PTEN tumor suppressor regulates p53 protein levels and activity through phosphatase-dependent and -independent mechanisms.

Abstract

We show in this study that PTEN regulates p53 protein levels and transcriptional activity through both phosphatase-dependent and -independent mechanisms. The onset of tumor development in p53(+/-);Pten(+/-) mice is similar to p53(-/-) animals, and p53 protein levels are dramatically reduced in Pten(-/-) cells and tissues. Reintroducing wild-type or phosphatase-dead PTEN mutants leads to a significant increase in p53 stability. PTEN also physically associates with endogenous p53. Finally, PTEN regulates the transcriptional activity of p53 by modulating its DNA binding activity. This study provides a novel mechanism by which the loss of PTEN can functionally control "two" hits in the course of tumor development by concurrently modulating p53 activity.

Full text

Page 1

A R T I C L E
PTEN tumor suppressor regulates p53 protein levels and
activity through phosphatase-dependent and -independent
mechanisms
Daniel J. Freeman,1,5Andrew G. Li,3,5Gang Wei,3Heng-Hong Li,3Nathalie Kertesz,1Ralf Lesche,1,6
Andrew D. Whale,3Hilda Martinez-Diaz,1Nora Rozengurt,2Robert D. Cardiff,4Xuan Liu,3,*
and Hong Wu1,*
1Howard Hughes Medical Institute, Department of Molecular and Medical Pharmacology
2Department of Pathology and Laboratory Medicine
UCLA School of Medicine, Los Angeles, California 90095
3Department of Biochemistry, University of California, Riverside, California 92521
4Department of Pathology, Center for Comparative Medicine, University of California, Davis, California 95616
5These two authors contributed equally.
6Present address: Epigenomics AG, Kastanienallee 24, 10435 Berlin, Germany.
*Correspondence: hwu@mednet.ucla.edu (H.W.), xuan.liu@ucr.edu (X.L.)
Summary
We show in this study that PTEN regulates p53 protein levels and transcriptional activity through both phosphatase-
dependent and -independent mechanisms. The onset of tumor development in p53?/?;Pten?/?mice is similar to p53?/?
animals, and p53 protein levels are dramatically reduced in Pten?/?cells and tissues. Reintroducing wild-type or phospha-
tase-dead PTEN mutants leads to a significant increase in p53 stability. PTEN also physically associates with endogenous
p53. Finally, PTEN regulates the transcriptional activity of p53 by modulating its DNA binding activity. This study provides
a novel mechanism by which the loss of PTEN can functionally control “two” hits in the course of tumor development by
concurrently modulating p53 activity.
Introduction
al., 1999). Many cancer-related mutations have been mapped
within the conserved catalytic domain of PTEN, suggesting that
the phosphatase activity of PTEN is required for its tumor sup-
pressor function. In addition to its phosphatase domain, PTEN
also contains a putative C2 regulatory domain, a PIP2 binding
motif, two PEST homology regions, and a consensus PDZ bind-
ing site. The functions of these structural domains include regu-
lating the phosphatase activity and controlling the stability or
subcellular localization of PTEN (Maehama et al., 2001). Disrup-
tion of the murine Pten locus further supports the importance
of PTEN as a bona fide tumor suppressor (Di Cristofano et al.,
1998; Podsypanina et al., 1999; Suzuki et al., 1998). Homozy-
gous deletion of Pten results in early embryonic lethality. Pten
heterozygous mice develop hyperplastic/neoplastic changes in
multipleorgansasearlyas4weeks,withabiastowardincidence
in lymphoid, endometrial, mammary, gastrointestinal, and adre-
nal gland tissues (Podsypanina et al., 1999; Suzuki et al., 1998).
The major function of the tumor suppressor PTEN relies
on its phosphatase activity and subsequent antagonism of the
PTEN (phosphatase and tensin homolog deleted on chromo-
some 10) tumor suppressor gene was the first phosphatase
identified to be frequently mutated/deleted somatically in vari-
ous human cancers (Li et al., 1997, 1998; Steck et al., 1997).
In addition, germline mutations in the PTEN gene have been
associated with Cowden syndrome and related diseases in
which patients develop hyperplastic lesions (harmatomas) in
multipleorganswithincreasedrisksofmalignanttransformation
(Liaw et al., 1997; Marsh et al., 1998; Nelen et al., 1997). PTEN
contains a sequence motif that is highly conserved in the mem-
bers of the protein tyrosine phosphatase family. PTEN has been
shown to possess phosphatase activity on phosphotyrosyl and
phosphothreonyl-containing substrates (Li et al., 1998; Myers
and Tonks, 1997) in vitro and on phosphatidylinositol (3,4,5)
trisphosphate, a product of phosphatidylinositol 3-kinase
(PI3K), both in vitro and in vivo (Lee et al., 1999; Maehama and
Dixon, 1998; Myers et al., 1998; Stambolic et al., 1998; Sun et
S I G N I F I C A N C E
It has been shown by many groups, including our own, that the major function of the PTEN tumor suppressor relies on its phosphatase
activity by antagonizing the PI3K/AKT pathway. This study provides evidence that PTEN can modulate p53 function independently
of its phosphatase activity, and that a physical interaction between PTEN and p53 is important for this modulation in vivo. We also
provide in vivo evidence for PTEN’s function inside of the nucleus. Consequently, our study provides a novel mechanism by which
PTEN regulates p53 levels and activities, and possibly provides an avenue for targeted therapy for cancers caused by PTEN or p53
loss.
CANCER CELL : FEBRUARY 2003 · VOL. 3 · COPYRIGHT  2003 CELL PRESS117

Page 2

A R T I C L E
PI3K/AKT pathway (Di Cristofano and Pandolfi, 2000; Maehama
et al., 2001; Simpson and Parsons, 2001). Loss of PTEN func-
tion, either in murine embryonic stem cells or in human cancer
cell lines, results in accumulation of PIP3 and activation of its
downstream effectors, such as AKT/PKB (Franke et al., 1997;
Stambolic et al., 1998; Sun et al., 1999; Wu et al., 1998). As a
serine/threonineproteinkinase,AKTfunctionsbyphosphorylat-
ing key intermediate signaling molecules, leading to increased
cell metabolism, cell growth, cell survival, and cell invasive-
ness—all hallmarks of cancer (for review, see Downward, 1998;
Hanahan and Weinberg, 2000; Hill and Hemmings, 2002). Ge-
netic studies also indicate that AKT plays an essential role in
PTEN-controlled signal transduction and tumorigenesis (Stiles
et al., 2002; Stocker et al., 2002).
Recent studies suggest that there is a tight link between
PTEN and p53, another major tumor suppressor. Both PTEN
and p53 regulate cell proliferation and cell death (for review,
see Bargonetti and Manfredi, 2002; Yamada and Araki, 2001).
Even though PTEN and p53 are frequently mutated in both
inherited and spontaneous tumors in humans, concomitant mu-
tations in both p53 and PTEN loci are not common (Kurose et
al., 2002; Fujisawa et al., 2000; Kato et al., 2000; Koul et al.,
2002; Rasheed et al., 1997). Several commonly used human
PTEN null tumor cell lines, such as U87 MG, 786-O, MCF-7,
and LNCap, are wild-type for p53. These genetic observations
suggest that loss of PTEN may reduce the selective pressure on
tumors during cancer progression to lose p53. Mechanistically,
several lines of evidence suggest that PTEN-controlled AKT
activation plays an essential role in modulating MDM2-depen-
dent p53 degradation (for review, see Mayo and Donner, 2002;
Zhouand Hung,2002). p53isa short-livedprotein andstabiliza-
tion of p53 is crucial for its tumor suppressor function. The p53
protein is maintained at low levels in normal cells by MDM2-
mediatedubiquitinationandsubsequentproteolysis.AKTphysi-
cally associates with MDM2 and phosphorylates MDM2 on ser-
ine residues, a key step involved in MDM2 nuclear translocation
andMDM2-mediatedp53degradation(MayoandDonner,2001;
Ogawara et al., 2002; Zhou et al., 2001). Therefore, it is conceiv-
able that activation of AKT in PTEN-deficient cells may cause
more rapid p53 degradation, which could contribute to tumori-
genesis caused by PTEN loss. Furthermore, p53 can bind to
the PTEN promoter region and activate PTEN transcriptionally
(Stambolic et al., 2001), although there is a debate about the
significance of such regulation (Sheng et al., 2002).
To test the functional significance of the crosstalk between
PTEN and p53 in vivo, we employed a genetic approach to
study tumorigenesis in Pten?/?;p53?/?compound heterozygous
mice on a 129/Balb/c genetic background. We found that the
onsetoflymphomadevelopmentin p53?/?;Pten?/?miceissimi-
lar to that seen in p53?/?animals. To further investigate the link
between PTEN and p53, we demonstrated that p53 protein
levels are dramatically reduced in Pten?/?cells and that PTEN
stabilizes p53 by increasing its half-life. Furthermore, ectopic
expression of PTEN phosphatase-dead mutants also leads to a
significantincreaseinp53proteinlevels,andPTENcanstabilize
p53, even in the absence of MDM2, suggesting that PTEN can
regulate p53 levels in a phosphatase-independent and MDM2-
independentmanner.Significantly,weobservethatPTENphysi-
cally associates with endogenous p53, and PTEN regulates the
transcriptional activity of p53 by modulating its DNA binding.
Thus, PTEN-p53 may represent a self-reinforced circuit impor-
tant for maintaining normal cellular functions, and disruption of
this circuit may lead to cancer formation.
Results
PTEN and p53: A genetic study
We have generated a Pten-deficient mouse line by homologous
recombination via embryonic stem cells (ES cells). Similar to
previous reports, mice lacking PTEN died early during embryo-
genesis, and mice with heterozygous Pten deletion developed
tumors in various tissues (Figure 1A). However, the conse-
quences of Pten inactivation on a 129/Balb/c genetic back-
ground are quite different from what was reported on a 129/
C57 background (R. Lesche et al., submitted). In particular,
most of the heterozygous mice are tumor free for the first six
monthsoftheirpostnatallife:10% ofPten?/?animalsdeveloped
focal lymphoid hyperplasia in the thymus and enlarged lymph
nodes near the neck and axilla (Figure 1A, left). Among aged
animals (7–14 months), 25% of them developed lymphoid hy-
perplasia/dysplasia (Figure 1A, right) and 10% of those pro-
gressed to early lymphoblastic lymphomaof T cell origin, similar
to previous reports (Podsypanina et al., 1999; Suzuki et al., 1998).
This prolonged latency in tumor development provided us
with a unique opportunity to genetically evaluate the possible
interaction between PTEN and p53. We generated mice hetero-
zygous for both p53 and Pten (compound heterozygous,
p53?/?;Pten?/?) on the 129/Balb/c background and followed a
cohort of compound heterozygous animals (n ? 33; male:
female ? 14:19) for six months. Seven p53?/?;Pten?/?mice
died between 2–6 months of age. The rest of the animals were
subjected to pathological analysis upon tumor development
from as early as 1 month of age. As shown in Figure 1B, the
onset (left panel) and lymphoma (right panel) development in
the p53?/?;Pten?/?mice are very similar to those of p53?/?(n ?
23) mice, but differ significantly from mice heterozygous for p53
(Macleod and Jacks, 1999) and for Pten (Figure 1A). Similarly
to what was reported in p53?/?animals, p53?/?;Pten?/?mice
developlymphomaspreferentiallyinthelymphnodes.Asshown
in Figure 1C, the enlarged lymph nodes are tightly packed with
a dense population of homogeneous small lymphocytes with a
few scattered germinal centers. Immunohistochemical staining
using anti-CD3 antibody reveals nodules of uniform, atypical,
immature T cells in the peri-follicular zones, suggesting early
lymphoblastic lymphoma of T cell origin. The majority of these
samples have well-defined cortex and medullary with some
highly vascularized areas. In 40% of cases, neoplastic lympho-
cytes are packed in the vessels and have infiltrated the sur-
rounding tissues (data not shown).
To ensure that the initiation of tumor development in
p53?/?;Pten?/?mice is not due to the loss of the second allele
of either the Pten or p53 gene (loss of heterozygosity, LOH),
immunohistochemical (IHC) analysis was performed on the tu-
mors derived from the compound heterozygous (p53?/?;
Pten?/?) mice. Figure 2A shows that PTEN protein is present in
these tumors (lower left) at levels similar to adjacent normal
tissues (lower right) and comparable to those of Pten?/?mice
(upper left). However, p53 is expressed at basal levels in most
of the tissues and tumor samples and cannot be detected by
conventional IHC analysis (data not shown). Thus, we con-
ducted PCR and Western blot analysis. The second alleles of
Pten and p53 are retained in the tumor samples (Figure 2B),
118
CANCER CELL : FEBRUARY 2003

Page 3

A R T I C L E
Figure 1. Tumor development in Pten?/?and
compound heterozygous p53?/?;Pten?/?mice
A: Tumor onset curve (left panel) and spectra
(right panel) show the tumor susceptibility in
Pten?/?mice on the 129/Balb/c background.
B: Tumor onset curve shows that the tumor sus-
ceptibility in p53?/?;Pten?/?mice is greatly en-
hanced over p53 or Pten heterozygous mice but
similar to p53?/?mice (left panel); the incidence
of lymphoma formation in the compound het-
erozygous animals is also similar to that of the
p53 null mice (right panel).
C: Low magnification images showing an early
lymphoblastic lymphoma in a lymph node; left,
H/E staining showing the architecture; middle,
B220-positivestainingBcells;andright,CD3-posi-
tive stained T cells. Note that the central mass of
the immature cells is most positive for CD3. Bar ?
200 ?M.
and a substantial amount of p53 protein can also be detected
in these tumors using an antibody that recognizes both WT and
mutated form of p53 (Pab-240, Figure 2C, upper panel). To
ensure that WT p53 is present in the mutant sample, we con-
ducted IP-Western blot analysis using an antibody that only
reacts with WT p53 (Pab-246, Figure 2C, lower panel). It ap-
pears, from this study, that WT p53 is present in the tumor
samples. These results lead to our conclusion that LOH is not
required for the initiation of tumor development in these com-
pound heterozygous mice. However,we cannot exclude (1) loss
of function mutations in either the p53 or Pten genes which
cannot be detected by our assay systems; (2) localized or single
cell-based allelic deletion or gene silencing events; and (3) that
LOH is involved in or associated with tumor progression and
local invasion.
PTEN modulates p53 protein stability
To investigate whether PTEN indeed controls p53 function in
vivo, we first measured endogenous p53 mRNA and protein
levels in isogenic Pten?/?, Pten?/?, and Pten?/?ES cells. p53
protein levels are significantly reduced in Pten heterozygous ES
cells and almost completely lost in Pten null ES cells (Figure
3A, upper panel), while these ES cells do express similar levels
of p53 mRNA (Figure 3A, lower panel), suggesting that PTEN
modulates p53 at the posttranscriptional level. To confirm that
PTEN is capable of controlling p53 protein levels in vivo and to
prove that this regulation is a common mechanism rather than
a cell type-specific event, we measured p53 levels in brain
tissues derived from Pten conditional knockout mice (Groszer
et al., 2001). Similarly to ES cell studies, p53 levels are reduced
in heterozygous animals and almost completely lost in Pten null
brains (Figure 3B). To develop a system that is amenable to
CANCER CELL : FEBRUARY 2003 119

Page 4

A R T I C L E
Figure 2. The presence of PTEN and p53 in lym-
phomas generated from compound heterozy-
gous p53?/?;Pten?/?mice
A: Immunohistochemical detection of PTEN pro-
tein. Positive signal can be detected in the brain
tissue of Pten?/?heterozygous mice (upper left),
the lymphoma sample of p53?/?;Pten?/?com-
pound heterozygous mice (lower left), the nor-
malmammaryglandadjacenttothelymphoma
seen in the lower left (lower right), but not in
thebraintissues frombrain-specificPtendeletion
mice (Groszer et al., 2001).
B: PCR analysis of tail DNA and tumor DNA from
representative p53?/?;Pten?/?compound het-
erozygous mice, indicating the presence of the
WT alleles of both p53 and Pten genes.
C: Upper panel, Western blot analysis using anti-
p53 antibody (Pab 240, Santa Cruz technology).
p53 proteins are present in the representative
tumor samples of p53?/?;Pten?/?compound het-
erozygous mice; lower panel, cell lysate from the
tumor tissue was immunoprecipitated with an
anti-p53 antibody which specifically reacts to
the WT p53 protein (Pab 246, Santa Cruz Bio-
technology) and immunoblotted with Pab 240
antibody.
biochemicalanalysis,wegeneratedseveralisogenicmouseem-
bryonic fibroblast (MEF) lines, which are wild-type for p53 but
differ in their Pten status, first by immortalizing MEF cells from
Pten conditional knockout mice (Lesche et al., 2002), Ptenloxp/loxp
(functional equivalent to WT [L/L]),and then by confirming PTEN
and p53 expression by Northern and Western blot analyses
(data not shown). Immortalized Ptenloxp/loxplines, which contain
both p53 and PTEN protein products, were treated transiently
with Cre recombinase. Three independent clones, which carry a
homozygous Ptendeletion(Figure3C,upperpanel; Pten?loxp/?loxp,
functional equivalent to null mutant [?/?]) but are wild-type for
p53, were generated and used for studies described below.
Similar to our analysis using ES cells and brain tissues, p53
protein levels, but not mRNA levels, were significantly reduced
in Pten?loxp/?loxpMEF cells (Figure 3C).
To determine the cause of the reduced p53 protein levels
in Pten null cells and tissues, we measured the half-life of p53
in Ptenloxp/loxpand Pten?loxp/?loxpMEF cells. Notably, the half-life
of p53 in Ptenloxp/loxpMEF cells was significantly longer than in
Pten?loxp/?loxpMEF (17 min versus 6 min, Figure 3D), providing
genetic evidence that PTEN controls endogenous p53 protein
levels by modulating its protein stability. Interestingly, the half-
life of p53 falls in between WT and Pten null cells (12 min) when
Pten?loxp/?loxpMEF cells are pre-treated with wortmannin, a PI3
kinase inhibitor (see Supplemental Data at http://www.cancercell.
org/cgi/content/full/3/2/117/DC1), suggesting that PTEN may
control p53 protein levels via both PI3 kinase/AKT-dependent
and -independent mechanisms.
PTEN affects p53 protein stability: phosphatase
and MDM2-independent mechanisms
In order to determine whether reduced p53 stability in Pten null
cells and tissues is due to a PTEN phosphatase-independent
activity,wereintroducedeitherwild-typeorPTENphosphatase-
deadmutantsintothe Pten?loxp/?loxpMEFcells.AsshowninFigure
4A, reintroducing WT PTEN led to a significant increase in p53
120
CANCER CELL : FEBRUARY 2003

Page 5

A R T I C L E
Figure 3. PTEN controls p53 protein stability
p53 protein levels in (A) ES cells and (B) brain tissues with different Pten status, indicating that PTEN regulates p53 protein posttranscriptionally.
C: PTEN regulates p53 protein levels in isogenic MEF cell lines (upper panel). The absence of Pten gene expression and presence of p53 mRNA in Pten?loxp/?loxp
MEFs are demonstrated by Northern blot analysis (lower panel).
D: PTENcontrols p53half-life. Ptenloxp/loxpand Pten?loxp/?loxpMEF cells werecultured to70% confluencebefore beingtransferred to serumfree mediumcontaining
50 ?g/ml cyclohexamide. Cells were directly lysed at the indicated time points and p53 proteins were detected by Western blot analysis. The half-life of
p53 protein in the presence (first row) and absence (second row) of endogenous PTEN proteins were calculated using Quantity One (Bio-Rad).
proteinlevels(comparinglanes2and3).Interestingly,twophos-
phatase-dead mutants, PTEN-CS and PTEN-GR, as well as the
PTEN-C2 domain alone, were also able to increase p53 protein
levels, even though they did not affect AKT phosphorylation, a
hallmark of PTEN’s phosphatase activity (Figure 4A, comparing
lane3withlanes4–6).Wefurthertestedtheabilitiesofwild–type
or PTEN-CS to modulate human p53 protein levels using p53
null human osteosarcoma SAOS2 cells. Again, both the WT and
PTEN-CS mutants led to a 5- to 6-fold increase in transfected
p53 protein levels but not its mRNA levels (Figure 4B). These
data suggest that (1) PTEN controls p53 protein levels in both
human and murine cells; and (2) PTEN controls p53, at least in
part, by a mechanism independent of its phosphatase activity.
The control of p53 protein levels is orchestrated by multiple
factors with MDM2 being one of the major players. Thus, it is
not surprising that increased AKT activity in Pten-deficient cells
will leadto increased MDM2phosphorylation andnuclear trans-
location, resulting in p53 degradation. Similar to previous re-
ports, MDM2 nuclear translocation is indeed increased in our
Pten?loxp/?loxpcells (see Supplemental Data on Cancer Cell web-
site). However, the MDM2-dependent pathway is probably not
the only mechanism by which PTEN can control p53. The fact
that the PTEN phosphatase-dead mutants could regulate p53
levels (Figures 4A and 4B) suggests that PTEN can control p53
by multiple mechanisms, one of which does not require its
phosphatase activity and thus may be MDM2 independent. To
genetically prove that PTEN can control p53 in a MDM2-inde-
pendent manner, we took advantage of a p53?/?;Mdm2?/?MEF
cell line. As illustrated in Figure 4C, the expression of PTEN led
to a near 7-fold increase in p53 protein levels in the absence
of MDM2 (Figure 4C, lanes1 and 2). Furthermore, PTEN expres-
sion could partially prevent p53 degradation caused by overex-
pression of MDM2 (Figure 4C, lanes 3 and 4). To further investi-
gate the mechanism involved in MDM2-independent p53
stabilization by PTEN, we measured the half-life of p53 in
p53?/?;Mdm2?/?MEF cells by transfecting p53 with and without
PTEN. The half-life of p53 is significantly prolonged in the PTEN
co-transfected cells (15 hr; Figure 4D, ?PTEN), as compared
top53alone(10hr;Figure4D,?PTEN).Again,PTEN-CSmutant
had similar effect on p53 protein levels (see Supplemental Data
at http://www.cancercell.org/cgi/content/full/3/2/117/DC1).
How PTEN modulates p53 half-life in the absence of MDM2
CANCER CELL : FEBRUARY 2003 121