Senescence and Cell Cycle Control

Chicago Avenue, Chicago, IL 60611, USA kiyokawa@northwestern.edu Abstract In response to various stresses, such as telomere shortening during continuous proliferation, oxidative stress,

P Kaldis: Cell Cycle Regulation

DOI 10.1007/001/Published online: 23 November 2005

© Springer-Verlag Berlin Heidelberg 2005

Senescence and Cell Cycle Control

Hiroaki Kiyokawa

Department of Molecular Pharmacology and Biological Chemistry,

Feinberg School of Medicine, Northwestern University, 303 E Chicago Avenue,

Chicago, IL 60611, USA

kiyokawa@northwestern.edu

Abstract In response to various stresses, such as telomere shortening during continuous proliferation, oxidative stress, DNA damage and aberrant oncogene activation, normal cells undergo cellular senescence, which is a stable postmitotic state with particular morphology and metabolism Signaling that induces senescence involves two major tu-mor suppressor cascades, i.e., the INK4a-Rb pathway and the ARF-p53 pathway Diverse stimuli upregulate these interacting pathways, which orchestrate exit from the cell cycle Recent studies have provided insights into substantial differences in senescence-inducing signals in primary cells of human and rodent origins This review is focused on recent advances in understanding the roles of the tumor-suppressive pathways in senescence.

1

Senescence

Senescence was originally defined as an “irreversible” state of cell cycle arrest that reflects consumed proliferative capacity of the cell (Hayflick and Moor-head 1961) In eukaryotic cells each chromosome shortens from telomeres during every round of DNA replication (Smogorzewska and de Lange 2004; Campisi 2001) The structure of telomeres with repetitive sequences func-tions as a cap to prevent chromosome end fusions and genomic instability (de Lange 1998; Sharpless and DePinho 2004) While germ cells express telom-erase, which resynthesizes the telomeric repeats to maintain the chromoso-mal length, most human somatic cells do not express telomerase In somatic cells proliferating continuously, attrition of telomeres beyond a threshold triggers a response leading to “replicative” senescence Recent studies have indicated that telomere attrition provokes DNA damage-responsive signaling pathways (d’Adda et al 2003; Gire et al 2004; Herbig et al 2004) In addition, senescent cells exhibit a particular flat morphology with enlarged cytoplasm, and also express particular biochemical markers, such as senescence associ-ated β-galactosidase activity (Dimri et al 1995) “Premature” senescence or

stasis could be triggered in cells without telomere attrition by ectopic onco-gene activation, DNA damage, oxidative stress and other stressful conditions (Serrano et al 1997; Chang et al 2002; Chen and Ames 1994) These two forms of senescence are morphologically indistinguishable, and are likely to

258 H Kiyokawa

depend on common signaling pathways leading to cell cycle arrest The path-ways that play key roles in senescence induction, i.e., the ARF-p53-p21Cip1 pathway and the p16INK4a-Rb pathway, are major tumor-suppressor cascades (Sherr and DePinho 2000) Thus, it has been postulated that senescence is

a potent tumor-suppressive mechanism, like programmed cell death or apop-tosis (Lowe and Sherr 2003) While there have been interesting discussions

on whether cellular senescence is involved in organismal aging (Pelicci 2004), this review focuses on the tumor-suppressor pathways and cell cycle control during cellular senescence

2

Role of the p53 Pathway in Senescence

The tumor-suppressor p53 plays a key role in induction of senescence as evidenced by studies using mouse embryonic fibroblasts (MEFs) from p53-deficient (knockout) mice (Livingstone et al 1992; Lowe et al 1994) Under standard culture conditions, MEFs from wild-type embryos proliferate up to 15–25 population doublings, followed by induction of senescence In con-trast, p53-null MEFs continue to proliferate without obvious cell cycle arrest

or senescence-like morphology The same immortal phenotype is seen in wild-type MEFs expressing dominant-negative p53 mutants, short interfer-ing RNA (siRNA) against p53, or the papilloma virus oncoprotein E6, which facilitates p53 degradation (Munger and Howley 2002) Thus, loss of p53 func-tion is sufficient to abrogate the senescence checkpoint in MEFs (Sharpless and DePinho 2002) The immortalization step dependent on p53 perturbation renders mouse cells susceptible for activated Ras-induced malignant trans-formation In contrast, Ras activation in wild-type MEFs induces premature senescence, as already described (Serrano et al 1997) Although the role of Ras in senescece induction remains to be fully understood, Ras activates the external transcribed spacer transcription factors (ets) through the mitogen-activated protein (MAP) kinase pathway, which may upregulate p16INK4a(see Sect 4 for details) Furthermore, Ras activation leads to accumulation of in-tracellular reactive oxygen species [ROS] (see Sect 5 for details) These cellu-lar changes are apparently involved in Ras-induced senescence response Hu-man diploid fibroblasts (HDFs) taken from p53-heterozygous patients with Li–Fraumeni syndrome show prolonged replicative life spans in culture, as-sociated with loss of heterozygosity (Boyle et al 1998) Studies using HDFs, keratinocytes, and mammary epithelial cells suggest that loss of p53 is not sufficient for cooperating with Ras activation to transform human cells (Dray-ton and Peters 2002) It has been described that a combination of SV40 large T-antigen (T-Ag), activated Ras and the telomerase catalytic subunit (hTERT) can transform HDFs and human keratinocytes (Hahn et al 1999) T-Ag inac-tivates the p53- and Rb-dependent senescence-inducing pathways and hTERT

eliminates telomere-mediated signaling, allowing cells to undergo transform-ation in response to Ras activtransform-ation Interestingly, human mammary epithelial cells seem to have higher requirements for transformation, being insensitive

to the combinatory treatment (Hahn et al 1999) These data suggest substan-tial diversity of senescence control in different types of cells and in different species

Activation of p53 could result in one of the two cell fates, senescence

or apoptosis Loss of p53 function could contribute to immortalization and enhanced survival, both of which are hallmarks of cancer cells Thus, p53

is a multifuctional tumor suppressor, playing a central role in preventing malignant transformation While the proapoptotic function of p53 involves

a number of genes, the only known mediator of its prosenescent function is

Fig 1 The p16INK4a-Rb and ARF-p53 tumor-suppressor pathways control senescence The INK4a/ARF locus on human chromosome 9q21 encodes two tumor-suppressor proteins,

p16INK4a and p14Arf (p19ARFin mice) While p16INK4a directly inhibits Cdk4, p14Arf in-duces p21Cip1, which inhibits Cdk2 Cdk inhibition results in repression of the E2F target

genes via reduced phosphorylation of the retinoblastoma (Rb) family proteins

260 H Kiyokawa

the Cdk inhibitor p21Cip1(Fig 1) (Sharpless and DePinho 2002; Vousden and Prives 2005; Xiong et al 1993; Harper et al 1993) In an early study, p21Cip1 was isolated as a senescence-related gene (Noda et al 1994) During induc-tion of senescence, p53 transactivates the p21Cip1 gene (el Deiry et al 1993), and p21Cip1protein binds to and inhibits G1-regulatory cyclin-dependent ki-nases (Cdk), especially Cdk2 in complex with cyclins E and A (Sherr and Roberts 1999) In addition to Cdk inhibition, p21Cip1appears to affect expres-sion and functions of various proteins, which may lead cells to senescence

in an orchestrated manner (Roninson 2002) Importantly, forced expression

of p21Cip1 results in cellular accumulation of ROS with undefined mechan-isms (Macip et al 2002) ROS could cause DNA damage, which then activates p53-dependent pathways, possibly forming a positive feedback Disruption

of p21Cip1in HDFs prolongs the replicative life span in culture (Brown et al 1997) These observations suggest a central role for p21Cip1 in senescence However, studies using knockout mice provide a conflicting view p21Cip1 -null MEFs undergo senescence normally (Pantoja and Serrano 1999), and p21Cip1-null mice display only limited susceptibility to spontaneous tumori-genesis (Martin-Caballero et al 2001) Thus, p21Cip1may not be essential for senescence of mouse cells An alternative possibility is that p21Cip1-null mice undergo developmental adaptation to the absence of p21Cip1/Waf1, for which other Cdk inhibitors and possibly p130 (Coats et al 1999) may compensate Studies using acute disruption of p21Cip1 by conditional gene targeting will

be informative for better understanding of the role of p21Cip1in senescence

of mouse cells

3

Role of the Rb Pathway in Senescence

The Rb-family pocket binding proteins, i.e., Rb, p107 and p130, also play critical roles in cell fate determination between senescence and immortaliza-tion These proteins bind to the E2F family transcription factors and main-tain the repressor function of E2F (Hatakeyama and Weinberg 1995; Stevaux and Dyson 2002) Phosphorylation of the pocket binding proteins by Cdk complexes, such as cyclin D/Cdk4 (or Cdk6) and cyclin E/Cdk2, results in

dissociation of the proteins from E2F complexes, and is thought to medi-ate derepression or transactivation of E2F target genes (Fig 1) Senescence of MEFs induced by p53 overexpression depends on the repressor activity of E2F (Rowland et al 2002), suggesting that the senescence-inducing signal from p53 converges to the Rb/E2F pathway This signaling crosstalk could result

from p21Cip1 inhibition of Cdk2 and possibly Cdk4/6 Inactivation of these

pocket binding proteins by the papillomavirus E7 oncoprotein, together with telomerase activation by hTERT expression, has been shown to immortalize primary human epithelial cells (Kiyono et al 1998), although

immortaliza-tion of this system may involve addiimmortaliza-tional mutaimmortaliza-tions In MEFs, disrupimmortaliza-tion

of Rb, p107 and p130 results in increased proliferation with a shortened G1 phase and immortalization (Dannenberg et al 2000), whereas MEFs with any one of the three proteins still exhibit senescence in culture (Sage et al 2000) These observations suggest that the pocket binding proteins have over-lapping functions in controlling senescence However, acute disruption of

Rb by the Cre-Loxp recombination system has been demonstrated to im-mortalize MEFs (Sage et al 2003) This apparent discrepancy suggests that germline disruption of one or two pocket binding proteins leads to devel-opmental adaptation, which helps cells retain the senescence checkpoint by

a compensatory mechanism For instance, p107 expression is upregulated

in Rb-null MEFs Interestingly, the same study showed that Rb disruption induced cell cycle reentry in a small fraction of apparently senescent cells, suggesting that Rb plays a key role in maintenance of the postmitotic status in senescent cells Indeed, Rb, but not p107 or p130, is found in the senescence-associated heterochromatic foci (SAHF) (Narita et al 2003), which may play

a critical role in long-term transcriptional repression specific in senescent cells Interestingly, similar cell cycle reentry from senescence has been de-scribed in MEFs infected with lentivirus for anti-p53 short hairpin RNA (Dirac and Bernards 2003) While these studies intriguingly suggest that senescence may not necessarily be an irreversible process, this notion awaits further investigations on regulation of SAHF and other characteristics of senescence

Immortalization requires aberrant activation of the cell cycle machin-ery Cdk4 activation plays a key role when cells overcome the senescence checkpoint, presumably via phosphorylation of the pocket binding proteins MEFs with targeted Cdk4R24C mutation, which express a constitutively ac-tive Cdk4 protein insensiac-tive to the INK4 inhibitors, exhibit an immortal phenotype in culture (Rane et al 2002) Activated Ras is sufficient to trans-form Cdk4R24C MEFs, and mice with the Cdk4R24C mutation spontaneously develop various tumors, such as endocrine and skin tumors (Sotillo et al 2001; Rane et al 2002) In contrast, MEFs from Cdk4-null mice are resis-tant to immortalization induced by a dominant negative p53 muresis-tant (DNp53)

or disruption of the INK4a/ARF locus (Zou et al 2002) Cdk4-null MEFs

undergo transformation poorly in response to Ras plus DNp53 or Myc Consistent with the resistance to transformation, Cdk4-null mice are

refrac-tory to skin carcinogenesis in response to the keratin-5-Myc transgene or

the tumor initiator 7,12-dimethylbenz[4a4]anthracene plus the tumor

pro-moter 12-O-tetradecanoylphorbol-13-acetate (Rodriguez-Puebla et al 2002;

Miliani de Marval et al 2004) Genetic alterations that activate Cdk4, such

as overexpression of Cdk4 or D-type cyclins and deletion of the Cdk4 in-hibitor p16INK4a, are observed in the majority of human cancers (Ortega

et al 2002) Thus, derepression of E2F target genes as a consequence of Cdk4 activation seems to be required for cellular immortalization While

262 H Kiyokawa

Cdk2- and Cdk6-null mice have been generated (Berthet et al 2003; Malum-bres et al 2004; Ortega et al 2003), it has not been determined yet whether Cdk6 or Cdk2 plays an indispensable role in immortalization, similarly

to Cdk4

4

The Role of the INK4A /ARF Locus in Senescence

The INK4A/ARF locus was originally identified as a major tumor-suppressor locus MTS1 (multiple tumor suppressor-1) on human chromosome 9q21

(Kamb et al 1994) This locus is very frequently deleted in a variety of human cancers Importantly, the locus encodes two proteins, p16INK4a and p14ARF (or p19ARF in the case of mice), which cooperate for induction of senes-cence (Fig 1) (Quelle et al 1995; Carnero et al 2000; Sherr and DePinho 2000) While the coding sequences of these proteins partly overlap owing to the use of alternative reading frames, each product has its unique promoter and first exon p16INK4a is an specific inhibitor of Cdk4 and Cdk6 (Serrano

et al 1993) p14ARF stabilizes p53 by interfering with ubiquitin-dependent degradation by the ring-finger protein MDM2 (Pomerantz et al 1998; Stott

et al 1998; Zhang et al 1998) p53 stabilization leads to induction of p21Cip1 expression Therefore, p16INK4a and p14ARF are positive regulators of the Rb- and p53-dependent pathways, respectively Both proteins are upregulated during the senescence process A variety of stimuli could be involved in con-trol of the p16INK4a and p14ARF promoters, although the regulation of the INK4a/ARF locus is yet to be completely elucidated The expression of

ac-tivated oncogenes, such as Ras (Serrano et al 1997), Raf (Zhu et al 1998)

or MEK (Lin et al 1998), upregulates the p16INK4a promoter It has been described that the transactivation involves the Ets family transcription fac-tors (Ohtani et al 2001) Ets-1 can directly transactivate p16INK4aand induce senescence, and this regulation is abrogated by Id-1 as a specific inhibitor Bmi-1, a transcriptional repressor in the polycomb group protein family, in-hibits the senescence-associated induction of p16INK4a (Jacobs et al 1999) The polycomb proteins and the antagonizing trithorax group proteins are in-volved in transcriptional control during development (Park et al 2004) In particular, Bmi-1 has been shown to be a critical factor for maintenance of the proliferative capacity in stem cells Overexpression of Bmi-1 abrogates the in-duction of both p16INK4aand p19ARF in MEFs, slowing down the senescence process In HDFs, Bmi-1 expression represses p16INK4aand slows down senes-cence until telomere erosion causes crisis Expression of a dominant negative mutant of Bmi-1, which lacks the ring-finger domain, induces p16INK4a ex-pression in HDFs, accelerating replicative senescence (Itahana et al 2003) Another polycomb protein, Cbx7, can also downregulate the INK4a/ARF

lo-cus (Gil et al 2004) Thus, Bmi-1 plays a role in the timing of senescence

induction by controlling p16INK4a (and p19ARF in MEFs), while the detail

of the promoter regulation requires further investigations In mouse cells, p19ARFexpression is increased by senescence-inducing signals such as onco-gene activation or cellular stresses The ARF promoter contains E2F-binding sites, and can be activated by overexpression of E2F-1 or c-myc (DeGregori

et al 1997; Bates et al 1998; Zindy et al 1998; Dimri et al 2000) Repression

of the ARF promoter by E2F, especially E2F-3, suggests that the Rb path-way could function upstream for the p53 pathpath-way, as a feedback or crosstalk mechanism that controls immortalization vs senescence

In contrast to p19ARF in MEFs, p14ARF is minimally upregulated when HDFs undergo replicative or premature senescence (Bates et al 1998; Palmero

et al 1998; Ferbeyre et al 2000; Wei et al 2001) These data imply that mouse cells depend more on p19ARF for senescence, compared with human cells This hypothesis is consistent with observations that MEFs lacking p19ARF with intact p16INK4aexpression are immortal in culture (Kamijo et al 1997),

so are MEFs lacking both p19ARF and p16INK4a (Serrano et al 1996) On the other hand, MEFs lacking p16INK4a with intact p19ARF undergo senes-cence apparently normally (Sharpless et al 2001; Krimpenfort et al 2001) These studies suggest that p19ARFis essential and p16INK4ais dispensable for senescence of primary mouse cells (Krimpenfort et al 2001) However, it is noteworthy that an antisense RNA construct against p16INK4acan extend pro-liferative life span in primary wild-type MEFs (Carnero et al 2000) This may suggest that acute loss of p16INK4a in MEFs impacts on senescence control more significantly than germline knockout of the gene does In contrast to p16INK4a-null MEFs, HDFs from patients with mutations of the INK4A/ARF

locus that disrupt only p16INK4aexpression show a prolonged proliferative life span in culture, followed by arrest between the senescence (M1) and crisis (M2) checkpoints (Brookes et al 2002, 2004) Therefore, germline disruption

of p16INK4a in human cells appears to affect senescence more significantly than in mouse cells

5

Mouse Cells vs Human Cells:

Roles of Reactive Oxygen Species and Telomere Attrition

Oxidative stress triggered by ROS plays a major role in cellular senescence,

as well as organismal aging (Itahana et al 2004) Intracellular accumula-tion of ROS, for instance, by treatment with H2O2induces premature senes-cence In contrast, culturing cells under 3–5% oxygen reduces ROS levels and prolongs replicative life span in HDFs This hypoxic condition is closer

to physiological conditions cells in vivo are exposed to than in atmospheric oxygen (approximately 20%) It is noteworthy that MEFs do not undergo replicative senescence under 3% oxygen (Parrinello et al 2003) MEFs

cul-264 H Kiyokawa

tured in atmospheric oxygen exhibit characteristics of oxidative damage to DNA, which is less obvious in HDFs These observations suggest that mouse cells are more sensitive to oxidative stress than human cells In other words, senescence of mouse cells induced by continuous culture largely depends

on oxidative stress-responsive pathways (Fig 2a) Ras activation is known

to result in ROS accumulation ROS could activate p53 function in a DNA damage-dependent manner, through activation of ataxia-telangiectasia mu-tated (ATM)/ATM and Rad3-related (ATR) kinases ROS can also upregulate

p19ARFeffectively in primary mouse cells Furthermore, the stress-responsive MAP kinase p38MAPK may play a key role in mediating ROS signals to in-duce senescence, especially by upregulating p16INK4a (Iwasa et al 2003) In addition, a protein named seladin-1 was isolated in a genetic screen for regu-lators of Ras-induced premature senescence in rat embryonic fibroblasts (Wu

et al 2004) Seladin-1 may function as an ROS effector that facilitates p53 stabilization and consequently p21Cip1induction

In contrast to the ROS-dependent senescence of mouse cells, human cells undergo replicative senescence largely in a manner dependent on telom-ere shortening (Fig 2b) Telomtelom-ere attrition after every round of DNA repli-cation could be an intrinsic mechanism that counts cumulative numbers

of cell division When telomeres get shorter than a threshold level, DNA damage-responsive pathways are activated to induce senescence A recent re-port demonstrated that telomere shortening triggers senescence in human fibroblasts through the ATM-dependent DNA damage-responsive pathway leading to activation of p53 and induction of p21Cip1 (Herbig et al 2004)

Fig 2 aSenescence-inducing pathways in primary mouse cells Reactive oxygen species

(ROS) play a critical role in induction of senescence in cultured mouse cells p19ARF is predominantly upregulated during senescence by oncogene activation and/or oxidative

stress

In human cells undergoing replicative senescence, p16INK4acould be induced independently of telomeres or DNA damage Nonetheless, a recent study pro-vided evidence that p16INK4a significantly contributes to p53-independent response to telomere attrition in HDFs (Jacobs and de Lange 2004) While the mechanism of p16INK4aupregulation in telomemediated senescence re-mains to be clarified, p16INK4alevels could affect cellular sensitivity to prema-ture or replicative senescence by somehow affecting oncogene- and damage-responsive pathways HDFs with high p16INK4a levels undergo premature senescence in response to activated Ras, whereas HDFs with low p16INK4a levels do not show Ras-induced senescence (Benanti and Galloway 2004) Compared with telomeres in human cells (5–15 kb), telomeres in mouse cells (40–60 kb) are remarkably longer, which may be associated with detectable levels of telomerase in many somatic cell types in mice Therefore, telomere attrition is unlikely to function as a rate-limiting factor for induction of senes-cence in mouse cells under normal conditions Studies using knockout mice deficient for the telomerase RNA component (mTR) showed that mTR–/–

mice develop significant telomere attrition and chromosomal instability only after five to six generations of breeding (Rudolph et al 1999) Interestingly, mTR–/– mice at the fifth generation show resistance to carcinogen-induced

skin tumorigenesis in a p53-dependent manner (Gonzalez-Suarez et al 2002), suggesting the critical role of the p53 pathway in senescence in this engi-neered mouse model for telomere shortening

Fig 2 b Senescence-inducing pathways in primary human cells Unlike mouse cells, telomere attrition during each round of DNA replication plays a key role in trigger-ing senecence-induction pathways in human cells Premature senescence induced by oncogene activation involves oxidative stress with ROS accumulation While the role of p16INK4a in human cell senescence is well established, the role of p14Arfin human cells seems more complex

266 H Kiyokawa

6

Conclusions

Cellular senescence is an orchestrated program in response to a cue of various stresses Senescence is also an intrinsic tumor-suppressor mechan-ism, forming a checkpoint barrier against malignant transformation Di-verse senescence-inducing signals converge to the p16INK4a-Rb and ARF-p53 tumor-suppressor pathways A number of studies using genetically engi-neered mouse and human cells have revealed how these pathways interact with each other to execute the senescence program p53 plays a key role in mediating DNA damage-responsive signals elicited by telomere attrition or ROS During the senescence program, the E2F repressive action of Rb could

be sustained through Cdk inhibition as a consequence of p53-dependent upregulation of p21Cip1, as well as independent upregulation of p16INK4a However, fundamental differences in senescence between mouse and human models exist ARF plays a more prominent role in mouse cells than in human cells In contrast, senescence-associated p16INK4a upregulation is generally more robust in human models than in mouse models This is also the case for the tumor-suppressor role of p16INK4a Further investigations are needed for better understanding the role of senescence in tumor suppression and possibly in age-related pathological changes

Acknowledgements I apologize to many colleagues for being unable to cite their papers critical for the field I thank Nissim Hay, Rob Costa, Pradip Raychaudhuri, Oscar Colam-onici, David Ucker and Xianghong Zou for helpful discussions, and the National Institutes

of Health, the Department of Defense and the American Cancer Society for grant support for my research.

References

Bates S, Phillips AC, Clark PA, Stott F, Peters G, Ludwig RL, Vousden KH (1998) p14ARF links the tumour suppressors RB and p53 Nature 395:124–125

Benanti JA, Galloway DA (2004) Normal human fibroblasts are resistant to RAS-induced senescence Mol Cell Biol 24:2842–2852

Berthet C, Aleem E, Coppola V, Tessarollo L, Kaldis P (2003) Cdk2 knockout mice are viable Curr Biol 13:1775–1785

Boyle JM, Mitchell EL, Greaves MJ, Roberts SA, Tricker K, Burt E, Varley JM, Birch JM, Scott D (1998) Chromosome instability is a predominant trait of fibroblasts from Li-Fraumeni families Br J Cancer 77:2181–2192

Brookes S, Rowe J, Ruas M, Llanos S, Clark PA, Lomax M, James MC, Vatcheva R, Bates S, Vousden KH, Parry D, Gruis N, Smit N, Bergman W, Peters G (2002) INK4a-deficient hu-man diploid fibroblasts are resistant to RAS-induced senescence EMBO J 21:2936–2945 Brookes S, Rowe J, Gutierrez DA, Bond J, Peters G (2004) Contribution of p16 INK4a to replicative senescence of human fibroblasts Exp Cell Res 298:549–559

Brown JP, Wei W, Sedivy JM (1997) Bypass of senescence after disruption of p21 CIP1/WAF1

gene in normal diploid human fibroblasts Science 277:831–834