ABSTRACT
Photocleavage of dsDNA by the fluorescent DNA stains oxazole yellow (YO), its
dimer (YOYO) and the dimer TOTO of thiazole orange (TO) has been investigated as a function of binding ratio. On visible illumination, both YO
and YOYO cause single-strand cleavage, with an efficiency that varies with the dye/DNA binding
ratio in a manner which can be rationalized in terms of free dye being an
inefficient photocleavage reagent and externally bound dye being more efficient
than intercalated dye. Moreover, the photocleavage mechanism changes with
binding mode. Photocleavage by externally bound dye is, at least partly, oxygen
dependent with scavenger studies implicating singlet oxygen as the activated
oxygen intermediate. Photocleavage by intercalated dye is essentially oxygen-independent but can be inhibitied by moderate concentrations of
[beta]
-mercaptoethanol-direct attack on the phosphoribose backbone is a possible
mechanism. TOTO causes single-strand cleavage approximately five times less efficiently than YOYO. No
direct double-strand breaks (dsb) are detected with YO or YOYO, but in both cases single-strand breaks (ssb) are observed to accumulate to eventually produce
double-strand cleavage. With intercalated YO the accumulation occurs in a manner consistent with random generation of strand lesions, while with
bisintercalated YOYO the yield of double-strand cleavage (per ssb) is 5-fold higher. A contributing factor is the slow dissociation of the bis-intercalated dimer, which allows for repeated strand-attack at the same binding site, but the observation
that the dsb/ssb yield is considerably lower for externally bound than for bis-intercalated YOYO at low dye/DNA ratios indicates that the binding
geometry and/or the cleavage mechanism are also important for the high dsb-efficiency. In fact, double-strand cleavage yields with bis-intercalated YOYO are higher than those predicted by simple models,
implying a greater than statistical probability for a second cleavage event to
occur adjacent to the first (i.e. to be induced by the same YOYO molecule).
With TOTO the efficiency of the ssb-accumulation is comparable to that observed with YOYO.
The dimers (YOYO and TOTO, Fig.
1
) of the asymmetric cyanine dyes oxazole yellow (YO) and thiazole orange (TO)
bind strongly to double-stranded DNA (dsDNA), and the fluorescence quantum yields of the bound
dyes are typically 1000-fold higher than when free in solution (
1
). As a result, the background fluorescence from free dye is extremely low which
makes these dyes excellent probes for high sensitivity quantification of DNA (
2
), and for imaging of individual DNA molecules (
3
-
5
). YOYO can also be used for imaging of single-stranded DNA (ssDNA) (
6
).
The use of these dimeric dyes for such applications is not without problems,
however. First, upon mixing, the strong binding results in a long-lived bimodal distribution of dye molecules between the DNA molecules in
the sample (
7
). Such inhomogeneous staining may produce serious artefacts in fluorescence-based DNA quantification, since the quantum yield is strongly dependent on
binding ratio (
8
). This dye distribution problem can be overcome, however, by a simple
equilibration protocol (
7
). A potentially more serious side-effect in many applications is the tendency of these dyes to photocleave
the stained DNA, especially under the intense illumination used for imaging of
individual molecules by fluorescence microscopy. By trial and error it has been
found that the cleavage problem can be minimized by exclusion of oxygen, either
by purging with argon (
5
) or by enzymatic methods (
3
), and usually [beta]-mercaptoethanol is also added, as is common in fluorescence
microscopy in order to minimize dye bleaching (
9
). However, the mechanisms of photocleavage by YOYO and TOTO and the manner in
which these agents provide protection are not known.
Supercoiled and nicked circular form [Phi]X174 DNA were from Pharmacia and New England Biolabs, respectively. The
negatively supercoiled sample contained 5-10% nicked circles. Linear form was produced by cleavage of a supercoiled sample with restriction enzyme
Xho
I (Pharmacia). YO-PRO-1, YOYO-1 and TOTO-1 (referred to here as YO, YOYO and TOTO) were from
Molecular Probes, (Eugene, OR, USA) and were used as received. Methylene blue (puriss) was
from Fluka, and was purified by column chromatography (
13
). Concentrations were determined by optical absorption, using the following exctinction coefficients: [epsilon](260 nm) = 6600 M
-1
cm
-1
for DNA bases (
14
), [epsilon](481 nm) = 66 000 M
-1
cm
-1
for YO, [epsilon](457 nm) = 96 100 M
-1
cm
-1
for YOYO, [epsilon](480 nm) = 97 900 M
-1
cm
-1
for TOTO
(
15
) and [epsilon](664 nm) = 81 600 M
-1
cm
-1
for methylene blue (
16
). Unless otherwise stated, the dye loading is given as the N/C ratio between
concentrations of DNA nucleotide and added dye chromophores (one for monomer
and two for dimer).
All samples contained 40 [mu]M DNA nucleotide and a dye concentration corresponding to the stated N/C,
except for the lowest N/C of 1.5 (of YO or YOYO) which contained 20 [mu]M DNA nucleotide. The lower DNA concentration was chosen to keep the maximum
absorption of the dye in the samples (at 480 nm) below ~0.1. All samples had a final total volume of 20 [mu]l in 50 mM TBE-buffer (pH = 8.3), and were prepared by adding 5 [mu]l DNA stock to the dye diluted in the buffer, and
subsequently incubating at 50oC for 2 h to reach an equilibrium in the dye distribution (
7
). For deoxygenation, 100 [mu]l (i.e. five aliquots) was bubbled vigorously with water-saturated argon for 1 min. The efficiency of this degassing procedure
was confirmed by the observation of a 64% reduction compared to air-saturated solution of the ssb induced in dsDNA by methylene blue (N/C = 15
and 50 mM ionic strength) (
17
). The rate of ssb formation with YO at different N/C was also investigated in
the presence of 15-50 mM concentrations of NaN
3
, in D
2
O, urea, thiourea or glycerol, and in the presence of 0.5-2.5 10
3
U/ml of superoxide dismutase and/or catalase.
Illumination was performed using a 150 W xenon lamp, a lens and a long-pass (320 nm) filter, with the sample forming a droplet (maximum path
length 0.1 cm) in the closed end of a horizontal Eppendorf tube sealed with a
glass lid. In the absence of dye no DNA cleavage was detected during the
longest illumination time used in this study (45 min). The stability of the
lamp over a 7 h period was confirmed by a +-3% variation in the relative concentrations of nicked and supercoiled
forms obtained by illumination of an air-saturated sample containing YOYO at N/C = 3 for 5 s.
The main experimental parameter in this study was the dye binding ratio. For
each binding ratio the time profiles for the conversion of the supercoiled
species (form I) into nicked circles (form II), linear DNA (form III), and
linear fragments (form IV) were determined. As the N/C was increased the
reaction rates decreased, and the range of illumination times therefore had to
be increased in order to monitor the formation of all DNA forms. At the highest
N/C of 40 some illuminations extended over 15 min, and in those cases
evaporated solvent that had condensed on the walls of the tube was recovered by
spinning the samples every 15 min in a bench-top microcentrifuge for 2 min.
Photocleavage products were analysed on 1% agarose gels (in cleavage buffer) at
3 V/cm for 4.5 h. The positions of the zones of the separated products in each
lane were determined from the fluorescence intensity profiles of YO (or YOYO)
remaining bound, measured by scanning the gel on the stage of a Zeiss Axioplan epifluorescence microscope (excitation filter: 450-490 nm, dichroic mirror: 510 nm, emission long-pass filter: 520 nm, 5* objective, numerical aperture = 0.15), equipped with a
Hamamatsu photomultiplier. The epifluorescence illumination provided excellent
signal-to-noise ratio, even for the highest N/C value (= 40), because of the
very low quantum yield of fluorescence for free compared with bound dye (
1
). As evidenced by comparison with non-illuminated samples stained at the same N/C, the order of migration was
always nicked circle, linear and supercoil forms. The DNA forms were baseline
separated except for a slight overlap of the zones of the supercoiled and
nicked circles at binding ratios close to N/C = 12 (due to unwinding effects from the
intercalated dyes), which were resolved on the basis of the concentration
profiles of the pure DNA forms. The zone of the intact linear form (III) and
the tailing smear of linear fragments (form IV) were resolved on the basis of
the observation that zones of pure form III are symmetric.
The quantity of each form of DNA was obtained by integrating the scanned
intensity over each zone. Repetitive scans along the same lane showed that no
detectable bleaching occured as a result of the scanning. Two different
approaches were used for staining the DNA. Mainly we have exploited the
intrinsic fluorescence of the YO or YOYO that remains bound after
electrophoresis, in order to make use of the very good signal-to-noise ratio for those dyes. This protocol also avoids the procedures
of removing these strongly bound dyes and subsequently purifying and restaining
the DNA, steps which could introduce artificial ssb, cleavage at alkali-labile sites or other damage. Some bound dye was probably lost during
electrophoresis, but the quantification of ssb and dsb cleavage rates only
requires knowledge of the relative concentrations of the different DNA forms
(see eq.
2
below), which were obtained by normalizing with respect to the total intensity
of the pertinent lane. For any experiment (i.e. at a certain N/C), as long as
the fraction of dye lost is the same for the different DNA forms, the relative
concentrations and the evaluated rates should not be affected. As with
conventional post-electrophoresis staining, corrections for the differential binding to
supercoiled and nicked circular/linear DNA have to be made if an intercalating
dye is employed. Prior to illumination the dye could be considered bound to the
supercoiled form at the nominal N/C, since the contamination of relaxed circles
in the supercoiled sample could be neglected. However, since large amounts of
relaxed circles were produced in the course of the cleavage experiments, dye
may have transferred between the newly created nicked circles and molecules
that remained supercoiled, before the two forms were electrophoretically separated. (No dye transfer will occur between nicked
circular and linear DNA since the affinity for an intercalator is the same.)
This possibility was tested by removal of the dye and post-electrophoresis staining with the groove-binder DAPI, the affinity of which is negligibly affected by DNA
topology. To keep DNA handling simple and gentle, destaining was performed
after the illuminated samples had been migrated a few millimeters into the gel
(at 3 V/cm for 0.5 h). The gel (150 ml) was soaked for 5 h in 1000 ml of 0.1 M
NaCl containing 10 mM SDS (CMC under these conditions is 2 mM), a step which removed all the YO or YOYO from the DNA, as evidenced by no detectable fluorescence from
the zones. The gel was washed twice in electrophoresis buffer for 4 h,
electrophoresis was resumed for 4 h at 3 V/cm, and finally the gel was soaked in 500 ml of 10 [mu]M DAPI in electrophoresis buffer for 4 h and washed in electrophoresis
buffer for 1 h. The DNA zones were again detectable, by the blue-white
fluorescence characteristic of DAPI bound to DNA, and the gels were scanned
using a 365 nm excitation band-pass filter, a 395 nm dichroic mirror and a 420 nm emission long-pass filter. [The fluorescence is from groove-bound DAPI, because of the quenching of the fluorescence of
the GC-intercalated DAPI (
20
), present in a minor fraction at the staining conditions employed;
18
,
19
.] The separation was better than with YO, where unwinding slows down and
broadens the zone of the supercoiled form compared with form II, but the S/N
ratio is considerably lower with the DAPI protocol. Therefore DAPI was used for
quantification only to investigate effects of differential staining by YO or
YOYO on apparent cleavage kinetics, by illuminating two sets of samples and
analysing them with both methods.
The consumption of form I DNA was analysed in terms of single exponential
kinetics with a time constant [tau]. The actual absorbance of each sample was not measured, so absolute
quantum yields have not been obtained. The rate of nicking per DNA molecule (1/
[tau]) was normalized with respect to the number of absorbed photons by dividing
by the number of dye molecules per DNA molecule
k
1
= (1/[tau]) * (N/C)
(
1
)
where
k
1
is the relative nicking efficiency per YO chromophore (in units of s
-1
) whether in a monomer or a dimer. Since cleavage rates are normalized with
respect to added chromophore,
k
1
is the average over intercalated, externally bound and free dye, which have
slightly different absorption cross-sections. Except at the highest N/C values used, where all dye
chromophores can be considered intercalated, there are potential effects on the
relative rates due to this simplification which are discussed further in the
main text.
The average numbers of ssb (
n
1
) and dsb (
n
2
) per DNA molecule were calculated from the equations of Povirk and co-workers (
21
,
22
)
.
If the supercoiled form is consumed by both ssb and dsb which are randomly
distributed among the DNA molecules, the fraction
f
I
of surviving supercoiled circles is related to the total number of breaks,
n
tot
=
n
1
+
n
2
,
via
n
tot
= -ln
f
I
(
2a
)
However, we will argue that in the case of YO and YOYO form I is consumed only
by ssb since no true dsb (i.e. as a result of a single absorption event)
occurs. In such a case
f
I
is given by
n
1
= -ln
f
I
(
2b
)
If the supercoiled fraction
f
I
decreases monoexponentially with time, eq. (
2b
) implies the number of ssb increases linearly in time.
For a random distribution of dsb over all the DNA molecules, the total number of
dsb can be calculated from the fraction
f
III
of the molecules which are in the linear form
f
III
=
n
2
exp(-
n
2
)
(
2c
)
which can be expressed in a more useful form (
22
)
n2 = 1 over
{ (fI + fI I + fI I I) / fI I I - 1}
(
2d
)
where
f
II
is the fraction of the molecules which are found as form II. The presented
values of
n
1
and
n
2
are based on data comprising <6% of form IV. In the case of YO at N/C = 40 it is only over a very limited
range of illumination times that forms I and III occur simultaneously, which is
necessary for eq.
2b
and
2d
to be applicable to the same sample. In this case the number of ssb for longer
times is estimated by extrapolating the average number of ssb per DNA molecule
linearly in time, since the supercoil nicking rates were mono-exponential.
The results of a typical cleavage experiment, in this case with YOYO at N/C = 3,
are presented in Figure
2
. The overall band pattern for the illumination time course (Fig.
2
a) is that expected if the first cleavage event is a single-strand break. Initially the amount of nicked circles (II) increases, at
the expense of the supercoil. With longer illuminations, II is consumed by
accumulation of ssb producing dsb, which results in the appearance and growth of an
intermediately positioned zone of linear molecules (III). Ultimately these
linear molecules are analogously degraded by accumulated ssb to form linear
fragments, manifest as a smear of weak intensity extending from the band of
intact linear molecules in the direction of migration. After normalization with
respect to the total lane intensity (Fig.
2
b), the relative concentrations of each component can be plotted versus
illumination time (Fig.
2
c). The error bars demonstrate that the variation between experiments is within
5% except when there is an appreciable fraction of fragmented linear molecules
(IV), the quantification of which requires very accurate baseline
determinations since the intensity is dispersed over a long distance on the
gel. All quantification of ssb and dsb has therefore been limited to data
containing <6% of form IV. With YO at N/C = 3 (results not shown) the product profiles of
all DNA forms are very similar to those observed with YOYO. Also TOTO at N/C =
3 exhibits product profiles similar to YOYO but cleaves about five times slower
than YOYO. This is exemplified by the maximum concentration of form II
occurring much later (125 s) than with YOYO (30 s, Fig.
2
c), but with similar amounts of nicked circular (85%) and linear DNA (10%) at
this point.
Nicking of the supercoil exhibited monoexponential kinetics in most cases (Fig.
4
). However, with YO at N/C = 40 (Fig.
4
, inset) an apparent lag in the disappearance of supercoiled circles gave rise
to non-monoexponential kinetics if YO was used for quantification (open symbol).
By contrast, mono-exponential kinetics (and no lag) were observed if the DAPI-protocol was employed for the same sample (closed symbols). With the
YO-protocol (but not with DAPI post-staining) an apparent lag is expected if YO is transferred from form II to form I in the period between
illumination and electrophoretic separation of the two components. This can
occur with YO because the negatively supercoiled DNA binds an intercalator more
strongly than the nicked circle, if the N/C is higher than the equivalence
binding ratio needed for complete removal of the supercoiled turns of the DNA (
23
). The equivalence point of our DNA sample is N/C = 10-11 for YO because the mobility of the supercoiled form has a minimum at ~0.09-0.1 YO per nucleotide, at which point the mobility is close
to that for the nicked circle (data not shown). The apparent lag was pronounced
only at N/C = 40, and as expected from the proposed mechanism disappeared
completely at N/C = 12 which is close to the equivalence point of our sample.
Below N/C = 10-11 transfer in the opposite direction should occur, but the semi-logarithmic plots remained linear, presumably since the driving
force for transfer becomes smaller as saturation of intercalation sites is
approached, and since externally bound dye present at low N/C should exhibit no
DNA-topology dependence of binding.
The apparent fraction of form I was lower with DAPI by a magnitude which was
reasonable in view of an estimated (
23
) ratio of 1.5 for the binding constants of YO at N/C = 40 (assuming an
unwinding angle of 20o) (
10
), which in equilibrium corresponds to a transfer of 10% of dye from II to I when half of the circles have been
converted. For YO such a transfer is likely to have enough time to occur before
the DNA forms are separated by electrophoresis ( >= 10 min), since the dissociation time constant of the similar intercalator
propidium (divalent, with a side chain similar to that of YO-PRO) is typically in the range of seconds or faster (
24
). With bis-intercalating YOYO (which did not exhibit a lag) the off-rate is on the order of hours (
7
,
25
), and control experiments on a mixture of stained and non-stained DNA indeed showed that no detectable transfer of YOYO occurred on
the time scale of the electrophoretic analysis (not shown).
The effect of binding mode on the single-strand cleavage was investigated by evaluating the rate constant (per
chromophore) for supercoil nicking (eq.
1
) at different binding ratios of nucleotide to added dye chromophore (Fig.
5
). At high N/C, YO (top) and YOYO (bottom) exhibit a nicking efficiency that is
essentially the same per chromophore for the monomer and the dimer. As N/C
decreases the rate increases gradually with YO, and goes through a maximum
around N/YO of 3, from which it falls rapidly for the lowest N/C. By contrast,
with YOYO (bottom) the nicking rate decreases gradually as N/C decreases from
40 down to N/C = 10, where from a shallow minimum, the rate grows rapidly to a
value per chromophore almost twice as large as the maximum level observed with
YO. No decrease is observed with YOYO even at the lowest N/C. It is clear that
the photo-cleavage by YO and YOYO cannot be explained in terms of a single process
for one binding mode, since the rate constants for supercoil nicking vary
strongly with the binding ratio.
The binding of YO and YOYO to double-stranded DNA has been studied (
8
,
10
) under the the same buffer and DNA conditions as employed here, except T2 DNA
was used. The population of the different binding modes should be very similar
with [Phi]X174 DNA, since YO exhibits very little sequence dependence of binding (
8
). Many of the results presented here can be understood in terms of YO and YOYO having two binding modes: (i) at high N/C, intercalated (bis-intercalated for YOYO) and strongly fluorescent, and (ii) at low N/C,
externally bound and more weakly fluorescent (
8
,
10
). Being structurally very similar to YOYO (Fig.
1
) a similar binding behaviour for TOTO is indicated by very similar spectral
changes in absorption and fluorescence (
1
) upon binding to dsDNA, and NMR studies (
28
) support bis-intercalation as the most stable binding mode.
With YOYO at N/C = 40 form III appears when 50% of the supercoil has been
converted (Fig.
3
). By contrast, with YO the linear form appears only when >95% of the
supercoiled form has been consumed, which is expected for random accumulation
of ssb created by a true single strand cleaving agent (
29
). We will return below to the enhanced efficiency of linearization with the dimer, but at
this stage we simply demonstrate that YOYO is not a true double-strand cleaving agent, in the sense that a dsb is not created as a result
of a single excitation event. The number of dsb per DNA molecule (Fig.
6
) grows quadratically with time (normalized to binding ratio by multiplication
by N/C), both for YOYO and YO. A linear growth in the number of dsb with time
is expected for true dsb (
21
,
30
), but inclusion of a linear term did not improve the quality of the fits. Thus,
the data for the photocleavage of DNA by YOYO and YO can be analysed assuming
that the first damaging event is always nicking of one strand.
Bound dye cleaves more efficiently than free
. The drop of ssb activity for YO below N/C = 3 (Fig.
5
, top) indicates that the free dye does not cause photocleavage, since
spectroscopic studies (
10
) show that free YO is present at these binding ratios. This interpretation is
further supported by the observation that release of bound dye by addition of
Mg
2+
decreases the photo-cleavage yield (results not shown). By contrast, with YOYO no decrease in
supercoil-cleavage efficiency is observed even at very low N/C (Fig.
5
, bottom). No data are available for the amount of free dye at N/C = 1.5, but no
free dye could be detected at N/C = 2.5 (
10
), where a significant fraction of the more weakly bound YO was free. We therefore
suggest that the absence of a decrease in ssb activity for YOYO at low N/C is
due to a low free dye concentration as a result of the strong dimer binding.
Attempts to create free dye by increasing the YOYO concentration even further,
resulted in aggregation of the DNA.
An apparent decrease in cleavage rate at low N/C ratios (e.g. 1.5 with YO) could
potentially arise from strong light absorption at the comparatively high dye
concentrations. However, the cleavage rate increases with YOYO but decreases
with YO when N/C is reduced from 3 to 1.5, even though the dye absorption is
similar in all four samples, so this is not a dominant effect.
Externally bound dye cleaves more efficiently than intercalated
. The decrease of the average cleavage rate per YO above N/C = 3, towards an
approximately constant level at high N/C (Fig.
5
, top) indicates that externally bound YO, present at low N/C, induces ssb more
efficiently than intercalated dye, which is the only bound form at high N/C.
For YOYO (Fig.
5
, bottom), the supercoil conversion rate is also considerably higher when
externally bound dye is present (N/C < 4) than when all dye molecules are bis-intercalated (N/C >> 4), consistent with intercalated chromophores being
less efficient at creating ssb also as dimers.
There are clear differences in the behaviour of YO and YOYO, however. YOYO
exhibits a shallow minimum in the cleavage rate (at N/C = 6), before the rapid
increase at lower N/C. By contrast, with YO there is a monotonous and more
gradual increase in cleavage rate as N/C is decreased. We tentatively ascribe these differences in
the cleavage profiles to YO and YOYO being differently partitioned between the
two binding modes at intermediate N/C values. With YOYO, intercalation sites
are essentially saturated (at N/C = 4) before external binding occurs (
10
), which explains the abrupt increase in cleavage efficiency for the dimer below
N/C < 4. For YO, the two binding modes overlap (
10
) and the transition to higher average cleavage rates due to the presence of
externally bound dye therefore occurs more gradually. The question, why with
YOYO there is an increase in the cleavage efficiency with increasing N/C above
N/C = 4, although all molecules should be in the same bis-intercalated binding mode, is returned to after discussion of the results
of mechanistic studies.
Mechanistic studies of the rate of ssb by YOYO.
As a first step towards understanding the chemistry underlying the cleavage, we
have investigated some key elements of the reaction mechanism, including the role of binding geometry. The fact that deoxygenation had a
stronger inhibiting effect on cleavage at N/C = 1.5 (both intercalated and
externally bound dye) than at N/C = 10 (intercalated dye only), shows that O
2
plays a more important role in cleavage by externally bound than by
intercalated dye. This observation suggests different cleavage pathways for the
different YOYO binding geometries, and it is thus likely that the large
difference in the rates of ssb formation in the two binding modes is due to the
existence of two different cleavage mechanisms, rather than only one mechanism
which is less efficient for the intercalated mode.
The accumulated evidence from the quencher experiments with NaN
3
, D
2
O, catalase and superoxide dismutase suggests that while externally bound dye
effects cleavage at least in part via photosensitized
1
O
2
production, intercalated dye reacts primarily via a pathway that does not
involve an activated oxygen species. (Some quencher experiments were performed
with YO rather than YOYO, but it is reasonable to assume that the YO chromophore cleaves by the same
mechanism in the monomer and the dimer for a given binding mode.) Lack of
oxygen involvement in cleavage by intercalated dye is not surprising, since it
is well known that the intercalation pocket usually protects fluorescent dyes
from quenching agents in the solution, such as dissolved oxygen (
31
). No evidence was found for involvement of hydroxyl radical species (either at high or low N/C), which suggests that intercalated dye reacts
directly with the DNA, perhaps
via hydrogen-abstraction from the sugar, but more detailed studies of the cleavage
mechanism by the intercalated YO chromophore are needed to clarify this point.
Cleavage efficiency at intermediate N/C of intercalated chromophore.
With YOYO there was an increase in cleavage efficiency with increasing N/C
(>4), which was surprising because in this range the dimer is reported to bind
only by bis-intercalation (
10
). YOYO may bind in a third mode as monointercalated dimer, which is expected to
occur when the intercalation sites are almost saturated
.
(
32
) i.e. just above N/C = 4. However, this seems an unlikely explanation for the
partial quenching of the cleavage in this range of binding ratios, since
externally bound YO-chromophores cleave more efficiently than intercalated ones.
Variations with N/C of the quantum yields for fluorescence of fully intercalated
dye have been reported for both YO and YOYO (
10
).
These observations suggest interactions between intercalated chromophores that
might also modulate other photophysical properties of the DNA-dye complex, such as cleavage efficiency. A notable similarity is that
for both fluorescence quantum yield (
8
) and the cleavage rate (Fig.
5
), the same value per chromophore is obtained for YO and YOYO, when measured at
such high N/C (= 40) that interactions between different dye molecules should
be negligible. Hence, the chromophores seem to act independently when
constrained in a dimer, which apparently contradicts the argument that
interactions between dyes are responsible for the variation of fluorescence
quantum yield and cleavage rates with N/C, i.e. the effect of interactions
between intercalated YO chromophores in close proximity would be expected to be
permanently manifest in YOYO.
To explain the variation of the fluorescence quantum yield with N/C, Larsson
et al
. (
10
) have suggested that the dye molecules interact either directly, by electronic
coupling of the chromophores, or indirectly, through alterations in the DNA
helix due to unwinding. In solution the absorption spectrum of YOYO is
different from that of YO, most likely because of electronic interactions
between stacked chromophores in the dimer (
33
). Bis-intercalation of YOYO with DNA ruptures chromophore-chromophore stacking, which essentially removes the electronic
interactions between the chromophores as evidenced by the YO and YOYO
absorption spectra being almost identical when the dyes are intercalated (
10
). This suggests that the interactions between intercalated chromophores are not
electronically coupled but mediated by a helix perturbation. Possibly, the
demands of accommodating the YOYO-linker by the extended and unwound DNA result in less pronounced perturbation of the helix between two YO
chromophores within a dimer than between adjacent monomers or dimers. Hence,
helix-mediated interactions within isolated dimers may be reduced compared with
interactions between neighbouring dimers or monomers, consistent with
fluorescence, cleavage and absorption observations.
At intermediate N/C, where interactions between different intercalated dye
molecules are observed to occur (
8
), there is less consistency between fluorescence quantum yield and cleavage. It
is only with YOYO that we observe an increase in cleavage rate with increasing
N/C above N/C = 4 (Fig.
5
), and in fluorescence it is instead the quantum yield of YO which increases
most strongly with increasing N/C (
8
). However, direct comparison is not trivial since with YO there is overlapping
of binding modes, and further studies of the excited state properties of YO and
YOYO in both binding modes are clearly needed in order to understand the
effects of N/C on single-strand cleavage.
Enhanced double-strand cleavage with the dimer.
Since with both YO and YOYO dsb occur as a consequence of ssb-accumulation, comparison of the dsb rates must take the nicking
efficiencies of YO and YOYO into account. Figure
5
shows that the rates of ssb formation per chromophore by YO and YOYO are indeed
very similar at N/C = 40 (0.35 s
-1
), and it is thus clear from Figure
3
that the dimer induces dsb considerably more efficiently (per chromophore) than
the monomer. This effect is better demonstrated by a plot of the number of dsb versus the number of ssb (Fig.
7
, inset), from which it is clear that at N/C = 40 the number of dsb induced per
ssb is much higher with YOYO than YO.
Both YO and YOYO cleave DNA by formation of ssb.
Free dye causes negligible cleavage of DNA, while both intercalated and
externally bound dye generate ssb, but by different cleavage mechanisms.
Externally bound dye cleaves comparatively more efficiently, in an oxygen-dependent manner.
Intercalated dye cleaves less efficiently, and by an essentially oxygen-insensitive pathway.
YOYO does not cause direct dsb, but the dsb-yield is still in clear excess of that expected (and observed with the
monomer YO) from accumulation of randomly distributed ssb, because the coupling
of the chromophores in YOYO leads to an enhanced accumulation of ssb compared
to the monomer.
Models for permanently bound and isolated dimers of an ssb-inducing agent predict an initially quadratic variation of the number of
dsb with the number of ssb, in agreement with observations for YOYO, but the
predicted dsb yields are considerably lower than those observed even if
repetitive cleavage by each chromophore is allowed.
YOYO gives higher yields of dsb/ssb at high N/C than at low N/C, which indicates
that bis-intercalation is important for realization of the dimer's potential for
double-strand cleavage.
The Magn. Bergwall (B.Å.) and Carl Trygger (E.T.) foundations are thanked for support.
REFERENCES
Return