Electrophoresis of DNA W
ith the advent of gene-sequencing techniques, there has heen an increasing demand for analysis and separation of large DNA fragments. One potential technique for this purpose is gel electrophoresis. Unfortunately, for large molecules, the method has not had the resolving power to he of use. However, the introduction of pulsedfield gel electrophoresis (PFGE) has dramatically extended the range of sizes of DNA molecules that can he separated. Here we compare the use of constant and pulsed fields in electrophoretic separations and summarize the recent experiments by Steven Smith and co-workers that describe the visualization of individual DNA molecules during these separations. Conventional agarose gel electrophoresis of DNA is a sieving technique. Separation of molecules smaller than the maximum pore size of the matrix occurs because the smaller the molecule, the larger the fraction of pores it can enter. The size limit for sieving of DNA is a few hundred base pairs (bp) as DNA lengths approach the size of the pores. Once DNA molecules become larger than the pores of the agarose, they have to distort to enter the gel and the sieving power of the gel becomes unimportant. Instead, these larger molecules move through the agarose matrix with a mobility that is dependent on both the electric field strength and the molecular chain length. Resolution is predicted to decrease with the square of the molecular size. Furthermore, at high constant field strengths, electrical orientation effects further reduce the size dependence of mobility and DNA molecules with sizes above approximately 20,000 base pairs (20 kilobase pairs, 20 kb) do not resolve well in conventional agarose gel electrophoresis. In 1983 Schwartz et al. ( I ) described a technique in which the DNA is subjected alternately to two electrical fields at different angles for a time called the pulse time. The molecules
presumably change direction prior to net translational motion each time tho field is switched. Larger molecules take longer to change direction and have less time to move during each pulse, so they migrate more slowly than the smaller molecules. Use of pulsed fields has led to the separation of DNA ranging from 10 to 10,000 kh, a remarkable increase of resolving power over cunventional gel electrophoresis. T h e electrophoretic mobility of DNA is expressed as the distance moved per unit field in unit time. In pulsed-field electrophoresis, only a nominal mobility can he estimated hecause DNA follows a zigzag path. Therefore the measured distance is less than that actually migrated. In PFGE, the relative or normalized mobility is the relative distance traveled per unit time in the average field direction when
FOCUS
compared with a standard size DNA marker. With short pulse times, the molecules probably migrate directly along the diagonal in response to an average field effect, whereas at intermediate pulse times, no net translational motion may occur for a significant proportion of the pulse, and so the elapsed time is an overestimate of the migration time. The size of the largest resolved DNA species increases with increasing pulse time. A t a given pulse time, an increase in the electric field should decrease the reorientation time and lead to the resolution of larger fragments. This has indeed heen found to he true. The resolution of pulsed-field gel electrophoresis is dramatically affected by the number and configuration of the electrodes used, because they alter the shape of the applied electrical fields. The initial description of the technique, referred to as pulsed-field gel electrophoresis, involved any separation in which the electric fields were
ANALYTICAL CHEMISTRY, VOL. 61, NO. 8, APRIL 15, 1989
-
551 A
alternated. Subsequently, numerous names have been given to pulsed-field techniques (2.3). Because each of these is simply a variation of the original electrode geometry, it would, perhaps, he more useful to describe the reorientation angle that was used and whether the fields were homogeneous or inhomogeneous. These names include OFAGE (orthogonal field alternating gel electrophoresis), FIGE (field inversion gel electrophoresis), TAFE (transverse alternating field electrophoresis), CHEF (contour-clamped homogeneous electric field), PHOGE (pulsed homogeneous orthogonal field gel electrophoresis), RFGE (rotating field gel electrophoresis), and PACE (programmable, autonomously controlled electrodes).
Theoreticalmodels Although use of pulsed fields has led to substantive improvements in the separation of much larger DNA molecules, the underlying motion of the molecules in both constant and pulsed fields has remained somewhat of a mystery. Two theoretical models have been proposed to describe the motion in constant electric fields: a “biased reptation” model (4-6) and a direct simulation incorporating many degrees of freedom of the DNA (7). Biased reptation makes use of the concept of a segmented tube through which a chain paeaes. A very long DNA strand snakes its way through gel pores with one end leading and the rest of the molecule following the same path. The path chosen by the head is assumed to he a semirandom walk, biased by the electric field force. The idealized motion, referred to as reptation, has been biased by the electric field. The reptation model saw early success when applied to diffusion-controlled processes in the absence of any electric field. Several groups (4,8) extended this theory to the case of a constant external field. Although several assumptions were made and the extension was not straightforward, experimental data for the constant field case appeared to support the biased reptation-tube theory. The direct simulation of Deutsch (7) represents the DNA as a set of charged beads connected by freely hinged links and the gel as a two-dimensional lattice of obstacles. There is a short-range repulsive force between each obstacle and all the beads on the chain that completely prevents the chain from crosaing an obstacle. Deutsch’s results are shown for DNA in a fairly low constant field (Figure 1).A chain of DNA moves downward in the direction of the electric field. Panel A shows the chain 552,.
Fluure 1. The evolution of a DNA molecule in a gel. The parameters In this run were qEl/kBT = 0.604. where q is Ihe charge on Uw bead, €is !he electric fiem. I is me persistence I-. kBis the Bolumann’s constam. and T 16 the absolute temperature. The relativetimes tor each hsme shown are A, 260; B, 2 7 0 C. 280; D. 290; E. 330; F. 340,G 350; and H. 360. The electric Held is VBmcally downward. (Adapted wim permission horn Referenc8 7: copyright 1966. AAAS.)
has “bunched up” at the leading edge. As time progresses, a hairpin forms (panel B) that grows in size. The location of the top of the hairpin remains fixed in space, but the entire chain has slid down, so that the location of the hairpin relative to the rest of the chaii has moved up. As time progresses. the hairpin envelops the trailing edge of the chain (panel C). The effect of this is to drastically alter the path the upper half of the chain takes. It now slides downward along the path of the hairpin, causing the chain to enter an inverted “U” shape (panel D). Eventually one end exhibits a greater mobility (panels E & F), and the chain unwinds around the obstacles. It ends up almost fully extended (panel G). In panel H, the chain ends up looking as it did in panel A but is further downstream. The cycle is ready to be repeated over and over again. The lower the electric field, the looser the coiling and the more open the reorientation. This kind of motion differs from the biased reptation model in that the DNA does not appear confined to a tube and forms hairpins. The difference between these two models is largest for pulsed field electrophoresis, where hairpin formation is crucial in predicting the chain motion (9). Deuteh’s direct simulation is able to reproduce several features of pulse
ANALYTICAL CHEMISTRY, VOL. 61, NO. 8, APRIL 15, 1989
field experiments not explainable by the biased reptation model (7). The first direct experimental evidence concerning molecular orientation during both constant and pulsedfield gel electrophoresis was described earlier this year hy S. B. Smith, P. K. Aldridge, and J. B. Callis a t the University of Washington, Seattle (IO).Their approach was based on the work of M. Yanagida and co-workers ( I I ) , who were the first to observe individual DNA molecules that had been stained with a dye under a fluorescent microscope. Smith and eo-workers used ethidium bromide t o stain DNA. The stained X-phage DNA (48.5 kb) was premixed into molten agarose, and then thin layers (10-20 pm) were cast under a cover slip onto a microscope slide. The sample was excited in green light and viewed in red. An image intensifier and vidicon camera recorded the image with infrared light from the excitation source blocked. The recorded video images were processed and photographed from the monitor screen. Fine platinum wires placed adjacent to the cover slip formed the electrophoresis electrodes. In the absence of an electric field, the DNA molecules emhedded in the agarose were globular in shape and appeared to be quite motionless. However, when an electric field was applied (6 V/cm), the series of
I
’
-.
Fig during OFAGE.
Flgure 2. Time sequence of a molecule of Xphagc tin 1.5% agarose with a field of 8 Vlcm. me h a m s me Wen 0.4 s apart. The Wntour lenglh of UnsIained &phage DNA Is 15 pm, but the inter calating dye lengthens it. Scale bar is 10 p m . (Adapted wim permission ham Reference lo: CopyrigM 1989, AAAS.)
images showed the A-phage DNA in the agarose migrating in a slow constrained motion, winding through the invisible pores in the agarose, stretching in the direction of the electric field, and contracting as the tails caught up with the heads (see Figure 2). The molecules often encountered obstacles such that both ends moved down field simultaneously, forming a “U”shape. Increasing the field strength caused the DNA to become more extended and better aligned. “How remarkably similar these observations were to the simulation of Deutach,” comments Smith. According to Smith, experiments with very long DNA chains such as yeast chromosomal DNA (Saccharomyces cereuisiae, 20I%1500 kb) indicate that the longer chain DNA aligned more exactly in the electric field than did the A-phage DNA. T o avoid any shearing of these DNA molecules, solid agarose beads containing yeast cells treated to remove the cell walls and protein were mixed with the molten agarose containing ethidium bromide and cast into a thin gel. When the electric field was applied, one could visualize the intact chromosomal DNA molecules streaming into the bulk of the gel from the beads. When the electric field was removed, these molecules contracted and filled the pores in the agarose, making the gel voids and larger pores visible for the first time. During pulsed-field gel electropho-
D
’
.
e
a a’..
A wntourzlamped hcincgenems electrlc field sysmn wan cmsbucted on a micrwcqe slide. A WmpUter with a dlgltal-tpanalog convener and mun~plexercontrolled the potential on 16 elecbodes adjacent to me edges of the square cover slip. The direcIion and magnitude of the electric field were controlled with a joystick M cycled automatically. The elecbic field of 4 Vlcm had changed dlrectlon by 120’ 5 s betwe mis picture was reccided. Scale bar 18 10 rm. (Adapted wlm permission from Reference IO,Copyright 1989. AAAS.)
resis, size separation appears to be decheck on the accuracy of the model and pendent on the way that molecules reshould lead to efficient molecular dyalign when the direction of the electric namics simulation that will predict the field is changed. Smith and co-workers mobilities of DNA species in various studied the electrophoretic separation PFGE conditions. of stained yeast DNA in an electric The revolution in separations based field that was suddenly rotated 120°. on gel electrophoretic techniques is (A CHEF system that produced a hojust beginning! Sharon Boots mogeneous OFAGE field was used.) The molecules reversed their direcRefeIellCeS tions, according to Smith, and the tails (1) Schwartz,D. C.; Saffran, W.; Welsh, J.; became the heads. These observations Haas, R.; Goldenberg, M.; Cantor, C. R. (Figure 3) correspond quite well with Quant. Bid. 1983,47,189-95. the switchback model of Southern (E), (2) Lai, E.; Birren, B. W.; Clark, S. M.; Simon, M. I.; Hood, L. BioTechniques 1989, who predicted this sort of motion and 7,3439. showed how the continual tacking of (3) Ferris, S.; Freehy, S.; ZoUer, P.; Ragsthe field would lead to the resolution of dale, C.; Stevens, A. Amer. Biotech Lab. 1989, 7,3&42. molecules based on chain length. A t the (4) Slater, G. W.; Rousseau, J.;Noolandi, J. same time, microscopy showed the adBiopolymers 1987,26,863-12. ditional affect of alternating chain (5) Lumpkin, 0. J.; Zimm, B. H. Biopolyelongation and contraction was also mers 1982,21,2315. (6)Lumpkin, 0. J.; Dejarden, P.; Zimm, important in the separation process. B. H. Biopolymers 1985.24,1575. Several computer models have re(7) Deutsch, J. M. Scienee 1988.240, 922cently been constructed using the rep24. tation tube theory or a variation of i t (8) Viovy, J. L. Biopolymers 1987,26,1929dn (IO, 13, 14). Tube-type models lack (Sjkeutseh, J. M.Phys.Rev.Lett. 1987,59, some of the generality of Deutach‘s 1255-5% ~ ~ . . model. T h e motion is more con(10) Smith, S. B.; Aldridge, P. K.; Callis, strained, and i t is not possible for a loop J. B.Science 1989.243,203-6. (11) Yanagida, M.;Hi,raoka, Y.; Katsura. I. to expand out of the center of the moleQuant. Biol. 1982,47, 177. cule in a tube model. However, tube (12) Southern, E. M.; Analid, R.; Brown, models run faster on a computer, reW.R.A.; Fletcher, D. S. Nucleic Acids quiring fewer calculations. This is imRes. 1987,15,5925. (13) Lalande, M.; N wlandi,J.;Turmel, C.; portant for megabase-sized molecules. Rousseau, J.; Slater, G. W , Proe. Nnt. Comparison of computer model graphAcad. Sci. 1987.84,8011 ics with microscopic motion pictures of (14) Zimm, B. H. Phys. Re actual DNA molecules will give a quick 2965-68. ANALYTICAL CHEMISTRY, VOL. 61, NO. 8. APRIL 15, 1989 * 5 5 3 A