ABSTRACT
We present experiments indicating that the SV40 large T-antigen (T-ag) helicase is capable of unwinding the third strand of DNA triple
helices. Intermolecular d(TC)
20
[middot]d(GA)
20
[middot]d(TC)
20
triplexes were generated by annealing, at pH 5.5, a linearized double-stranded plasmid containing a d(TC)
27
[middot]d(GA)
27
tract with a
32
P-labeled oligonucleotide consisting of a d(TC)
20
tract flanked by a sequence of 15 nt at the 3
'
-end. The triplexes remained stable at pH 7.2, as determined by agarose gel
electrophoresis and dimethyl sulfate footprinting. Incubation with the T-ag helicase caused unwinding of the d(TC)
20
tract and consequent release of the oligonucleotide, while the plasmid
molecules remained double-stranded. ATP was required for this reaction and could not be replaced by
the non-hydrolyzable ATP analog AMP-PNP. T-ag did not unwind similar triplexes formed with
oligonucleotides containing a d(TC)
20
tract and a 5
'
flanking sequence or no flanking sequence. These data indicate that unwinding
of DNA triplexes by the T-ag helicase must be preceded by binding of the helicase to a single-stranded 3
'
flanking sequence, then the enzyme migrates in a 3
' ->
5
'
direction, using energy provided by ATP hydrolysis, and causes release of the
third strand. Unwinding of DNA triplexes by helicases may be required for
processes such as DNA replication, transcription, recombination and repair.
Polypurine[middot]polypyrimidine sequences, which are highly dispersed in eukaryotic
genomes (
1
-
8
), may undergo transitions into unusual structures containing DNA triple helices
(
9
-
18
; for recent reviews see
19
,
20
). Formation of these structures could be inhibitory to vital processes which
require DNA strand separation, such as DNA replication and transcription. To
overcome the inhibition eukaryotic cells might produce enzymes that
specifically recognize and unwind DNA triplexes. Alternatively, helicases which
unwind DNA duplexes (
21
-
26
) might also be capable of unwinding the third strand in DNA triplexes.
The latter possibility was suggested by studies recently carried out on
interactions of the SV40 large T-antigen (T-ag) helicase with d(TC)
i
[middot]d(GA)
i
[middot]d(TC)
i
triplexes. We have found that unwinding of DNA substrates containing such
triplexes by the large T-ag was inhibited relative to unwinding of similar `control' substrates
which could not form triplexes (
27
). Notwithstanding this observation, it became apparent that the T-ag helicase may unwind triplexes, because inhibition was not complete,
particularly at neutral pH. However, in the substrates used for these earlier
studies the third strand of the triplexes was covalently linked to one of the
strands in the duplexes. Hence, in those experiments unwinding of the third
strand was not assayed independently of unwinding of the duplexes. For this
reason we were unable to rule out the possibility that unwinding of the third
strand occurred spontaneously, but was only seen when the duplexes were unwound
by the helicase. Others have reported that the bacteriophage T4-encoded dda helicase unwound a different type of DNA triplex (
28
). In those studies too, unwinding of the third strand of the triplex has not
been assayed independently of duplex unwinding (
28
).
To address the question of whether helicases are capable of unwinding the third
strand in DNA triple helices we have prepared substrates containing
intermolecular d(TC)
20
[middot]d(GA)
20
[middot]d(TC)
20
triplexes with a radioactively labeled third strand. Here we report studies
carried out with these substrates in which the SV40 large T-ag was found to possess such a triplex unwinding activity.
The following three oligonucleotides were purchased from Biotechnology General
(Israel) and were used for preparation of helicase substrates: (i) d[(TC)
20
TGACGCTCCGTACGA], designated 3'- tailed oligonucleotide; (ii) d[AGCATGCCTCGCAGT(TC)
20
], designated 5'-tailed oligonucleotide; (iii) d(TC)
20
.
A 2904 bp plasmid was constructed by insertion of a 210 bp rat cell DNA fragment
containing a d(GA)
27
[middot]d(TC)
27
tract into the
Kpn
I site of plasmid pUC18 (
29
). This plasmid, which we have designated pMA73, was digested with either the
single cut restriction enzyme
Nde
I or the single cut restriction enzyme
Xba
I. A sample of 150 ng linearized double-stranded plasmid were annealed with 0.75 ng of either one of the three
oligonucleotides described in the previous section, which had been 5'-end-labeled with
32
P at a specific radioactivity of 2-5 * 10
8
c.p.m./[mu]g (
30
). The molar ratio plasmid/oligonucleotide was 2:1. For some assays the
linearized plasmid was 3'-end-labeled with
32
P as described below. The annealing was performed in 10 [mu]l buffer containing 33 mM Tris-acetate, 66 mM potassium acetate, 100 mM NaCl, 10 mM MgCl
2
and 0.40 mM spermine. The pH of the buffer was 5.5 in all assays, except those
shown in Figure
1
, in which the pH values were as indicated. The samples were incubated at 56oC for 1 h and then slowly cooled to 25oC. Association between the oligonucleotide and the linearized plasmid
duplex was determined by agarose gel electrophoresis, as described in the next
section.
These assays were carried out at 37oC in 20 [mu]l of a solution containing 36 mM Tris-acetate pH 7.2, 73 mM potassium acetate, 10 mM NaCl, 11 mM MgCl
2
, 0.40 mM spermine, 0.50 mM 1,4-dithiothreitol (DTT), 0.05 mg/ml bovine serum albumin, 2 mM ATP, 5%
glycerol, 0.037 ng/[mu]l DNA substrate and 24 ng/[mu]l SV40 large T-ag prepared as previously described (
27
). The reactions were terminated by addition of 5 [mu]l of a solution containing 15% glycerol, 3% sodium dodecyl sulfate, 80 mM Na
2
-EDTA, 8 mM Tris-HCl, pH 8.0, 0.80 mM bromophenol blue, 1 mM xylene cyanol and 3.0 nM
unlabeled oligonucleotide. The mixtures were electrophoresed for 20 h at 4oC in a 1% agarose gel at 30 V. The running buffer consisted of 40 mM Tris-acetate, 5 mM sodium acetate, 1 mM MgCl
2
, pH 5.5. The gels were dried and autoradiographed as described (
27
).
The pMA73 plasmid containing the d(GA)
27
[middot]d(TC)
27
tract was first prepared as described (
31
) and was further purified by CsCl-ethidium bromide equilibrium centrifugation (
30
). The plasmid was digested with the single cut restriction enzymes
Xba
I and
Hin
dIII, whose recognition sites are 24 bp apart. The cleaved DNA was 3'-end-labeled with
32
P (specific radioactivity 4 * 10
6
c.p.m./[mu]g) by filling the 5' staggered
Xba
I cuts using Klenow polymerase, such that the labeled end was located 104 bases
beyond the d(GA)
27
tract. Molecules containing the other labeled strand consisted of 24 bp and did
not interfere with the footprinting assays. An aliquot of 5.7 ng 3'-tailed oligonucleotide was annealed with 250 ng radioactively
labeled duplex plasmid DNA, constituting a 2-fold molar excess of the oligonucleotide. The annealed molecules and
plasmid molecules that had not been annealed with the oligonucleotide were
exposed to dimethyl sulfate (DMS) in a 100 [mu]l solution containing 36 mM Tris-acetate, 73 mM potassium acetate, pH 7.2, 10 mM NaCl, 11 mM MgCl
2
, 0.40 mM spermine, 0.50 mM DTT, 2.0 mM ATP and 2.5 ng/[mu]l DNA. DMS concentrations were either 0, 0.08 or 0.16%. The samples were
incubated for 15 min at 22oC and the reactions were then terminated by addition of [beta]-mercaptoethanol to a final concentration of 0.60 M. The samples
were ethanol precipitated, dissolved and cleaved in 1 M piperidine at 90oC and electrophoresed in a 6% Long Ranger sequencing gel (AT Biochemical),
as described (
32
).
Previous studies have shown that d(TC)
i
[middot]d(GA)
i
[middot]d(TC)
i
triplexes could be generated by annealing linear double-stranded DNA molecules containing d(TC)
n
[middot]d(GA)
n
tracts with oligonucleotides containing d(TC)
i
sequences (
33
,
34
). We used this approach to prepare a DNA substrate that was suitable for our
helicase assays. A long (2904 bp) linearized double-stranded plasmid, designated pMA73, which contained a d(TC)
27
[middot]d(GA)
27
tract was annealed with a short (55 nt)
32
P-labeled oligonucleotide designated 3'-tailed oligonucleotide. This oligonucleotide consisted of a
d(TC)
20
sequence flanked by a sequence of 15 nt at the 3'-end. The molar ratio plasmid:oligonucleotide was 2:1. Figure
1
shows an experiment in which annealing of these two molecules was carried out
at several pH values under the conditions specified in Materials and Methods.
Then the annealed molecules were electrophoresed in a 1% agarose gel at 4oC and pH 5.5 and the gel was autoradiographed. It can be seen that
electrophoresis separated the unbound oligonucleotide from the complex formed
between the oligonucleotide and the linear duplex plasmid, which co-electrophoresed with the plasmid molecules (see Fig.
4
below). No binding of the oligonucleotide to the linear plasmid occurred at pH
8.0 and 7.5. Some binding occurred at pH 7.0. The binding increased as the pH
was lowered to 6.5 and reached a maximal value between pH 6.0 and 5.5.
The pH dependence of the association between the oligonucleotide and the duplex
was a strong indication that this association was due to formation of d(TC)
20
[middot]d(GA)
20
[middot]d(TC)
20
triplexes, in which the d(TC)
20
repeats in the oligonucleotide were bound by Hoogsteen hydrogen bonds to d(TC)
20
[middot]d(GA)
20
repeats in the double-stranded plasmid molecule (
35
); for one of the two Hoogsteen bonds between C residues in the third strand and
G residues in the duplex could only be generated if those C residues were
protonated (
9
-
13
,
33
,
34
). The possibility that association between these two molecules was due to
strand displacement appears unlikely in view of the large difference in the
lengths of the two molecules. Also, strand displacement would not be expected
to exhibit the observed pH dependence. Furthermore, oligonucleotides of
comparable lengths which were homologous to other regions of the plasmid did
not associate with the plasmid molecules (not shown).
Figure
2
shows footprinting assays which more directly demonstrated that the association
between the plasmid and the oligonucleotide molecules was due to formation of
DNA triplexes. In this experiment the plasmid DNA was cut with a restriction
enzyme near the d(TC)
27
[middot]d(GA)
27
tract and the strand containing the d(GA)
27
sequence was end-labeled with
32
P. Samples of this
32
P-labeled plasmid DNA were annealed with the oligonucleotide at pH 5.5 at a
molar ratio plasmid:oligonucleotide of 1:2. The mixtures were then brought to
pH 7.2, the pH at which helicase assays were performed (see below) and exposed
to two different concentrations of DMS. Other samples of the linearized plasmid
duplex to which no oligonucleotide was added were similarly treated. Finally,
these samples and samples that were not exposed to DMS were treated with
piperidine and electrophoresed in a sequencing gel (
36
). It can be seen that binding of the oligonucleotide protected the N7 atoms of
25 out of 27 G residues in the d(GA)
27
repeats against methylation by DMS (
36
). This result indicates that the 25 G residues were bound to C residues of the
d(TC)
20
repeats in the oligonucleotide by Hoogsteen hydrogen bonds. In view of the
excess oligonucleotide in these samples the protection of 25, rather than 20, G
residues may have been due to binding of two oligonucleotide molecules to one
plasmid duplex. Alternatively, this result could have been due to slippage of
the d(TC)
20
repeats along the d(TC)
27
[middot]d(GA)
27
tract and a consequent statistical distribution of the d(TC)
20
repeats in the oligonucleotide over the d(TC)
27
[middot]d(GA)
27
tract within the duplex DNA.
Figure
3
shows an experiment in which substrate molecules containing d(TC)
20
[middot]d(GA)
20
[middot]d(TC)
20
triplexes were prepared at pH 5.5, as shown in Figure
1
. Then the pH was raised to 7.2 and the samples were incubated with the SV40
large T-ag at 37oC for various time periods in the presence of ATP (see below).
Following incubation the samples were electrophoresed in an agarose gel. A
sample that was similarly incubated for 30 min in the absence of large T-ag and a sample that was heated for 5 min at 90oC were also electrophoresed in the same gel. It can be seen that in
this experiment: (i) all detectable oligonucleotide molecules formed a complex
with the linearized plasmid molecules; (ii) no oligonucleotide molecules were
released from the complex in the sample incubated for 30 min at 37oC in the absence of T-ag; (iii) a time-dependent release of oligonucleotide was observed in samples
that were incubated with T-ag. Release was apparently due to unwinding of the d(TC)
20
[middot]d(GA)
20
[middot]d(TC)
20
triplexes, which accounted for association between the two molecules.
Microdensitometric analysis of these data (not shown) revealed that the extent
of oligonucleotide released increased linearly during the period of the
experiment, such that after 30 min 70% of the triplexes were unwound. It can
also be seen that all oligonucleotide molecules were released by heating the
complex to 90oC (lane marked den.).
Since the SV40 large T-ag helicase has been reported to unwind long duplex molecules (
37
), it was interesting to find out whether at the end of the triplex unwinding
reaction the linearized double-stranded plasmid DNA molecules were also unwound. For this purpose DNA
triplexes were generated by annealing unlabeled or
32
P-labeled linearized plasmid molecules with unlabeled or
32
P-labeled 3'-tailed
oligonucleotide. These complexes and
32
P-labeled plasmid molecules which had not been annealed with the
oligonucleotide were incubated with or without large T-ag and were electrophoresed in an agarose gel, as described in the
previous sections. Figure
4
shows the data obtained in these assays. Clearly, electrophoresis resolved
denatured single-stranded plasmid molecules from the native double-stranded plasmid molecules (which co-electrophoresed with the complex formed between the plasmid
and the oligonucleotide) and from the released oligonucleotide. Inspection of
these data reveals that unwinding of the d(TC)
20
[middot]d (GA)
20
[middot]d (TC)
20
triplexes by large T-ag helicase resulted in release of the oligonucleotide, while the long
linear duplex molecules remained double-stranded.
To find out whether release of the oligonucleotide from the complex with the
linearized plasmid was caused by T-ag helicase activity and not by a different mechanism we sought to
determine whether certain parameters of the reaction were compatible with a
helicase mechanism. Helicases, including the SV40 large T-ag, require ATP as an energy source for migration along single-stranded DNA and for unwinding duplex DNA (
21
,
24
-
26
,
38
). Therefore, we tested the requirement for ATP for release of the
oligonucleotide from the complex. As Figure
5
shows, omission of ATP eliminated release of the oligonucleotide by the
helicase. Inclusion of the non-hydrolysable ATP analog AMP-PNP in the reaction, instead of ATP, did not restore the activity.
These results indicate that ATP hydrolysis is required for unwinding of the
third strand of the triplex by T-ag helicase. It should also be noted that in this experiment association
of the
32
P-labeled oligonucleotide with the linearized plasmid also generated, in
addition to complexes co-migrating with the plasmid molecules, slower migrating complexes. These
species, which were also occasionally seen in other assays (see Fig.
6
below), might represent aggregates including two or more plasmid molecules
bound to one oligonucleotide molecule.
The data presented in this paper indicate that SV40 large T-ag helicase is capable of unwinding the third strand of d(TC)
i
[middot]d(GA)
i
[middot]d(TC)
i
triplexes. The requirements for this activity of T-ag helicase are apparently similar to the previously established
requirements for unwinding of double-stranded sequences within DNA molecules lacking an SV40 origin of DNA
replication. To be efficiently unwound by T-ag double-stranded DNA in such molecules must be flanked by single-stranded DNA containing a 3'-end (
37
,
39
). Similarly, unwinding of the third strand of the d(TC)
20
[middot]d(GA)
20
[middot]d(TC)
20
triplexes occurred in the presence of a single-stranded 3'-tail attached to the third strand, but not in the presence of
a 5'-tail or in the absence of a single-stranded tail. In addition, ATP was found to be required for
unwinding of the third strand and could not be replaced by the non-hydrolysable ATP analog AMP-PNP. Furthermore, since AMP-PNP stimulates binding of T-ag to single-stranded DNA to a similar extent to ATP (
40
), it is clear that mere binding of T-ag to the 3'-tail of the third strand does not cause unwinding of this
strand. Thus the triplex unwinding reaction is apparently coupled to ATP
hydrolysis, like duplex unwinding by T-ag helicase (
40
). We have not yet carried out detailed studies of the efficiencies of triplex
unwinding versus duplex unwinding by T-ag helicase. Nevertheless, a comparison of the present data with previous
data on duplex unwinding (
27
) indicates that the concentrations of T-ag needed for unwinding a duplex substrate and the third strand in a
triplex substrate are within the same order of magnitude.
Figure
7
presents a graphic illustration of the triplex unwinding reaction suggested by
the data discussed above. Apparently T-ag first binds the 3'-tail of the oligonucleotide in the complex with the linear
duplex molecules and then migrates in a 3' -> 5' direction, using energy provided by ATP hydrolysis. As it
encounters the triplex it causes unwinding of the third strand, such that at
the end of the reaction the oligonucleotide is released and the duplex remains
intact. Unwinding of the third strand by T-ag helicase indicates that the enzyme is not only capable of disrupting
Watson-Crick hydrogen bonds, but may also cause disruption of Hoogsteen hydrogen
bonds. This inference has gained further support from more recent experiments
which indicated that T-ag helicase can also disrupt Hoogsteen hydrogen bonds in DNA quadruplexes
which, unlike DNA triplexes, do not contain any Watson-Crick hydrogen bonds (Pukshansky, Baran and Manor, unpublished data).
We thank Dr J. Borowiec for helpful comments. This research was supported by a
grant from GIF, the German-Israeli Foundation for Scientific Research and Development, by grant 93-00156 from the United States-Israel Binational Science Foundation and by a project grant
from the Israel Cancer Research Fund.
REFERENCES
Return