Stem–loop oligonucleotides as tools for labelling
double-stranded DNA
Be
´
ne
´
dicte Ge
´
ron-Landre, Thibaut Roulon and Christophe Escude
´
Laboratoire ‘Re
´
gulation et Dynamique des Ge
´
nomes’, De
´
partement ‘Re
´
gulations, De
´
veloppement et Diversite
´
Mole
´
culaire’, Muse
´
um
National d’Histoire Naturelle, Paris
Triple-helix forming oligonucleotides (TFOs) repre-
sent an interesting tool for the sequence-specific
recognition of double-stranded DNA. They can be

pH [4,5].
Keywords
triple helix; DNA labeling; stem–loop
oligonucleotide; sequence specificity;
padlock oligonucleotide
Correspondence
C. Escude
´
, Laboratoire ‘Re
´
gulation et
Dynamique des Ge
´
nomes’, De
´
partement
‘Re
´
gulations, De
´
veloppement et Diversite
´
Mole
´
culaire’, USM 0503 Muse
´
um National
d’Histoire Naturelle, CNRS UMR5153,
INSERM U565, Case Postale 26, 43 rue
Cuvier, F-75231 Paris Cedex 05, France

their target by photocrosslinking of psoralen–oligo-
nucleotide conjugates [6,7] or by using oligonucleotides
conjugated to alkylating agents such as chlorambucyl
[8]. We have described a triple-helical complex in which
the third strand is topologically linked to its target [9].
This was achieved by circularization of the TFO after it
had wound around its double-stranded DNA target
thanks to triple-helix formation. When the target was
carried by a circular DNA, i.e. a plasmid, the TFO was
irreversibly linked to the plasmid. We have shown that
the stability of the triple helix made by the topologically
linked TFO, also called padlock oligonucleotide, was
enhanced compared to that formed with a linear TFO
[10]. For example, we showed that such a padlock oligo-
nucleotide strongly inhibits DNA digestion by a restric-
tion endonuclease, and that the complex is strong
enough to inhibit the elongation of transcription by an
RNA polymerase [11]. The triple helix used in these
studies involved a third strand containing G and T,
binding of which was stabilized by the use of a triplex
specific intercalator. This made possible the use of a
third strand which did not contain many Gs. Moreover,
as the triple helix was not stable in the absence of inter-
calator, it was possible to switch easily from conditions
where the triple helix was very stable to conditions
where it was totally unstable. A derivative of this
approach was developed in which the ends of the TFO
hybridized to each other and were ligated to either a
short stem–loop oligonucleotide or to a DNA fragment
that had a complementary sticky end (Fig. 1). The

The first type contains G and T in the loop of the
AB
Fig. 1. Scheme of the padlock structures. The central part of the
TFO forms a triplex with the target dsDNA. The 5¢-and3¢-part of
the TFO hybridize to each other, thereby forming a short double-
stranded stem, and leaving a four nucleotide single-stranded dan-
gling end. This end hybridizes to the complementary extremity of
either a short hairpin oligonucleotide (A) or a long DNA fragment
(B). A ligation reaction results in the formation of a closed dumbell-
like oligonucleotide or a very long stem–loop structure. In both
cases, the TFO encircles the target. Rupture of the topological link
requires cleavage of the circular oligonucleotide (A) or denaturation
of the double-stranded DNA (B).
Fig. 2. Sequence of targets and TFOs used in this study. The cen-
tral part of TG and TC TFOs as well as the target sequences inser-
ted in plasmids pY, pY1m and pY2m are shown. The pY plasmid
contains a 20-bp oligopurineÆoligopyrimidine target sequence,
shown in bold. The pY1m and pY2m plasmids have the same
target sequence as pY except for one or two mismatch(es),
respectively, which are shown underscored. This sequence can
form a triplex made of TÆAxT and CÆGxC
+
triplets with a parallel TC
TFO or a triplex made of TÆAxT and CÆGxG triplets with an antiparallel
TG TFO.
Stem–loop padlock oligonucleotides for dsDNA B. Ge
´
ron-Landre et al.
5344 FEBS Journal 272 (2005) 5343–5352 ª 2005 FEBS
TFO and can form a triple helix in the presence of the

mated from the amount of labelled oligonucleotide
that comigrates with the plasmid. No labelling was
observed when the samples were not heated (Fig. 3B,
lane 4) or when ligase was omitted (data not shown).
The length of the stem was varied from 6 to 11 bp (see
Table 1 for sequences). The maximal yields were
achieved for lengths between 8 and 10 bp (Fig. 3A,
lanes 1–4). We tried to vary the cooling rate between
80 °C and 30 °C. The best yields were achieved at
a rate of 0.25 °CÆmin
)1
(Fig. 3B). A cooling rate of
0.25 °CÆmin
)1
was therefore preferred for subsequent
experiments with GT TFOs.
Padlock formation with TC oligonucleotides
The use of TC TFOs has not been previously described
for the formation of topologically linked complexes
where a TFO encircles its double-stranded DNA tar-
get. Our previous attempts came up with the fact that
formation of the triple helix requires a pH that is too
low to enable efficient ligation by T4 DNA ligase. To
circumvent this problem we decided to perform the
incubation step at acidic pH, and then to neutralize
the sample before addition of T4 DNA ligase. An aci-
dic pH should favour triplex formation without affect-
ing the stability of the stem. After cooling the sample,
the stem–loop structure should remain stable upon
increasing the pH. Therefore, addition of the ligase

TG8
CGGTCCTAGTACTCGACGCTAGCAAAAGTTTTGGTGGTTTGTGTTTTAAAACACGTGGAGCTGTACTAGG
TG10
CGGTCCTAGTACGCTCGACGCTAGCAAAAGTTTTGGTGGTTTGTGTTTTAAAACACGTGGAGCTGCGTACTAGG
TG11
CGGTCCTAGCTACGCTCGACGCTAGCAAAAGTTTTGGTGGTTTGTGTTTTAAAACACGTGGAGCTGCGTAGCTAGG
B. Ge
´
ron-Landre et al. Stem–loop padlock oligonucleotides for dsDNA
FEBS Journal 272 (2005) 5343–5352 ª 2005 FEBS 5345
0.25 °CÆmin
)1
(Fig. 3B), as observed for the GT oligo-
nucleotides. However, we noticed that a long heating
in the acidic buffer resulted in plasmid nicking. There-
fore, a cooling rate of 1 °CÆmin
)1
was prefered for
subsequent experiments with TC TFOs in order to
preserve the supercoiled conformation of the target
plasmid.
Specificity of padlock formation
Triple-helix formation is a sequence specific process
and it has been reported that the presence of mis-
matches between the third strand and the target
duplex decreases the stability of triple helical
complexes [15,16]. We wondered whether the use of
stem–loop TFOs would affect the specificity of com-
plex formation. We therefore constructed two other
plasmids (pY1m and pY2m) containing the 20-bp

M) of pY (3.0 kb) and 170 ng (5 nM)of
pY1m (4.4 kb). Controls were performed with pY alone (lanes 1
and 7) or pY1m alone (lanes 2 and 8). Padlocks were formed with
radiolabelled GT TFOs (lanes 1–6) or TC TFOs (lanes 7–11) and the
short hairpin oligonucleotide. These TFOs differ in the length of the
double-stranded stem, as indicated by the number in their names.
The relative rate of padlock formation for the perfectly matched
and mismatched sequences is shown below the gels.
A
B
Fig. 3. Yields of padlock formation. (A) Influence of triple-helix motif
and stem length. (B) Influence of the cooling rate. Padlocks were
formed on the pY plasmid using radiolabelled GT TFOs (lanes 1–4)
or TC TFOs (lanes 5–7) and the hairpin oligonucleotide. The TFOs
differ in the length of the double-stranded stem, as indicated by
the number in their names. During padlock formation, the samples
were heated to 80 °C and cooled to 30 °C at various rates, as indi-
cated. A minus sign (–) indicates that the samples were not heated
at all (lanes 4 and 8). After the ligation reaction, complexes were
analysed on an agarose gel, which was dried and autoradiographed.
The intensities of both bands were quantified and summed. The
yield of padlock formation is shown below the gels, relatively to
the best yield for each experiment.
Stem–loop padlock oligonucleotides for dsDNA B. Ge
´
ron-Landre et al.
5346 FEBS Journal 272 (2005) 5343–5352 ª 2005 FEBS
The pY plasmid was incubated in the presence of the
nonlabelled 5¢-phosphorylated TFO (20 nm) and the
radiolabelled fragment (30 nm), and the sample was

of the linear DNA. The sharpness of the different
bands and the different migration rates suggest that
the short DNA fragment remains tightly associated
with its target sequence during electrophoresis.
Discussion
The aim of this work was to investigate the formation
of topological complexes where a stem–loop oligo-
nucleotide encircles a double-stranded DNA molecule.
Formation of these structures uses triple-helix forma-
tion to wind an oligonucleotide around a double-stran-
ded DNA target. The TFO is then ligated to a short
hairpin oligonucleotide or to a longer double-stranded
DNA, which results in a topological link between the
TFO and the target duplex. The influence of various
parameters such as TFO stem length and heating tem-
perature have been studied, as well as the sensitivity
and specificity of this approach. Heating the samples is
necessary for efficient complex formation. Stem length
and heating conditions have a great influence on the
labelling yield for both GT and TC TFOs. An optimal
Fig. 5. Plasmid detection with a radiolabelled DNA fragment. A radio-
labelled DNA fragment was ligated to a TFO (TG8) in the presence of
different amounts of the pY plasmid. The quantity of plasmid is indi-
cated, as well as the signal to noise ratio (S ⁄ N).
Fig. 6. Padlock stability on linear DNA. Plasmid pY was digested
with the restriction enzymes XmnI, DraIII, or XbaI, which cut at
1040, 498, and 74 bp, respectively, from the triplex target site. Pad-
locks were formed on the undigested plasmid (lanes 1 and 5) or on
the linearized plasmids (lanes 2–4 and 6–8) with the TFO TG8
(lanes 1–4) or TC8 (lanes 5–8) and a radiolabelled DNA fragment.

triple-helix formation (Fig. 7). We have previously sug-
gested that in the absence of heating, the TFO
becomes ligated while it is not wound around the dou-
ble-stranded DNA target. Whether this circular TFO
can still form a triple helical structure after ligation
remains an unresolved question, as the formation of
triple helices by linear TFOs could not be detected by
gel electrophoresis under our experimental conditions.
But the length of the linkers between the region that
forms a triple helix and the stem (14 nucleotides on
both sides) is probably too short to permit binding of
the ligated TFO to more than 10 base pairs of the target
(i.e. one turn around the double helix). In contrast, the
topological link provides an enhanced stability, which
results in a band shift that can be clearly detected.
The proposed model can be further exploited in
order to explain the results observed in the present
study (Fig. 7). For long stem lengths, the stem can
dissociate only at very high temperatures and is likely
to reassociate at a temperature higher than the one at
which triplex formation occurs. The ligation reaction
will result in a complex that does not encircle the tar-
get DNA. The decreased efficiency observed with very
short stem lengths requires a different explanation.
The short double-helical structure has a low stability.
Therefore, hybridization of the short single-stranded
ends may be inhibited by the tension exerted on them
by the triple helical structure. We observed indeed that
ligation was inhibited in the presence of an excess of
plasmid (not shown). This hybridization may occur

´
ron-Landre et al.
5348 FEBS Journal 272 (2005) 5343–5352 ª 2005 FEBS
strategies have been described for designing DNA
probes with an enhanced specificity. For example,
molecular beacons are stem–loop oligonucleotides
which are used in a specific single-stranded DNA
detection assay based on fluorescence quenching
[19,20]. Competition between formation of the stem–
loop structure and hybridization to the target results
in an increased specificity. The stem–loop oligonucleo-
tides presented in this paper differ from molecular bea-
cons as the stem–loop structure and the triple-helical
complex exist simultaneously, in contrast to molecular
beacons. Formation of the stem–loop structure reinfor-
ces the strength and specificity of the triple-helical
structure.
Several oligopurineÆoligopyrimidine sequences have
been used as targets for attaching stem–loop padlock
oligonucleotides ([12,13], B. Ge
´
ron-Landre, T. Roulon,
M. Bello-Roufaı
¨
& C. Escude
´
, unpublished results).
Their length varies from 12 to 20 base pairs. We present
here the first systematic study of stem length for a 20-bp
target sequence and two derivatives with one or two mis-

offer an interesting alternative to irreversible triplexes
such as those obtained for example by irradiation of
psoralen–TFO conjugates [6]. It is also possible to cir-
cularize an oligonucleotide around a locally denatured
DNA target. This can be achieved using peptide
nucleic acids openers, forming a so-called earring com-
plex that can be used for DNA labelling or isolation
of specific sequences from genomic DNA [24]. In such
complexes, it was believed that the fact that the circle
was threaded between both DNA strands was required
in order to inhibit sliding of the circular oligonucleo-
tide. The present report shows that the circular TFO
did not slide during gel electrophoresis even under con-
ditions that were not favourable to triplex stability,
such as the absence of the triplex stabilizing agent, or
a pH that does not favour triple-helix formation. Pre-
vious experiments, in which triplex formation competed
with cleavage by a restriction enzyme, had shown that
a circular TFO could remain tightly locked on its target
sequence or leave the restriction site accessible, depend-
ing on the presence of a triplex stabilizing agent [10].
Therefore, conditions that promote local mobility of
the circular TFO are not sufficient to make it slide
freely along the target DNA, and some forces must be
exerted on the padlock oligonucleotide in order to
make it move. Such forces may be provided by the
movements of enzymes that translocate on DNA [25]
or by processive enzymes like RNA polymerases [11].
In this regard, our system may provide an interesting
tool for the study of protein movements on DNA.

labels in order to enhance detection sensitivity. The
stem–loop structure of the oligonucleotide displays
unique and undescribed characteristics in terms of
probe–target interactions which represent a new
approach for enhancing the specificity of nucleic acid
hybridization.
Experimental procedures
Oligonucleotides and chemicals
Sequences of the TFOs are given in Table 1. The sequence of
the short hairpin oligonucleotide was 5¢-ACCGTCCGG
ATTGGCTTTTGCCAATCCGGA-3¢. This oligonucleotide
was 5¢-phosphorylated during synthesis. The sequence of the
primers used for PCR were 5¢-CGGTATCAGCTCACTC
AAAG-3¢ (fw), and 5¢-ATGCTGGTCTCTACCGGCGAT
AAGTCGTGTCTTAC-3¢ (rv). The rv primer was biotinyl-
ated at the 5¢ end during synthesis. All these oligonucleotides
were obtained from Eurogentec (Seraing, Belgium); their
concentration was calculated using a nearest-neighbour
model for absorption coefficients.
TFOs were radiolabelled by incubating 10 pmol TFO in
20 lL T4 polynucleotide kinase buffer (New England Bio-
labs, Beverly, MA, USA) with 10 lCi [
32
P]ATP[cP] (Amer-
sham, > 5000 CiÆmmol
)1
) and 5 U T4 polynucleotide kinase
(New England Biolabs) for 1 h at 37 °C. Unincorporated
[
32

of a 0.5 kb starting from the pBluescript SK+ plasmid
(Stratagene). The sequence of the fw primer was chosen in
order to introduce a cleavage site for the BsaI restriction
enzyme. The PCR was carried out by mixing in 50 lL Taq
buffer (Promega) supplemented with 2 mm MgCl
2
1.6 lm
of each primer, 200 lm of each dNTP, 50 lCi
[
32
P]dCTP[aP] (10 lCiÆ lL
)1
, 3000 CiÆmmol
)1
) (Amersham),
10 pgÆlL
)1
pBluescript and 0.1 UÆlL
)1
Taq DNA poly-
merase (Promega). After 30 cycles of 30 s at 94 °C, 30 s at
61 °C and 1 min at 72 °C, the concluding extension was
carried out for 10 min at 72 °C. Primers and unincorporat-
ed dNTP were removed using Qiagen (Valencia, CA, USA)
PCR purification kits, using the standard protocol. Then
the PCR products were digested overnight at 50 °C with
50 U of BsaI (New England Biolabs). The biotinylated
extremities and the nondigested biotinylated PCR products
were removed using streptavidin-coated magnetic beads
(Dynabeads, Dynal, Oslo, Norway). The labelled fragments

)1
in a
MJResearch thermocycler. It was then cooled on ice and
2 lL of an ice-cold neutralizing solution (150 mm Tris
pH 8.0, T4 DNA ligase 100 U ÆlL
)1
) were added. The
sample was incubated overnight at 16 °C.
Stem–loop padlock oligonucleotides for dsDNA B. Ge
´
ron-Landre et al.
5350 FEBS Journal 272 (2005) 5343–5352 ª 2005 FEBS
Padlock formation was assessed by electrophoresis on a
1% agarose gel in 0.5· TBE buffer at room temperature.
The gels were then dried, autoradiographed using a Typhoon
apparatus (Amersham) and analysed with the imagequant
software (Molecular Dynamics, Sunnyvale, CA, USA).
Acknowledgement
BGL was supported by a grant from Ministe
`
re de la
Recherche.
References
1 Giovannange
´
liC&He
´
le
`
ne C (2000) Triplex-forming

Meyer RB & Gamper HBJ (1997) Factors influencing
the extent and selectivity of alkylation within triplexes
by reactive G ⁄ A motif oligonucleotides. Nucleic Acids
Res 25, 4123–4131.
9 Escude
´
C, Garestier T & He
´
le
`
ne C (1999) Padlock
oligonucleotides for duplex DNA based on sequence-
specific triple helix formation. Proc Natl Acad Sci USA
96, 10603–10607.
10 Roulon T, He
´
le
`
ne C & Escude
´
C (2001) A ligand-modu-
lated padlock oligonucleotide for supercoiled plasmids.
Angew Chem Int Ed Engl 40, 1523–1526.
11 Bello-Roufai M, Roulon T & Escude
´
C (2004) Ligand-
mediated transcription elongation control using triplex-
based padlock oligonucleotides. Chem Biol 11, 509–516.
12 Roulon T, He
´

e M, Montenay-Garestier T,
Barcelo F, Chomilier J & He
´
le
`
ne C (1991) Sequence
specificity in triple-helix formation: experimental and
theoretical studies of the effect of mismatches on triplex
stability. Biochemistry 30, 9791–9798.
17 Ryan K & Kool ET (1998) Triplex-directed self-assem-
bly of an artificial sliding clamp on duplex DNA. Chem
Biol 5, 59–67.
18 Rouge
´
e M, Faucon B, Mergny JL, Barcelo F, Giovann-
angeli C, Garestier T & He
´
le
`
ne C (1992) Kinetics and
thermodynamics of triple-helix formation: effects of ionic
strength and mismatches. Biochemistry 31, 9269–9278.
19 Tyagi S & Kramer FR (1996) Molecular beacons:
probes that fluoresce upon hybridization. Nat Biotechnol
14, 303–308.
20 Bonnet G, Tyagi S, Libchaber A & Kramer FR (1999)
Thermodynamic basis of the enhanced specificity of
structured DNA probes. Proc Natl Acad Sci USA 96,
6171–6176.
21 Goni JR, de la Cruz X & Orozco M (2004) Triplex-

(2004) DNA mapping using microfluidic stretching and
single-molecule detection of fluorescent site-specific tags.
Genome Res 14, 1137–1146.
28 Nilsson M, Malmgren H, Samiotaki M, Kwiatkowski
M, Chowdhary BP & Landegren U (1994) Padlock
probes: circularizing oligonucleotides for localized DNA
detection. Science 265, 2085–2088.
29 Zain R, Marchand C, Sun J, Nguyen CH, Bisagni E,
Garestier T & He
´
le
`
ne C (1999) Design of a triple-helix-
specific cleaving reagent. Chem Biol 6, 771–777.
Stem–loop padlock oligonucleotides for dsDNA B. Ge
´
ron-Landre et al.
5352 FEBS Journal 272 (2005) 5343–5352 ª 2005 FEBS