I-'~[~,Av
TRENDS in Cell Biology Vo1.11 No,ll November 2001
,q27
C e l l u l a r s e n e s c e n c e as a tumor-suppressor mechanism Judith Campisi Organisms with renewable tissues had to evolve mechanisms to prevent the development of cancer. One such mechanism is cellular senescence, which irreversibly arrests the growth of cells at risk for neoplastic transformation. Recent findings have revealed the complexities of the senescence phenotype and unexpected possible consequences for the organism. Ceil division is essential for the survival o f multicelhilar
we have learned that this process, n o w k n o w n as replica-
organisms that contain renewable tissues. However, cell
tive senescence, is driven by telomere shortening.
division also puts organisms at risk for developing cancer.
Telomeres - the repetitive DNA sequence (TTAGGG
Genomes are continually damaged by environmental in-
in vertebrates) and specialized proteins that cap the ends
suits, oxidative metabolism, and, in dividing cells, errors in
o f linear c h r o m o s o m e s - are essential for chromosomal
DNA replication and mitosis. Depending on the level and
integrity. O w i n g to the biochemistry o f DNA replication,
type o f damage, cells can attempt repair, or die. In dividing
5 0 - 2 0 0 bp o f telomeric DNA are not replicated during
cells, the major risk from genomic damage is mutations,
each S phase. Because telomerase, the enzyme that can
w h i c h are generated by failures or mistakes in repair. If a
synthesize telomeric DNA de n0v0, is not expressed by most
mutation confers a growth or survival advantage, or causes
h u m a n cells, telomeres shorten with each cell cycle.
the g e n o m e to become unstable (and thus hypermutable),
W h e n the telomeres erode from their m a x i m u m size o f
the stage is set for the development o f cancer (oncogenesis).
10-15 kb (in the germ line) to an average size o f 4 - 6 kb,
Complex organisms have evolved at least two cellular
h u m a n cells irreversibly arrest growth, producing a char-
mechanisms to suppress the proliferation (used here inter-
acteristic (senescent) phenotype z'a. Notably, the senes-
changeably with ' g r o w t h ' ) o f cells at risk for oncogenic
cence arrest occurs before the telomeres become short
transformation: apoptosis or p r o g r a m m e d cell death, and
e n o u g h to c o m p r o m i s e chromosomal integrity. The strin-
cellular senescence or the senescence response. Cellular
gency o f the senescence response, and w h e t h e r short
senescence irreversibly arrests cell growth and is a major
telomeres or other factors (discussed below) induce the
barrier that cells must overcome in order to progress to
response, is highly species dependent.
full-blown malignancy 1,2. In this regard, cellular senescence
Two points regarding replicative senescence are note-
is similar to apoptosis. However, whereas apoptosis kills
worthy. First, it is very likely that cells respond to dis-
and eliminates potential cancer ceils, cellular senescence
r u p t i o n o f the telomere structure, rather than respond to
irreversibly arrests their growth.
I ATRENDS Guidtoe 1 Cancer Biology
telomere shortening per se4. Second, telomerase is ex-
Recent findings have shed n e w light on the causes o f
pressed in the germ line, early embryonic and a few adult
cellular senescence, the complexity of the senescence pheno-
cells, and most t u m o r ceils. Expression o f telomerase
type and the potential consequences o f cellular senescence
is the most c o m m o n mechanism by w h i c h cancer ceils
for the organism. These findings are discussed here, with
stabilize their telomeres and hence avoid replicative
an emphasis on the senescence response o f h u m a n cells.
senescence 3. Recently, stimuli having little or no impact on telo-
Judith Campisi Life Sciences Division,
Potentially oncogenic events cause cellular senescence
meres were s h o w n to induce normal ceils to arrest growth
Lawrence Berkeley
w i t h a senescence phenotype z. These stimuli include DNA
National Laboratory,
Cellular senescence was first recognized more than 40 years
damage, chromatin remodeling and strong mitogenic sig-
1 Cyclotron Road,
ago as a process that prevented normal h u m a n fibroblasts
nals. Thus, replicative senescence is an example o f a m o r e
Berkeley, CA 94720, USA.
from growing indefmitely in culture 1,2. In the past decade,
general process, termed here as 'cellular senescence'.
e-mail:
[email protected]
http://tcb,trends.com 0962-8924/01/$-seefrontmatter©2001EIsevierScienceLtd.AIIrightsreserved. PI]:S0962-8924(01)02151-1
$28
TRENDS in Cell B i o l o g y Vo1.11 No.ll
November 2001
shorteningt0 J2, Moreover, ceils that express telomerase Senescence signals
E2F, DNAdamage ' i z , z /
~
\
nonetheless senesce in response to stimuli such as onco-
, Ras, short telomeres
\
,
genic mutants of Ras 1°. Thus, diverse stimuli, not solely telomere shortening, induce a senescence response.
\
What do these stimuli have in common? All have the potential to cause or contribute to cancer. Telomere eroCBP/p300
sion iaevitably leads to genomic instability, and thus hypermutability. Likewise, DNA damage can cause mutations (in oncogenes or tumor-suppressor genes), chromosomal aberrations and genomic instability. Chromatin disrnp tion, particularly loss of silencing, can derange normal differentiation, causing unregulated growth, invasiveness and other properties typical of tumor cells. In addition,
~'
p21
supraphysiological mitogenic signals can, of course, drive
Cell cycle/ senescence arrest
unregulated growth. Thus, cellular senescence appears to
TRENDS in Cell Biology
be a mechanism for irreversibly arresting the growth of cells at risk for tumorigenesis.
Figure 1. Control of cellular senescence by the p53 pathway Shown are the consequences of senescence-inducing signals on the oncogenes (highlighted in green) and tumor-suppressor genes (highlighted in red) in the p53 tumor-suppressor pathway. Broken arrows indicate effects that are presumed or hypothesized. Solid arrows indicate effects that are supported by experimental evidence. Senescence-inducing signals such as overexpression of E2F or DNA damage, are presumed to decrease the activity of TBX2, a transcription factor that represses the p14ARFpromoter. The decline in TBX2 activity causes pl 4ARFexpression to rise. pl 4ARF sequesters MDM2, a protein that facilitates p53 degradation, which leads to an increase in p53 activity. Senescence-inducing signals such as oncogenic Ras, short telomeres and possibly other signals, increase the expression of promyelocytic leukemia protein (PML), which interacts with two acetyltransferases - CBP and p300. The PML-acetyltransferase complex acetylates p53, which stimulates its activity, p53 induces the transcription of several genes, including the cyclin-dependent protein kinase (CDK) inhibitor p21, that cause or facilitate the senescence arrest.
Tumor suppressors control cellular senescence Consistent with its role in suppressing cancer, cellular senescence is controlled by several tumor-suppressor genes 13,14.The most crucial of these encode the p53 and pgB proteins, which lie at the heart of two major tumorsuppressor pathways. Together, p53 and pRB are the most commonly lost functions in mammalian cancers, p53 is a transcriptional activator and repressor that controls the expression of genes that cause cell-cycle arrest or apoptosis in response to genomic damage, pRB regulates transcription indirectly, by interacting with transcription factors and recruiting chromatin-remodeling proteins to genes that control cell-cycle progression and differentiation. The pathways controlled by p53 and pRB are essential for cells to establish and maintain the senescence growth
DNA damage - double strand breaks or oxidation - can
arrest in response to diverse stimuli.
induce cellular senescence2.This might explain why mouse cells senesce after many fewer doublings than human cells, despite having longer telomeres and, frequently, being able
p53 activity, and in some cases protein levels, increases
to express telomerase: mouse cells might be more sensitive
w h e n cells senesce 1s. The mechanisms responsible for this
to the 20% oxygen in which cells are typically cultured. In
activation are incompletely understood, but some mol-
addition, agents that open or decondense chromatin induce
ecular details are emerging (Fig. 1). One cause of p53
cellular senescence 2's. Because these agents perturb chro-
activation appears to be an increase in the expression of
matin structure, they might abolish chromatin-mediated
p14 ape, a tumor suppressor encoded by the INK4a locus.
gene silencing. Loss of gene silencing can derange normal
p 14~ (p 19m# in mouse) stimulates p53 activity because
differentiation, a common feature of cancer cells. Finally,
it sequesters MDM2, a protein that facilitates p53 degra-
normal h u m a n and mouse cells senesce in response to
dation. Thus, p14 Ae# prevents negative-feedback regu-
intense mitogenic signals; for example, overexpression of
lation of p53 by MDM213. p14 AR~ is induced by E2F1,
the growth-stimulatory transcription factor E2F1 (Ref 6),
oncogenic Ras and DNA damage. It is repressed by TBX2,
or activated forms of the growth-factor signal-transducing
a transcription factor and potential oncogene ~6. The
proteins Ras (Ref. 7), Raf (Ref. 8) or MEK (Ref. 9).To the
mechanisms that alleviate repression of p 14 aRFby TBX2 in
extent it has been examined, these stimuli induce senes-
response to senescence-inducing signals are not known.
cence after only a few cell divisions and without telomere
Another cause for the increase in p53 activity might be
http://tcb.trends.com
A TRENDS Guide to Cancer Biology
evie~
TRENDS in Cell B i o l o g y Vo1.11 No.ll
November 2001
S29
the promyelocytic leukemia (PML) tumor suppressor. Senescence signals
PML is induced by replicative senescence and oncogenic
/
t~asl 7,18 by as-yet-unknown mechanisms. PML interacts with
/
CBP/p300 acetyltransferase (CBP/p300), which acetylates p53 and stimulates its activity 17 (Fig. 1).
i \\\
Ras Short telomeres
\
//
,
/
E S
\
-1
/
pRB exists only in its active (hypophosphorylated) growthinhibitory form in senescent cells. This is because senescent cells express high levels of p21, p 16 and, in some cases, p27 (Ref. 13). These proteins inhibit the cyclindependent protein kinases (CDKs) that phosphorylate and inactivate pRB during cell-cycle progression. It is not known why p27 increases in senescent cells, but it might
Cell cycle/ senescence arrest
be a consequence of increased activity of the PTEN tumor suppressor ~3. p21 is elevated at least partly because the gene is a direct target of p53 transactivation ~s, although
TRENDS in Cell Biology
p53-independent, posttranscriptional mechanisms also contribute to the rise in p 2 i (Re£ 19). p16, a second tumor suppressor encoded by the INK4a locus 13,~4, in-
function in order to overcome the proliferative barrier
creases in part because Etsl, a transcription factor that
imposed by cellular senescence. This can occur by
stimulates p 16 expression, accumulates in senescent cells,
mutation or epigenetic silencing of one or more key
whereas Idl, which negatively regulates Ets activity,
components of the pathways.
declinesz°. The resulting increase in Ets activity presumably and member of the Polycomb family of chromatin-
Cellular senescence suppresses tumorigenesis in vivo
remodeling proteins 2~. Oncogenic Ras might induce cel-
Although m u c h of the evidence that links cellular senes-
overcomes the repression of p 16 by BMI-1, an oncogene
Iular senescence by activating the mitogen-activated protein
cence and tumor-suppressor pathways derives from cell
kinase (MAPK) cascade, which stimulates Ets activity z°
cultures, there is substantial supporting evidence from
(Fig. 2). It is not known how other senescence inducers
intact organisms. Perhaps the best evidence derives from
stimulate Ets activity, or how Idl is repressed in response
mice in which genes encoding p53 or INK4a proteins are
to senescence signals. Nonetheless, these findings identify
inactivated in the germline. Cells derived from these ani-
p16 expression as an important target for senescence-
mals fail to senesce in response to multiple stimuli. In all
inducing signals that engage the pRB pathway.
cases, the animals develop cancer at an early age z4. There are several other genetically modified mice in which cells
i~';eracti;~$ 9atnwags
resist or fail to respond to senescence signals. In general,
From the above, it is clear that cellular senescence
these animals are highly cancer prone z4. By contrast, a
entails the activation of several tumor-suppressor pro
genetic manipulation that causes premature senescence of
teins and inactivation of several oncoproteins, each of
mammary epithelial cells suppresses the development of
which ultimately engages either the p53 or pRB path-
breast cancer in young mice exposed to the mouse mam-
way (Figs 1 and 2). This is not to say that the p53 and
mary tumor virus 2s.
pRB pathways are independent; rather, they interact
Human cells are markedly more resistant to neoplastic
at multiple levels la,l
