A new approach to overcome potassium-mediated inhibition of triplex formation
A new approach to overcome potassium-mediated inhibition of triplex formation
Fedor
Svinarchuk
1,3,
*
,
Dmitry
Cherny
2,4
,
Arnaud
Debin
1
,
Etienne
Delain
2
and
Claude
Malvy
1
1
Laboratoire de Biochimie-Enzymologie and
2
Laboratoire Microscopie Cellulaire et Moleculaire, CNRS URA 147, Institute
Gustave Roussy, rue Camille Desmoulins, 94805
Villejuif
Cedex,
France
,
3
Department of Biochemistry, Novosibirsk Institute of Bioorganic Chemistry, 8
Prospect Lavrenteva,
Novosibirsk
630090,
Russia
and
4
Institute of Molecular Genetics, RAS, Kurchatov's sq.,
Moscow
123182,
Russia
Received April 16, 1996;
Revised and Accepted August 9, 1996
ABSTRACT
G,A-containing purine oligonucleotides of various lengths form extremely
stable and specific triplexes with the purine-pyrimidine stretch of the
vpx
gene [Svinarchuk,F., Monnot,M., Merle,A., Malvy,C. and Fermandjian,S. (1995)
Nucleic Acids Res
., 22, 3742-3747]. The potential application of triple-helix- forming oligonucleotides (TFO) in gene-targeted therapy has prompted us to study triplex
formation mimicking potassium concentrations and temperatures in cells. Triplex
formation was tested by dimethyl sulphate (DMS) footprinting, gel-retardation, UV melting studies and electron microscopy. In the presence
of 10 mM MgCl
2
, KCl concentrations up to 150 mM significantly lowered both efficiency (triplex
: initial duplex) and rate constants of triplex formation. The KCl effect was
more pronounced for 11mer and 20mer TFOs than for 14mer TFO. Since the
dissociation half-life for the 11mer TFO decreases from 420 min in the absence of monovalent cations to 40 min in the presence of 150 mM KCl, we suggest that the negative effect could be explained by a decrease in triplex stability. In contrast, for the 20mer TFO no dissociation of the triplex was observed during 24 h
of incubation either in the absence of monovalent cations or in the presence of
150 mM KCl. We suppose that in the case of the 20mer TFO the negative effect of
KCl on triplex formation is probably due to the self-association of the oligonucleotide in competitive structures such as
parallel duplexes and/or tetraplexes. This negative effect may be overcome by
the prior formation of a short duplex either on the 3
'
- or 5
'
-end of the 20mer TFO. We refer to these partial duplexes as `zipper' TFOs.
It was demonstrated that a `zipper' TFO can form a triplex over the full length
of the target, thus unzipping the short complementary strand. The minimal
single-stranded part of the `zipper' oligonucleotide which is sufficient to
initiate triplex formation can be as short as three nucleotides at the 3
'
-end and six nucleotides at the 5
'
-end. We suggest that this type of structure may prove useful for
in vivo
applications.
INTRODUCTION
Short homopurine-homopyrimidine regions in DNA have attracted a great deal of attention in
connection with their possible role in gene regulation in eukaryotes (
1
,
2
). These regions raise the possibility of manipulating gene expression, gene-targeted mutagenesis and inhibition of virus propagation through
artificial triple helix formation (
3
-
5
). Recognition of DNA by triplex forming oligonucleotides (TFO) occurs by
hydrogen bonding between oligonucleotide bases and purine bases in the major
groove of duplex DNA. The triple helix formation arises in either of two
patterns, termed the pyrimidine motif and the purine motif (
6
,
7
). To promote triplex formation with cytosine-containing oligonucleotides a slightly acidic environment is required. In
contrast, the purine motif is pH-independent, so has been used far more often for successful
in vivo
inhibition of transcription. However, a growing number of publications report
that triplexes involving guanine-rich oligonucleotides are inhibited by physiological ionic conditions,
particularly by the presence of K
+
cations (
8
-
10
). It can be assumed that this effect is caused by the formation of the
competitive structures, such as guanine quartets and parallel-stranded homoduplexes (
8
-
10
).
We have previously investigated the triplex formation by G,A-containing oligonucleotides of various lengths (from 11mer to 20mer) with
a sequence of the c-
pim
promoter region and a highly conserved 20 bp-long purine-pyrimidine tract of the
vpx
gene of SIV and HIV-2 at 50 mM Na
+
concentration (
11
,
12
). Despite the high efficiency of triplex formation
in vitro
, we were unable to detect triplex formation inside the cell (
13
). Taking this into account, in the present work,
in vitro
studies have been extended to include physiological K
+
concentrations. It was found that an increase in Na
+
or K
+
concentration leads to a decrease in the rate of triplex formation and the
level of guanine protection judged by DMS footprint experiments. This effect is
more pronounced in the case of K
+
. Based on the stability measurements of the corresponding triplexes we suggest
that the reasons for the decrease in guanine protection are different for the
11mer versus the longer oligonucleotides (14-, 17- and 20mer). For the 11mer TFO, the triplex is significantly
destabilised by the increase of K
+
concentration, leading to less protection of guanines. For longer
oligonucleotides, the data indicate that the increase of potassium
concentration favours the formation of inter- and/or intramolecular structures which are stable at physiological
temperature thus decreasing the effective concentration of TFOs. This
consequently results in a lower efficiency of triplex formation. To avoid these
self-association of the oligonucleotides the 20mer TFO was annealed with
complementary oligonucleotides (from 10- to 17mers) matching either the 5'- or the 3'- end of the TFO. Thus preformed duplexes, named
`zipper' TFOs were able to form triplexes with the target sequence under our
experimental conditions. The rate of triplex formation with `zipper'
oligonucleotides is practically independent of the K
+
or Na
+
concentration over a range of 0 to 150 mM. The `zipper' TFOs containing 8 or 10
residues in a single-stranded form at the 3'-end have displayed the maximum efficiency and rate of triplex
formation. Moreover these oligonucleotides form practically `perfect' triplexes
with the target sequence as evaluated by DMS footprinting. We suggest that
`zipper' TFOs may prove useful in gene targeted therapy based on triple helix
formation.
MATERIALS AND METHODS
Oligonucleotide preparation
Oligonucleotides were synthesised on an Applied Biosystems 391A DNA synthesiser
using the solid phase phosphoramidite procedure. They were precipitated with 10
vol of 3% solution of LiClO
4
in acetone and the pellets were washed with acetone, dried and dissolved in
water. Concentrations were determined spectrophotometrically. Extinction
coefficients were calculated by the extinction coefficients for the nucleotides
and dinucleotide phosphates according to the equation given in Puglisi
et al.
(
14
). For gel-retardation assays oligonucleotides were purified by electrophoresis in a
20% denaturing polyacrylamide gel. After electrophoresis the oligonucleotides
were eluted from the gel in 1 ml 0.2 M LiClO
4
solution during 12 h at 37 oC followed by a precipitation with 10 vol acetone. The 5'-end of the pyrimidine strand of the duplex was radiolabelled with [[gamma]-
32
P]ATP (Amersham) by T4 polynucleotide kinase (New England Biolabs) as per manufacturer's instructions. The oligonucleotide was labelled at a specific activity of 20 Ci/mmol. `Zipper' oligonucleotides were prepared
by heating of a 20mer TFO with an excess of the corresponding complementary oligonucleotide (molar ratio 1/1.2) at 95oC for 15 min and then the mixture was allowed to cool slowly down to room
temperature.
Plasmid construction
The plasmid pVpx1 containing the polypurine stretch of the SIV
vpx
gene was made by inserting the oligonucleotides 5'-CTAGACCTGGAGGGGGAGGAGGAGGAGGTCCG-3'/5'-GATCCGG-ACCTCCTCCTCCTCCCCCTCCAGGT-3' into the
Xba
I-
Bam
HI sites of the vector pBluescript II (Stratagene). All plasmids were grown in
the XL1 Blue bacterial strain (Stratagene) and purified by CsCl gradient
centrifugation (
15
).
DMS footprinting
To prepare a DNA fragment for modification by DMS the pVpx1 plasmid was cut with
the
Cla
I restriction enzyme, 3' labelled with the Klenow fragment of DNA polymerase I, and digested with
the
Xho
I restriction enzyme. A larger labelled fragment (1.0 pmol) was dissolved in a
20 [mu]l volume containing 50 mM MOPS, pH 7.2, 10 mM MgAc
2
and monovalent cations at the concentrations indicated in the legends of
Figures
2
and
7
. After addition of the TFO, the mixture was brought to 37oC for the time specified in the figure legends. Then 2 [mu]l 5% DMS was added and the reaction was performed for 2 min at 25oC. The reaction was stopped by the addition of a 5 [mu]l solution containing 10% mercaptoethanol, 1 mM EDTA and 0.1 M
NaAc. After two precipitations in ethanol the samples were treated with 50 [mu]l 10% piperidine at 95oC for 20 min, and the cleavage products were separated on a 6%
denaturing polyacrylamide gel.
The level of guanine protection in DMS footprinting experiments for VPX20 and
Zip2 TFO was measured by using a FUJIX BAS 1000 phosphoimager. Two regions of
the targeted sequence, i.e. GGGGG and GGAGG, marked I
T
and I
t
on Figure
5
A, respectively, were chosen for this purpose. A sequence close to the target
sequence, marked I
c
on Figure
5
A, was chosen as a control. For each sequence the total intensities of five
bands were measured in two runs of experiment: in the presence of TFO (A
T
, A
t
and A
c
, respectively) and in its absence (C
T
, C
t
and C
c
, respectively). These were further used for the calculation of guanine
protection, X
T
and X
t
, respectively, according to the formula (given for I
T
region):
X
T
= 1 - (A
T/
A
c
)/(C
T/
C
c
)
where the term C
T/
C
c
represents the normalisation factor used to take into account possible
variation in the quantity of the radioactivity, loaded on the gel. X
T
and X
t
were determined for various times for VPX20 and Zip2 TFOs (lanes 1-6 and 8-13 in Fig.
5
, respectively). The same equation was used to determine the level of guanine
protection in the sequence GGAGG.
UV spectroscopic temperature dependent melting studies
Absorbance of the oligonucleotide mixtures was measured at 258 nm as a function
of temperature with an Uvicon 941 spectrophotometer equipped with a Bioblock
Ministat cryothermostat and a Huber PD415 temperature programmer through software developed for
T
m
recording. The rate of temperature increase was 0.5oC/min. The buffer composition and oligonucleotide concentration are
specified in the text. Before performing melting studies, all the samples were
heated at 95oC for 15 min and then allowed to return slowly to room temperature.
Co-migration assay
After triplex formation a 50% glycerol solution was added to bring the solution
to a 5% final concentration in glycerol, and the samples were loaded on a 1%
agarose gel. Electrophoresis was performed at 3.5 V/cm at room temperature for
4 h in the presence of 20 mM Tris-acetate, 50 mM NaAc, 10 mM MgAc
2
and 1 [mu]g/ml ethidium bromide. The quantity of the radioactivity bound to the
targeted DNA was determined using a FUJIX BAS 1000 phosphoimager with flow-dried gel.
Rate of triplex formation as determined by a co-migration assay
To determine the rate of triplex formation, 2 [mu]g plasmid pVpx1 (1 pmol) was incubated for various times with 10 pmol [gamma]-
32
P- labelled TFO (specific activity 20 Ci/mmol). The reaction was carried out
in a 15 [mu]l solution containing 10 mM MgAc
2
, 20 mM Tris-acetate, pH 7.5, and 200 pmol of the unrelated oligonucleotide 5'-GAGGCGGCAGGGCGAGAGGC-3' in order to avoid non-specific adsorption of the TFO to the tube
walls. The concentration of K
+
is specified in the text. After co-migration, the amount of radioactively labelled oligonucleotide bound to
the plasmid was determined using a FUJIX BAS 1000 phosphoimager.
Determination of the dissociation rate of the triplexes
Triplexes were preformed by overnight incubation at 37oC of 10 pmol pVpx1 plasmid DNA, linearised with
Cla
I, with 2 pmol of the TFO (specific activity 1000 Ci/mmol) in the presence of 10
mM MgAc
2
, and 20 mM Tris-acetate, pH 7.5. The preformed triplex (1 pmol of the plasmid) was further
incubated for different lengths of time with an excess (500 pmol) of the same
non-labelled oligonucleotide in a 15 [mu]l solution containing the same buffer and Na
+
or K
+
at the concentrations specified in the text. After performing a co-migration assay, the amount of radioactive labelled oligonucleotide bound
to the plasmid was determined using a FUJIX BAS 1000 phosphoimager.
Electron microscopy (EM)
The oligonucleotide 5'-GGAGGAGGAGGAGGGGGAGG-3'- biotin (bio-VPX20) was purchased from GenSet (France).
pVpx1 plasmid DNA was linearized with the
Kpn
I. For EM a procedure similar to that described by Cherny and co-workers (
16
,
17
) was used: 0.1 [mu]g pVpx1/
Kpn
I plasmid DNA was incubated
at 37oC
for the desired time with bio-VPX-20 in a 10 [mu]l volume, either in 10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 10 mM MgAc
2
(buffer A) or in 20 mM Tris-HCl, pH 7.5, 10 mM NaCl, 150 mM KCl and 10 mM MgAc
2
(buffer B). Then the mixture was passed through a Superose 6 column
equilibrated with buffer A, and streptavidin (Sigma) was added to DNA-containing fractions to a final concentration of 5-20 [mu]g/ml (80-300 nM); after 10 min incubation at room temperature the
gel filtration step was repeated. For the kinetic experiments aliquots were
taken from the mixture and diluted 10-fold in buffer A; then streptavidin was immediately added at a 10-fold molar excess in comparison with the concentration of
oligonucleotide. After 10 min incubation at room temperature the mixture was
subjected to gel-filtration as above. A 5 [mu]l aliquot was then applied to a carbon film glow-discharged in the presence of pentylamine vapours as described
by Dubochet
et al
. (
18
), stained with an 0.5-1% aqueous solution of uranyl acetate and rotary shadowed with
Tantalum/Tungsten with the electron gun of a Balzers MED 010 apparatus. The
samples were observed with a Zeiss/LEO CEM-902 electron microscope in the annular dark-field mode using the technique described by Delain
et al
. (
19
,
20
). Image recording and length measurements of DNA molecules were performed with
the built-in Kontron image analyzer system and software. The efficiency of triplex
formation was measured as a ratio of the number of DNA molecules with bound
streptavidin in the right position to the total number of scored DNA molecules.
Usually, 500-700 DNA molecules were scored for each experiment.
RESULTS
Purine-purine-pyrimidine triplex formation is affected by temperature and
monovalent cations
DMS footprinting.
The oligonucleotides used for the targeting of the
vpx
gene in the present experiments are listed in Figure
1
. Triplex formation with the pVpx1 plasmid DNA in the buffer containing 50 mM Na
+
was monitored by a DMS footprinting assay (Fig.
2
). DMS modifies the N7 position of guanines leading to phosphate backbone chain
cleavage after treatment with piperidine. This reaction does not occur with the
purines of the double-stranded DNA within a purine-purine-pyrimidine triplex due to formation of Hoogsteen base
pairing (
21
). Figure
2
shows that the guanines located within the target sequences of all
oligonucleotides tested (VPX11, VPX14, VPX17 and VPX20) are less affected by
DMS treatment in comparison to those external to the corresponding targets
suggesting that triplex is formed with each of the oligonucleotides under the
experimental conditions used. In accordance with our previously published
results (
12
), guanines within triplexes formed at 37oC were less protected over the full length of the TFO by longer
oligonucleotides, especially at the 3'-end of the targeted sequence (Fig.
2
, lanes 3-5). The pattern of protection changed when triplexes preformed at 37oC in the buffer containing 50 mM Na
+
were heated to 70oC for 3 min followed by DMS treatment at room temperature. In this case the
observed protection corresponded well to the length of TFOs (Fig.
2
, lanes 6-9).
VPX20 structure: UV melting studies
To study possible inter-and/or intramolecular interactions, we performed melting studies of VPX20
TFO under conditions of different ionic strengths. In the presence of 10 mM
MgAc
2
and 20 mM Tris-acetate, pH 7.5, the
T
m
was concentration-dependent, ranging from 36.0oC at 0.5 [mu]M of oligonucleotide concentration to 42.5oC at 8 [mu]M (Fig.
4
). The addition of 50 mM KCl in the buffer produces melting profiles without
pronounced transition. The shape of the curves are independent of the KCl
concentration from 50 to 150 mM KCl (Fig.
5
).
VPX20 structure: a non-denaturing gel assay
In order to determine whether self-association of VPX20 is responsible for the inhibition of triplex
formation by monovalent cations we investigated the state of the
oligonucleotide under triplex forming conditions by an electrophoretic mobility
shift assay.
Figure
6
shows that heating of the oligonucleotide solution with 10 mM MgAc
2
and 20 mM Tris-acetate, pH 7.5 at 95oC for 15 min followed by slow cooling to room temperature leads to
the appearance of two bands upon non-denaturing gel electrophoresis (Fig.
6
A, lanes 1-3). The mobility of the faster band coincides with the control pyrimidine-rich oligonucleotide 5'-CCTCCTCCTCCTCCCCCTCC-3' position which is indicated by an arrow
SS 20 in Figure
6
A. The second, slower band, migrates with the mobility corresponding to the
duplex of VPX20 with its complementary strand (marked Du 20 in the Figure
6
). The pattern remains the same when the concentration of the TFO increased from
0.5 to 8 [mu]M (Fig.
6
A, lanes 1-3). The appearance of the slowly migrating band can be attributed to the
formation of the parallel purine/purine duplex previously described for GGA-containing oligonucleotides (
10
).
VPX20 structure: a 12mer complementary oligonucleotide decrease self-association in the VPX20 TFO
In order to avoid possible inter- and/or intramolecular interactions, VPX20 was annealed to a 12mer
oligonucleotide, 5'-TCCTCCTCCTCC-3', which is complementary to the 5'-end of the TFO. Heating at 95oC for 15 min followed by slow
cooling to room temperature of the purine TFO with a complementary strand,
either in buffer without monovalent cations or in the presence of 50 mM NaAc,
results in the formation of one major band most probably corresponding to the
duplex formed between VPX20 and the 12mer (Fig.
6
A and B, lane 4). Moreover, the addition of the 12mer to the VPX20 in buffer
containing 50 mM NaAc eliminated the bands migrating more slowly than the
duplex (Fig.
6
B, lane 4). In the presence of 50 mM KAc and the highest concentration of the
TFO tested (8 [mu]M) the 12mer oligonucleotide substantially decreases the self-association of the oligonucleotide. This is shown by the decreased
intensity of the band located in the well and the appearance of the major band
with the mobility close to that of the marker duplex (Fig.
6
C, lane 4).
Melting temperature studies showed that the addition of this 12mer
oligonucleotide to VPX20 to the buffer containing 10 mM MgAc
2
, 150 mM KAc and 20 mM Tris-acetate, pH 7.5 triggers a transition in the melting curves with
T
m
varying from 47.5 to 55.0oC for concentrations of oligonucleotides ranging from 0.5 to 8.0 [mu]M (Fig.
5
). Because these melting temperatures coincide with the melting temperatures of
the duplex 5'-GGAGGAGGAGGA-3'/5'-TTCTTCTTCTTC-3' (data not shown), we
suggest that addition of the 12mer to VPX20 leads to the Watson-Crick base pairing between these two oligonucleotides and thus prevents
VPX20 self-association.
These results demonstrate that duplex formed between the 5'-end of VPX20 TFO significantly eliminate the self-association of VPX20 TFO. We named this partial duplex a
`zipper' TFO.
Kinetic analysis of triplex formation by VPX20 and its `zipper' forms
As noted above the dissociation of the triplex formed by VPX20 is very slow (t
1/2
>100 h). This allowed us to use as a first approximation the following equation
of triplex formation:
d[Tr]/dt = -d[D]/dt =
k
in
[D][O],
1
where [Tr], [D] and [O] are the triplex, duplex and TFO concentrations,
respectively. Since the oligonucleotide concentration in our experiments is 10
times the concentration of the target sequence, this second order reaction can
be solved as a pseudo first order one:
[Tr]/[Do] = Tr
[infinity]
{1 - exp(-
k
in
[O]t)},
2
where [Do]
is the initial concentration of the target site, Tr
[infinity]
is the fraction of the targeted sites occupied by the TFO when the reaction
reaches the plateau. Estimates for
k
in
and for Tr
[infinity]
were obtained by least squares fitting of kinetic data to the equation
2
. The kinetic data for VPX20 and `zipper' oligonucleotide Zip2 were obtained by
three methods.
(i) Footprinting experiments (Fig.
7
)
. We estimated the rate of triplex formation for different parts of the target
sequence: for the first stretch of five Gs and for the following sequence
GGAGG. In Figure
7
C one can see that equation
2
describes the process of triplex formation by `zipper' TFO both for the stretch
of five Gs and for the sequence GGAGG. The corresponding values for
k
in
and Tr
[infinity]
are given in Table
1
A. These values are lower for the GGAGG sequence in comparison to those for the
GGGGG sequence. The
k
in
and Tr
[infinity]
values for VPX20 determined by the footprint method for the five Gs sequence
were two times lower than the corresponding values for Zip2 TFO (Table
1
A). However we did not see any footprint for the sequence GGAGG in the case of
the VPX20 oligonucleotide (Fig.
7
A and C).
(ii) Co-migration assay (Table
1
B)
. The value of
k
in
determined on the basis of the co-migration assay indicates the total amount of oligonucleotide bound to the
targeted plasmid but does not permit the observation of non-perfect triplexes. Nevertheless the footprinting and co-migration methods yield similar values for
k
in
.
(iii) Electron microscopy experiments
.
k
in
estimated in the experiments with bio-VPX20 at 2 [mu]M TFO concentration was found to be about 0.3 +- 0.05 * 10
3
M
-1
s
-1
. Comparison of the Tr
[infinity]
values for bio-VPX20, either pure (Tr
[infinity]
= 0.15) or after duplex formation with the complementary 12mer oligonucleotide
(Tr
[infinity]
= 0.80), also confirmed the difference in triplex formation efficiency.
K
in
and Tr
[infinity]
for different types of `zipper' oligonucleotides, estimated by the co-migration assay are given in Table
1
B. Two generalizations can be made from these data. First, some of the `zipper'
oligonucleotides (numbers 1, 2 and 4) possess higher efficiency of triplex
formation in comparison to VPX20. Second, `zipper' oligonucleotides with
complementary strand matching the 5'-end of TFO are faster in triplex formation when compared to those
with the complementary strand matching the 3'-end. However there is no direct correlation between the rate of
triplex formation
k
in
and its efficiency Tr
[infinity]
. For example, the value of Tr
[infinity]
for Zip4 is very close to the unit but the rate is seven times less than for
VPX20.
Zip3 and Zip5 TFOs give the lowest values as for
k
in
as well for Tr
[infinity]
. Nevertheless, the three nucleotides in a single-stranded form at the 3'-end (Zip3) or six nucleotides at the 5'-end (Zip5) promote triplex formation detectable
by the co-migration assay.
DISCUSSION
Previously we found that the G,A-containing oligonucleotides targeted to the c-
pim
promoter region and to the
vpx
gene of SIV and HIV2 viruses form very stable triplexes at 50 mM Na
+
concentration. Melting temperatures of these triplexes were found to be higher
than the corresponding values for double-stranded target sequences. Despite the high efficiency of triplex
formation
in vitro
, we were unable to detect triplex formation inside the cell (
13
). Recent investigations have indicated that K
+
promotes the formation of oligonucleotide species refractory to triplex
formation (
8
,
10
). Taking this into account, in the present work,
in vitro
studies have been extended to include physiological K
+
concentrations. An increase in Na
+
or K
+
concentration decreases the rate of triplex formation and the level of guanine
protection as judged by DMS footprint experiments. These phenomena are more
pronounced for K
+
. We suggest that at least three processes might be responsible for these
phenomena.
First, elevated concentration of the monovalent cations may change the
parameters of interaction between the target sequence and TFOs at the triplet
level. Because the stability of the triplex can be considered as a function of
the stability of each triplet in the triplex, these parameters determine the
minimum length of the stable triplex. In our case only the stability of the
11mer TFO was substantially decreased by the addition of K
+
up to 150 mM thus leading to less guanine protection. For the longer TFO,
however, we did not see any substantial decrease in the stability of the
triplexes at increased monovalent cation concentrations.
Second, DMS footprinting indicates that TFOs of a length of 14mer or more can
simultaneously form a triplex with the targeted sequence and participate in
additional intra- and/or intermolecular interactions. This in turn results in the
appearance of non-perfect or `degenerate' triplexes, in which only a part of the TFO
interacts with the target sequence. These non-perfect triplexes are probably less stable than the correctly formed ones,
but they could prevent `perfect' triplex formation by other TFOs over the full
target length. Ten times excess of initial TFO concentration in our experiments
may result in full occupancy of the target site by the complexes of various
stability while only a part of them yield the triplexes registered by any
methods. This effect could explain the low value of Tr
[infinity]
for VPX20 TFO in our kinetic measurements. The coexistence of perfect together
with non-perfect triplexes has been already described for PNA oligomers (
22
).
Third, for longer oligonucleotides, the data indicate that the increase of
potassium concentration favours the formation of inter- and/or intramolecular structures which are stable at physiological
temperature. This, in their turn, can change the parameters of interaction
between TFOs and their targets lowering the efficiency of triplex formation.
The negative effect of K
+
on triplex formation by guanosine-rich oligonucleotides has been described (
8
,
10
,
23
). Various explanations have been suggested to explain this phenomenon including
formation of some oligonucleotide structures which are refractory to triplex
formation (
8
,
10
) or which do not involve the self-association of TFO (
23
). In the latter case the authors tested self-association of the TFO by a co-migration assay in the presence of EDTA in the gel, when the triplex
formation took place in the presence of 10 mM Mg
2+
. From our experience, and others (
10
) the presence of Mg
2+
may be crucial for the stability of the self-associated TFO structures. This may explain why the authors did not see
oligonucleotide aggregates by their method. Also the difference in the TFO
behaviour might be attributed to differences in oligonucleotide sequences.
Parameters of the kinetic of triplex formation for various TFOs
VPX20
zip1
zip2
zip3
zip4
zip5
A
GGGGG
K
in
0.43 +- 0.05
1.01 +- 0.10
Tr
[infinity]
0.48 +- 0.02
0.84 +- 0.03
GGAGG
K
in
0.00
0.81 +- 0.05
Tr
[infinity]
0.76 +- 0.01
B
K
in
0.72 +- 0.18
0.58 +- 0.05
0.96 +- 0.05
>0.002
0.10 +- 0.02
0.002 +- 0.0005
Tr
[infinity]
0.53 +- 0.05
0.90 +- 0.03
0.86 +- 0.02
-
0.94 +- 0.05
K
in
is expressed in 10
3
M
-1
s
-1
. Tr
[infinity]
is expressed as a ratio of triplex concentration after 3 h of incubation to the
initial concentration of the target. (A) Data were obtained from footprinting
experiments for two regions of the target GGGGG and GGAGG (see also Fig. 7).
(B) Data were obtained from co-migration experiments.
For triplexes which we classified as `stable' in our previous work (with
T
m
identical or even higher than the
T
m
of the corresponding target duplex) (
12
,
24
) it was reasonable to suggest that oligonucleotides complementary to a part of
the TFO could be pushed out during the interaction of the TFO with the targeted
duplex. Moreover, one can expect that this partial duplex can protect purine-rich oligonucleotides from self-association. We have named these structures `zipper'
oligonucleotides. We have tested different oligonucleotides, matching different
parts of the 20mer TFO. It was found that even three residues in a single-stranded form at 3'-end of the `zipper' TFO are sufficient to provoke slow
triplex formation with the targeted sequence. Increase of the length of the
single-stranded part leads to a substantial increase in the rate of triple-helix formation. As one sees in Figure
5
A, triplexes formed by the 20mer alone and by `zipper' oligonucleotide are
qualitatively different. In the case of the 20mer alone, only the stretch of
the first five guanines is protected in DMS footprint experiments. We were
unable to get full protection of guanines even at 10 [mu]M concentration of VPX20. The footprint of the triplex formed by the
`zipper' oligonucleotide shows a much more uniform protection of the guanines
within the targeted sequence. This indicates that the reaction of triplex
formation proceeds as predicted: the single-stranded part of the `zipper' oligonucleotide finds the target followed by
the quick zipping of the triplex-forming oligonucleotide with the targeted DNA and the pushing-out of the complementary part of the `zipper' oligonucleotide.
Recently, Gee
et al.
(
25
) and Olivas and Maher (
7
) suggested an alternative approach to overcome potassium-mediated inhibition of triplex formation. They have developed
oligonucleotides, containing 6-thioguanine bases substituted for part of the guanines in the TFOs. These
TFOs were resistant to potassium-mediated aggregation, yet were still able to form stable triplexes. The
principal difference between this and our approach is the absence of modified
bases in our TFO.
Based on these results one may explain the `optimal length' for target binding
obtained with a 12mer purine TFO in DNAse footprint experiments (
8
). It seems very likely that in the case of long oligonucleotides the decrease
in their ability to form triplexes might be due to the oligonucleotide self-association. The `zipper' oligonucleotide approach presents the first
attempt to overcome potassium inhibition of triplex formation for
oligonucleotides composed of unmodified purine bases. It seems likely that this
strategy may prove useful for triple-helix based gene targeted therapy.
ACKNOWLEDGEMENTS
We thank Dr E. Lescot for oligonucleotides synthesis. We are very thankful to Dr
J. Paoletti for the software of
T
m
recording and Drs T. O'Conner and L. Pritchard for the helpful discussion. This
work was supported by the `Agence Nationale de Recherches sur le SIDA' research
fellowship to F.S.; by INSERM grant no. 94 EO 08 and by le Ministère de l'Enseignement Supérieur et de la Recherche and l'Institut de Formation Supérieure Biomédicale research fellowship to A.D. and by the
Programme Internationale de Cooperation Scientifique PICS no. 227 to D.C.
*To whom correspondence should be addressed at: Laboratoire de
Biochimie-Enzymologie, CNRS URA 147, Institut Gustave Roussy, rue Camille
Desmoulins, 94805 Villejuif Cedex, France
