In vitro expansion of DNA triplet repeats with bulge
binders and different DNA polymerases
Di Ouyang, Long Yi, Liangliang Liu, Hong-Tao Mu and Zhen Xi
State Key Laboratory of Elemento-Organic Chemistry and Department of Chemical Biology, Nankai University, Tianjin, China
Triplet repeats are the most abundant simple sequence
repeats in the coding and non-coding sequences of all
known eukaryotic genomes [1]. The frequency of spe-
cific types of triplet repeats and their localization in
genes vary significantly between genomes, reflecting
their important role in genome evolution [1,2]. Expan-
sions of DNA triplet repeat sequences are associated
with  16 inherited neurological disorders known as
triplet repeat expansion diseases [3–5], which can lead
to total disability and death. The severity of a triplet
repeat expansion disease is increased anticipatively
and the age of onset is reduced with each successive
generation [6,7]. The high mutation rate of triplet
repeats makes them a rich source of quantitative
genetic variation [8–11]. The tendency for repeating
DNA strands to form hairpin loops or slipped confor-
mations, and their inherent conformational properties,
for example their high degree of flexibility, writhing
and the stability of the hairpin formation, are impor-
tant in the investigation of DNA slippage phenomena
[3,11,12].
Among the non-B-form DNA conformations formed
by triplet repeats, simple bulged structures (one or
more unpaired bases) have been postulated as inter-
mediates in the synthesis of slipped DNA and are
associated with the unstable expansion of triplet
repeats on the basis of their entropy [13]. Several

synthesized a pair of simple chiral spirocyclic compounds [Xi Z, Ouyang D
& Mu HT (2006) Bioorg Med Chem Lett 16, 1180–1184], DDI-1A and
DDI-1B, which mimic the molecular architecture of the enediyne antitumor
antibiotic neocarzinostatin chromophore. Both compounds strongly stimu-
lated slippage in various DNA repeats in vitro. Enhanced slippage synthesis
was found to be synchronous for primer and template. CD spectra and UV
thermal stability studies supported the idea that DDI-1A and DDI-1B
exhibited selective binding to the DNA bulge and induced a significant
conformational change in bulge DNA. The proposed mechanism for the
observed in vitro expansion of long DNA is discussed.
Abbreviations
DDI, Double Deck Intercalater; NCS-chrom, neocarzinostatin chromophore.
4510 FEBS Journal 275 (2008) 4510–4521 ª 2008 The Authors Journal compilation ª 2008 FEBS
Molecular studies have shown that the affinity of
NCSi-gb for DNA bulges is mostly dependent on the
spirocyclic ring junction being at an appropriate angle,
the pendant aminosugar group that enhances binding
at the bulged site, and the two discrete aromatic moie-
ties for p-stacking that mimic a base pair. Molecules
that mimic the wedge-shaped natural product have
been designed and synthesized, with the expectation
that they may be used to study the role of bulged
structures in nucleic acid function [16]. For example,
the compound double deck intercalater (DDI), which
has an spirocyclic backbone almost identical to that of
NCSi-gb (Scheme 1B), was able to enhance slippage
synthesis of various repeat DNA strands [16,20,28].
Analogs of NCSi-gb, the aminoglucose in a-glycosidic
linkage or the natural sugar N-methylfucosamine in
b-glycosidic linkage to the backbone, were found to

32
P-labeling primer or template
in the presence or absence of DNA-binding agents
(DDIs and doxorubicin). The DNA bulge binding
of both compounds was detected by CD and UV
melting experiments. Possible slippage mechanisms are
discussed.
Results and Discussion
Effect of DDI-1A and DDI-1B on repeats
expansion
DDI-1A and DDI-1B were tested for their effect on
the expansion of several doublet and triplet repeats in
the presence of the Klenow fragment of DNA poly-
merase I. The reaction contained 5¢-
32
P-end-labeled
9-mer primer, unlabeled template, dNTPs and the
Klenow fragment. Figure 1 shows the extension prod-
ucts on a denaturing polyacrylamide gel. Band intensi-
ties in each lane were measured using a Phosphor
Imager. In the control reaction (Fig. 1, lane 2), the
9-mer primer with different sequences was extended to
different lengths. Sequences with relatively unstable
secondary structures, such as the triplet repeats
(AAT)
3
⁄ (ATT)
5
and (ATT)
3

Scheme 1. (A) DNA bulge-specific compound derived from NCS-
chrom upon base catalysis. (B–D) Synthetic compounds mimicking
NCS-chrom, which showed selectivity for binding to DNA bulge site
[16], and strongly enhanced the repeat nucleotide slippage during
in vitro DNA synthesis [20].
D. Ouyang et al. Expansion of DNA repeat sequences
FEBS Journal 275 (2008) 4510–4521 ª 2008 The Authors Journal compilation ª 2008 FEBS 4511
of DDI-1A was better than that of DDI-1B, presum-
ably because of the different conformation of the agly-
con moiety. DDI-1A has a right-handed aglycon helix
with geometry mimicking the DNA helix, and is there-
fore more effective at intercalating into DNA base
pairs [20]. Although DDI-1B also mimics the structure
of NCSi-gb, it has a left-handed aglycon helix and
may be less effective at base pair intercalation. It
should be noted that 2-deoxy-2-aminoglucose (with
concentrations of 10–1000 lm) or the aglycon back-
bone (concentrations of 10–100 lm) of DDI-1A and
DDI-1B did not affect DNA slippage (data not
shown). Interestingly, there was a similar hierarchy of
intensities for the three bands in the (AAT)
3
⁄ (ATT)
5
and (ATT)
3
⁄ (AAT)
5
systems (Fig. 1A,B), each appar-
ently separated by two nucleotides, and this was

5
and (CTG)
3
⁄ (CAG)
5
, were
too short to generate a similar pattern on the gel.
Doxorubicin, an anthracycline glycoside that inter-
calates between DNA base pairs [30], inhibited the
expansion of all the repeat sequences used (Fig. 1,
lane 5). When both DDI-1A or DDI-1B and doxo-
rubicin were present, similar inhibition was found at
experimental concentrations (data not shown).
In vitro studies show that single-stranded tracts con-
taining (CTG)
n
repeats have a higher propensity to
form hairpin structures than similar tracts containing
the complementary (CAG)
n
repeats [31]; possibly
accounting for the orientation-dependent behavior of
these repeats in replication. Hairpin stability is attrib-
uted to the TÆT mismatch which stacked more effi-
ciently on the CTG strand than the AÆA mispair on
the complementary CAG strand, resulting in expanded
CTG fragments that are shorter than those of the
CAG strand (Fig. 1E,F). This rule is also the same for
other repeat sequences. As a result, the slippage effects
of AAT and CA repeats (Fig. 1A,D) are better than

under similar reaction conditions to those in Fig. 1. In
the control reaction, the 15-mer template of various
sequences (Fig. 2, lane 2) was extended to different
lengths depending on the stability of the secondary
structure formed between the primer and template
(Fig. 1). Enhancement of sequences with less stable
secondary structures was stronger than that with rela-
tively stable secondary structures. After addition of
DDI-1A and DDI-1B, slippage synthesis was greatly
enhanced for all sequences, as reflected by the presence
of much longer products (Fig. 2, lanes 3 and 4) in
comparison with the control. The stimulation effect of
DDI-1A in the template extension was obviously better
than that of DDI-1B, which was similar in the primer
extension reaction. As expected, doxorubicin inhibited
template expansion for all the repeated sequences
chosen (Fig. 2, lane 5). Again, the gel band pattern of
the synthesized DNA products reflected the particular
nucleotide repeat unit. A similar band pattern in both
the labeled primer and the template expansion system
implied that template and primer extension took place
synchronously.
Time course of DNA expansion
A time course for the extension of the repeat sequences
was performed (Table 1) in the assays shown in Figs 1
and 2. In the control, longer DNA fragments were
generated with the increase in reaction time, indicating
that primer and template slippage occurred during
DNA synthesis. In the presence of DDI-1A and DDI-
1B, radioactivity bands (both primer and template)

merase (Fig. 3C, lane 2), Taq DNA polymerase
(Fig. 3E, lane 2) and pfu DNA polymerase (Fig. 3F,
lane 2). The addition of DDI-1A and DDI-1B strongly
increased the slippage effect in these polymerase
systems. Among these, Sequenase showed the weakest
ABCDEF
Fig. 3. Effect of different polymerases on the stimulation of a triplet repeat expansion. A standard reaction (23 °C, 24 h) containing 5¢-
32
P-
end-labeled (AAT)
3
and unlabeled template (ATT)
5
was catalyzed by different prokaryotic polymerases (indicated). The concentration of pri-
mer–template and deoxynucleoside triphosphates in the reaction system is 4 l
M and 1 mM, respectively. The amount of polymerase used
was almost equal, i.e. 0.0177 unitÆlL
)1
of each enzyme. (A–F) Lane 1, control to which no DNA polymerase was added; lane 2: control reac-
tion system lacking drug, but with an equal volume of dimethyl sulfoxide; lanes 3 and 4, reaction system to which DDI-1A or DDI-1B (60 l
M)
was added; lane 5, reaction system to which 40 l
M doxorubicin was added. Products were resolved on a 15% sequencing gel. The numbers
indicate size markers of 26 and 41 nucleotides (random sequence) in length.
Table 1. Time course of primer ⁄ template expansion in the presence or absence of DDI-1A or DDI-1B. The concentration of DDI-1A or DDI-
1B is 60 l
M. Data are from experiments similar to those described in Figs 1 and 2 using
32
P-labeled primer ⁄ templates. After gel analysis of
the products, the band intensities were quantitated by Phosphor Imager (Molecular Dynamics). *5¢-

10.3 12.9 8.2 15.0 25.9 21.4 18.5 33.6 28.7
(CAG)
3
⁄ (CTG)
5
* 2.2 25.1 12.3 3.5 38.7 19.4 5.2 49.6 26.8
(CTG)
3
* ⁄ (CAG)
5
0 7.8 1.0 0 15.8 2.0 0 23.2 4.5
(CTG)
3
⁄ (CAG)
5
* 6.9 37.5 18.1 9.4 52.0 28.2 10.9 67.2 44.3
(GT)
4
G* ⁄ (CA)
7
C 0 32.7 9.4 1.5 67.7 12.0 12.0 89.7 36.8
(GT)
4
G ⁄ (CA)
7
C* 32.9 70.4 55.7 46.3 93.1 83.7 52.5 95.2 90.7
(CA)
4
C* ⁄ (GT)
7

extended the primer to some extent, and excised the 3¢
overhung nucleotides from the duplex to give smaller
fragments. Because of the presence of excised short
oligomers, the extension bands in these lanes were
much lighter than the others, and various types of
duplex were formed by the primer and template. We
did not observe any strong stimulation to DNA slip-
page synthesis in the gel pattern by the addition of
DDI-1A and DDI-1B in these cases. These results may
be due to DNA polymerases with strong 3¢ to 5¢
exonuclease activity (including T7 DNA polymerase
and DNA polymerase I) degrading the product. To
our surprise, the addition of doxorubincin did not
obviously inhibit expansion, but did inhibit the exonu-
clease activity of DNA polymerase I to some extent;
the excised short oligomers were obviously less
(Fig. 3A, lane 5) than in the control and drug-addition
reactions.
Again, a triple band pattern was apparent through-
out the gel. Although the pattern in the Taq and
pfu DNA polymerase system differed from that in the
Escherichia coli DNA polymerase I-based system, the
expanded primary bands were almost all seen in
the three-nucleotide unit, which indicated that the
in vitro DNA strand slippage synthesis of (ATT)
3
⁄
(AAT)
5
tract was mainly a triplet expansion pattern. It

of the complex, are also presented, assuming that the
conformation of the drug was not significantly altered
because the molecular models of DDI-1A and DDI-1B
are fairly rigid. The differential CD spectra of the
complex formed between DNA and the drugs are
shown in Fig. 4. The observed CD spectrum of the
native DNA (solid line) consists of a distinct positive
band at 280 nm caused by base stacking and a negative
band at 250 nm caused by helicity [33], which is char-
acteristic of DNA in the right-handed B-form. CD
spectra of DNA with DDI-1A (dashed line) and DDI-
1B (dotted line) consistently revealed an isoelliptic
point at  260 nm, except for the oligomer without a
bulge structure (Fig. 4A), suggesting formation of a
drug–DNA complex. For oligomer with a hairpin
structure (HT3AT), the band at 252 nm shifted to
241 nm (Fig. 4A), whereas for DNA with simple bulge
structures (one to three unpaired bases), the band at
252 nm shifted to 244 nm for DDI-1A and to 248 nm
for DDI-1B. There was no overall change in ellipticity
detected from the differential spectra of DNA
(Fig. 4B) for the oligomer HT3AT. In this case, the
binding of DDI-1A or DDI-1B to DNA might be via
simple groove binding and ⁄ or electrostatic interaction
that showed fewer or no perturbations on the base
stacking and helicity bands [34], ruling out the possi-
bility of conformational change.
D. Ouyang et al. Expansion of DNA repeat sequences
FEBS Journal 275 (2008) 4510–4521 ª 2008 The Authors Journal compilation ª 2008 FEBS 4515
AB

behaviors obviously different to that of the bulge
DNA host. The addition of DDI-1A and DDI-1B to
the two-base bulge (HT3AGTT) and three-base bulge
(HT3AATTT and HT3AAATT) caused the DNA
spectrum to be altered significantly. The trend and
characteristics of the conformational transformation
were similar to that of the one-base bulge oligomer.
However, the aglycon unit of DDI-1A and DDI-1B
(10–500 lm), lacking any CD signal itself, did not
affect the conformation of DNA (data not shown).
From the CD results, both compounds can interact
with oligomers containing a simple bulge and induce
significant conformational change. Therefore, the addi-
tion of DDI-1A and DDI-1B may induce formation of
the bulge or stabilize the bulge structure. The UV
melting temperature (T
m
) of oligonucleotides with a
three-base bulge increased upon intercalating with
DDI-1A and DDI-1B (Table 2), implying that DNA
secondary structures were stabilized by interaction with
the drug. For example, the change in T
m
(DT
m
) for the
ATT bulge increased by 3.4 and 1.7 °C in the presence
of DDI-1A and DDI-1B, respectively. The increase in
DT
m

a process of stimulated slippage synthesis (Scheme 2).
After denaturing and annealing, the primers and tem-
plates form various types of duplex DNA. The small
DNA primer–template may have gone through multi-
ple rounds of slippage to reach the large expanded
products observed. Each cycle is initiated by the disso-
ciation of polymerase to re-associate at a new inter-
mediate. The intermediate is a combination of various
DNA strands with an unorthodox structure, such as
hairpin, bulged and slipped DNA, and may be the
main contributor to expansion. Under the experimen-
tal conditions used, various combinations of these
unstable intermediates are in homeostasis. When one
round of extension finishes, the extended primer and
template separate and realign to form new intermedi-
ates for the next round of replication, and longer
extended products are obtained through multiple
rounds of replication. For example, following
bulge ⁄ hairpin formation on the AAT strand of an
AAT ⁄ ATT repeat tract, replication extends the fore-
shortened AAT strand. The AAT bulge ⁄ hairpin may
then come apart to allow the complementary ATT
strand to be extended by DNA polymerase along the
previously extended AAT strand, and vice versa. In
fact, template extension is the same as primer exten-
sion. We call it template extension to distinguish the
Table 2. T
m
values of oligomers (P3 and P4) and DT
m

pins in a length- and orientation-dependent manner
under physiological conditions [41–43]. Once the non-
B structure has formed, it is difficult for the CTG or
CAG strand to re-anneal to its complementary strand,
nor would realignment of primer and template and
further extension be easy. Thus, the expanded frag-
ments are relatively short.
Scheme 2. Mode for primer and template extensions stimulated by drug. The crooked region of two swallow-tailed shapes represent the
unstable intermediates that are composed of bulge, hairpin and slipped DNA etc. The compound formula represents DDI-1A or DDI-1B. One
cycle of simple extension and drug stimulation is shown for each pathway. It is assumed that multiple cycles through these pathways are
required to reach the dramatic expansion.
Expansion of DNA repeat sequences D. Ouyang et al.
4518 FEBS Journal 275 (2008) 4510–4521 ª 2008 The Authors Journal compilation ª 2008 FEBS
Once simple extension of the primer and template is
accomplished, slippage synthesis in the presence of
DDI-1A and DDI-1B becomes more pronounced as
incubation proceeds. In our experiment, the more
DDI-1A and DDI-1B were added and the longer incu-
bation time, the longer the expanded products
obtained; this may be due to two or more bulged inter-
mediates formed or induced by the additional drug
(Scheme 2). Compared with stimulation slippage and
bulge binding specificity, it is proposed that the associ-
ation and disassociation of the compound with the
bulged structure is also in a dynamic equilibrium,
whereas a molecule with moderate binding affinity and
binding dynamics to the bulged structure would facili-
tate further slippage, and yield a good stimulation
result [29].
In summary, DDI-1A and DDI-1B were designed

and both disease severity and age of onset, treatment
that interferes with triplet expansion or the generation
of ineffectual DNA triplet templates, might make sense
for RNA regulation and prevent the formation of toxic
proteins, such as polyglutamine [44] and polyalanine
tract [45].
Experimental procedures
Materials
Oligodeoxyribonucleotides were synthesized on a EXPE-
DITEÔ 8909 nucleic acids synthesis system (Applied
Biosystems, Foster City, CA, USA), and purified by elec-
trophoresis on a denaturing polyacrylamide gel using a
standard procedure [46]. The product was recovered from
the gel by phenol ⁄ chloroform extraction and ethanol pre-
cipitation. T4 polynucleotide kinase, E. coli DNA polymer-
ase I, the Klenow fragment of E. coli DNA polymerase I
lacking 3¢ to 5¢ exonuclease activity, Taq DNA polymerase
and pfu DNA polymerase were from Takara Biotechnology
(Dalian City, China). T7 DNA polymerase was from MBI
Company (Tangshan City, China). Sequenase Version 2.0
DNA polymerase was from U.S. Biochemical Corporation
(London, UK). Radioactive materials were from Beijing
Furui Biological Engineering Company (Beijing, China).
Other chemicals were from Sigma (St Louis, MO, USA).
The oligonucleotides were 5¢-
32
P-end labeled using
[
32
P]ATP[cP] and polynucleotide kinase.

sities were quantitated on a Phosphor Imager (Molecular
Dynamics, Sunnyvale, CA, USA).
UV melting experiments
Ultraviolet absorptions of 2 lm oligonucleotides were mea-
sured using a Cary-Bio100 UV-Visible spectrophotometer
D. Ouyang et al. Expansion of DNA repeat sequences
FEBS Journal 275 (2008) 4510–4521 ª 2008 The Authors Journal compilation ª 2008 FEBS 4519
with heating at 0.5 °CÆmin
)1
in phosphate buffer containing
10 mm phosphate, 10 mm NaCl, pH 7.0. The T
m
for DNA
in the presence of DDI-1A or DDI-1B was determined
when the concentration of drug was 10-fold that of DNA,
and calculated using the derivative method supplied in the
cary winuv software package for T
m
calculation.
CD spectropolarimetry
CD spectra were performed on a Jasco-715 spectropolarim-
eter, using a cylindrical quartz cell of 1 mm path length.
The cell compartment was purged continuously with dry
N
2
. Data were recorded at a bandwidth of 1.0 nm and mea-
sured every 0.2 nm over 210–325 nm at 20 ± 1 °CinTE
buffer (10 mm Tris, 1 mm EDTA, pH 8.0) containing
10 mm NaCl. All oligonucleotides were heated to 95 °Cin
the same buffer for 5 min and then cooled slowly to room

6 Ashizawa T & Zoghbi HY (1997) Diseases with trinu-
cleotide repeat expansion. In Current Neurology, Vol.
17 (Appel SH, ed.), pp. 79–135. IOS Press, Amsterdam.
7 Kunkel TA (1993) Slippery DNA and diseases. Nature
365, 207–208.
8 Kashi Y, King D & Soller M (1997) Simple sequence
repeats as a source of quantitative genetic variation.
Trends Genet 13, 74–78.
9 Napierala M, Bacolla A & Wells RD (2005) Increased
negative superhelical density in vivo enhances the genetic
instability of triplet repeat sequences. J Biol Chem 280,
37366–37376.
10 Parniewski P, Jaworski A, Wells RD & Bowater RP
(2000) Length of CTGÆCAG repeats determines the
influence of mismatch repair on genetic instability.
J Mol Biol 299, 865–874.
11 Wells RD (1996) Molecular basis of genetic instability
of triplet repeats. J Biol Chem 271, 2875–2878.
12 Perutz MF (1996) Glutamine repeats and inherited
neurodegenerative disease: molecular aspects. Curr Opin
Struct Biol 6, 848–858.
13 Harvey SC (1997) Slipped structures in DNA triplet
repeat sequences: entropic contributions to genetic
instabilities. Biochemistry 36, 3047–3049.
14 Kappen LS, Lin Y, Jones GB & Goldberg IH (2007)
Probing DNA bulges with designed helical spirocyclic
molecules. Biochemistry 46, 561–567.
15 Xi Z, Jones GB, Qabaja G, Wright J, Johnson FS &
Goldberg IH (1999) Synthesis and DNA binding of
spirocyclic model compounds related to the neocarzino-

ment of new simple molecular probes of DNA
bulged structures. Bioorg Med Chem Lett 16, 2895–
2899.
23 Hwang GS, Jones GB & Goldberg IH (2004) Stereo-
chemical control of small molecule binding to bulged
DNA: comparison of structures of spirocyclic
enantiomer-bulged DNA complexes. Biochemistry 43,
641–650.
24 Kappen LS & Goldberg IH (1993) DNA conformation-
induced activation of an enediyne for site-specific cleav-
age. Science 261, 1319–1321.
25 Yang CF, Stassinopoulos A & Goldberg IH (1995) Spe-
cific binding of the biradical analog of neocarzinostatin
chromophore to bulged DNA: implications for thiol-
independent cleavage. Biochemistry 34, 2267–2275.
26 Gao X, Stassinopolous A, Ji J, Kwon Y, Bare S &
Goldberg IH (2002) Induced formation of a DNA bulge
structure by a molecular wedge ligand-postactivated
neocarzinostatin chromophore. Biochemistry 41, 5131–
5143.
27 Xi Z & Goldberg IH (1999) Comprehensive Natural
Products Chemistry, Vol. 7 (Barton DHR & Nakanishi
K, eds), pp. 553–592. Elsevier Science, Oxford.
28 Kappen L, Xi Z, Jones GB & Goldberg IH (2003) Stim-
ulation of DNA strand slippage synthesis by a bulge
binding synthetic agent. Biochemistry 42, 2166–2173.
29 Liu L, Yi L, Yang X, Yu Z, Xi Z & Wen X (2008) Syn-
thesis and spectroscopic characterization of binaphthol
aminosugars for stimulation of DNA strand slippage
synthesis. Tetrahedron, 64, 5885–5890.

r
ˇ
V
& Kypr J (2001) A-like guanine-guanine stacking in the
aqueous DNA duplex of d(GGGGCCCC). J Mol Biol
307, 513–524.
36 Schlotterer C & Tautz D (1992) Slippage synthesis
of simple sequence DNA. Nucleic Acids Res 20
, 211–
215.
37 Lyons-Darden T & Topal MD (1999) Effects of temper-
ature, Mg
2+
concentration and mismatches on triplet-
repeat expansion during DNA replication in vitro.
Nucleic Acids Res 27, 2235–2240.
38 Pearson CE & Sinden RR (1998) Slipped strand DNA,
dynamic mutations, and human disease. In Genetic
Instabilities and Hereditary Neurological Diseases (Wells
RD & Warren ST, eds), pp. 585–626. Academic Press,
New York, NY.
39 Ohshima K, Kang S, Larson JE & Wells RD
(1996) TTAÆTAA triplet repeats in plasmids form a
non-H bonded structure. J Biol Chem 271, 16784–
16791.
40 Trotta E, Grosso ND, Erba M & Paci M (2000) The
ATT stand of AATÆATT trinucleotide repeats adopts
stable hairpin structures induced by minor groove
binding ligands. Biochemistry 39, 6799–6808.
41 Tam M, Montgomery ES, Kekis M, Stollar DB,