54
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A
[6]
for genetic diseases. The methods described here supplement existing techniques and provide alternative approaches to identifying other yet unknown genes. By obtaining a N ot I restriction map utilizing clones from linking and jumping libraries, it may be possible to isolate contiguous genomic DNA fragments by the "fishing" protocol. This would facilitate microanalysis of any macrochromosomal region of interest. Although the affinity capture method needs further refinement to realize its potential for isolating megabase size fragments, modifications of the basic technique suitable for such purposes are being explored. These methods provide yet another tool to investigate the molecular structure of mammalian chromosomes. Acknowledgments We wish to thank Mary Wallace for excellent technical preparation of the manuscript. This work was supported by an Outstanding Investigator Grant CA 42556 of the National Cancer Institute to Sherman M. Weissman, and a National Institutes of Health Grant GM 40633 to David C. Ward.
[6] T r a p p i n g E l e c t r o p h o r e s i s o f E n d - M o d i f i e d D N A in P o l y a c r y l a m i d e G e l s B y L E V Y ULANOVSKY
Introduction and Principle of Method Field inversion gel electrophoresis (FIGE) dramatically improves the separation of large double-stranded (ds) DNA fragments (e.g., in the size range of yeast chromosomes) on agarose gels, as compared to constant field electrophoresis. 1 However, relatively little improvement by FIGE has been found for short single-stranded (ss) DNA (hundreds of bases) on polyacrylamide gels. 2-4 This chapter discusses the mechanism of a manyfold improvement in short ssDNA separation achieved by attaching a small protein, streptavidin, to one end of the DNA. Why does FIGE work well for long DNA fragments, but not for short I G. F. Carle, M. Frank, and M. V. Olson, Science 232, 65 (1986). 2 E. Lai, N. Davi, and L. Hood, Electrophoresis 10, 65 (1989). 3 B. W. Birren, M. I. Simon, and E. Lai, Nucleic Acids Res. 18, 1481 (1990). 4 D. L. Daniels, L. Marr, R. L. Brumley, and F. R. Blattner, in "Structure and Methods" (R. H. Sarma and M. H. Sarma, eds.), Vol. 1, p. 29. Adenine Press, New York, 1990.
METHODS IN ENZYMOLOGY, VOL. 216
Copyright © 1992 by Academic Press, Inc. All rights of reproduction in any form reserved.
[6]
TRAPPING ELECTROPHORESIS OF END-MODIFIED D N A
55
ones? Although the nature of the FIGE phenomenon is still largely a mystery, there is little disagreement that the FIGE mechanism should require some kind of orientation of the DNA by the electric field. Optical techniques have made it possible to measure how much the electric field orients DNA in gels. 5 It was observed that the degree of the DNA orientation in constant-field gel electrophoresis decreases with the fragment size. The degree of orientation is a compromise between the influences of the electric field and of the thermal motion, kT. While the electric field tends to orient (to align) the DNA chain along the field, the thermal motion tends to disorient it into a random coil. The actual shape of the DNA chain is in a dynamic equilibrium controlled by these two factors. The transition from a completely oriented configuration to a completely random one requires that a certain "orientation energy" be overcome. The longer the DNA, the larger is the orientation energy. Therefore, for large molecules the equilibrium is shifted toward the elongated configuration, while for small ones the shape is closer to the random coil. The DNA orientation (elongation) during constant-field gel electrophoresis is probably what gives rise to the well-known phenomenon of "limiting mobility." The essence of this phenomenon is that above a certain size of DNA, the mobility of the fragment no longer depends on its length and the electrophoresis loses its resolution. These large fragments of different sizes comigrate and are usually seen on gels as one thick band above the bands of shorter fragments resolved below it (e.g., see Fig. 1, lane C). Existing theories of DNA electrophoresis 6,7 invoke a transition from "small" to "large" DNA to explain the limiting mobility phenomenon. For small DNA molecules the thermal motion energy, kT, is higher than the orientation energy and thus randomizes the configuration of the DNA. The electrostatic force component that pulls a DNA molecule through the pores of the gel is proportional to the end-to-end distance of the fragment projected onto the field direction. For randomly coiled small fragments, the end-to-end distance is proportional to the square root of the DNA length L. This predicts a 1/L DNA size dependence for the mobility of small molecules, as described by the "biased reptation" theory (see Refs. 8 and 9; for a simple explanation see the appendix at the end of this article). At a large enough size, however, naked DNA moves through the gel in an elongated shape, kT being too low to disrupt the configuration. For these 5 G. Holtzwarth, C. B. McKlee, S. Steiger, and G. Crater, Nucleic Acids Res. 15, 10031 (1987). 6 0 . J. Lumpkin, P. Dejardin, and B. H. Zimm, Biopolymers 24, 1573 (1985). 7 G. Slater and J. Noolandi, Biopolymers 25, 431 (1986). 8 L. Lerman and H. Frisch, Biopolymers 21, 995 (1982). 9 O. J. Lumpkin and B. H. Zimm, Biopolymers 21, 2315 (1982).
56
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A
[6]
large fragments, extended along the field, the end-to-end distance is proportional to their contour length. Therefore, both the electrostatic force and the opposing friction force applied to the large fragments are proportional to the fragment length, and thus the (limiting) mobility is size independent but voltage dependent. An interesting and less understood fact regarding the DNA orientation gave rise to the idea behind the present work. It is that, even within the size range of the limiting mobility (where the fragments are all presumably oriented along the electric field to some extent) the degree of orientation falls rapidly with decreasing DNA size.5 It indicates that while the phenomenon of the limiting mobility may occur regardless of the degree of orientation of DNA (as long as it is oriented at all, rather than randomized), the FIGE effect is probably dependent on this degree. If the DNA orientation is what facilitates the separation by FIGE, we may wish to consider why the electric field has such a hard time orienting small DNA chains. One of the physical reasons is that along its length DNA has a rather uniform distribution of charge, mass, and other physical characteristics determining the electrophoretic behavior. Then a question arises: can one make FIGE work by introducing an artificial nonuniformity, an asymmetry into the molecule? The asymmetry can be achieved by making one end substantially different from the other, for example by binding a protein or another entity to one end of the DNA, but not to the other. The difference between the two ends would determine which end leads the way, and which end trails behind. Then the orientation is no longer a matter of chance. Indeed, it has been found that such an asymmetric modification, achieved by binding streptavidin to one end, results in a dramatic change in the pattern of DNA mobility. ~0 Surprisingly, the change was observed not only in FIGE, but even in constant field. This phenomenon is discussed below and interpreted as DNA trapping in the gel (see Results and Discussion, below). It should be noted that the whole phenomenon of the electrophoretic trapping of end-modified ssDNA has not yet been investigated thoroughly enough for practical purposes, in particular for DNA sequencing. Therefore, the present chapter should perhaps be regarded as a guide for further experiments, rather than as an established standard for practical applications. Technical Notes, Materials, and Methods Streptavidin can be attached to one end of ssDNA by means of a noncovalent but very strong bond between streptavidin and biotin. Biotin l0 L. Uianovsky, G. Drouin, and W. Gilbert, Nature (London) 3437 190 (1990).
[6]
TRAPPING
ELECTROPHORESIS
OF END-MODIFIED
DNA
57
is a small molecule (Mr 244) that can be covalently linked to a nucleotide in a DNA fragment. There are at least three ways to make end-biotinylated ssDNA molecules: (I) filling in the recessed 3' end with the incorporation of a biotinylated nucleotide in one position and then denaturing; (2) enzymatic extension of 5' end-biotinylated primer and then denaturing; and (3) 3' end extension by terminal transferase. We will discuss the first reaction in detail and then the other two briefly. Filling in Recessed 3' End. An example of such a reaction would be filling in by Sequenase (USB, Cle~,eland, OH) of the recessed 3' end left by HindlII restriction cleavage: 5'--
. . . . .
NNN--3'
IIIIIIII 3'--
. . . . .
N NNCTAG--5'
For a final reaction volume of 24/~1, take 10/~1 of 1.0/zM DNA ends in TE buffer. Add 3/zl 5 x reaction buffer for Sequenase (200 mM Tris-HCl, pH 7.5, 100 mM MgCI2, 250 mM NaCI) and 1.5/zl 0.I M dithiothreitol (DTT). The first base to be filled in, guanine, is currently not available as a biotinylated triphosphate (biotinylated dGTP), but the other three are. So, unmodified dGTP (0.5 p.1, 1.0 mM) is added for filling in this position. Next comes dATP, which in contrast to dGTP is available in a biotinylated form; so we add 1.0/xl of 0.4 mM biotinylated dATP. Radioactive dTTP (6.0/~1, 3.3/zM being the sum of both hot and cold dTTP) is incorporated next. Its concentration is that of the commercially available stock. If no radioactive labeling is required, cold dTTP (0.5/zl, 1.0 mM) is used instead. The last nucleotide should be an unmodified dNTP, in this case dCTP (0.5/zl, 1.0 mM). This is recommended in view of the exonuclease activity of potymerases (even a low level), because unmodified dNTPs are available in a high concentration. The radioactive nucleotide must follow the biotinylated one and not vice versa. Otherwise, there is a chance that the radioactive label may be incorporated without biotinylation of the DNA. This could happen due to a relatively low incorporation rate of biotinylated dNTP by polymerases, as compared to the unmodified dNTP. If [32p]dNTP is used as a label and incorporated in the 3' direction of the biotinylated nucleotide, as above, the incorporation of the biotinylated dNTP can be determined by the following test. First, as a control for the biotinylation reaction, it is advisable to carry out a reaction, parallel to the above, except that the biotinylated dNTP is replaced by the unmodified dNTP. Then a drop of 1 to 3/zl of each of the two reaction products is put on a small Whatman (Clifton, N J) DE-81 paper circle. The radioactivity can be measured with any Geiger counter sensitive to [32p]. If the counter goes off scale, the measurement is performed with the DE-81 circle kept at a fixed distance from the detector tube, e.g., 5 cm. The DNA immedi-
58
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A
[6]
ately binds to the paper. After washing three or four times with 0.5 M Naphosphate buffer, pH 7.0 (3 min for each wash), the unincorporated dNTPs are removed while the DNA remains bound to the paper. Then, after these washes are completed, another reading is taken with the same Geiger counter at the same distance as before. The reading of the biotinylated DNA should be compared to that in the control experiment, in which the unmodified dNTP is used instead of the biotinylated one. A comparison of the two readings shows the incorporation efficiency of the biotinylated dNTP. Another control is also recommended, in which no polymerase is added and thus no incorporation occurs. This control shows the efficiency of washing out the unincorporated dNTPs. The unwashed dNTP should give less than 1% of the counts before washing. The acceptance of a biotinylated dNTP by DNA polymerases is much lower than that of an unmodified dNTP. Even the most efficient polymerases, like Sequenase, require much higher concentrations of biotinylated dNTP than that of unmodified dNTP to match their incorporation efficiencies. Therefore, one should keep the biotinylated dNTP concentration as high as possible. In practice, commercially available biotinylated dNTPs are supplied in 0.3 to 0.4 mM concentration, the final concentration in the reaction being even lower. In contrast to the biotinylated dNTPs (in the above example biotinylated dATP), unmodified dNTPs (in the above example dGTP and dCTP) are available in concentrations that are higher by orders of magnitude. However, the concentration of the unmodified dNTP (dGTP and dCTP above) in the reaction should not exceed that of biotinylated dNTP (dATP above) by more than two orders of magnitude. Otherwise, the incorrect incorporation of unbiotinylated dGTP or dCTP in place of biotinylated dATP by the DNA polymerase may be too frequent. If one follows these rules, the biotinylation should be on the order of 100%. Unbiotinylated DNA molecules should not be detected on the autoradiogram in any event (if the base order is kept as recommended above), because they lack the radioactive label. Enzymatic Extension of Biotinylated Primer. Most techniques for biotinylated primer synthesis produce a mixture ofbiotinylated and nonbiotinylated primers of the same sequence. These can be separated on polyacrylamide gel electrophoresis (PAGE), the mobility difference being equivalent to about one nucleotide difference in length (if only one nucleotide per primer is biotinylated). The biotinylated primer is then purified and used for a regular primer extension reaction and, in particular, an enzymatic sequencing reaction with the dideoxy terminators. Apart from a dideoxy sequencing ladder, one can produce a size marker with one biotinylated end by enzymatically extending the same biotinylated primer with no dideoxy terminators, and then cutting the DNA with a restriction
[6]
TRAPPING ELECTROPHORESIS OF END-MODIFIED D N A
59
enzyme of a given fragment size pattern. This technique is an alternative to the method of filling in the sticky ends described above. 3' End Extension by Terminal Transferase. This reaction can result in an incorporation of more than one biotin molecule into a single end. To avoid this "overbiotinylation" one can use a very low ratio of biotinylated to nonbiotinylated dNTP. However, in this way a relatively low percentage of the DNA molecules may become biotinylated. To overcome this problem, one possibility would be to use radioactive streptavidin, thus leaving the unbiotinylated molecules undetectable. Storage and Streptavidin Binding. After the DNA is end biotinylated by one of these three methods, or in some other way, it can be stored at 20 °. After thawing, the binding of streptavidin is performed by incubation of the biotinylated DNA with streptavidin at room temperature for 20 min. Excess of Streptavidin. How can the binding of more than one DNA fragment to a streptavidin molecule be avoided? Streptavidin consists of four identical subunits. It has a tetrahedral symmetry, and thus four equivalent binding sites for biotin. To avoid multiple binding of biotinylated DNA fragments to streptavidin one can use a large molar excess (30-100 times) of streptavidin over biotin, both incorporated and unincorporated. This leaves the three extra binding sites unoccupied in each streptavidin molecule. These sites can probably be used for binding various biotinylated groups and biotinylated small molecules to streptavidin. It may be done to vary the parameters of the trapping process and thus change the electrophoretic behavior of the complex, as well as to attach fluorescent, radioactive, or similar label. Denaturing ofDNA. How can the DNA be denatured without disrupting the biotin-streptavidin bond? In the presence of urea or formamide, boiling for only a few minutes may disrupt the biotin-streptavidin bond. Therefore, to denature the DNA in 70-90% formamide, the samples should be heated at 65° for about 5 min. This way the DNA is well denatured and the biotin-streptavidin link is preserved. However, with no denaturing agent, such as urea, formamide, or sodium dodecyl sulfate (SDS), boiling for a few minutes does not damage the biotin-streptavidin link. Diffusion of Streptavidin in Gel. In the polyacrylamide gel, streptavidin can migrate across the lanes and bind to biotinylated DNA in adjacent lanes. In view of this phenomenon, the problem arises of how to run control lanes that are supposedly carrying nonstreptavidinated DNA. If this control DNA carries biotin, then one way to circumvent the problem is to keep the control lane at a distance from the lanes carrying free streptavidin. If, on the other hand, the control lane should be next to a streptavidinated one, the control DNA should carry no biotin. The latter -
60
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A
[6]
A B C D
861
~
738
o
615 526 O 492
....
447 6 369
246 .....
861 738 615 -~An,, 526 = ~ - ~ - - ~ 447
FIG. 1. Autoradiogram of constant-field electrophoresis of end-streptavidinated and unmodified ssDNA on 6% (a) and 8% (b) polyacrylamide gels. l° The electric field strengths are 60 and 100 V/cm, respectively. In (a) and (b) the samples in the lanes are as follows. Lane A, end-streptavidinated SV40/HindlII DNA digest; lane B, end-streptavidinated 123-base ladder; lane C, unmodified 123-base ladder; lane D, unmodified SV40/HindlII DNA digest. The numbers on the left- and right-hand sides show the lengths (in bases) of the streptavidinated and unmodified DNA fragments, respectively. The arrows indicate the size markers in the two outer lanes. The gels were 13.5 cm long (slot to bottom) and 0.5 mm thick. Cooling by circulating water resulted in a temperature of 15° during the 8-hr run (a), and in a rise from 14 to 20° during the 60-min run (b).
[6]
TRAPPING ELECTROPHORESIS OF END-MODIFIED D N A
ABCD
61
b
738 615
~i
~.492 526 447 369
246 --
215
FIG. 1. (continued) is preferable and implies that an unbiotinylated sample should be prepared in parallel to the biotinylated one. Results and Discussion
Constant-Field Electrophoresis of End-Streptauidinated ssDNA. The autoradiograms in Fig. 1 show how streptavidin attached to one end of
62
ISOLATION,
SYNTHESIS, I
I
DETECTION I
[
OF
DNA
I i i ii
AND
[6]
RNA
I
t000
500
O A ~
200
...1 0
~
50 ~NAKED V
SA+ CH DNA 30 - - * -
DNA --A--
59 100
•- o -
t0
1
02
I
I
I
0,5
I
II
t,0
2,0
10
D N A size (kb)
FIG. 2. Plots of electrophoretic velocity (cm/hr) per unit electric field (V/cm) in constantfieldelectrophoresis, versus ssDNA fragment length for end-streptavidinated and unmodified ssDNA. I° The polyacrylamide gel concentration is 6%. The three types of curves relate to three different electric field strengths (see inset). SA, streptavidin. ssDNA retards this complex in constant electric field P A G E in comparison with the same ssDNA lacking streptavidin. Figure 2 shows plots of D N A size d e p e n d e n c e of the mobility of both streptavidinated and unmodified DNA. At small D N A sizes, the s t r e p t a v i d i n - D N A complex is only moderately retarded as c o m p a r e d to the unmodified D N A of the same size. The mobility curves in the size range between 120 and 500 bases look like almost parallel straight lines. This parallel appearance, however, arises from the log-log nature of the plot. But in fact, in this size range, the absolute mobility gap narrows with increasing size, as it should, because the relative contribution of streptavidin to the total friction of the streptavid i n - D N A complex decreases with D N A size. One would expect this mobility gap between the streptavidinated and unmodified D N A to keep narrowing. This is what, indeed, happens to end-streptavidinated d s D N A in agarose gels, where the gap eventually vanishes (for large DNA) at the sizes approaching the limiting mobility plateau (L. Ulanovsky, 1989, unpublished data).
[6]
TRAPPING ELECTROPHORESIS OF END-MODIFIED D N A
63
~S~eptavidin GelFiber
~=\~'X 4IElectri~c ~ ~ \\ \~
ssDNA field
FIG. 3. One possible type of DNA trapping model (schematic, for illustration only). Because the DNA charge is negative, the electric force direction shown here is opposite to the actual electric field direction.
Contrary to these expectations, however, no merging of the two curves occurs for ssDNA in PAGE (Fig. 2). Instead, the streptavidinated DNA mobility curve drops down steeply at about the size where the unmodified DNA approaches the limiting mobility. Above this size, the separation between bands of streptavidinated DNA is several times wider than between bands of unmodified DNA (compare lanes B and C in Fig. la or b). Moreover, above a certain DNA size the streptavidin-DNA complex does not seem to migrate in the gel at all (e.g., above roughly 1000 bases in Fig. la). An interesting aspect of this effect is related to the voltage dependence of the mobility pattern. The mobility curves for three different voltage gradients [30, 59, and 100 V/cm (constant in time)] are shown in Fig. 2. As the voltage grows, the size of the DNA fragments at which the steep drop of the mobility curve occurs decreases. Trapping Model. A physical model that seems to agree with all these data, and which also finds support in FIGE experiments (see below), is the DNA trapping model. In this model, the electric field pulls the endstreptavidinated DNA into a trap formed by the fibers of the polyacrylamide gel. Figure 3 shows schematically one possible type of trap. The total charge of the hanging DNA chain and thus the energy of trapping depend on the DNA length, each nucleotide carrying one electron charge. In other words, the longer the DNA (and/or the higher the electric field strength), the larger is the energy required to lift the streptavidin-DNA
64
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A
EABCD
[6]
a
ii,i~
S243
984 861 738 615 526
=
447
= 369
246
738 615
492 526 = 447 369
215 FIG. 4. Autoradiogram of FlOE of end-streptavidinated and unmodified ssDNA, l° The electric field strength is 60 V/cm. The forward/reverse pulse durations are 800/200 msec in (a), and 800/20 msec in (b). The samples in lanes A - D are as in Fig. 1. The samples in lanes E and F are SV40/BamHI (5243 bases) and pUC/HindIII (2686 bases), respectively. The polyacrylamide gel concentration is 6%.
[6]
TRAPPING ELECTROPHORESIS OF END-MODIFIED D N A
FABCD
65
b
26861
1107
984 861 738
615 861
526 492 447
738 615
369
492 246
526 447
369 FiG. 4. (continued
complex and release it from the trap. If the thermal energy can do that to a smaller streptavidinated DNA, it may have a harder time with a longer one. Therefore, the probability of the release of the complex in a unit of time depends on the DNA size. A long streptavidinated DNA may spend most of its time in the traps. Depending on the DNA size, the complex may or may not be able to jump from one trap to another. A small DNA would jump much more often, and thus spend more time moving down
66
[6]
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A 1000
I
I
~
~
L
I
~
I iiI
I
I
500
% --
200
x
° > >.N 0
t00
5O
20
iO O,i
FORgARD/REVERSEI (.SEO I 800/200 j --o-- 800/20 I 80/20 I --v-8/2 I 0,2
[
t 0,5
; t,O
DNA
~ , ', I
\ / \1 /I /I
'
Y
L
I 2,0
I
I il
iO
size (kb)
FIG. 5. Plots of electrophoretic velocity (cm/hr) per unit electric field (V/cm) versus ssDNA fragment length in different field inversion regimes (see inset) for end-streptavidinated and unmodified ssDNA. ]° The 6% (w/v) polyacrylamide gels were run at 60 V / c m at a constant temperature of 15°. The four types of curves relate to different pulse durations (see inset). For the purpose of velocity calculations, the net running time was the sum of the duration of all forward pulses minus that of all reverse pulses in the run.
the field from trap to trap. This mechanism can explain the sharp DNA size dependence of protein-DNA complex mobility on the gel plotted in Fig. 2. The effect of voltage on the DNA size dependence of the mobility (Fig. 2) also fits this trapping model. The higher the voltage, the smaller the DNA size above which the trapping of the streptavidin-DNA complex begins. This effect is somewhat similar to the trapping of open circular dsDNA in agarose gels. 1~ In a more schematic manner one can imagine a coat hanging on a hook. The force of gravity would represent the electric field. A gusty wind would represent the thermal motion. If the coat is very light, a gust may have enough energy to blow the coat off the hook. If, however, the coat is heavy, the probability of its release by the wind is much lower. This setup can be viewed as a method of "coat separation by weight." II S, D. Levene and B. H. Zimm, Proc. Natl. Acad. Sci. U.S.A. 84, 4054 (1987).
[6]
TRAPPING ELECTROPHORESIS OF END-MODIFIED D N A •" . • "'".
67
"Tube" _DNA FElectri c
• " "" • •
I eld VF o r c e
FIG. A.1. Reptation model of DNA migration in gel electrophoresis. According to the model, the DNA is moving through the gel as a snake through grass, in the sense that the obstacles (gel fibers) confine the motion of the molecule to its own path, and the sideways motion is not allowed by the obstacles (gel fibers). The obstacles, therefore, form an imaginary " t u b e " along the DNA path. Regarding the field direction, see Fig. 3 caption.
Field Inversion Gel Electrophoresis of End-Streptavidinated DNA. The trapping model is supported by streptavidinated DNA behavior in FIGE as well. Indeed, three different observations indicate this. First, the reverse pulse enhances the mobility of the streptavidinated DNA as compared to the constant field regime at the size range of the trapping (where the steep drop of the mobility curve occurs in the constant field; compare Fig. 2 with Fig. 5). The interpretation is that the reverse pulse helps to release the streptavidin-DNA complex from the trap by lifting the complex for a while. Second, the longer the reverse pulse, the larger the size of the streptavidinated DNA that is allowed to migrate in the gel (Figs. 4 and 5). This behavior can be expected of an elastic chain hanging on a hook in terms of the probability of release. Third, a comparison of the 800/20- and 80/20-msec curves on Fig. 5 shows that streptavidinated DNA of the same size migrates faster when the 20-msec reverse pulse is applied every 0.01 sec (the release occurs more often), than when it is every 0.1 sec. Appendix: Biased Reptation Theory
Simplified Derivation of l/L Dependence of DNA Mobility on Fragment Size Until 1982 it was a mystery why DNA was separated by size in gel electrophoresis. Simple logic suggested that the DNA migration velocity should depend on the ratio of the charge to the friction coefficient of the fragment. Because both are proportional to the DNA length, there seemed to be no reason why the velocity should be size dependent. Indeed, DNA
68
ISOLATION, SYNTHESIS, DETECTION OF D N A AND R N A
[6]
electrophoresis in solution, in contrast to that in gel, exhibits a sizeindependent velocity. The breakthrough in our understanding came in 1982 with the emergence of the biased reptation theory. 8'9 The basic assumption of the biased reptation theory is that the DNA molecule in a gel "reptates" through the pores, i.e., moves in a snakelike fashion. The DNA chain moves along its curving path, its " t u b e , " but not sideways (Fig. A. 1). For short enough DNA fragments, both their shape in the gel and their movement are dominated by the thermal motion, to which the electric field provides only a small bias, Therefore, the shape of such a molecule is essentially a random coil. The molecule is pushed back and forth along its random path in a Brownian manner with a small net velocity in the direction of the electric field force. This velocity will be calculated here. The total electric force acting on the DNA chain is QE, where Q is the charge of the DNA fragment, and E is the electric field strength. The longitudinal component of this force that pulls the fragment along its reptation path is Q E H / L , where H is the projection of the end-to-end distance on the field direction, and L is the total length of the fragment. The corresponding longitudinal velocity u (along the tube axis) is u = QEH/Lx
where x is the friction coefficient of the fragment. When H = 0, there is no movement, except Brownian. When H is not zero, then the net movement due to the field is toward the end that is downstream relative to the field force. This process may be schematically visualized as a flow of liquid in a capillary in Fig. A. 1. As the DNA chain moves ahead by one nucleotide, it is equivalent to a transfer of one nucleotide from the tail to the head by length L along the chain, or by distance H in projection on the field direction. Therefore, to convert the longitudinal velocity u into the vertical velocity V, we should add another H / L factor: V = u H / L = QEHE/L2x
While Q E / x does not depend on the chain length L, because both Q and x are proportional to L, the other factor, H 2 / L 2, does. Averaged over time, H 2 is proportional to L, due to the random walk properties of the DNA coil. Therefore, the averaged migration velocity V is proportional to 1/L, as it is indeed observed. Acknowledgments I thank Dr. WalterGilbert for support, guidance, and stimulatingdiscussions.