High-resolution separation of DNA restriction fragments using capillary electrophoresis with Near-IR, diode-based, laser-induced fluorescence detection

Article abstract of DOI:10.1021/ac960975v

The near-IR dye thiazole green (TAG) was used as a monomeric nuclear staining dye for the low-level detection of DNA restriction fragments separated via high-performance capillary electrophoresis with near-IR laser-induced fluorescence detection. TAG poss

Full text of DOI:10.1021/ac960975v

Anal. Chem. 1997, 69, 1256-1261
Hig h -Re s o lu t io n S e p a ra t io n o f DNA Re s t ric t io n
Fra g m e n t s Us in g Ca p illa ry Ele c t ro p h o re s is w it h
Ne a r-IR, Dio d e -Ba s e d , La s e r-In d u c e d Flu o re s c e n c e
De t e c t io n
Clyde V. Ow e ns , Yola nda Y. Da vids on, Sa tya jit Ka r, a nd Ste ve n A. Sope r*
Department of Chemistry, Louisiana State University, Baton Rouge, Louisiana 70803-1804
stringent requirements associated with the detection protocol are
necessary in order to detect modest concentration levels of
dsDNAs. While native UV detection at 254 nm can be imple-
mented, it suffers from poor sensitivity due to the short optical
path length associated with capillary electrophoresis. Therefore,
laser-induced fluorescence (LIF) with intercalating dyes has
proven to be the method of choice in critical applications requiring
low levels of detection.8-13
The near-IR dye thiazole green (TAG) was used as a
monomeric nuclear staining dye for the low-level detection
of DNA restriction fragments separated via high-perfor-
m ance capillary electrophoresis with near-IR laser-
induced fluorescence detection. TAG possessed an ab-
sorption maximum at 7 3 5 nm and an emission maximum
at approximately 765 nm and, in the presence of dsDNAs,
showed a fluorescence enhancement ratio of approxi-
mately 1 0 2 , with a binding constant to dsDNAs deter-
LIF detection has typically been undertaken using intercalating
dyes which possess absorption and emission maxima in the visible
region of the electromagnetic spectrum (450-630 nm). The
detection is affected by monitoring the perturbation in the
spectroscopic property of the dye when in the bound state.
Cationic dyes which have planar aromatic or heteroaromatic rings
and exhibit enhancements in their fluorescence emission upon
complexation with dsDNA include the monointercalating dyes
such as ethidium bromide (EtBr)14 thiazole orange (TO),15 and
oxazole yellow (YO).15 EtBr shows a fluorescence enhancement
of 20 when it interacts with dsDNA, and the dye has been found
to show a binding constant of 1.1 × 106 M-1 to dsDNAs.16 In
addition, the dimeric forms of these monointercalating dyes have
been used for the analysis of dsDNAs.17-19 The dimeric staining
dyes consist of two chromophores covalently linked through a
polycationic chain and show larger binding constants to dsDNA
when compared to their monomeric counterparts. Because the
fluorescence quantum yields of the dimeric form of the dyes are
significantly improved in the bound state compared to those in
free solution, the background fluorescence from free dye is very
low, which makes these dyes excellent probes for the low-level
quantification of dsDNA.
mined to be 6 .1 × 1 0 6 M-1
. The high-resolution sepa-
ration of the Ha eIII restriction digest of ΦX1 7 4 was
carried out using capillary electrophoresis on the native,
ethidium bromide-stained, and TAG-stained DNA frag-
ments. The TAG-stained DNA fragments resulted in
higher plate numbers compared to the native and EtBr-
stained restriction fragments as well as enhanced resolu-
tion; however, the 2 7 1 / 2 8 1 fragments could not be
resolved using these CE conditions. To investigate the
detection sensitivity of the TAG-stained DNA in capillary
electrophoresis, an all-solid-state diode-based, laser-
induced fluorescence detector was constructed, which
consisted of a GaAlAs diode laser, with a principal lasing
line at 7 5 0 nm and an avalanche photodiode. Using a
running buffer composed of an entangled polymer (HPMC)
and 1 µM TAG with no prestaining of the dsDNA prior to
the electrophoresis, the limit of detection was found to
be 2 0 fg (SNR ) 3 ) of DNA per electrophoretic band. In
addition, using the LIF system, the 271/ 281 bp fragments
were nearly baseline resolved, with plate numbers ex-
ceeding 1 × 1 0 6 plates/ m.
(8) Schwartz, H. E.; Ulfelder, K. J. Anal. Chem. 1 9 9 2 , 64, 1737-1740.
(9) McCord, B. R.; McClure, D. L.; Jung, J. M. J. Chromatogr. 1 9 9 3 , 652, 75-
82.
(10) Srinivasan, K.; Girard, J. E.; Williams, P.; Roby, R. K.; Weed, V. W.; Morris,
S. C.; Kline, M. C.; Reeder, D. J. J. Chromatogr. 1 9 9 3 , 652, 83-91.
(11) Kim, Y.; Morris, M. D. Anal. Chem. 1 9 9 4 , 66, 1168-1174.
(12) Zhu, H.; Clark, S. M.; Benson, S. C.; Rye, H. S.; Glazer, A. N.; Mathies, R.
A. Anal. Chem. 1 9 9 4 , 66, 1941-1948.
Capillary electrophoresis (CE) in conjunction with entangled
polymer solutions as a sieving matrix has been shown to be an
attractive fractionating method for dsDNAs due to the high
efficiencies and resolution that are attainable and the speed
associated with the separation technique.1-7 However, due to the
low mass-loading levels associated with capillary electrophoresis,
(13) Figeys, D.; Arriaga, E.; Renborg, A.; Dovichi, N. J. J. Chromatogr. 1 9 9 4 ,
669, 205-216.
(1) Heiger, D. N.; Cohen, A. S.; Karger, B. L. J. Chromatogr. 1 9 9 0 , 516, 33-
48.
(2) Schwartz, H. E.; Ulfelder, K.; Sunzeri, F. J.; Busch, M. P.; Brownlee, R. G.
(14) Le Pecq, J. B.; Paoletti, C. Anal. Biochem, 1 9 6 6 , 17, 100-107.
(15) Rye, S.; Yue, S.; Wemmer, D.; Quesada, M.; Haugland, R.; Mathies, R.;
Glazer, A. Nucleic Acids Res. 1 9 9 2 , 20, 2803-2812.
(16) Gaugain, B.; Barbet, J.; Capelle, N.; Roques, B.; Le Pecq J. Biochem. 1 9 7 8 ,
17, 5078-5088.
(17) Benson, S.; Singh, P.; Glazer, A. Nucleic Acids Res. 1 9 9 3 , 21, 5727-5737.
(18) Rye, S.; Quesada, M.; Peck, K.; Mathies, R.; Glazer, A. Nucleic Acids Res.
1 9 9 0 , 19, 327-332.
J. Chromatogr. 1 9 9 1 , 559, 267-283.
(3) Strege, M.; Lagu, A. Anal. Chem. 1 9 9 1 , 63, 1233-1236.
(4) Grossman, P. D.; Soane, D. S. J. Chromatogr. 1 9 9 1 , 559, 257-266.
(5) Pariat, Y. F.; Berka, J.; Heiger, D. N.; Schmitt, T.; Vilenchik, M.; Cohen, A.
S.; Foret, F.; Karger, B. L. J. Chromatogr. 1 9 9 3 , 652, 57-66.
(6) Barron, A. E.; Soane, D. S.; Blanch, H. W. J. Chromatogr. 1 9 9 3 , 652, 3-16.
(7) Grossman, P. D. J. Chromatogr. 1 9 9 4 , 663, 219-227.
(19) Glazer, A.; Peck, K.; Mathies, R. Proc. Natl. Acad. Sci. U.S.A. 1 9 9 0 , 87,
3851-3855.
1256 Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
S0003-2700(96)00975-4 CCC: $14.00 © 1997 American Chemical Society

In capillary electrophoresis, the monomeric dyes are typically
added to the running buffers due to their smaller binding
constants, and the dsDNA need not be prestained prior to the
electrophoresis. Because the binding constants to dsDNA for the
dimeric forms of the dyes are much higher, the DNA is typically
prestained with the bis-intercalating dye, and it is unnecessary to
add the dye to the running buffer during the electrophoretic
separation. The limit of detection using the dimers is typically
improved compared to that obtained using the monomers, but it
suffers from poor resolution due to the multiple binding modes
of the dye.11,13
The instrumental difficulty associated with utilizing visible LIF
detection is that the excitation source is a gas-ion laser, for
example an Ar ion or He-Ne laser, which has a limited operational
lifetime and, in some cases, can be expensive and difficult to
operate. In addition, visible excitation can encounter interferences
in the form of scattering and fluorescence impurities, which can
degrade the limit of detection, especially when using bis-intercalat-
ing dyes.
An attractive alternative to visible fluorescence is the use of
near-IR fluorescence detection. Near-IR fluorescence has recently
been demonstrated to be a viable detection strategy in liquid
chromatography,20-22 free solution capillary electrophoresis,23-27
and capillary gel electrophoresis for DNA sequencing applica-
tions.28 One of the principal advantages of using near-IR fluores-
cence is that simple diode lasers with lasing lines >700 nm can
be used as the excitation source for the LIF detector. These diode
lasers are compact, employ simple turn-key operation, have
relatively high powers (20 mW), provide long operational lifetimes
(>40 000 h), and reduce the cost of setting up such a detection
system. In addition, near-IR fluorescence typically displays fewer
interferences from the background matrix resulting from impurity
fluorescence and/ or a smaller scattering contribution as a result
of the 1/ λ4 dependence on the scattering cross section, improving
the limit of detection.28 This potentially lower background will
be particularly attractive when using dimeric staining dyes in
conjunction with CE, when the dsDNA is prestained and the dye
not required to be present in the running buffer.
Fig u re 1 . Near-IR LIF system for CE analysis of DNA restriction
fragments. M1, M2, and M3, mirrors; L, laser focusing lens; C,
capillary tube; BD, beam dump; MO1 and MO2, microscope objectives
(see text for details); S, spatial filter; F, optical filters (see text); SPAD,
single-photon avalanche diode; C/T, counter/timer; PC, computer.
state and easily operated, producing an LIF detection system that
has a potentially longer operational lifetime and is very inexpensive
but displays detection sensitivities for dsDNAs comparable to
those of the visible systems that use gas ion lasers.
EXPERIMENTAL SECTION
Instrumentation. Electrophoretic separations of native, ethid-
ium bromide-stained, and TAG-stained ΦX174 HaeIII-digested
DNA were performed on a Beckman electrophoresis apparatus
with UV detection at 254 nm (Foster City, CA). The electro-
phoresis system was operated in the reverse polarity mode
(negative potential at the injection end of the capillary), and the
data were acquired using the chromatographic software on a
personal computer for displaying and analyzing the data. The
near-IR LIF detection system was constructed in-house and is
schematically shown in Figure 1. It consisted of a 7 mW, 750 nm
diode laser (GaAlAs, Melles Griot, Irvine, CA) and a diode laser
driver (Model 06 DLD 201, Melles Griot). The laser head
contained an anamorphic prism pair to produce a circular beam
that was single mode and also a thermoelectric cooler to maintain
the diode temperature to prevent mode-hopping due to temper-
ature fluctuations. Incorporation of external cavity optics improves
the quality of the laser beam, allowing tighter focusing for use in
electrophoresis devices with small inner diameter columns. The
laser beam was focused onto the capillary detection window with
a planoconvex lens (25 mm diameter, 25.0 mm focal length,
Edmund Scientific, Barrington, NJ) and produced a beam waist
(1/ e2) of 7 µm. The capillary was affixed horizontally (parallel to
the optical bench) onto a home-made Plexiglas capillary holder,
which was mounted on top of an X-Y micropositioner for
positioning the capillary with respect to the laser beam and
collection optics. The fluorescence was collected at right angles
with respect to the laser beam using a 40× epifluorescence
microscope objective (Melles Griot) with a numerical aperture of
0.65 and was imaged onto a slit serving as a spatial filter to reduce
the amount of scattered photons generated at the air/ glass and
glass/ liquid interfaces of the capillary from reaching the detector.
The fluorescence was further isolated from the scattering photons
In this work, we report on the first demonstration of near-IR,
diode-based LIF detection of restriction fragments separated by
CE with an entangled polymer solution and staining using a
monomeric intercalating dye which shows absorption and emis-
sion properties in the near-IR (>700 nm). The staining dye,
thiazole green (TAG), was prepared by extending the methine
chain in the common monointercalator TO-PRO 3, resulting in a
shift in the absorption maxima into the near-IR.29 The LIF system
consisted of a GaAlAs diode laser operating at 750 nm and a single-
photon avalanche diode (SPAD). Both of these devices are solid-
(20) Sauda, K.; Imasaka, T.; Ishibashi, N. Anal. Chim. Acta 1 9 8 6 , 187, 353-
356.
(21) Sauda, K.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1 9 8 6 , 58, 2649-2653.
(22) Williams, R. J.; Lipowska, M.; Patonay, G.; Strekowski, L. Anal. Chem. 1993,
65, 601-605.
(23) Higashijima, T.; Fuchigami, T.; Imasaka, T.; Ishibashi, N. Anal. Chem. 1992,
64, 711-714.
(24) Fuchigami, T.; Imasaka, T. Anal. Chim. Acta 1 9 9 3 , 282, 209-213.
(25) Bachteler, G.; Drexhage, K.-H.; Jacob, J.; Han, K.-T.; Kollner, M.; Muller,
R.; Sauer, M.; Seeger, S.; Wolfrum, J. J. Lumin. 1 9 9 4 , 62, 101-108.
(26) Fuchigami, T.; Imasaka, T. Anal. Chim. Acta 1 9 9 4 , 291, 183-188.
(27) Flanagan, J. H.; Legendre, B. L.; Hammer, R. P.; Soper, S. A. Anal. Chem.
1 9 9 5 , 67, 341-347.
(28) Williams, D. C.; Soper, S. A. Anal. Chem. 1 9 9 5 , 67, 3427-3432.
(29) Glazer, A.; Zeng, Z.; Benson, S. Anal. Biochem. 1 9 9 5 , 231, 247-255.
Analytical Chemistry, Vol. 69, No. 6, March 15, 1997 1257

by a 780 nm band-pass filter (Oriel, Stratford, CT) and a 780 nm
long-pass filter (Edmund Scientific, Barrington, NJ). The fluo-
rescence was then focused onto the photoactive area of the
detector with a 20× microscope objective (Melles Griot). The
detector was a SPAD (Model SPCM-AQ-141, EG&G Optroelec-
tronics Canada, Vaudreuil, Canada) with a 200 µm diameter
photoactive area and possessed a dark and maximum light count
rate, before saturation, of approximately 30 and 16 × 106 counts/
s, respectively. To increase the linear dynamic range, the SPAD
was operated in an actively quenched mode. The LIF signals were
acquired on a personal computer (Gateway 2000, Model P5-120)
using a 16-bit counter/ timer board (Model CYRCTM 05, Cyber-
Research Inc., Brandford, CT). The CE data were not subjected
to any type of filtering algorithm prior to presentation. The high-
voltage power supply was obtained from Spellman (CZ1000R,
Plainview, NY).
S c h e m e 1
Reagents. Tris-HCl, γ-[(methoxyacryl)oxy]propyltrimethoxy-
silane (MAPS), and ethidium bromide were obtained from
Amresco (Solon, OH) and used as received. Hydroxypropylmeth-
ylcellulose (MW ) 86 000), anhydrous ethylenediaminetetraacetic
acid (EDTA), and N,N,N′,N′-tetramethylethylenediamine (TEMED)
were obtained from Sigma Chemical Co. (St. Louis, MO). The
ΦX174/ HaeIII digest was received from Life Technologies (Gaith-
ersburg, MD) and was stored in a freezer at -20 °C until used
for the electrophoresis. Calf thymus DNA, used for studies of
the binding of TAG to dsDNA, was obtained from United States
Biochemical (St. Louis, MO) and was sonicated in an ultrasoni-
cator prior to use. During sonication, the DNA was placed on
ice, and following sonication, it was stored in ultrapure water at
-20 °C. The nucleotide base concentration was determined by
measuring the absorbance at 260 nm and assuming an extinction
coefficient of 6000 cm-1 M-1 per nucleotide base.
Sample and Sieving Buffer P reparation for Electrophore-
sis. HaeIII restriction digest samples of ΦX174 were diluted (1:
4; total [DNA] ) 100 µg/ mL) using 18 MΩ Milli-Q water prior to
electrophoresis and injected directly onto the CE capillary using
electrokinetic injection. The sieving matrix consisted of a 0.25%
HPMC solution, which was made by adding the appropriate
amount of polymer to the TAE running buffer (40 mM TRIS, 20
mM acetate, 2 mM EDTA, pH ) 7.61). The sieving buffer was
placed in a water bath and stirred with heating to completely
dissolve the polymer. Prior to electrophoresis, the sieving buffer
was filtered twice with 0.45 µm filter paper to remove particulates
and undissolved polymer.
Capillary P reparation. Fused silica capillaries (75 µm i.d.,
375 o.d.) were purchased from Polymicro Technologies, Inc.
(Phoenix, AZ) and were coated with a linear acrylamide (1%T)
according to published procedures.30 Briefly, the capillary was
rinsed first with 1.0 M NaOH and then 1.0 M HCl, each followed
with copious rinsing with doubly distilled H2O. The capillary was
rinsed overnight with a 50:50 MAPS/ MeOH solution, after which
the column was dried in an oven for several hours at 80 °C. A
1%T linear acrylamide solution was made in 1× TBE (TRIS, borate,
EDTA) from a 40%T/ 0%C stock solution. The solution was
thoroughly degassed by water aspiration for approximately 2.5 h.
Twenty-five microliters of fresh 10% ammonium persulfate (APS)
in H2O was added to the degassed acrylamide solution, followed
by 5 µL of TEMED. The capillary was quickly filled with the
acrylamide solution and allowed to rest horizontally until complete
polymerization occurred, after which high pressure was used to
remove excess linear acrylamide. The total capillary length used
for both conventional and LIF CE was 39 cm (31 cm from injection
to detection).
The capillary and cathodic and anodic reservoirs were filled
with the running buffer solution, which contained the TAE buffer,
0.25% (w/ v) hydroxypropylmethylcellulose, and the appropriate
concentration of the staining dye. Before injection of sample, the
column was preconditioned by applying a field strength of 170
V/ cm for 15 min. Nonstained dsDNA samples were injected into
the capillary electrokinetically using an applied voltage of 2 kV
for 5 s for the LIF system (30 s for conventional CE).
Synthesis of Near-IR Cyanine Dye. The near-IR nuclear
staining dye TAG was prepared following modifications of previ-
ously described procedures.29 The synthetic steps are outlined in
Scheme 1. 2-Methylbenzathiazole (0.067 mol) was added to 3
equiv of methyl iodide and refluxed for 18 h in ethanol at 72 °C
to form 2 . The iodoalkyl derivative 3 was prepared by adding
0.034 mL of lepidine to 5 equiv of 1,3-diiodopropane (0.176 mol)
and refluxed in toluene at 110 °C for 18 h. The yellow-brown
precipitate was filtered with extensive washing using ethyl ether
and collected. Malonaldehyde dianil hydrochloride was prepared
from malonaldehyde bis(dimethyl acetal) and aniline. The inter-
mediate was produced by refluxing 2 (0.0121 mol) and 4 (0.100
mol) in a 1:1 mixture of acetic acid and acetic anhydride. The
reaction progress was monitored via UV absorption and judged
(30) Hjerten, S. J. J. Chromatogr. 1 9 8 5 , 347, 191-198.
1258 Analytical Chemistry, Vol. 69, No. 6, March 15, 1997

Fig u re 3 . Spectrofluorometric titration of TAG with calf thymus DNA.
In all cases, the DNA concentration was measured in terms of the
nucleotide bases and was varied from 5 × 10-6 to 1 × 10-4 M. The
fluorescence was excited at 710 nm and the emission monochromator
scanned from 720 to 900 nm. The dye concentration used was 1.0
× 10-6 M.
Fig u re 2 . Absorption spectra of TAG in methanol (solid line), TRIS/
acetate buffer with no DNA (dot/dashed line), and a 100-fold molar
excess, in nucleotide bases, of calf thymus DNA (dotted line). In all
cases, the dye concentration was 2 × 10-5 M.
Construction of a modified Benesi-Hildebrand plot33 for these
data (see insert to Figure 3) indicated that the binding constant
of TAG to dsDNA was approximately 6.1 × 106 M-1, nearly 50×
greater than the binding constant of EtBr to dsDNA, which has
complete when the 286 nm band nearly vanished. Reaction of 3
(0.64 mol) and 5 (0.0059 mol) in ethanol with sodium acetate after
30 min of refluxing produced 6 (thiazole green, TAG). This
reaction was also monitored by UV/ vis absorption, which indicated
the presence of the symmetrical (665 nm) and unsymmetrical (735
nm) dye. TAG was purified by flash chromatography using a
chloroform/ methanol solvent (9:1 v/ v). The NMR and mass
spectral analysis revealed the identity and purity of 6 .
been determined to be 1.5 × 105 M-1 16
.
Integration of the spectra
in the buffer only and in a 100-fold molar excess (in nucleotide
bases) of dsDNA indicated that the fluorescence enhancement
ratio of bound to free dye was 102.
CE Analysis of Native and Stained DNA. In order to make
a comparison of the separation efficiency of native, EtBr-stained,
and TAG-stained DNA, the HaeIII restriction fragments of ΦX174
were analyzed using a conventional CE system with UV detection
at 254 nm. In Table 1 are shown the plate numbers (N, m-1) and
the apparent electrophoretic mobilities (µapp) of these restriction
fragments under different staining conditions. Generally, the TAG-
stained restriction fragments produced higher plate numbers than
the EtBr-stained DNA or the nonstained DNA. Also seen in this
data is that the apparent electrophoretic mobilities of the TAG-
stained DNA fragments are smaller than those associated with
the EtBr-stained fragments. This result is supported by the
binding constant measured for the TAG/ DNA complex, which
indicated a value larger than that associated with EtBr to dsDNA.
Therefore, the increased stability of the dye/ DNA complex would
result in a lower mobility due to reductions in the negative charge
of the biopolymer and increases in the frictional coefficient
resulting from elongation of the double helix produced from
intercalation and the increased molecular mass associated with
the complex. It should be noted that, in all cases, we were unable
to resolve the 271/ 281 bp fragments in this restriction digest under
these CE conditions.
Absorbance and Fluorescence Measurements. The ab-
sorption spectra were acquired on a Perkin-Elmer Lambda 3B UV/
vis spectrophotometer. The fluorescence spectra were obtained
on a SPEX Fluorolog spectrofluorometer (SPEX, Edison, NJ)
equipped with a 75 W xenon lamp. The emission gratings were
blazed for 750 nm, and the photomultiplier tube was a Ham-
mamatsu R636 red-sensitive photomultiplier tube.
RESULTS AND DISCUSSION
Absorption and Emission P roperties of TAG. In Figure 2
are shown the absorption spectra of TAG in methanol, TAE buffer,
and TAE with a 100-fold molar excess (in nucleotide bases) of
calf thymus DNA. In methanol, the dye gave a single band with
an absorption maximum at 735 nm. This band could be assigned
to the monomer, since it is expected that the dye shows little
tendency to aggregate in this solvent.31 In TAE buffer, the
spectrum shows two broad bands at approximately 705 and 533
nm, superimposed on a diffuse background. These spectral
properties are indicative of extensive dye aggregation, forming
dimers and other higher order aggregates, which is a typical
phenomenon for the extended cyanine dyes.32 In the presence
of double-stranded calf thymus DNA, the monomer band is
partially restored (λmax ) 740 nm), and another prominent band
appears at 650 nm, which could arise from a dimeric form of the
dye.
Near-IR LIF Detection. Figure 4 shows the CE separation
of the HaeIII digest of φX174 in a polyacrylamide-coated capillary
using near-IR LIF detection. From the apparent mobility and
electrokinetic injection conditions (2 kV for 5 s), it was estimated
that approximately 3.1 pg of the 603 bp fragment was inserted
onto the capillary. The SNR for this peak, which was determined
by integrating the area under the peak and dividing by the square
root of the average background integrated over the same time
In Figure 3 is shown a series of fluorescence spectra which
were obtained at various concentrations of calf thymus DNA (dye
concentration 1 µM). As can be seen, drastic enhancements in
the fluorescence were observed upon addition of calf thymus DNA.
(31) Soper, S. A.; Mattingly, Q. L. J. Am. Chem. Soc. 1 9 9 4 , 116, 3744-3752.
(32) West, W.; Pearce, S. J. Phys. Chem. 1 9 6 5 , 69, 1894-1903.
(33) Green, M. D.; Potanay, G.; Ndou, T.; Warner, I. M. Appl. Spectrosc. 1 9 9 2 ,
46, 1724-1728.
Analytical Chemistry, Vol. 69, No. 6, March 15, 1997 1259

Ta b le 1 . CE An a lys is o f t h e Ha e III Re s t ric t io n Dig e s t o f t h e Na t ive ΦX1 7 4 , Et Br, a n d TAG-S t a in e d Fra g m e n t s a
native
EtBr
TAG
fragment
size
µ
app (cm2 V-1 s-1
)
µ
app (cm2 V-1 s-1
)
µ
app (cm2 V-1 s-1
)
N (m-1)b (×105)
(×10-4
)
N (m-1) (×105)
(×10-4
)
N (m-1) (×105)
(×10-4
)
N (m-1)b (×105)c
72
118
194
234
271
281
310
603
872
4.77 ((0.08)
3.84 ((0.05)
3.16 ((0.09)
4.60 ((0.06)
d
4.26 ((0.04)
4.16 ((0.03)
3.98 ((0.03)
3.88 ((0.06)
d
5.63 ((0.07)
3.88 ((0.05)
3.98 ((0.05)
4.19 ((0.04)
d
4.04 ((0.05)
3.97 ((0.04)
3.82 ((0.03)
3.74 ((0.05)
d
5.11 ((0.08)
5.01 ((0.09)
5.17 ((0.07)
5.07 ((0.08)
d
3.99 ((0.04)
3.89 ((0.03
3.72 ((0.07)
3.62 ((0.05)
d
22.2 ((0.3)
23.5 ((0.4)
26.2 ((0.5)
27.6 ((0.3)
29.3 ((0.4)
29.6 ((0.3)
7.68 ((0.01)
10.6 ((0.2)
12.5 ((0.3)
13.7 ((0.4)
14.8((0.2)
d
d
d
d
d
d
6.00 ((0.08)
4.44 ((0.08)
4.49 ((0.10)
5.27 ((0.11)
5.43 ((0.11)
3.70 ((0.05)
3.22 ((0.05)
2.95 ((0.04)
2.82 ((0.03)
2.70 ((0.06)
4.50 ((0.04)
4.47 ((0.05)
4.11 ((0.06)
4.42 ((0.04)
4.77 ((0.03)
3.59 ((0.04)
3.18 ((0.03)
2.94 ((0.03)
2.81 ((0.04)
2.69 ((0.04)
6.29 ((0.10)
7.13 ((0.08)
8.28 ((0.06)
9.45 ((0.08)
9.87 ((0.09)
3.44 ((0.03)
2.99 ((0.02)
2.73 ((0.03)
2.60 ((0.03)
2.48 ((0.03)
1,078
1,353
a
Detection was accomplished using UV absorption at 254 nm. The restriction fragments were injected electrokinetically onto the CE column
for 30 s at -5 kV. The standard deviations, which were determined from three replicate measurements, are given in parentheses. b Plate numbers
were calculated, assuming a Gaussian peak, from the fwhm of the electrophoretic peak and the migration time. c Plate numbers determined for the
LIF system. d The 281 fragment comigrated with the 271 fragment.
reduction in the fluorescence intensity resulting from free dye
due to photobleaching of the staining dye, since many near-IR
dyes have been shown to exhibit poor photochemical stability as
compared to the visible fluorescent dyes.31 Since the dye travels
from the anodic to cathodic reservoirs due to its cationic nature
and the DNA travels in the opposite direction, the bleached dye
can compete with unbleached dye for the limited number of
binding sites associated with the DNA in each electrophoretic
band. Therefore, lower field strengths can degrade the limit of
detection in situations where the staining dye is added to the
running buffer. In the present case, we observed a SNR for the
603 bp fragment of 191 when the electrophoresis was run using
a field strength of -135 V/ cm; when the field strength was
increased to -270 V/ cm, the SNR increased to 231 when the DNA
Fig u re 4 . Electropherogram of ΦX174 DNA restriction fragments
with LIF detection. The dye concentration added to the running buffer
was 1.0 µM, and the running buffer was composed of 40 mM TRIS,
20 mM sodium acetate, 2.0 mM EDTA (pH 7.6), and 0.25% HPMC.
The sample was electrokinetically injected for 6 s at -2.0 kV onto
the column. The electric field strength used for the separation was
175 V/cm. The laser power was set at approximately 8.5 mW.
sample was injected onto the column using identical conditions
and similar DNA concentrations for both cases.
From the electropherogram depicted in Figure 4, it is seen
that the 271/ 281 bp fragments are nearly baseline resolved, with
a resolution determined to be 1.3, significantly better than that
observed for the conventional CE results where the native, EtBr-
stained, and TAG-stained 271/ 281 bp fragments comigrated (see
Table 1). Since the capillary length and conditions were similar
in both cases and the sieving buffer was identical, the enhanced
resolution in the LIF system most likely results from improve-
ments in the efficiency of the separation and not differences in
selectivity. This supposition is supported by our data, which
showed that the plate numbers for these restriction fragments
when stained with TAG were found to be significantly better for
the LIF system compared to the conventional CE case. For the
conventional system, the plate numbers for the 603 fragment were
4.13 × 105 plates/ m, while for the LIF system, the plate numbers
for this same fragment was 1.06 × 106 plates/ m. The significant
improvement in efficiency most likely results from the fact that
the zone variances arising from the finite injection volume and
detection volume are smaller in the LIF case compared to the
conventional case since, in the LIF system, the beam length was
significantly smaller and the amount of sample inserted onto the
column was reduced. The need for extended injection conditions
in the case of UV detection resulted from the need to obtain
sufficient signal strengths for analysis by inserting more material
onto the column.
interval, was 465. The on-column mass limit of detection for this
fragment under these electrophoresis conditions was estimated
to be 20 fg (SNR ) 3), comparable to the visible LIF detection of
this same digest stained with TO.13
The magnitude of the background signal, which in part
determines the limit of detection, depends on several experimental
parameters, including the concentration of the dye in the running
buffer, the applied electric field strength, the magnitude of Raman
and Rayleigh scattering, and the intensity of the laser beam. Since
the running buffer consisted of the staining dye, one of the major
contributions to the background is fluorescence from free dye
and scattering in the form of Raman and/ or Rayleigh scattered
photons. Using a 1 µM dye solution added to the running buffer,
we observed an increase in the background signal of ap-
proximately 10-15% with respect to the HPMC containing TAE
solution only. In addition, we noticed that significant decreases
in the background signal in the presence of free dye was observed
at lower electric field strengths. Using a dye concentration of 1
µM, the background signal decreased by 9.3% when the electric
field was decreased from -270 to -135 V/ cm. We attributed this
1260 Analytical Chemistry, Vol. 69, No. 6, March 15, 1997

helix. As a result, it is likely to result in broad peaks with longer
migration times, consistent with our data.
CONCLUSIONS
We have prepared a near-IR, monomeric nuclear staining dye
(TAG) which forms stable complexes to dsDNA and exhibits a
fluorescence enhancement ratio of bound to free dye of ap-
proximately 102, making it an ideal probe for the detection of DNA
restriction fragments or other dsDNAs, such as PCR products.
In addition, this dye has an absorption maximum near the principal
lasing line of a GaAlAs diode laser, allowing the construction of
an LIF detector for capillary electrophoresis which consists of all-
solid-state components, reducing cost and simplifying operation.
The detection limit of the dsDNA fragments was determined to
be 20 fg of DNA per electrophoretic band, comparable to those
obtained with visible LIF systems using Ar ion laser excitation.
Using capillary electrophoresis with an entangled polymer sieving
matrix, TAG-stained dsDNA fragments were found to have
improved efficiency compared to EtBr-stained fragments, most
likely resulting from the higher affinity of TAG to dsDNA
compared to EtBr.
The attractiveness of near-IR LIF detection also arises from
the fact that the inexpensive diode laser and avalanche diode can
easily be miniaturized and integrated directly onto a microfabri-
cated electrophoresis system for the analysis of dsDNAs, produc-
ing a fully miniaturized apparatus. Work in our laboratory is
currently underway toward developing miniaturized, high-sensitiv-
ity LIF detectors for microelectrophoresis systems. In addition,
we are synthetically preparing nuclear staining homodimers which
will also have absorption and emission properties associated with
the near-IR region in order to improve detection limits.
Fig u re 5 . Electropherograms of ΦX 174 DNA restriction fragments
obtained using (a) 4.0, (b) 1.0, and (c) 0.2 µM dye in the running
buffer. See Figure 3 for experimental details.
We next investigated the effect of dye concentration placed in
the running buffer on CE detectability and efficiency for TAG-
stained DNA restriction fragments. The results of this investiga-
tion for three different dye concentrations are shown in Figure 5
(0.2, 1.0, and 4.0 µM). The fluorescence intensity of the DNA
fragments decreased significantly when the dye concentration in
the buffer was decreased to 0.2 µM due to less dye incorporated
into the dsDNA fragment, which is based on thermodynamic
considerations (see Figure 5c). In addition, it was found that the
peaks broadened as well when the dye concentration was reduced
to 0.2 µM as compared to the case when 1 µM dye was used (N
) 1.38 × 106 plates/ m for 0.2 µM dye and N ) 1.06 × 106 plates/ m
for 1.0 µM dye for the 603 bp fragment). Increasing the dye
concentration in the running buffer to 4.0 µM deteriorated the
peak shapes of all DNA fragments, causing significant peak
broadening and tailing (see Figure 5a, N ) 6.80 × 105 plates/ m
for the 603 bp fragment). In comparison to the data obtained
with 1.0 µM dye, the peak areas were found to be approximately
4 times larger for the 4.0 µM dye solution. However, the peak
heights were nearly equivalent (slightly higher and lower for 72-
310 and 603-1353 bp fragments, respectively), indicating that the
peaks were broadened under these conditions, as apparent from
the reduced plate numbers for the 4 µM dye case. Schwartz et
al.2 postulated that the excess dye possibly binds to ssDNAs as
well. If this binding process becomes thermodynamically prefer-
able, the presence of excess dye could destabilize the DNA double
ACKNOWLEDGMENT
The authors thank the donors of the Petroleum Research Fund
(ACS-PRF) and the Whitaker Foundation for financial support of
this research. In addition, the assistance of James Flanagan on
the preparation of the nuclear staining dye and Daryl Williams
for his guidance in preparing the polymer-coated capillary columns
is deeply appreciated. Y.Y.D. thanks the National Institute of
Standards and Technology (NIST) for a predoctoral fellowship.
Received for review September 23, 1996. Accepted
January 6, 1997.X
AC960975V
X Abstract published in Advance ACS Abstracts, February 15, 1997.
Analytical Chemistry, Vol. 69, No. 6, March 15, 1997 1261