a large excess of dye is not required to maintain the complex during electrophoresis.'8-2' Several bisintercalating dyes have been reviewed by Glazer and Rye18 with slab gel electrophoresis. Ethidium homodimer (EthD), thiazole orange homodimer (TOTO), and oxazole yellow homodimer (YOYO) all form highly stable fluorescent dsDNA/dye complexes (K= 108-109 M-I). In addition, the fluorescence quantum yield of bound bis-intercalating dye is much higher than that of the free dimer and is largely independent of base composition or sequence. Monomeric intercalating dyes have been used in both CGE and slab gel electrophoresis. Monomeric ethidium is added to running buffers in CGE even when UV absorption is used for detection. Longer migration times and better peak resolution of DNA fragments are obtained in both cross-linked polyacrylamideZ2and (hydroxypropylmethyl)cellulose.6 Similar dye/buffer systems have been used with LIF detection.23 Although the binding constants to dsDNA are relatively small, monomeric intercalating dyes might still be attractive for use
S.;Karger, B. L. J. Chromatogr. 1990,516,3348. (2) Milofsky, R. E.; Yeung, E. S. Anal. Chem. 1993, 65, 153-157. (3) Baba, Y.; Tsuhako, M.; Sawa, T.; Akashi, M.; Yashima, E. A w l . Chem. 1992, 64, 1920-3925. (4) Smith, L. M. Nature 1991, 349, 812-813. ( 5 ) Tietz, D.; Chrambach, A. Electrophoresis 1993, 14, 185-190. (6) Schwartz, H. E.; Ulferder, K.; Sunzeri, F. J.; Bush, M. P.; Brownlee,R. G. J . Chromatogr. 1991, 559, 267-283. (7) Guttman, A.; Cohen, A. S.;Heiger, D. N.; Karger, B. L. A w l . Chem. 1990, 62, 137-141. (8) Huang, X.C.; Quesada, M. A.; Mathics, R. A. Anal. Chem. 1992,64,21492154. (9) Swerdlow, H.; Gesteland, R. Nucleic Acids Res. 1990, 18, 1415-1419. (IO) Drossman, H.; Luckey, J. A.; Kostichka, A. J.; D'Cunha, J.; Smith, L. Anal. Chem. 1990,62,900-903. (11) Guttman, A.; Horvath, J.; Cooke, N. Anal. Chem. 1993,65, 199-203.
(12) Strege, M.; Lagu, M. A w l . Chem. 1991.63, 1233-1236. (13) Grossman, P. D.; Sonan, D. S.J. Chromatogr. 1991, 559, 257-266. (14) MacCrehan, W. A.; Rasmussen, H. T.;Northrop, D. M. J . Lip. Chromotogr. 1992, 15, 1063-1080. (15) Grossman, P. D.; Sonan, D. S . Biopolymers 1991, 31, 1221-1228. (16) Toulas, C.; Hernandez, L. LC-GC 1992,10,471476. (1 7) Zhang, J. Z.; Chen, D. Y.; Wu, S.; Harke. H. R.; Dovich, N. J. J. Chromatogr. 1991,559, 237-246. (18) Glazer, A. N.; Rye, H. S.Nature 1992, 359. 859-861. (19) Rye, H. S.;Yue, S.;Wemmer, D. E.; Quesada, M. A,; Haugland, P. R.; Mathies, R. A.; Glazer, A. N. Nucleic Acids Res. 1992, 20, 2803-2812. (20) Smith, A. B.; Aldridge, P. K.; Callis, J. B. Science 1989, 243, 203-206. (21) Glazer, A. N.; Peck, K.; Mathies, R. A. Proc. Natl. Acad. Sei. U.S.A. 1990, 87, 3851-3855. (22) Guttman, A.; Cooke, N. Anal. Chem. 1991, 63, 2038-2042. (23) Demana, T.;Lanan, M.; Morris, M. D. Anal. Chem. 1991, 63, 2795-2797.
(1) Heiger, D. N.;Cohcn,A.
1168
Analytical Chemistry, Vol. 66,No. 7, April I , 1994
0003-2700/94/03661168$04.50/0
0 1994 Amerlcan Chemical Societv
Br
-
A
B
H3C,
% ,
H3C,
,%
: W : m Flgure 1. Chemical structure of bis- and mono-Intercalating dyes: (A) ethidium bromide(EB); (B) propidiumiodide (PI): (C) ethidium homodimer 1 (EthD-1); (D) ethidium homodimer 2 (EthD-2).
anionic phosphate ions of nucleic acids. The binding modes of dimeric intercalators are somewhat more complicated. The dye’s planar aromatic rings are intercalated into nonadjacent base pairs of dsDNA following the neighbor exclusion principle: every second site along the double helix remains unoccupied.26 In addition to intercalation, two other modes of binding may occur. Noncooperative binding involves the formation of an ion pair between the nonintercalated ring and the anionic phosphate ion of DNA. There may also be dyedye or cooperative binding involving nonintercalated rings. Noncooperative and cooperative binding of bisintercalating dyes was demonstrated by Gaugain et al. using bothviscometric and fluorometric experiment^.^^ Their viscosity measurements showed no evidence for simultaneous intercalation of both dye ring systems. Subsequent experiments have confirmed intercalation of one ring only. Even though bis-intercalating dyes have been frequently used in slab gel electrophoresis and yield high detection sensitivity, there appears to be no systematic study of their use in CGE with polymer additives or comparison of their effects to those of the more familiar mono-intercalating dyes. Therefore, we have examined the electrophoresis of dsDNA fragments with both classes of intercalating dyes. The bisintercalating dyes EthD-1 and EthD-2 have been chosen for our study because their excitation maxima are better matched to our green He-Ne laser wavelength than those of TOTO and YOYO.
with LIF if the proper laser is chosen to yield intense fluorescence against the background of excess dye in the running buffer. Figure 1 shows the structure of two monomeric dyes, ethidium (Figure 1A) and propidium (Figure lB), and two homodimers,ethidium homodimer 1 (Figure 1C) and ethidium homodimer 2 (Figure 1D). Ethidium, propidium, and the ethidium homodimers have excitation maxima ranging from 526 to 536 nm, making them particularly attractive for use with green laser-excited fluorescence detection. Ethidium homodimer 1(EthD-1) has two phenanthridium rings and two bridging secondary amine nitrogens, written here as ammonium ions. Ethidium homodimer 2 (EthD-2) has two more methyl groups on each bridging nitrogen. The permanent positive charges on the quaternary bridging nitrogens in EthD-2 result in a higher binding affinity to dsDNA (KEthD-1 = 2 X lo8M-l, KEthD-2 = 1 X lo9M-l). The binding constant for monomeric ethidium (K= 1.5 X lo5 M-l) is about 1000 times less than that of EthD-1, because it has only one phenanthridiumring and a charge of 1 at pH 7-8. Propidium has an additional quaternary ammonium ion and a charge of +2 at pH 7-8. Consequently, propidium has a higher DNA affinity than ethidium. The additional methyl and ethyl groups on EthD-2 and propidium shift the excitation maxima 7-10 nm to the red of those of EthD-1 and ethidium. The interaction between intercalating dyes and dsDNA determines the migration behavior of dsDNA during CGE.24 Monomeric dyes such as ethidium bind primarily by intercalation to nucleic acid bases.25 In addition, there is some weak electrostatic binding between the cationic dye and the
EXPER I MENTAL SECTION Capillary electrophoresiswas performed in 75-pm i.d. fusedsilica capillaries (Polymicro Technologies, Inc.) with the cathode on the injection side and the anode on the detection side. The total capillary length was 33.0 cm with entrance detection distance of 28.5 cm. The fused-silica capillaries were conditioned for at least 15 min in 1 M NaOH and then coated with linear polyacrylamide. Following the method of Hjerten,28 (methacry1oxy)propyltrimethoxysilane (MAPS) was first covalently bound to the capillary wall, and then a He-saturated acrylamide solution with 3.5% T, 0% C was polymerized in the capillary using ammonium persulfate (APS) and NflJ’fl’-tetramethylethylenediamine (TEMED). A f20-kV amplifier (Model 20/20, Trek, Inc.) driven by a digital/analog converter was used to generate the voltage drop across the capillary. A laboratory fan was used to cool the capillary by forced air convection. On-column detection was by laser-induced fluorescence with a 2.5-mW, 543.6-nm He-Ne laser (PMS Electro-Optics) source. A 580-nm sharp cutoff filter and a 620-nm 10-nm band-pass interference filter were employed to isolate the emission. The optical signal was detected with a photomultiplier (Hamamatsu, R928), amplified and digitized to 14 bits at 7 points/s and stored on a 3 86SX-based computer. The pH 8.2 running buffer (1 X TBE) consisted of 89 mM Tris, 89 mM boric acid, and 2 mM EDTA. The buffer was filtered twice through a 0.25-pm pore size membrane filter prior to use. (Hydroxypropylmethy1)cellulose (HPMC) (Al-
(24) Watson, J. D.;Hopkins, N. H.; Roberts,J. W.; Steitz, J. A.; Weiner, A. M. Molecular Biology of the Gene, 4th 4.Benjamin/Cummings: ; Menlo Park, CA, 1987. (25) Lerman, L. S. J . Mol. Biol. 1961, 3, 18-30.
(26) Crothers, D. M. Biopolymers 1968, 6, 575-584. (27) Gaugain, B.; Barbet, J.; C a p e k N . ; Roques, B. P.; Le Pccq, J. E.Biochemistry 1978, 17, 5078-5088. (28) Hjerten. S.J. J.Chromatogr. 1985, 347, 191-198.
+
Analytical Chemistty, Vol. 00, No. 7, April 1, 1994
1169
Time b i n ) Flgure 2. Electrophoretic separation of 4X 174 RF DNA HeeIII digest in 1X TBE containing (hydroxypropylmethyl)celiuiose (HWC) using LIF detection wlth EthD-1: (A) 0.3% HPMC; (B) 0.4% HPMC; (C10.5% HPMC. Conditions: electric field strength, 180 V/cm; DNA:dye, 451; sample concentratbn, 53 ng/pL; sample injection, 4 kV for 12 s.
drich Chemical, 4000 CPat 2% solution) was added to the buffer to yield 0.3%, 0.4%, or 0.5% (w/w) HPMC solutions. Monointercalator separations were performed with either ethidium bromide or propidium iodide (Molecular Probes, Inc., Eugene, OR) added to the running buffer prior to CE a t concentrations of 3 and 5 pg/mL, respectively. For separations with bis-intercalating dyes, EthD-1 or EthD-2 (Molecular Probes) was added to DNA fragments a t DNA to dye mole ratios ranging from 5 : 1 to 90: 1. The DNA/dye mixture was then equilibrated in the dark for 30 min to. 1 h at room temperature. A g5X 174 R F DNA HueIII digest (Bathesda Research Labs) was used as the DNA fragment mixture for all experiments. In most experiments, the DNA concentration was 53 ng/pL. Samples were injected electrokinetically, typically for 12 s at 4 kV. Safety Considerations. Ethidium and propidium ions are potential mutagens, and the homodimers must be considered as potentially toxic as well. Caution should be taken to prevent contact of these materials or their solutions with the skin and ingestion or inhalation of the powders. The materials should be handled with gloves and protective clothing and eyewear. Manipulation of these materials in a fume hood is recommended.
RESULTS AND DISCUSSION Figure 2 shows the separation of HueIII digest of 4X 174 RF DNA in a linear polyacrylamide-coated capillary at 180 V/cm with 0.3% 0.4% and 0.5% (hydroxypropylmethy1)cellulose (HPMC) using the bis-intercalating dye EthD- 1. The peaks were identified by their increasing peak area. The assignmentsagree with the pattern of ref 6. The sieving ability of HPMC for various sizes of DNA fragments was dependent 1170 Analytical Chemistry, Vol. 66, No. 7, April 1, 1994
upon polymer concentration. The lowest concentration of polymer solution, 0.3%, provided better resolution for the longer fragments, but 0.4% was better for the small fragments. At 0.5%HPMC, the small fragment bands were broader than at 0.4%. At no HPMC concentration studied were the 271and 281-bp fragments resolved. It is known that pore size distribution of a polymer solution is important for the separation of dsDNA and that the average pore size of an entangled polymer is proportional to W 4 , where C is the polymer c o n ~ e n t r a t i o n Migration .~~ behavior of DNA fragments is usually adjusted by changing the polymer concentration. Interestingly, a t 0.3%, the 72-bp fragment is broad and barely visible above the background noise. The poor resolution may be caused by an effective poor size large enough that there is no sieving of the smallest fragments. Thus, 0.4% HPMC was used for further studies. The effect of DNA to dye ratio on the migration behavior of the fragments is shown in Figure 3 for 0.4% HPMC and EthD- 1. As the DNA/EthD- 1ratio is increased the migration time decreases and all of the bands become narrower. The improvement is dramatic as the DNA/dye ratio increases from5:l to45:l. Forshortfragments(
