The single-stranded DNA end binding site of p53 coincides with the C-terminal regulatory region
The single-stranded DNA end binding site of p53 coincides with the C-terminal regulatory region
Galina
Selivanova*
,
Violetta
Iotsova
1,+
,
Elena
Kiseleva
2,[sect]
,
Marika
Ström
,
Georgy
Bakalkin
3
,
Roland C.
Grafström
1
and
Klas G.
Wiman
Microbiology and Tumor Biology Center,
1
Department of Environmental Medicine,
2
Department of Cell and Molecular Biology and
3
Department of Drug Dependence Research, Karolinska Institute, S-171 77
Stockholm
,
Sweden
Received May 30, 1996
;
Revised and Accepted July 24, 1996
ABSTRACT
p53 is a transcription factor that binds double-stranded (ds) DNA in a sequence-specific manner. In addition, p53 can bind the ends of single-stranded (ss) DNA. We previously demonstrated that ssDNA
oligonucleotides interact with the C-terminal domain of p53 and stimulate binding to internal segments of long
ssDNA by the p53 core domain. Here we show that the p53 C-terminal domain can recognize staggered ss ends of dsDNA. We have mapped
the binding site for ssDNA ends to residues 361-382 in human p53 using a p53 deletion mutant (p53-
[Delta]
30) lacking the 30 C-terminal amino acid residues and a series of 22mer peptides. The binding
site for DNA ends coincides with a region previously implicated in regulation
of sequence-specific DNA binding by the core domain. The interaction of the C-terminal regulatory domain with the ends of ssDNA or with the
protruding ends of dsDNA stimulates both sequence-specific and non-specific DNA binding via the core domain. Electron microscopy
demonstrated the simultaneous binding of p53 to dsDNA and a ssDNA end. These
results suggest a model in which interaction of the p53 C-terminal tail with DNA ends generated after DNA damage causes activation
of sequence-specific p53 DNA binding
in vivo
and may thus provide a molecular link between DNA damage and p53-mediated growth arrest and apoptosis.
INTRODUCTION
The p53 protein plays a crucial role in the cellular response to DNA damage (for
a review, see
1
). p53 is normally expressed at low levels but accumulates by post-transcriptional mechanisms in cells exposed to DNA damaging agents (
2
-
6
). The increase in p53 protein levels causes G1 arrest, allowing DNA repair
prior to S phase entry and DNA replication. Alternatively, p53 may trigger cell
death by apoptosis (
7
-
9
). In both cases, p53 acts to prevent the propagation of cells with genomic
injury that may lead to tumour development.
p53 is a transcription factor that has sequence-specific DNA binding activity (
10
,
11
). Genes containing p53 consensus binding sites in their promoter regions
include WAF1/p21 (
12
), MDM2 (
13
,
14
), GADD45 (
15
), cyclin G (
16
) and bax (
17
). The specific DNA binding activity resides in the core domain of p53,
approximately corresponding to amino acid residues 102-290 (
18
-
21
). A vast majority of p53 mutations found in human tumours are clustered within
this domain (
22
). The transactivating activity is located in the N-terminal region of p53 (
23
,
24
), whereas the C-terminal domain harbours a region required for oligomerization (
20
,
21
,
25
,
26
).
Wild-type p53 is structurally flexible and appears to reversibly adopt distinct
conformations
in vivo
during cell division (
27
). Under normal conditions, p53 is present in cells in a latent form which is
unable to bind DNA specifically (
28
). Cryptic specific DNA binding of p53 can be activated
in vitro
by several modifications of the very C-terminal region, including binding of the monoclonal antibody PAb421,
deletion of the 30 C-terminal residues, interaction with dnaK and phosphorylation by casein
kinase II and protein kinase C (
29
,
30
). The activation of cryptic specific DNA binding was suggested to involve an
allosteric mechanism (
28
). The recent finding that a small peptide, corresponding to the C-terminal residues 369-383 in p53, can activate latent p53 for specific DNA binding (
31
,
32
) lends further support to this notion. Thus, the C-terminal region of p53 appears to play an important role as regulator of
specific DNA binding.
The C-terminal domain of p53 can bind DNA (
20
,
21
,
33
) and re-anneal complementary DNA strands (
34
,
35
). We found that human p53 can bind single-stranded (ss) DNA ends through the C-terminal residues 320-393 (
35
,
36
) and showed that short ssDNA oligonucleotides can stimulate binding of the p53
core domain to internal segments of ssDNA (
35
). Studies by Jayaraman and Prives (
37
) demonstrated that sequence-specific binding of p53 to supercoiled double-stranded (ds) DNA can be activated by short ssDNA oligonucleotides,
as shown by DNase I footprinting. Moreover, the C-terminal domain of p53 can bind insertion/deletion (IDL) mismatches in DNA
(
38
) and DNA damaged enzymatically or by ionizing radiation (
39
). Taken together, these findings suggest that the recognition of damaged DNA by
the C-terminus can lead to activation of specific DNA binding of p53
in vivo
.
In order to further characterize the biological role of the interaction between
the C-terminus of p53 and ssDNA ends, we have precisely mapped the region within
the C-terminal domain of p53 that interacts with ssDNA ends and found that it
coincides with the negative regulatory domain of p53. In addition, we show that
ssDNA end binding stimulates both sequence-specific and non-specific dsDNA binding through the p53 core domain. This supports
the hypothesis that the direct interaction with the ends of damaged DNA may
trigger an allosteric shift in p53 after DNA damage
in vivo
.
MATERIALS AND METHODS
Plasmids
The plasmids encoding the GST-human wild-type p53 fusion protein and the GST-p53-[Delta]30 deletion mutant fusion protein lacking amino
acid residues 364-393 were constructed by PCR amplification of human wild-type p53 cDNA using appropriate primers and subsequent cloning in-frame in the
Bam
HI and
Eco
RI sites of the vector pGEX-2T (Pharmacia, Sweden). DNA sequencing confirmed that no mutations had
been introduced in the p53 coding sequence during PCR. The PG-CAT plasmid containing 13 repeats of a p53 consensus site and the MG-CAT plasmid containing mutated p53 binding sites were provided by Dr
Bert Vogelstein and have been described elsewhere (
40
).
Production of GST-p53 fusion proteins in bacteria
GST-p53 proteins were produced in
Escherichia coli
and purified as described by Smith and Johnson (
41
), except that bacteria were grown at room temperature for 2 h after induction
with 0.2 mM IPTG and lysed by sonication in 1% Triton X-100 in PBS.
Oligonucleotides
Synthetic oligonucleotides were purchased from Eurogenthec (Belgium). The
following ss oligonucleotides were used: 5'-CTGATCCATG-3' (10mer); 5'-AAGAGAGGTCCGAGGAGGGG-3' (20mer); 5'-GGACGAATGCGCCGCATGCGAATATAGCGTTTGT-3'
(37mer); 5'-TCAACGTCCATTACAGTGC- ATCAAAGTCCATTACAGTGCGCACATTAACGCGCATTAATGTGCATCTTCAACACT-3' (75mer). The BC blunt end dsDNA oligonucleotide
5'-CCGGGCATGTCCGGGCACGTCCGGGCACGT-3' contains a p53 consensus binding site (
42
). The ds oligonucleotide 5'-CCTTAATGGACTTTAATGG-3' (
40
) was used as control. The sequence of the dsDNA oligonucleotide MN with
protruding ss ends of 2 nt was 5'-GAAGTGGGCGTGGTTTAAAGTATATAAGCA-3' (+ strand) and 5'-GTTGCTTATATACTTTAAACCACGCCCACT-3' (- strand).
Cell culture and preparation of cell extracts
MCF-7 human breast carcinoma cells and HL60 human promyelocytic leukaemia
cells were grown in IMDM medium (Gibco BRL) supplemented with 10% fetal bovine
serum (Gibco BRL). Subconfluent cell cultures were used for preparation of
nuclear extracts as described (
43
).
DNA binding assays
Band shift assays were performed as described (
31
) except that 5 mM MgCl
2
was added to the binding reaction. Circular pUC19 plasmid DNA (20-100 ng/reaction) was used as non-specific competitor. Typically, 0.5 ng
32
P-end-labeled probe, 10 ng purified GST-p53 protein and 1 [mu]g nuclear extract proteins were mixed in a 20 [mu]l binding reaction. The reactions were incubated for
30 min at room temperature when purified proteins were used or on ice when
nuclear extracts were added. The samples were analysed on 4% native
polyacrylamide gels containing 0.1% Triton X-100 and run at 200 V (constant voltage) for 1 h 45 min at 4oC. The gels were fixed in 10% acetic acid and 10% methanol (v/v),
dried and autoradiographed.
GST pull-down assays were performed with PG-CAT plasmid DNA linearized with
Bam
HI and labelled by filling in 5' overhangs with Klenow polymerase in the presence of radiolabelled dATP. Binding reactions (400 [mu]l total volume) containing 40 ng GST protein or GST-wild-type p53 fusion protein immobilized on glutathione-Sepharose beads and 100 ng labelled PG-CAT plasmid DNA were performed in the
buffer for band shift assays. After 30 min incubation at room temperature,
beads were collected by centrifugation and washed twice at 0oC. DNA was eluted from beads by addition of 0.2% SDS and 25 mM EDTA
followed by 10 min incubation at 50oC. The radiolabelled DNA was subjected to agarose gel electrophoresis and
visualized by autoradiography.
For filter binding assay, overlapping peptides (10 mg) covering the p53 C-terminus were immobilized on a PVDF filter (BioRad) blocked with 10 mg/ml
denatured salmon sperm DNA. The filters were incubated for 30 min at room
temperature with the labelled ss 37mer (10 ng/ml) in the binding buffer used
for band shift assays containing 2% BSA. Filters were then washed five times
and autoradiographed.
Antibodies
The p53-specific mouse monoclonal antibodies PAb421 and PAb1801 were from Oncogene
Science Inc. (New York, NY).
Electron microscopy
Analysis of p53/DNA complexes by electron microscopy was performed essentially
as previously described (
35
,
36
). The PG-CAT plasmid was digested either with
Pvu
II to generate blunt ends or with
Bam
HI to generate 5' protruding ends.
Synthetic peptides
22mer peptides covering the C-terminal domain of p53 (residues 337-393) were synthesized using the Merrifield solid phase method (
44
). Each peptide overlapped the previous peptide by 14 residues. Peptides were
dissolved in dimethylsulfoxide at a concentration of 1 mg/ml and stored at -20oC. Purification by HPLC was performed on a Super Pac pep-S column.
RESULTS
The C-terminal 30 amino acid residues of p53 are required for recognition of
ssDNA ends or protruding ends of dsDNA
Studies of complex formation of wild-type GST-p53 protein with different DNA substrates in a band shift assay
revealed two major types of p53/DNA complexes. A slowly migrating complex was
formed upon incubation of p53 protein with the labelled ds oligonucleotide BC
containing a p53 consensus binding site (Fig.
1
A, lane 5). This complexing, which is due to specific interaction of the core
domain with DNA, was enhanced by addition of PAb421 antibody and could be
competed by the BC oligonucleotide (data not shown). In contrast, interaction
of p53 with the labelled 37mer ss oligonucleotide gave rise to faster migrating
complexes (Fig.
1
A, lane 3). p53 interacts with the ends of the ss 37mer oligonucleotide through
the C-terminal domain, since these complexes were competed by short ssDNA (
35
; data not shown). p53 did not form any detectable complexes with the 10mer and
20mer ssDNA oligonucleotides (lanes 1 and 2), although complexes between p53
and the 20mer could be seen after longer exposure (data not shown). Thus, the
minimal length of ssDNA oligonucleotide that can be stably bound by the ssDNA
end binding site in the C-terminal domain of p53 is at least 20 nt.
Electron microscopy analysis of the interaction of full-length p53 and p53-
[Delta]
30 with ssDNA
Since band shift experiments showed that wild-type p53 and p53-[Delta]30 interacted with DNA in a different manner, we compared
their binding to ssDNA using electron microscopy. Our previous electron
microscopy analyses demonstrated that both full-length wild-type p53 and the C-terminal domain alone (residues 320-393) can bind to the ends of ssDNA, whereas the core
domain (residues 90-307) binds to internal ssDNA segments, producing a beads-on-a-string-like pattern (
35
,
36
). Both the full-length wild-type p53 and p53-[Delta]30 proteins were incubated with
Pvu
II-digested and heat-denatured MG-CAT plasmid and analysed by electron microscopy. As can be
seen in Figure
2
G and H, wild-type p53 bound to the ends of ssDNA. The p53-[Delta]30 mutant protein, in contrast, associated with internal
ssDNA in a manner indistinguishable from that of the core domain and did not
show any binding to ssDNA ends (Fig.
2
B-F). Up to 10 p53-[Delta]30 molecules were bound along one DNA molecule (Fig.
2
B). This confirms the data obtained by the band shift assays, demonstrating that
deletion of the very C-terminal 30 residues causes loss of ssDNA end binding. In addition, C-terminal truncation leads to activation of non-specific binding to internal segments of ssDNA by the core
domain.
Mapping of the DNA end binding site using synthetic peptides
In order to map the ssDNA end binding site in p53 in closer detail, a series of
overlapping 22mer peptides spanning most of the C-terminal domain (residues 337-393) were examined for their ability to bind ssDNA oligonucleotides
in band shift assays. The ss 37mer and the ss 75mer oligonucleotide were used
as labelled probes. As can be seen in Figure
3
A, lanes 4 and 10, peptide 46, containing amino acid residues 361-382 of p53, binds to both the ss 37mer and the ss 75mer.
Stimulation of specific DNA binding
in vitro
by ss ends of DNA oligonucleotides
The localization of the ssDNA end binding site to the C-terminal region of p53, previously implicated in modulation of specific
DNA binding by the core domain (
29
,
31
), raised the possibility that the interaction of ssDNA ends with the C-terminal region positively regulates sequence-specific DNA binding by inducing an allosteric shift in the p53
molecule. Jayaraman and Prives (
37
) have shown that short ssDNA oligonucleotides can indeed stimulate the specific
binding of p53 to supercoiled plasmid DNA using DNase I footprinting. To
explore this idea further, we analysed the specific DNA binding of bacterially produced GST-wild-type p53 in band shift experiments in the presence of short ssDNA
(the 37mer). Since the binding of p53 to a consensus binding motif can be
stabilized by nuclear extracts (
45
), suggesting that some unknown cellular factor or factors are needed for
specific binding, HL60 nuclear extracts were added to the reaction mixture. In
our initial experiments we observed only weak stimulation of specific DNA
binding by the ss 37mer (data not shown). However, addition of the ssDNA
oligonucleotide together with the activating antibody PAb421 had a synergistic
effect on p53 DNA binding. The amount of complexes between p53 and the
consensus BC oligonucleotide increased in the presence of ss 37mer DNA in a
dose-dependent manner (Fig.
4
A, lanes 3-5). According to Western blot analysis, the GST-p53 protein levels did not change during incubation with the HL60
nuclear extract in the presence or absence of ssDNA (data not shown).
Electron microscopy analysis of p53 binding to dsDNA containing tandem consensus
p53 binding sites in the presence of ssDNA ends
To analyse the effect of ssDNA ends on the sequence-specific DNA binding of p53 in closer detail, we studied the binding of
bacterially produced GST-full-length wild-type p53 protein to linearized plasmid dsDNA in the presence
of the ss 75mer oligonucleotide by electron microscopy. This technique allows
direct visualization of individual protein/DNA complexes at a given moment. As
can be seen in Figure
5
B, p53 bound very poorly to dsDNA. In agreement with our previous results (
35
), only ~2% of the dsDNA molecules were found in complex with p53. In contrast, p53
bound to the ends of the ss 75mer oligonucleotide (Fig.
5
C-E), which is consistent with the results from our band shift experiments
(data not shown). Sometimes p53 could be observed joining two ss 75mer DNA
molecules together (Fig.
5
F). Around 30% of the p53 molecules were bound to the ss 75mer oligonucleotide.
DISCUSSION
We have previously shown that p53 has affinity for ssDNA ends (
36
). We suggested that the interaction of p53 with ssDNA ends that appear in cells
exposed to DNA damaging agents may trigger a conformational change in the p53
molecule, leading to p53 activation and subsequent cell cycle arrest and/or
apoptosis (
1
). Here we have characterized the interaction of p53 with ssDNA ends in more
detail. Our observation that p53 not only binds to the ends of ssDNA, but also
efficiently interacts with dsDNA containing ss protruding ends supports the
idea that p53 can directly recognize DNA damage in the form of staggered DNA
strand breaks
in vivo
.
Deletion of the C-terminal 30 amino acid residues of p53 abolishes recognition of ssDNA
ends. The efficient competition of p53-[Delta]30/MN complexing by the BC oligonucleotide indicates that p53-[Delta]30 interacts with DNA through the core domain.
Moreover, electron microscopy revealed that p53-[Delta]30 interacts with ssDNA in a manner similar to that of the protein
representing the p53 core domain, i.e. p53-[Delta]30 binds internal segments of ssDNA but not ssDNA ends (Fig.
2
;
35
). In contrast, the full-length p53 protein binds to the ends of ssDNA (Fig.
2
;
36
).
Mapping of the ssDNA end site using overlapping peptides allowed a more precise
localization of this site to residues 361-382 (peptide 46) in the C-terminus of p53. The basic region of the p53 C-terminal domain was previously shown to bind DNA (
20
,
21
,
33
). Our data suggest that the biological function of this C-terminal DNA binding site in p53 is to recognize DNA damage
in vivo
.
The C-terminal region of p53 has several features of interest (Fig.
6
). Hupp
et al
. (
29
) showed that various modifications of the C-terminus can activate cryptic specific DNA binding by the p53 core domain.
The allosteric model for negative regulation of specific DNA binding by the C-terminal p53 domain proposed by Hupp and Lane (
28
) has been further substantiated by the finding that a small peptide
corresponding to amino acid residues 369-383 of the p53 C-terminus can activate specific DNA binding of p53
in vitro
, presumably by disrupting contacts between the regulatory domain and another
region in p53 (
31
). Thus, the ssDNA binding site, localized to residues 361-382 in p53, coincides with a region implicated in regulation of specific
DNA binding. Peptide 46, representing the ssDNA end binding site, can also activate latent p53 for specific DNA binding (G. Selivanova, unpublished results). Taken together, these observations
suggest that ssDNA end binding to the C-terminal site may unleash the sequence-specific DNA binding activity of the p53 core domain from the
negative control imposed by the C-terminal domain.
ACKNOWLEDGEMENTS
We thank Dr Tatjana Yakovleva (Karolinska Hospital) for providing the 75mer
ssDNA oligonucleotide and Dr Bert Vogelstein (Johns Hopkins Oncology Center)
for the PG-CAT and MG-CAT plasmids. We are grateful to all members of K. Wiman's laboratory for help and discussions, with special thanks to Sergei
Zotchev. This work was supported by grants from the Swedish Cancer Society,
Magnus Bergvalls Stiftelse and Åke Wibergs Stiftelse.
REFERENCES
1 Selivanova,G. and Wiman,K.G. (1995) Adv. Cancer Res., 66, 143-179.MEDLINE Abstract
*
To whom correspondence should be addressed
Present addresses:
+
Department of Molecular Biology, Bristol-Myers Squibb Pharmaceutical Research Institute, Princeton, NJ 08543-4000, USA and
[sect]
Structural Cell Biology, Paterson Institute for Cancer Research, Christie
Hospital NHS Trust, Wilmslow Road, Manchester M20 9BX, UK
