DNA modification with cisplatin affects sequence-specific
DNA binding of p53 and p73 proteins in a target
site-dependent manner
Hana Pivon
ˇkova
´
1
, Petr Pec
ˇinka
1
, Pavla C
ˇes
ˇkova
´
2
and Miroslav Fojta
1
1 Institute of Biophysics, Academy of Sciences of the Czech Republic, Brno, Czech Republic
2 Masaryk Memorial Cancer Institute, Brno, Czech Republic
The tumor suppressor protein p53 is known as a tran-
scription factor involved in cell cycle control [1–3]. It
plays a crucial role in preventing malignant transfor-
mation of a cell via induction of cell cycle arrest or
programmed cell death in response to stress conditions
(e.g. DNA damage). The functions of p53 are closely
related to sequence-specific recognition of response ele-
ments [p53 DNA-binding sites (p53DBSs)] in promot-
ers of downstream genes such as p21
WAF1 ⁄CIP1
(involved in cell cycle arrest), Bax (apoptosis), and
mdm2 (negative feedback regulation of p53) [1–3].
Using chromatin immunoprecipitation combined with
a paired-end ditag DNA sequencing strategy, Wei
et al. have recently established a global map of p53-
binding sites encompassing over 540 loci in the human
genome [4]. A typical p53DBS consists of two tandem
copies of the motif RRRCWWGYYY (where R ¼A
or G, Y ¼C or T, and W ¼A or T), which may be
separated by one or more base pairs [4,5]. Natural p53
response elements exhibit surprisingly high sequence
variability and may contain one or several nucleotides
not fitting the above formula [6,7]. The p53 protein
binds to the response elements as a tetramer via its
core domain. The importance of p53 sequence-specific
Keywords
cisplatin; DNA damage; protein p73;
sequence-specific DNA recognition; tumor
suppressor protein p53
Correspondence
M. Fojta, Institute of Biophysics, Academy
of Sciences of the Czech Republic,
Kra
´lovopolska
´135, CZ-612 65 Brno,
Czech Republic
Fax: +420 541211293
Tel: +420 541517197
E-mail: fojta@ibp.cz
(Received 25 April 2006, revised 18 July
2006, accepted 17 August 2006)
doi:10.1111/j.1742-4658.2006.05472.x
Proteins p53 and p73 act as transcription factors in cell cycle control, regu-
lation of cell development and ⁄or in apoptotic pathways. Both proteins
bind to response elements (p53 DNA-binding sites), typically consisting of
two copies of a motif RRRCWWGYYY. It has been demonstrated previ-
ously that DNA modification with the antitumor drug cisplatin inhibits
p53 binding to a synthetic p53 DNA-binding site. Here we demonstrate
that the effects of global DNA modification with cisplatin on binding of
the p53 or p73 proteins to various p53 DNA-binding sites differed signifi-
cantly, depending on the nucleotide sequence of the given target site. The
relative sensitivities of protein–DNA binding to cisplatin DNA treatment
correlated with the occurrence of sequence motifs forming stable bifunc-
tional adducts with the drug (namely, GG and AG doublets) within the
target sites. Binding of both proteins to mutated p53 DNA-binding sites
from which these motifs had been eliminated was only negligibly affected
by cisplatin treatment, suggesting that formation of the cisplatin adducts
within the target sites was primarily responsible for inhibition of the p53 or
p73 sequence-specific DNA binding. Distinct effects of cisplatin DNA
modification on the recognition of different response elements by the p53
family proteins may have impacts on regulation pathways in cisplatin-
treated cells.
Abbreviations
cisPt-DNA, cisplatin-modified DNA; CTDBS, C-terminal DNA-binding site; EMSA, electrophoretic mobility shift assay; fl, full length;
IAC, intrastrand crosslink; oligo, oligonucleotide; p53DBS, p53 DNA-binding site.
FEBS Journal 273 (2006) 4693–4706 ª2006 The Authors Journal compilation ª2006 FEBS 4693
DNA binding is underlined by the fact that most of
the cancer-related point mutations of p53 are located
in its core domain and the mutants are typically
unable to recognize the p53 response elements [1,8,9].
Besides the nucleotide sequence, binding of p53 to
the p53DBSs appears to depend on conformational
features of its target sites. It has been proposed that
intrinsic bending of the p53DBSs contributes signifi-
cantly to the stability of the p53–DNA complexes [10].
In addition, interactions of the p53 protein with cer-
tain response elements can be controlled by changes in
DNA topology inducing formation of non-B DNA
structures within the binding sites [11,12]. Interactions
of p53 with DNA are regulated mainly via post-trans-
lational modifications (phosphorylation, acetylation)
within the protein C-terminal domain [3,13,14]. Trun-
cated forms of p53 lacking a negative-regulating seg-
ment at the protein C-terminus (residues 369–383 [15])
are constitutively active for sequence-specific DNA
binding [7,16]. On the other hand, the C-terminus of
p53 was shown to be critical for its conformation-
selective DNA binding [11,12,17,18] and to favor p53
interactions with p53DBSs within long DNA molecules
[19,20].
The p73 protein has been identified as a p53 homo-
log exhibiting 63% amino acid sequence identity in the
DNA-binding domain [21–23]. In agreement with this
homology, the p73 protein can recognize the same
response elements as the p53 protein and activate an
analogous set of downstream genes. Multiple splice
isoforms of the p73 protein have been found that differ
in the structure of their N-terminal and ⁄or C-terminal
domains [21,22]. Although it was originally supposed
that the p53 homologs have redundant functions in the
regulation of gene expression, more recent data suggest
that p73 and p63 proteins do not act as ‘classic’ tumor
suppressors, but rather play important roles in the
regulation of cell development and differentiation
[21,23]. Nevertheless, some observations suggest that
p73 is involved in the cellular response to DNA dam-
age and in apoptosis control [24,25].
Cisplatin [cis-diamminedichloroplatinum(II)] is a
clinically used anticancer agent [26,27]. The drug binds
covalently to DNA, forming several kinds of adduct,
among which the most abundant are intrastrand cros-
slinks (IACs) between neighboring purine residues.
The spectrum of cisplatin adducts identified in globally
modified chromosomal DNA comprises about 50%
of 1,2-GG IACs, 25% of 1,2-AG IACs, 10% of
1,3-GNG IACs and interstrand crosslinks, and another
2–3% of monofunctional adducts. It has been found
that cisplatin cytotoxicity is related mainly to the IACs
that induce significant changes in the DNA
conformation, including bending and unwinding of the
DNA double helix [26,28]. The lesions are selectively
bound by a variety of nuclear proteins, and it was pro-
posed that these interactions are important for the
anticancer activity of the drug [26,29,30].
Interactions of the p53 protein with cisplatin-modi-
fied DNA (cisPt-DNA) have recently been studied [31–
36]. In the absence of the p53DBS, enhancement of
p53 sequence-nonspecific DNA binding due to DNA
cis-platination was observed [33–36]. On the other
hand, the same DNA treatment resulted in inhibition
of p53 sequence-specific binding [31,32]. An analogous
inhibitory effect was observed with the anticancer tri-
nuclear platinum complex BBR3464 but not with the
clinically ineffective transplatin. Quite recently, it has
been shown that DNA modification with a transplatin
analog, trans-[PtCl
2
NH
3
(4-hydroxymethylpyridine)],
inhibits p53 binding to the same p53DBS similarly as
does cisplatin [37]. It has been proposed that the inhib-
itory effects of the anticancer platinum complexes are
due to the formation of platinum adducts within the
p53DBS [31,32]. To our knowledge, no analogous
studies of the p73 protein interactions with chemically
damaged DNA have been reported yet.
In this work, we investigated the effects of global
DNA modification with cisplatin on sequence-specific
binding of p53 and p73 proteins to different target
sites. We demonstrated that the sensitivity of the pro-
tein–DNA interactions to cisplatin DNA treatment
correlated with the occurrence of sequence motifs
forming the cisplatin IACs (namely GG and AG dou-
blets) within the given p53DBS. Binding of both pro-
teins to mutated target sites not containing these
motifs was not significantly affected by the DNA cis-
platination. Formation of the cisplatin adducts outside
the p53DBSs did not apparently influence p53
sequence-specific DNA binding.
Results
To analyze the sequence-specific DNA binding of p53
and p73 proteins, we designed 50-mer oligonucleotide
substrates bearing various p53DBSs (Fig. 1). In most
experiments, we used a C-terminally truncated, consti-
tutively active p53(1–363) to eliminate the sequence-
nonspecific p53 interactions with the cis-platinated
DNA, which have been shown to be mediated pri-
marily by the p53 C-terminal DNA-binding site
(CTDBS) [34]. In the presence of competitor nonspe-
cific DNA, sequence-specific binding of the p53(1–
363) protein to the
32
P-labeled 50-mer targets resulted
in the appearance of a distinct retarded band R
53
in
the polyacrylamide gel (Fig. 2). Binding of the p73b
Cisplatin effects on p53 ⁄p73 DNA recognition H. Pivon
ˇkova
´et al.
4694 FEBS Journal 273 (2006) 4693–4706 ª2006 The Authors Journal compilation ª2006 FEBS
Fig. 1. Scheme of DNA substrates used in this work. All p53 DNA-binding sites (p53DBSs) were placed in the center of 50-mer oligonucleo-
tides (oligos), being flanked with the sequences shown on the top (the same stretches flank the p53DBSs in the pPGM1 and pPGM4 plas-
mids). The left part of the scheme shows two p53DBSs derived from natural p53 response elements in p21 (5¢-promoter) and mdm2
promoters, as well as the synthetic p53DBS PGM1. Motifs forming bifunctional adducts with cisplatin are highlighted (GG doublets are in
bold and underlined, AG doublets are in bold, and GNG triplets are marked by brackets). The p53DBSs shown on the right are derivatives of
p21 (p21a and p21b) or pPGM1 (pPGM4). In the latter targets, the incidence of the cisplatin-reactive sites was reduced or eliminated. Bases
not fitting the ‘canonical’ p53DBS [5] are denoted by lower-case letters.
AB
Fig. 2. Electrophoretic mobility shift assay of sequence-specific binding of p53 or p73 proteins to a 50-mer oligonucleotide (oligo) involving
the p53 DNA-binding sites (p53DBSs). (A) The
32
P-labeled p21 target was incubated with the given protein in presence of competitor calf
thymus DNA, and this was followed by electrophoresis on 5% polyacrylamide gel. Lane 1 contains only DNA without any protein; lanes 2, 3
and 4 correspond to DNA complexes with p53(1–363), p73dand p73b, yielding retarded bands R
53
,R
73d
and R
73b
, respectively. In lanes 5–7,
the protein–DNA complexes are supershifted with monoclonal antibodies DO-1 (p53) or anti-HA (both p73 isoforms; the respective super-
shifted bands are denoted as SR
53
,SR
73d
and SR
73b
; the presence of two supershifted bands in each of the lanes 5–7 corresponds to two
possible stoichiometries of the antibody–protein complexes). (B) Sections of an autoradiogram showing retarded bands due to binding of
p53(1–363) or p73dproteins to 50-mer target oligos containing PGM1, PGM4, mdm2, p21, p21a and p21b sites. Other details as in (A), lanes
2 and 3.
H. Pivon
ˇkova
´et al. Cisplatin effects on p53 ⁄p73 DNA recognition
FEBS Journal 273 (2006) 4693–4706 ª2006 The Authors Journal compilation ª2006 FEBS 4695
and p73dproteins to the DNA targets caused the for-
mation of analogous retarded bands (denoted as R
73b
or R
73d
, respectively; lanes 3 and 4 in Fig. 2A) whose
mobilities reflected different molecular weights of the
p73 isoforms. To verify the specificity of the band
shifts for DNA complexes with the proteins studied,
we used the band supershift assay with antibodies
against the p53 or p73 proteins. Addition of the DO-
1 antibody [17,38,39] mapping to the N-terminus of
the p53 protein resulted in further retardation of the
specific p53–DNA complexes (lane 5 in Fig. 2A),
producing two supershifted bands (SR
53
; Fig. 2A).
Formation of the two bands corresponded to two
possible stoichiometries of the antibody–p53 complex,
involving either one or two antibody molecules bound
per p53 tetramer [16,39]. For supershifting of DNA
complexes with the p73 constructs, which were tagged
with hemagglutinin (HA), we used antibody to HA
and obtained analogous band patterns to those
obtained with p53 (Fig. 2; lanes 6–7, bands SR
73b
and SR
73d
), confirming the specificity of the observed
protein–DNA complexes. All 50-mer substrates used
in this work were efficiently bound by the p53 and
p73 proteins [shown in Fig. 2B for p53(1–363) and
p73d], although their affinities for the proteins dif-
fered to some extent (which was manifested by differ-
ent intensities of the R bands). To eliminate these
differences, the effects of DNA cis-platination on the
protein–DNA interactions were always normalized
with the intensity of the retarded band resulting from
protein binding to the same but unmodified p53DBS.
Effects of cisplatin DNA modification on
sequence-specific binding of the p53 protein
Previously, it has been shown [31,32] that DNA modi-
fication with cisplatin causes dose-dependent inhibition
of the full-length (fl) p53 sequence-specific DNA bind-
ing to the synthetic target site PGM1 (Fig. 1). Here,
we studied the effects of DNA treatment with cisplatin
on p53(1–363) binding to the p53DBSs PGM1, p21
and mdm2 (Fig. 1) within the 50-mer oligonucleotides
(oligos) (Fig. 3A). All targets were treated with the
drug in excess of nonspecific calf thymus DNA. Inter-
action of the protein with any of these targets was sig-
nificantly affected by the cisplatin treatment, but the
levels of inhibition observed with individual p53DBSs
at the same degree of global DNA cis-platination dif-
fered significantly. The steepest decrease in p53–DNA
binding with degree of DNA modification was exhib-
ited by the mdm2 target. The R
53
band due to the
p53–mdm2 complex exhibited only 10% intensity for
r
b
¼0.02, compared to the R
53
band due to protein
binding to the same but unmodified substrate (the r
b
value refers to the number of platinum atoms per total
DNA nucleotide). In contrast, the PGM1 and p21 tar-
gets retained 75% and 53% of the p53-binding capa-
city at r
b
¼0.02, respectively (Fig. 3A). Increasing the
DNA modification degree to r
b
¼0.04 resulted in a
decrease of p53–p21 binding to 42%, whereas the
PGM1 site bound only 16% of the protein, compared
to the same but unmodified p53DBS. At r
b
¼0.06, all
mdm2, PGM1 and p21 targets exhibited very weak
p53 binding (about 4% for mdm2 and PGM1 and
10% for p21).
Sensitivity of the sequence-specific p53 DNA
binding to DNA cis-platination depends on the
incidence of cisplatin-reactive motifs within the
p53DBSs
The mdm2, PGM1 and p21 target sites (Fig. 1) differ
significantly in the occurrence of sequence motifs
known to form the cisplatin IACs [26,27]. The p21 site,
showing the weakest sensitivity of p53 binding to cisp-
latin treatment, contains only one GGG triplet within
the p53DBS. The PGM1 site possesses two AGG tri-
plets in one strand and two AG steps in the other. The
mdm2 target, whose interaction with p53 was most
strongly affected by DNA cis-platination, contains
GG, GGG and AG motifs in one strand and GG and
GTG motifs in the other, thus offering not only the
highest total number of reactive motifs among the
p53DBSs tested, but also the highest number of sites
known to be modified most frequently (i.e. the GG
doublets).
For the subsequent experiments, we designed
mutated p53 target sites from which the cisplatin-react-
ive motifs were eliminated. Two p53DBSs were derived
from the p21 target site (Fig. 1); in p21a, the GGG
triplet in the bottom strand was mutated into GAG.
This exchange resulted in elimination of the most
reactive GG doublets and the introduction of less
reactive AG and ⁄or GNG motifs [26]. In p21b, the
GGG triplet in the bottom strand was replaced by
GAA, which contains neither RG nor GNG motifs
(Fig. 1); owing to this mutation, all sites suitable for
formation of the bifunctional cisplatin adducts were
removed from the p53DBS. In addition, we derived
another ‘unreactive’ p53DBS from the PGM1 target
(PGM4; Fig. 1) by replacing all guanine residues,
except for those at the strictly conserved positions
[4,5], by adenines. All of these mutated p53DBSs
(when cisplatin-unmodified) exhibited sequence-specific
p53 binding comparable to that of the parent targets
(Fig. 2B).
Cisplatin effects on p53 ⁄p73 DNA recognition H. Pivon
ˇkova
´et al.
4696 FEBS Journal 273 (2006) 4693–4706 ª2006 The Authors Journal compilation ª2006 FEBS
We studied how the cisplatin treatment influences
interaction of the p53(1–363) protein with the
mutated target sites. The 50-mer oligos containing
sequences p21a, p21b or PGM4 were treated with
cisplatin as above. DNA modification to r
b
¼0.02
resulted in a decrease of p53 binding to the p21a tar-
get by about 15%, which represented weaker inhibi-
tion than observed with the p21 target (25% decrease;
Fig. 3). More conspicuous differences between the
p21a and p21 targets appeared at r
b
¼0.04 (35% or
58% inhibition, respectively). At r
b
¼0.06, the p21a
target retained 45% of the p53 binding, thus exhibit-
ing at least four times higher binding capacity than
the natural p21 p53DBS treated in the same way.
Binding of p53 to the mutated target p21b exhibited
even more remarkable resistance to the cisplatin treat-
ment. For r
b
values of 0.02, 0.04 or 0.06, 100%, 91%
or 85% of the p21b target was bound by the protein,
respectively, when compared to the untreated p21b.
The behavior of the PGM4 site was similar to that of
p21b, showing practically no inhibition of p53–PGM4
binding for r
b
¼0.02 or 0.04 and about 10% inhibi-
tion for r
b
¼0.06. The PGM4 site also exhibited
practically no loss of its p53-binding capacity due to
the DNA cis-platination when located within a
474 base pair fragment of the pPGM4 plasmid (not
shown), in contrast to the behavior of the analogous
pPGM1 fragment [31]. These data revealed a clear
correlation between the sensitivity of the p53
sequence-specific DNA binding to DNA treatment
with cisplatin and the ability of the particular p53
target site to accommodate the cisplatin IACs. The
higher the probability of formation of the cisplatin
IACs within the p53DBSs due to the occurrence of
the GG, AG and ⁄or GNG motifs, the stronger the
inhibition of the p53 sequence-specific DNA binding
to these targets caused by the DNA treatment with
cisplatin.
AB
Fig. 3. Effects of DNA modification with cisplatin on p53(1–363) binding to various target sites: (A), natural p53 DNA-binding sites (p53DBSs)
mdm2 and p21, and the synthetic PGM1 sequence; (B) mutated p53DBSs PGM4, p21a and p21b (Fig. 1). The top panels show sections of
autoradiograms showing the R
53
bands corresponding to complexes of p53 with the 50-mer target oligonucleotides (oligos) (Fig. 2). The
extents of DNA modification with cisplatin (r
b
) are indicated. Other details are as in Fig. 2. The graphs show the dependence of relative p53
binding to the targets on the degree of DNA modification (data obtained from densitometric tracing of the autoradiograms; for each target
site, the intensity of the R
53
band resulting from p53 binding to unmodified DNA was taken as 1.0, and the intensities of bands correspond-
ing to p53 binding to the same but cisplatin-treated substrate were normalized to this).
H. Pivon
ˇkova
´et al. Cisplatin effects on p53 ⁄p73 DNA recognition
FEBS Journal 273 (2006) 4693–4706 ª2006 The Authors Journal compilation ª2006 FEBS 4697