Real-time imaging of the reorientation mechanisms of YOYO-labelled DNA molecules during 90
o
and 120
o pulsed field gel electrophoresis
Real-time imaging of the reorientation mechanisms of YOYO-labelled DNA molecules during 90 o and 120 o pulsed field gel electrophoresis
Sergio
Gurrieri
1,2,3,
*
,
Steven B.
Smith
3
,
K. Sam
Wells
4
,
Iain D.
Johnson
4
and
Carlos
Bustamante
3,5
1
Dipartimento di Scienze Chimiche, Universita' di Cantania, Viale A. Doria 6, 95125
Catania
,
Italy
,
2
Instituto per lo Sudio delle Sostanze Naturali di Interesse Alimentare e Chimico-Farmaceutico, CNR, Via del Santuario 110, 95028
Valverde
(CT),
Italy
,
3
Department of Chemistry and Institute of Molecular Biology, University of
Oregon,
Eugene
, OR 97403,
USA
,
4
Molecular Probes, Inc., 4849 Pitchford Avenue, Eugene, OR 97402,
USA
and
5
Howard Hughes Medical Institute, University of Oregon,
Eugene
, OR 97403,
USA
Received August 5, 1996;
Revised and Accepted October 10, 1996
ABSTRACT
Pulsed field gel electrophoresis (PFGE) techniques have been developed to
overcome the limitations of conventional electrophoresis and to increase the
separation to DNA chromosomes of few megabase pairs in size. Despite of the
large success of these techniques, the various separation protocols employed
for PFGE experiments have been determined empirically. However, a deep
understanding of the molecular mechanisms of motion responsible for DNA separation becomes necessary for the rational optimization of these techniques. This paper shows the first clear observations of individual
molecules of DNA during the reorientation process in 90
o
PFGE and 120
o
PFGE. Real-time visualization of the DNA dynamics during PFGE was possible with the
use of an epi-illumination fluorescence microscope specifically equipped to run these experiments and by staining the DNA with YOYO-1 (1,1
'-(4,4,7,7-tetramethyl-4,7-diazaundecamethylene)- bis-4-[3-methyl-2,3-dihydro-(benzo-1,3-oxazole)-2- methyl-idene]-quinolinium tetraiodide). This dye forms a very stable, highly fluorescent complex with double- stranded DNA and dramatically improves the quality of the DNA images. The results of computer simulations used to reproduce the molecular mechanisms of motion as well as the DNA separation features are also discussed.
INTRODUCTION
Conventional (steady-field) agarose gel electrophoresis is the most widely used method for
separating small DNA molecules by size. However, for DNA chains approaching ~10-20 kb the electrophoretic mobility [mu] becomes increasingly independent of molecular size (
1
-
2
). The DNA size at which [mu] becomes independent of the length of the nucleic acid depends strongly on
the electrophoretic conditions employed and, in particular, on the electric
field strength and the agarose concentration (
3
-
4
).
To overcome this limitation, a whole series of pulsed field gel electrophoresis
(PFGE) techniques have been developed (
5
-
17
). In typical PFGE experiments, DNA molecules are present in an ordinary agarose
gel matrix in conventional low-ionic strength buffers. An electric field of ~10 V/cm, but sometimes considerably lower, is applied alternately in
two directions. Pulse times can range from less than a second to several
minutes or more, depending on molecular size. The various techniques differ
mainly in the configuration of the electrodes around the gel to create the two
alternating electric fields and in the angle between these fields. Since most
of these techniques are characterized by similar separation features, in this
paper we will refer to as 90o and 120o PFGE, simply indicating the angle between the electric fields,
independently on the way these fields are generated. Using PFGE, separation of DNA molecules as long as ~5 Mb pairs can be achieved (
18
).
However, even if these techniques turned out to be so successful, the various
separation protocols employed for PFGE experiments have been determined empirically, while descriptions of the molecular mechanisms of motion responsible for DNA separation have
lagged behind (
19
-
26
). Since the discovery of the original PFGE methods, several mechanisms of DNA reorientation have been proposed in order to provide a physical explanation of the success of
these techniques (
27
-
29
). In fact, in order to migrate through the gel matrix, the DNA molecules have
to change their direction of motion in response to the changing electric field,
and size separation must depend on the way the molecules reorient when the
electric field changes direction. Most of these reorientation mechanisms
predict a linear dependence of reorientation time on DNA size. As long as the
reorientation time is shorter than the pulse time, the mobility becomes
linearly dependent on size. In fact, according to these ideas (
29
), the DNA mobility should depend only on the fraction of each pulse remaining
for net migration. Short molecules will reorient very rapidly in the direction
of the new field and will spend a good fraction of the pulse time migrating
through the gel before the electric field is switched again (Fig.
1
a). On the contrary, long DNA molecules will spend most of the pulse time for
the reorientation process, thus having only a short time available for actual
migration (Fig.
1
b). As will be shown later, this idea is much too simple and only partially
correct because other features of the reorientation mechanisms become more
important in determining size separation.
MATERIALS AND METHODS
Spectroscopic and binding affinity measurements
Measurements were performed using an IBM instruments 9420 spectrophotometer.
Polystyrene cuvettes were employed to minimize surface adsorption of the
polycationic YOYO-1. To determine the molar extinction coefficients ([epsilon]
max
) for the DNA-bound dyes, 100 [mu]l of a concentrated Calf thymus DNA (Sigma) stock solution (5 * 10
-3
moles of bp/l) were added to 2.5 ml of a 4 * 10
-6
M dye solution in 10 mM Tris-HCl, pH 7.4 to yield a molar ratio of ~50 DNA bp per dye molecule. The DNA-dye complex was allowed to equilibrate for at least 90 min.
The change in the maximum absorbance relative to the aqueous free dye, after
correction for the dilution due to DNA addition was converted to give [epsilon]
max
for the complex.
Either a PTI Alphascan or an SLM-500C spectrofluorometer were employed to record fluorescence measurements.
Solutions typically contained 4 * 10
-7
M dye in 10 mM Tris-HCl, 50 mM NaCl, 1 mM EDTA, pH 7.4 and 2 * 10
-5
moles of bp/l Calf thymus DNA, equilibrated as above. Fluorescence quantum
yields ([Phi]
F
) for these samples were determined relative to fluorescein in 0.1 M NaOH (
38
). Appropriately oriented polarizers were used to eliminate the possible effects of non-isotropic fluorescence emission from the DNA-dye complexes. The fluorescence enhancement on DNA binding, [Psi], (the ratio of the fluorescence quantum yield for a dye
molecule bound to DNA, [Phi]
bound
, over the fluorescence quantum yield of a dye molecule free in solution, [Phi]
free
) was determined by measuring and integrating the respective fluorescence
emission spectra with the excitation at the isosbestic point between the
corresponding absorption spectra. Careful subtraction of background signals
measured using a dye-free blank was necessary for accurate quantitation of the very weak
emission of YOYO-1 free in solution.
The binding affinities of ethidium bromide and acridine orange were determined
by measuring the fluorescence enhancement ([Delta]F), at a fixed low concentration of dye, as function of DNA concentration [DNA] (expressed in moles of bp/l). The x-intercept of a linear plot of 1/[Delta]F versus 1/[DNA] is equal to the binding affinity,
K
a
. Double-stranded Calf thymus DNA was titrated against a fixed concentration of fluorescent dye ([EB] = [AO] = 10
-7
M) in phosphate buffer (10 mM Na
2
PO
4
, pH 7.0, containing 1 mM EDTA and 100 mM NaCl). The binding affinity for YOYO-1 was obtained from reference (
39
).
Preparation of samples for imaging experiments
All the samples used for imaging were prepared using T2 phage genomic DNA (164
kb) purchased from Sigma Chemical Co. A stock solution of T2 phage DNA (1.95 * 10
-4
moles of bp/l = 125 [mu]g/ml) was prepared by dissolving the lyophilized DNA in 2 ml of distilled water
and was stored refrigerated at 5oC. In the preparation of the solutions for our experiments, micro-pipettes with cut tips were used to minimize DNA breakage. All dyes-YOYO-1, ethidium bromide and acridine orange-were provided by Molecular Probes, Inc. Their
stock and working solutions and also the DNA-dye solutions were stored in plastic vials at 5oC in the dark. It was found that YOYO-1 partitions from aqueous solution, even when complexed with
DNA, onto glass surfaces. Aqueous solutions of YOYO-1 appear to be indefinitely stable when kept in plastic containers.
Storage in glass is practical for DMSO solutions.
Stock solutions of YOYO-1 (1.0 * 10
-3
M) were prepared by dissolving solid YOYO-1 in dimethyl-sulphoxide (DMSO). The concentration was obtained by measuring the
absorbance in methanol using [epsilon]
481
= 125 800. Working solutions of YOYO-1 (1 * 10
-5
M) were prepared by dilution in 0.5* Tris-borate-EDTA (TBE) buffer (45 mM Tris, 45 mM Borate, 1 mM EDTA, pH 8.2). Staining of T2 DNA was accomplished by incubating overnight T2 DNA
(5 * 10
-6
moles of bp/l) and YOYO-1 (1 * 10
-6
M) at a DNA bp/YOYO-1 ratio of 5:1. This solution was then diluted 1000-fold in 0.5* TBE containing 2% (v/v) [beta]-ME ([beta]-mercaptoethanol). The final concentrations were: 5 * 10
-9
moles of bp/l T2 DNA, 1 * 10
-9
M YOYO-1, 2% (v/v) [beta]-ME. Eight [mu]l of this final solution were put on a microscope glass slide and a
22 * 22 mm coverslip was placed on top and sealed with fingernail polish or molten candle wax. For the electrophoresis experiments a suspension of 1% (w/v) ME SeaKem agarose gel (FMC Bioproducts) in 0.5* TBE buffer (45 mM Tris, 45 mM borate, 1 mM EDTA) was prepared.
After 5 min this suspension was gently boiled for 5 min and then cooled down to
50oC. Distilled water was used to restore the water lost by evaporation. [beta]-ME was added to a concentration of 2%. The 5:1 DNA/YOYO-1 concentrated solution (5 * 10
-6
moles of bp/l T2 DNA, 1 * 10
-6
M YOYO-1) was diluted 1000-fold in the molten gel. The final concentrations were: 1% agarose, 5
* 10
-9
moles of bp/l T2 DNA, 1 * 10
-9
M YOYO-1, 2% [beta]-ME (v/v). Ten [mu]l of this mixture were put on top of a microscope slide
already equipped with electrodes, and pre-heated at 50oC. A 22 * 22 mm coverslip was placed on top and sealed at the four
corners with fingernail polish. The gel was allowed to solidify for 10 min at
room temperature. The electrical connection was completed by dropping molten
gel over the electrodes and 0.5* TBE buffer was only occasionally added to keep it wet.
Electrophoresis instrumentation
For 90o PFGE experiments a custom made microscope slide was equipped with 20 point
electrodes arranged around a square contour, as shown in Figure
2
a in a schematic diagram. This electrode geometry produces two uniform and homogeneous electric fields oriented at 90o with respect to one another. A switch-box connected to the 20 electrodes and to two different power
supplies was used to switch the field between two orthogonal directions. A
similar electrode configuration was used for 120o PFGE experiments where 12 point electrodes were employed, arranged as
shown in Figure
2
b. This electrode geometry produces two electric fields oriented at an angle of
90o at the top of the slide and ~150o near the bottom. This way, the reorientation behaviour of the
DNA molecules could be studied at any arbitrary angle between 90o and 150o by simply observing DNA molecules in different positions of the
slide. Also angles <90o could be used by simply inverting the polarity of one of the electrode
pairs.
.
Properties of fluorescent dyes upon binding to dsDNA
Ethidium bromide
Acridine Orange
YOYO-1
[lambda]
ex
max
(nm)
526
502
491
[epsilon]
max
(M
-1
cm
-1
)
3200
53000
99000
[lambda]
em
max
(nm)
604
538
509
Quantum yield ([Phi])
0.15
0.43
0.52
Fluorescence enhancement ([psi])
21
1.5
460
Binding affinity (M
-1
)
1.5 * 10
5
3.1 * 10
4
6 * 10
8
Signal/background
1.3
1.2
3.9
Imaging instrumentation
An epi-illumination fluorescence microscope (Axiovert 35M, Zeiss) equipped with a 100* / 1.3 N.A. oil immersion Plan-Neofluar objective, and a high pressure Mercury lamp was used
for all electrophoresis experiments. The microscope was coupled, through a high gain image intensifier (KS-1380 Videoscope International, LTD), to a high sensitivity video camera
(VS2000N Videoscope International, LTD) designed specifically for critical low-light imaging. The DNA images presented here were recorded in real time using an 8 mm
video camera recorder (CCD-V701, Sony Corporation). When acridine orange or YOYO-1 were used to fluorescently label the DNA a blue incident light
filter set was employed (excitation filter BP 450-490, dichroic mirror FT 510, emission filter LP 520). In the case of
ethidium bromide a green incident light filter set was used (excitation filter
BP 510-560, dichroic mirror FT 580, emission filter LP 590). All observations
were made immediately after preparing the glass slides and for no longer than 1
h. The molecules selected for study were all at least 5 microns distant from
the glass/gel interface. The pictures were taken directly from a video screen
by pausing the video-tape on a single frame. No image processing system was employed. Length
measurements were performed directly from the screen, previously calibrated using a fluorescent grid of known dimensions. The photographs presented here are illustrations of the typical behavior
encountered in a large number of observations.
In order to quantify signals from the DNA-dye complexes and the background, a charge-coupled device (CCD) camera (Star-1, Photometrics, Ltd) was used. This device made it possible
to obtain linear signals over a dynamic range of ~4000. All instrumental parameters were kept constant to obtain accurate
relative signals between samples and over time. Full CCD images were
transferred to a MATROX image processor (Image-1/AT, Universal Imaging) for analysis.
RESULTS AND DISCUSSION
Improvement of signal/background using YOYO
In early attempts of imaging the dynamic behaviour of single DNA molecules by
fluorescence microscopy it has been very difficult to obtain high-quality images using standard fluorescent dyes like ethidium bromide or
acridine orange (
19
-
21
). In fact, the binding and spectroscopic characteristics of these dyes are such
that relatively high DNA and dye concentrations must be employed, leading not
only to confusion between different molecules, but also to a high and
inhomogeneous fluorescence background. This drawback has clearly limited the
application of this technique for the study of the physical properties and the
dynamic behaviour of individual molecules of DNA under various experimental
conditions.
In this context, our first goal was to understand the relationship between the
DNA-dye complex and the corresponding image contrast, and, subsequently, to
search for new fluorescent dyes that would improve the signal/background of the
DNA images. Thus, we started out by characterizing the most important physical
and spectral properties of several fluorescent dyes including the members of a
new family of cyanine dyes recently designed and synthetized by Molecular
Probes Inc. Some of these results are summarized in Table
1
. In fact, the quality of fluorescence images of individual DNA molecules is
controlled by those factors that determine the difference in fluorescence
intensity between the molecules and the surrounding background. Clearly, the
binding affinity of the dye for the DNA and the enhancement in the dye
fluorescence upon binding to DNA are the critical factors determining the
contrast, i.e. the intensity of the signal over the background. Instead, the
extinction coefficient and the fluorescence quantum yield are the critical
factors determining the overall brightness of a fluorescently labeled DNA
molecule. Unfortunately, these conditions are poorly met by standard
fluorescent dyes such as ethidium bromide or acridine orange (Table
1
). On the other hand, it was found that a new dimeric cyanine dye, YOYO-1 (1,1'-(4,4,7,7-tetramethyl-4,7-diazaundecamethylene)-bis-4-[3-methyl-2,3-dihydro-(benzo-1,3-oxazole)-2-
methyl-idene]-quinolinium tetraiodide), is characterized by a substantial increase
in electrostatic binding affinity for DNA (due to the presence of four positive
charges on the molecule) about three orders of magnitude higher than ethidium
bromide and acridine orange (Table
1
). The very high binding affinity of YOYO-1 for DNA greatly minimizes the amount of free dye in solution that,
together with a high fluorescence enhancement upon binding to DNA (Table
1
), leads to an extremely low fluorescence background (YOYO-1 is virtually non-fluorescent when free in water solution). Moreover, YOYO-1 bound to double-stranded DNA has 30-fold higher extinction coefficient and a 3-fold higher fluorescence quantum yield
compared to ethidium bromide (Table
1
). Therefore YOYO labeled DNA molecules result much brighter with a
significantly lower background signal. Instead, acridine orange has, compared
to ethidium bromide, lower binding constant and lower fluorescence enhancement
upon binding, while it has a higher extinction coefficient and fluorescence
quantum yield (Table
1
). Therefore, DNA staining with acridine orange produces intrinsically brighter
molecules than ethidium bromide (about as bright as YOYO-stained molecules) but with a lower contrast.
Molecular mechanisms of DNA motion during steady-field electrophoresis
Only recently fluorescence microscopy has been used to image individual DNA
molecules undergoing agarose gel electrophoresis (
19
-
26
). As shown in Figure
3
, under the application of a steady electric field the dynamics of the molecules
turned out to be rather complex. The molecules migrate through the gel by a
cyclic motion of alternating elongated and compact conformations, resembling
the motion of a caterpillar. Elongated states form because the molecules remain hooked around gel fibers, extending both of the arms downfield, here named U-shapes (Fig.
3
D), and then slipping off, with the longer arm driving the chain (Fig.
3
F). The `tail' then catches up with the `head' and the molecule starts another
caterpillar cycle (Fig.
3
I). Increasing the electric field strength, the molecules become, on average,
better aligned in the direction of the field and more elongated. However, the
molecules never stretch totally to their theoretical contour length. This is
due to the intrinsic entropic elasticity of the molecules and also to the zig-zag path through the agarose network that they are constrained to follow. The leading end is usually bunched up, appearing brighter than
other parts of the molecule, since it has to find its way through the gel
pores. On the other hand, the body of the DNA chain is usually stretched
because the trailing end often remains wrapped around some gel fibers, and,
since it is already threaded through pores, just follows along the same path of
the `head'. Also, computer simulations (
24
,
36
-
37
) have clearly shown that in a caterpillar cycle, the head of the molecule moves
at approximately constant speed. Thus, chain retardation is not controlled by
the time required to form and slip out of the U-shapes, but by the sieving of the leading end through the gel pores. Since
this process is clearly independent of molecular size it explains the failure
of conventional electrophoresis for DNA molecules longer than 10-20 kb (
1
-
2
).
Here, the use of YOYO-1 has allowed a major improvement in the quality of the images. During
electrophoresis, the DNA molecules labeled with ethidium bromide or acridine
orange appear relatively bright under the microscope only if they are in a
compact state. However, when the molecules assume elongated conformations they
appear so dim that is barely possible to follow their contour, especially if
part of them are out of the plane of focus. The quality of the images is also
affected by the high and inhomogeneous background due to the molecules out of
the plane of focus and to the presence of an excess of free dye in solution. On
the contrary, DNA molecules labeled with YOYO-1 appear very bright, also in conditions of maximum extension (Fig.
3
C-E), against a very dark and homogeneous background.
Molecular mechanisms of separation in 90
o
PFGE and 120
o
PFGE
Fluorescence microscopy and computer simulations have shown that in DNA
molecules undergoing 90o
PFGE, the `head' of the chain (that is bunched) usually remains the leading end
of the molecule and turns the corners in the direction of the new field,
entering new gel pores. Figure
4
shows a time sequence of a T2 DNA molecule reorienting by the leading end. Only
a few short kinks (Fig.
4
B-C) (where the molecule is doubled over on itself) are observed at the
beginning of the reorientation, and these disappear very rapidly. Also, both ends of the molecule can extend and turn in the new direction (Fig.
4
D-F) but eventually the leading end keeps driving the chain (Fig.
4
G). This reorientation mechanism is also schematically shown in Figure
5
A, where for simplicity the DNA molecule is drawn in a straight conformation.
When the field is switched by an angle <= 90o the same end leads the molecule during the reorientation because the
component of the new electric field along the previous path direction is either along the same direction the molecule was moving before the field was switched (for
angles <90o) or it is zero (for angles = 90o). In this case, since the same end leads the molecule during the
reorientation, the overall DNA motion is not very different from that observed
during conventional (steady-field) electrophoresis (Fig.
3
). Here, as well, chain retardation is controlled by the sieving of the leading end through the gel pores (a process independent of molecular
size). Therefore, this reorientation mechanism provides a clear explanation for
the poor separation obtained when angles <= 90o are employed in PFGE experiments. Only occasionally the `tail' end of
DNA molecules undergoing 90o
PFGE was observed to turn in the direction of the field and sometimes even
succeed in driving the chain.
REFERENCES
1 McDonell,M.W., Simon,M.N. and Studier,F.W. (1977) J. Mol. Biol., 110, 119-146.MEDLINE Abstract
29 Cantor,C.R., Smith,C.L. and Mathew,M.K. (1988) In Engelman,D.M, Cantor,C.R. and Pollard,T.D. (eds) Annual Reviews of Biophysics and Biophysical Chemistry. Annual Reviews, Inc., Palo Alto, CA, Vol. 17, pp. 287-304.
33 Larsson,A., Carlsson,C., Jonsson,M. and Albinsson,N. (1994) J. Am. Chem. Soc., 116, 8459-8465.
34 Netzel,T.L., Nafisi,K., Zhao,M.,Lehnard,J.R. and Johnson,I. (1995) J. Phys. Chem., 99, 17936-17947.
35 Carlsson,C., Larsson,A. and Jonsson,M. (1996) Electrophoresis, 17, 642-651.MEDLINE Abstract
36 Smith,S.B., HellerC. and Bustamante,C. (1990) Biochemistry, 30, 5264-5274.
37 Smith,S.B., Gurrieri,S., Bustamante,C. (1990) In Lai.E. and Birren,B.W. (eds) Current Communications in Cell and Molecular Biology. Electrophoresis of Large DNA Molecules: Theory and Applications. Cold Spring Harbor Press, Cold Spring Harbor, New York, Vol. 1, pp. 55-79.
38 Weber,G. and Teale,F.W.G. (1957) Trans. Farad. Soc., 53, 646-655.
39 Haughland,R.P. (1992) In Larison,K.O. (ed.) Fluorescent Probes and Research Chemicals. Molecular Probes, Inc., Eugene, OR, pp. 224.
