ABSTRACT
The p53 tumour suppressor protein is a potent transcription factor which plays a
central role in the defence of cells against DNA damage and the propagation of
malignant clones. We have previously shown that phosphorylation of serine 386
in mouse p53 by the growth-associated protein kinase, casein kinase II (CKII), plays an important
role in the ability of p53 to block the proliferation of drug-resistant colonies. In this paper we show that blocking phosphorylation of
serine 386 through an alanine substitution, or placing a constitutive negative
charge at this position in the form of aspartate, had no significant influence
on p53-dependent transcriptional activation of a promoter containing 13 copies of
a p53 consensus binding sequence, or of the p21
WAF1
promoter which is a natural target for p53. In contrast, the alanine mutant
showed a weak reduction in the ability of p53 to repress expression from the c-
fos
promoter, which is a target for p53-dependent repression
in vivo
. Strikingly, when the repression of the SV40 early promoter was examined, a
reduction in the repression capacity of up to 5-fold was observed. Moreover, repression of the SV40 promoter could be
partially restored by aspartic acid substitution at the phosphorylation site.
These data indicate that phosphorylation at a specific C-terminal site can selectively regulate p53-dependent repression, but not transactivation.
The p53 tumour suppressor protein (reviewed extensively by Donehower and
Bradley;
1
) is a potent transcription factor which is activated in response to a variety
of DNA-damaging agents. Induction of p53 occurs at least in part by a post-translational mechanism leading to stabilisation of the normally
rapidly-degraded protein (
2
). Activation of p53 leads to cell growth arrest at the G1/S boundary (
3
,
4
) or the induction of apoptosis (
5
,
6
), thereby blocking the survival of genetically damaged cells. Loss of p53
suppressor function through mutation is a common event in the development of a
wide variety of human cancers and may contribute to an increase in the number
of genetic abnormalities (
7
).
p53 transactivates a range of promoters through site-specific binding to a
cis
-acting DNA sequence element (
8
-
11
) and there is a clear correlation between the transactivation and growth
suppression functions of p53 (
12
,
13
). Genes whose expression is stimulated by p53 include GADD45 (
14
),
mdm
2 (
15
,
16
), and
bax
(
17
). The most well characterised target for p53-dependent transactivation is p21
WAF1
, an inhibitor of cyclin dependent kinases (
18
), the induction of which prevents cell cycle progression by blocking
phosphorylation-dependent inactivation of the retinoblastoma protein (
19
-
21
).
p53 can also repress a wide variety of cellular and viral promoters (
22
-
25
) including the c-
fos
, c-
jun
, b-
myb
, DNA polymerase [alpha] and SV40 early region promoters (
23
,
24
,
26
,
27
). Repression is not a general effect but shows promoter selectivity (
23
,
27
) and can be cell type specific (
28
,
29
). Like transactivation, repression is a function of wild type, but not mutant
p53 proteins (
26
,
30
,
31
), suggesting that loss of p53-dependent repression may have significance in tumour development.
Transactivation and repression are separable activities within the p53
polypeptide (
32
,
33
); the transactivation domain lies within the first 42 amino acids of p53 while
the transrepression function is encoded within two regions, one of which
overlaps with the transactivation domain, while the other lies within the C-terminal 66 amino acids. Other lines of evidence support the division of
the transactivation and repression functions. For example, three oncoproteins
(adenovirus E1B-19K, E1A and
bcl
2) specifically control repression but not transactivation (
34
-
36
). Moreover, the transactivation function of p53 appears to be dispensible for
p53-dependent apoptosis (
37
,
38
) but recent evidence has suggested that its transrepression function may be
important (
35
).
p53 is phosphorylated at multiple sites
in vivo
and by several different protein kinases
in vitro
(reviewed recently by Meek;
39
) including the protein kinase CKII (casein kinase II;
40
). CKII is a ubiquitous cyclic nucleotide-independent serine/threonine protein kinase which targets many cellular
proteins including several nuclear proteins involved in growth regulation (for reviews see
41
,
42
). Phosphorylation by CKII potently activates the sequence-specific DNA binding function of p53
in vitro
(
43
). In addition, we have shown previously that mutation of p53 at the CKII
phosphorylation site to encode alanine (which cannot be phosphorylated)
abolishes the growth suppressor activity of p53 when this mutant is expressed in several non-transformed or SV40-transformed mammalian cell lines, while partial suppressor activity
can be restored by encoding aspartic acid at this position, suggesting that the
additional negative charge may mimic, albeit weakly, the effect of
phosphorylation (
44
). These data were confirmed using a temperature-sensitive mutant p53 which permitted the experiments to be internally
standardised (
44
). Other groups have also examined the effects of introducing mutations at the
CKII site on various activities encoded within the p53 polypeptide (
12
,
45
,
46
). These experiments indicated that phosphorylation by CKII has little effect on
the regulation of p53-dependent transcriptional activation or the ability of p53 to block S
phase entry. However, loss of the CKII site significantly reduced the ability
of p53 to block cellular transformation by dominant oncogenes (
12
). The phosphorylation of p53 by CKII may therefore selectively regulate
specific activities of p53.
In this paper we have explored further the role of phosphorylation in regulating p53 biological function, in particular p53-dependent transcriptional modulation. Our data show that loss of the CKII
phosphorylation site has no detectable effects on the ability of p53 to
transactivate promoters containing p53 repsonsive DNA elements including the
p21
WAF1
promoter which is a physiological target of p53. However, we demonstrate clear
effects of the loss of phosphorylation at serine 386 on p53-dependent transcriptional repression. These results support a model in
which transactivation and repression are not only separable activities within
the p53 polypeptide, but are also regulated differentially by oncoproteins and
by phosphorylation.
The mammalian cell lines used in the study were as follows; SAOS-2 cells, which are a p53 null line derived from a human osteosarcoma (
47
) and a non-clonal population of p53-null murine fibroblasts which were obtained from David Lane (Dundee)
and originally derived from p53 knockout mice (
48
). The human fibroblast cell line GM701 was a kind gift from M. Jacobson
(University College, London); these cells express the SV40 large T antigen and
grow well at low passage. All cell lines were maintained in Dulbecco/Vogt
Modified Eagle's Medium with the addition of 10% foetal bovine serum and
supplemented with 2 mM glutamine, and 100 IU/ml each of penicillin and
streptomycin. The cells were grown as monolayers on plastic culture dishes at
37oC, 5% atmospheric CO
2
in a humidified incubator.
In all cases the reporter plasmids used to measure p53-dependent transcriptional effects employed the chloramphenicol acetyl
transferase (CAT) gene. The plasmids used to measure p53-dependent transactivation were as follows. pPG13CAT, which consists of the
polyomavirus early promoter and 13 upstream copies of a consensus p53 binding
site (PG or polygrip) (
9
). A control plasmid containing 15 copies of a mutated version of the p53
binding site in place of the PG sequences was used in parallel experiments.
This construct (pMG15CAT) is not activated by wild type p53. The other plasmid
was pWAF1CAT, containing the WAF1 promoter which is a natural target for p53
transactivation (
49
). These plasmids were a kind gift from Dr B. Vogelstein (John Hopkins Oncology
Center, USA).
The plasmids used to measure p53-dependent repression were as follows. pSV40CAT (Promega) contains the
simian virus 40 (SV40) immediate early promoter/enhancer fused upstream of the
CAT gene. This transcriptional unit has been shown to be repressed by wild type
p53
in vivo
(
24
,
50
). The plasmid pSVTKCAT was constructed by cloning the SV40 enhancer region from
pSV40CAT (as a
Hin
cII fragment) upstream of the thymidine kinase promoter from herpes simplex
virus in the p41X (pTKCAT) plasmid (a kind gift from Dr M. Jackson, Beatson
Institute, Glasgow). The plasmid pc-
fos
CAT contains 0.7 kb of the human c-
fos
promoter region cloned upstream of the CAT gene and was a kind gift from Dr G.
Glenn, Salk Institute, San Diego). The human c-
fos
promoter has also been shown to be repressed by wild type p53 (
23
,
26
). The plasmid pCMV[beta]gal expresses [beta]-galactosidase from the CMV early promoter and was obtained
from Dr B. McStay (Biomedical Research Centre, Dundee).
The plasmids expressing wild type or mutant p53 proteins have been reported
previously (
44
). These were pCMVNc9 (encoding wild type p53) and pCMVc5 (encoding a
transforming mutant of p53); both plasmids were obtained from M. Oren,
(Weizmann Institute, Israel). pCMVdel has p53 codons 1-330 deleted from pCMVNc9, while pCMVp53SA and pCMVp53SD have point
mutations encoding alanine and aspartic acid residues, respectively, at the
CKII site (serine 386) (
44
). The coding sequences of these plasmids have been checked to ensure that only
the desired mutations are present and the expression of the p53 proteins from
these plasmids has been checked in Rat1 and BHK cell lines (
44
). We have previously shown that the plasmids encoding the changes at the CKII
phosphorylation site express similar levels of p53 to the wild type plasmid
(pCMVNc9) in a number of cell lines (
44
).
The plasmids used in the transfections were all prepared by caesium chloride
density centrifugation. Cells were seeded at 2 * 10
5
cells per 5 cm dish (or 5 * 10
5
cells for the slower growing lines such as SAOS-2) and the DNAs were transfected into the cells by calcium phosphate
precipitation. A typical precipitation (to transfect a 5 cm diameter dish)
contained a total of 10 [mu]g DNA comprising 5 [mu]g of reporter plasmid and 5 [mu]g of a p53-encoding plasmid under control of the CMV early promoter (or vector-control plasmid). In some experiments 1 [mu]g of pCMV[beta]-gal was also included, allowing
measurement of [beta]-galactosidase activity as an internal standard for transfection
efficiency (see below). All precipitations were carried out in triplicate or
quadruplicate. The cells were harvested 48 h after transfection and the
monolayers were washed twice with chilled PBS to remove excess serum. The cells
were lysed using Reporter Gene Lysis Buffer (0.2 ml per plate; Promega), and
the lysates were heated to 65oC for 10 min to inactivate any endogenous deacetylase activities. The
lysates were cleared by microcentrifugation and the soluble fractions were
routinely stored at -80oC before assaying for CAT activity.
In order to examine the efficiency of transfection in this system, a [beta]-galactosidase reporter plasmid was co-transfected, and a single plate from each quadruplicate
transfection was stained with 1 mg/ml X-gal, 2 mM MgCl
2
, 5 mM K
3
Fe(CN)
6
, 5 mM K
4
Fe(CN)
6
in phosphate-buffered saline. The percentage of the cells in the dish whose nuclei stained blue gave an indication of the efficiency of the transfection method. Routinely, 3-10% of cells took up enough of the [beta]-gal plasmid to stain an intense blue (the fraction of
cells stained was always consistent within a set of transfections but varied
between experiments).
Incubations for the measurement of CAT activity contained: 100 [mu]l cell lysate, 3 [mu]l [
14
C]chloramphenicol (ICN; 100 [mu]Ci/ml, 105 mCi/mmol in 25 mM Tris pH 7.4), 14 [mu]l 3.5 mg/ml acetyl coenzyme A and 10 [mu]l distilled water. (The amounts of lysate in the reactions were
corrected for protein concentration as described by others;
24
,
34
.) Reactions were incubated at 37oC for periods varying from 10 to 19 h, depending on the levels of CAT
activity obtained from the promoter construct under investigation and the cell
type used. Reactions were extracted with 500 [mu]l ethyl acetate to partition the chloramphenicol into the solvent layer, and
480 [mu]l of the solvent phase was dried under vacuum. The dried reaction products
were resuspended in 30 [mu]l ethyl acetate and were applied to a TLC plate and allowed to dry
completely. The products were separated by chromatography in a 95:5
chloroform:methanol mixture in a pre-equilibrated tank. After drying, the plates were subjected to autoradiography to visualise the reaction
products, and to phosphorimage analysis using the Biorad GS-250 Molecular Imager and Molecular Analyst sofware to allow quantitation.
In the graphs, all points are displayed as the mean of triplicate assays
together with the standard deviation over the mean.
To examine the requirement for phosphorylation by CKII in the transcriptional
activation function of p53, plasmids expressing wild type or mutant p53
proteins encoding alanine or aspartic acid at the CKII phosphorylation site
were co-transfected in triplicate into p53-null fibroblasts or SAOS-2 cells together with an equal amount of pPG13CAT or pMG15CAT.
Following transfection, the cells were incubated for 48 h after which lysates
were prepared and analysed for CAT activity. The data (Fig.
1
A) show that expression of wild type p53 (lanes W) in the p53-null mouse fibroblasts gave rise to efficient transactivation of the
pPG13CAT construct containing a multiple-site p53-responsive element, but not the pMG15CAT construct which contains a
mutated version of the site. The negative control plasmid pCMVc5 (lanes M)
failed to cause transactivation of the reporter, confirming that the effect
observed is due to wild type p53 protein, and not a non-specific effect of p53 overexpression or of the CMV promoter. The p53
proteins altered at the CKII phosphorylation site to encode alanine or aspartic
acid (lanes A and D, respectively) were also able to transactivate the reporter
plasmid to an extent comparable to wild type p53 (see lanes for A and D co-transfected with pPG13CAT). These data indicated that, at the levels of
DNA transfected, alteration of the CKII phosphorylation site did not
significantly influence the ability of p53 to transactivate an artificial
reporter construct containing multiple copies of the p53 consensus binding
site. These results were confirmed using the human SAOS-2 cell line (data not shown) indicating that the effect was reproducible
in other cell types.
One possible explanation for the lack of activity changes observed with the
phosphorylation site mutants was that the levels of DNA transfected gave
saturating levels of p53 in the cells which could mask subtle changes in the
level of transcriptional activity. For example, others have shown that the
effects of deletion of the C-terminus of p53 are masked when high levels of plasmid are transfected and
only become apparent when the amount of DNA transfected is reduced accordingly
(
51
). In order to examine this possibility, the co-transfection experiments described above were repeated in the mouse p53-null fibroblasts using variable amounts of the pCMV-p53 constructs. The total amount of DNA transfected was
equalised with the pCMV-del construct to avoid any potential confusion from the number of copies
of the CMV promoter present in each cell. A range of amounts of the p53
constructs were tested, from 1 ng up to 5 [mu]g. In a preliminary transfection using pCMVNc9 (encoding wild type p53), it
was observed that maximal CAT activity was obtained using 0.5-1 [mu]g of this plasmid and that higher levels of plasmid led to a slight
reduction in the level of transactivation observed (data not shown), probably
due to sequestration of the basal transcription factors machinery by excess p53
molecules (
33
,
51
). The results of the dose-response analysis for the wild type and mutant proteins are shown in
Figure
2
. Each data point shows the mean of triplicate transfections together with the
standard deviation. The data for both pPG13CAT and pWAF1CAT indicate that the
levels of transactivation observed are dependent on the amount of p53-expressing plasmid transfected, with approximately background levels being
observed at the lowest dose (1 ng) and levels increasing up to the maximum of
100 ng shown here. When pPG13CAT was used as the reporter (Fig.
2
A), the responses of wild type p53 and the two phosphorylation site mutants at
each level of p53-expressing plasmid were essentially similar. A slight decrease in the
level of transactivation from the alanine 386 mutant (in comparison with the
wild type p53) was observed with the pPG13CAT plasmid (Fig.
2
A) but when the errors in the measurements were taken into account, this
difference was not significant. Moreover, when the pWAF1CAT reporter plasmid was examined (Fig.
2
B), there were no significant differences in the behaviour of the wild type p53
in comparison with the alanine or aspartic acid 386 mutants. These data
therefore confirm that the serine 386 mutations do not significantly influence
the ability of p53 to transactivate transcription mediated by a p53 responsive element, even at suboptimal levels of
p53 in the cell.
To determine whether phosphorylation by CKII played a role in transcriptional
repression by p53, reporter constructs with the c-
fos
promoter or the SV40 early promoter fused to the CAT gene were used; both of
these promoters are repressed by p53
in vivo
(
23
,
24
,
26
,
50
). The effects of the CKII phosphorylation site mutations on p53-dependent repression of the c-
fos
promoter in p53-null murine fibroblasts are shown in Figure
3
A. The data show that repression is mediated (although weakly) by wild type p53
(W) but not by the CMV vector alone (del). Lack of repression by the
transforming p53 mutant (M), indicated that the effect was not due to non-specific squelching by an excess of p53 protein. The alanine and aspartic
acid 386 mutants behaved much more like the transforming mutant in this assay,
showing essentially no repressive activity. These data therefore suggested that
loss of the phosphorylation site significantly reduced the ability of p53 to
repress the c-
fos
promoter. The effects of these p53 proteins on repression of a hybrid promoter
comprising the human herpes simplex virus thymidine kinase basal promoter with
the SV40 enhancer element were were also examined in p53-null fibroblasts. Although this was a composite rather than a natural
promoter, the level of transcription was higher (3-4-fold) than with the c-
fos
promoter (the level of transcription, and hence repression, was barely
detectable with the thymidine kinase promoter alone; Fig.
3
B). Once again, the phosphorylation site mutants were significantly less able to
repress transcription than wild type p53. (Owing to errors in measuring the low
levels of transcription observed with the Fos and SVTK promoter plasmids, it
was difficult to discern differences between the alanine and aspartic acid
mutants.) Similar results were obtained using SAOS-2 cells and when the complete SV40 early promoter was examined in both
cell types (data not shown).
In this paper, we describe the effects of introducing mutations at the CKII site
in mouse p53, on the ability of the protein to behave as a transcriptional
regulator. Two mutations were examined: the first of these replaced serine 386
with alanine which cannot be phosphorylated; the other change introduced an
aspartic acid, and therefore a constitutive negative charge at this position.
When expressed in murine fibroblasts or the human osteosarcoma line SAOS-2, both of which lack endogenous p53, the phosphorylation site mutant
proteins were indistinguishable from wild type p53 in their abilities to
transactivate a promoter containing tandem synthetic p53 DNA responsive
elements, or the WAF1 promoter which is a natural target for p53 (Fig.
1
). Moreover, even at progressively lower levels of p53 expression, where the
amount of p53 available for transactivation becomes limiting, the mutant
proteins were still able to mimic wild type p53 (Fig.
2
), indicating that the change of residue at the phosphorylation site had no
significant role in transcriptional activation. Several other groups have reported similar findings (
12
,
45
,
46
) and our data are therefore in full agreement with these publications. This
result is perhaps surprising because activation of the sequence-specific DNA binding function of p53 through phosphorylation by CKII (
43
) might be expected to stimulate p53 as a transcriptional activator. However, it
is possible that other mechanisms in the cell (e.g. phosphorylation of p53 at
other sites by different enzymes) may be able to fully activate p53-dependent transactivation and overcome any possible contribution from
phosphorylation of p53 at the CKII site. The CKII mutants have also been
reported to efficiently mediate G1 growth arrest (
12
,
45
,
46
), an activity which is closely linked to the transactivation capacity of p53 (
12
,
13
).
In contrast to the lack of effects on transcriptional transactivation, the data
in this paper consistently show that the presence of serine at position 386 is
an important factor in achieving a high level of p53-dependent transcriptional repression (Figs
3
and
4
). The magnitude of this effect (and indeed of the level of repression with wild
type p53) appears to be promoter and cell type specific (other researchers have
made similar observations;
28
). Although the experimental errors were generally high in examining repression
of the (weak) c-
fos
promoter, the general trend of the results was consistent, indicating that
mutation of the CKII phosphorylation site resulted in lower repression activity
of the p53 (Fig.
3
). In contrast, the high levels of expression of the SV40 promoter in the GM701
cells and the potency of p53-dependent repression in this system allowed accurate quantitative
measurement of the effects of the p53 mutants (Fig.
4
). Moreover, the intermediate effect of the aspartic acid mutant was observed
much more clearly and reproducibly in the GM701 cells (Fig.
4
) and is consistent with an important role for phosphorylation of this residue.
One potential contributing factor to the transcriptional measurement in the
GM701 cells is the presence of SV40 which has a well-established role in the stabilisation and accumulation of p53 in the cell
(including p53 uncomplexed with T antigen). This is a phenomenon which occurs
as part of the activation of p53 in response to DNA damaging agents (
2
,
3
,
52
). SV40 may therefore stimulate molecular pathways which activate p53 and the
effects of phosphorylation at the CKII site may well be intensified under such
conditions. It is also possible that T antigen may itself have effects on
repression of the SV40 promoter through its ability to block wild type p53
function. However, although T antigen is present in the GM701 cells, the
differences in the levels of p53-mediated repression are still observed. Moreover, since the p53-mediated repression is stronger in these cells, we were able to
accurately measure the effects of the loss of phosphorylation at serine 386.
The mechanism by which phosphorylation of p53 at the CKII site contributes to
efficient repression is not known, but may be mediated (in this case) at least
partly through the interation of T antigen and p53 (our previously published
data indicated that complex formation between p53 and T antigen was unaffected
by phosphorylation of serine 386;
44
). Similarly, the oligomerisation status of p53 is not affected by
phosphorylation at serine 386 (data not shown). p53 is also able to bind
directly to a number of basal and other transcription factors (including TFIID
components, the CAAT-binding factor, Sp1 and AP-2;
31
,
53
-
57
) and may repress transcription by preventing the interaction of certain factors
with the promoter region (
30
,
55
). It is therefore possible that phosphorylation of the C-terminal tail of p53 signals a conformational shift which enables the
protein to bind more efficiently to other transcription factors. A similar
mechanism is thought to unmask the specific DNA binding domain of p53 in
response to CKII phosphorylation (
43
) and it is possible that movement of this tail could additionally stimulate interaction with other transcription factors. The data presented in
this paper, indicating that phosphorylation can regulate repression
independently of transactivation, are also consistent with the finding that
domains of the p53 protein responsible for transactivation and repression
overlap, but are not identical (
32
,
33
).
The finding that phosphorylation of a C-terminal site can selectively regulate repression but not transactivation
is striking but not unprecedented. For example the repression function of the
Fos protein is also controlled through the phosphorylation of C-terminal sites while the transactivation capacity remains unaltered (
58
). Moreover, other mechanisms exist which can control p53-dependent repression independently of transactivation. For example, three
oncoproteins (adenovirus E1B-19K, E1A and
bcl
2) specifically control repression but not transactivation (
34
-
36
). The finding that viral oncogene products, cellular oncogene products and
phosphorylation each permit independent modulation of transactivation and repression implies a strong evolutionary
pressure for the selective regulation of different activities within the p53
protein.
We and others have previously studied the effects of p53 proteins with
alterations at the CKII phosphorylation site of p53. A number of different assays measuring separate activities within the p53 protein have
been examined in these studies. Thus, the mutants were reported to be unaltered
in specific DNA binding (
59
), transcriptional transactivation of various promoters (
12
,
45
,
46
), G1 growth arrest (
45
,
46
), inhibition of drug-resistant colony formation (
46
) and cellular transformation by mutant forms of p53 (
60
). In one study however, a minor reduction in inhibition of the growth of SAOS-2 cells was observed while there was a significant (5-10-fold) reduction in the ability to block transformation by
activated
ras
and E7 oncogenes (
12
). Moreover, our own work has attributed an important role for phosphorylation
at the CKII site to the ability of p53 to block the growth of G418-resistant colony formation, particularly in SV40-transformed cell lines (
44
) (this assay was used as a measure of the anti-proliferative activity of p53, but it did not discriminate between G1
arrest or apoptosis). Additionally, we have shown in the present study that
repression is also dependent on the integrity of the CKII site. Taken together,
these studies suggest that phosphorylation by CKII can selectively regulate
particular functions within the p53 protein and that the magnitude of
phosphorylation-dependent changes in activity may vary according to factors such as cell
type and growth or transformation status. It will be important to determine the
relationship between these regulated functions and to show whether
phosphorylation by CKII
per se
is responsible for controlling p53 in the cell. Experiments are underway in our
laboratory to address these issues.
We thank Frances Fuller-Pace and Uwe Knippschild for helpful comments. We are also grateful to M.
Jackson, B. Vogelstein and M. Jacobson for the gifts of reagents. DWM is the
recipient of a Medical Research Council Senior Fellowship.
REFERENCES
Return