**4. Self-renewal regulators in adult HSCs**

Stem cells are crucial for the homeostatic maintenance of mature, functional cells in many tissues throughout the lifetime of the animal, and this pool of stem cells must itself be maintained (Muller-Sieburg and Sieburg, 2008). This is achieved by self-renewal, a specialized cell division in which one or both daughter cells remain undifferentiated and retain essentially the same replication potential of the parent. The self-renewal program must involve the activity of dedicated regulatory genes (Gazit et al., 2008); but although the phenotypic and functional properties of HSCs have been characterized extensively, we have only just begun to understand how self-renewal is regulated.

Insights Into Stem Cell Aging 251

manner to maintain gene silencing (Pietersen et al., 2008) (Table 1). PRC2 initiates silencing, whereas PRC1 maintains and stabilizes gene repression. PRC2 contains histone methyltransferases (HMTs) that methylate lysines 9 and 27 on histone H3 and lysine 26 on histone H1. Deletion of PRC2 genes in mice results in early embryonic death, underscoring their importance in development. PRC1 recognizes the H3 lysine 27 methyl group added by PRC2 (Valk-Lingbeek et al., 2004), and subsequently the monoubiquityl-ligase activity of the PRC1 proteins Bmi1 or Ring1A/B toward histone H2A generates uH2AK119, which prevents access of the transcription machinery and facilitates chromatin compaction (Wang et al., 2004). Mouse mutants of most PRC1 members, in spite of displaying homeotic transformations, survive until birth as a result of partial functional redundancy provided by

PRC2 is recruited to target genes by the cofactor jARID2 (jumonji/ARID domain-containing 2). Paradoxically, jARID2 also seems to inhibit PRC2 methyltransferase activity and may therefore regulate both the targeting and fine-tuning of PRC2 activity in stem cells and during differentiation (Panning, 2010). Once PRC1 recognizes and binds the H3K27me3 mark added by PRC2, it recruits additional proteins to establish the repressed chromatin configuration (Jones and Baylin, 2007). Gene promoters marked with H3K27me3 in ESCs are significantly more likely than other promoters to become methylated in cancer (Schlesinger et al., 2007). Moreover, the PcG targets in normal prostate cells are the same as those that become methylated in prostate cancer (Gal-Yam et al., 2008). Thus, altered chromatin structure does not always result in changes in gene expression associated with disease. Rather, disease results from the replacement of PcG repressive histone marks with methylation directly on DNA, which locks the chromatin in an inactive state, a process called epigenetic switching (Gal-Yam et al., 2008). Although the mechanism underlying predisposition of PcG targets to DNA methylation is not fully understood, the PRC1 component Cbx7 (chromobox homologue 7) was recently shown to interact directly with DNA (cytosine-5)-methyltransferase (DNMT)1 and DNMT3B at PcG target genes,

homologues, an exception being Ring1B-deficient mice (Voncken et al., 2003).

establishing a link between histone and DNA methylation (Mohammad et al., 2009).

Among PcG proteins, the PRC1 component Bmi1 is a fundamental self-renewal regulator, being required for self-renewal of all postnatal stem cell populations studied to date (Molofsky et al., 2003; Park et al., 2003; van der Lugt et al., 1994). Bmi1 was originally described as a proto-oncogene that induces B and T cell leukemias (van Lohuizen et al., 1991), and is overexpressed in several human cancers, including mantle cell lymphoma, colorectal carcinoma, liver carcinomas, non-small-cell lung cancer, and cerebral tumors such as medulloblastomas (Martin-Perez et al., 2010). This evidence has strongly influenced cancer research, supporting the above-mentioned theory that cancer is essentially a stem cell disorder (Reya et al., 2001). The self-renewal function of Bmi1 in adult stem cells relies largely on the silencing of one of its targets, the locus encoding the p16INK4a and ARF tumor suppressors (Molofsky et al., 2006). Deletion of p16INK4a and/or ARF partially rescues the self-renewal defects observed in various stem cell populations from Bmi1-null mice. Nevertheless, as described by the authors, this rescue is incomplete and thus other major Bmi1 regulated genes must exist. Candidates for additional Bmi1 targets in the context of self renewal are the Hox (homeobox) genes. A subset of Hox genes has been implicated in mammalian brain development, and several of them are highly expressed in neurospheres formed in vitro from cultured neural stem cells of the subventricular zone of

Fig. 2. Potential stem cell stage: interplay between aging and cancer. During normal aging, stem cells accumulate DNA damage as the consequence of endogenous (telomere dysfunction, oxidative stress) or exogenous (oxidative stress, g-irradiation, UV light, and others) attacks. This provokes subsequent stress-dependent changes (for example, accumulation of the products from the *INK4a/ARF* locus or telomere shortening), which activates checkpoint responses that result in apoptosis or cellular senescence. If these events occur in stem/progenitor cells, there is a decrease in the overall number and/or functionality of both stem and progenitor cells, leading an alteration of tissue homeostasis and regenerative capacity–a phenomenon that might contribute to aging and aged-related pathologies. If, instead, DNA mutations that inactivate these checkpoint pathways accumulate (for instance, loss of p16INK4a or reactivation of telomerase), then cancer can arise.

Polycomb complex in the maintenance of stemness. PcG proteins regulate self-renewal and lineage restriction in stem cells by inducing reversible chromatin modifications. PcG proteins have attracted increasing attention in stem cell and cancer stem cell research, given that it is now widely recognized that dynamic reprogramming of cells, for instance during differentiation, requires alterations to the epigenetic status of genes (Valk-Lingbeek et al., 2004). These features makes them interesting subjects for stem cell research, since it is conceivable that dynamic reprogramming of cells, for instance during differentiation, requires alterations in the epigenetic state of gene expression programs. The two major multiprotein PcG complexes identified to date, PRC1 and PRC2, function in a cooperative

Fig. 2. Potential stem cell stage: interplay between aging and cancer. During normal aging,

functionality of both stem and progenitor cells, leading an alteration of tissue homeostasis and regenerative capacity–a phenomenon that might contribute to aging and aged-related

Polycomb complex in the maintenance of stemness. PcG proteins regulate self-renewal and lineage restriction in stem cells by inducing reversible chromatin modifications. PcG proteins have attracted increasing attention in stem cell and cancer stem cell research, given that it is now widely recognized that dynamic reprogramming of cells, for instance during differentiation, requires alterations to the epigenetic status of genes (Valk-Lingbeek et al., 2004). These features makes them interesting subjects for stem cell research, since it is conceivable that dynamic reprogramming of cells, for instance during differentiation, requires alterations in the epigenetic state of gene expression programs. The two major multiprotein PcG complexes identified to date, PRC1 and PRC2, function in a cooperative

stem cells accumulate DNA damage as the consequence of endogenous (telomere dysfunction, oxidative stress) or exogenous (oxidative stress, g-irradiation, UV light, and others) attacks. This provokes subsequent stress-dependent changes (for example, accumulation of the products from the *INK4a/ARF* locus or telomere shortening), which activates checkpoint responses that result in apoptosis or cellular senescence. If these events

occur in stem/progenitor cells, there is a decrease in the overall number and/or

pathologies. If, instead, DNA mutations that inactivate these checkpoint pathways accumulate (for instance, loss of p16INK4a or reactivation of telomerase), then cancer can

arise.

manner to maintain gene silencing (Pietersen et al., 2008) (Table 1). PRC2 initiates silencing, whereas PRC1 maintains and stabilizes gene repression. PRC2 contains histone methyltransferases (HMTs) that methylate lysines 9 and 27 on histone H3 and lysine 26 on histone H1. Deletion of PRC2 genes in mice results in early embryonic death, underscoring their importance in development. PRC1 recognizes the H3 lysine 27 methyl group added by PRC2 (Valk-Lingbeek et al., 2004), and subsequently the monoubiquityl-ligase activity of the PRC1 proteins Bmi1 or Ring1A/B toward histone H2A generates uH2AK119, which prevents access of the transcription machinery and facilitates chromatin compaction (Wang et al., 2004). Mouse mutants of most PRC1 members, in spite of displaying homeotic transformations, survive until birth as a result of partial functional redundancy provided by homologues, an exception being Ring1B-deficient mice (Voncken et al., 2003).

PRC2 is recruited to target genes by the cofactor jARID2 (jumonji/ARID domain-containing 2). Paradoxically, jARID2 also seems to inhibit PRC2 methyltransferase activity and may therefore regulate both the targeting and fine-tuning of PRC2 activity in stem cells and during differentiation (Panning, 2010). Once PRC1 recognizes and binds the H3K27me3 mark added by PRC2, it recruits additional proteins to establish the repressed chromatin configuration (Jones and Baylin, 2007). Gene promoters marked with H3K27me3 in ESCs are significantly more likely than other promoters to become methylated in cancer (Schlesinger et al., 2007). Moreover, the PcG targets in normal prostate cells are the same as those that become methylated in prostate cancer (Gal-Yam et al., 2008). Thus, altered chromatin structure does not always result in changes in gene expression associated with disease. Rather, disease results from the replacement of PcG repressive histone marks with methylation directly on DNA, which locks the chromatin in an inactive state, a process called epigenetic switching (Gal-Yam et al., 2008). Although the mechanism underlying predisposition of PcG targets to DNA methylation is not fully understood, the PRC1 component Cbx7 (chromobox homologue 7) was recently shown to interact directly with DNA (cytosine-5)-methyltransferase (DNMT)1 and DNMT3B at PcG target genes, establishing a link between histone and DNA methylation (Mohammad et al., 2009).

Among PcG proteins, the PRC1 component Bmi1 is a fundamental self-renewal regulator, being required for self-renewal of all postnatal stem cell populations studied to date (Molofsky et al., 2003; Park et al., 2003; van der Lugt et al., 1994). Bmi1 was originally described as a proto-oncogene that induces B and T cell leukemias (van Lohuizen et al., 1991), and is overexpressed in several human cancers, including mantle cell lymphoma, colorectal carcinoma, liver carcinomas, non-small-cell lung cancer, and cerebral tumors such as medulloblastomas (Martin-Perez et al., 2010). This evidence has strongly influenced cancer research, supporting the above-mentioned theory that cancer is essentially a stem cell disorder (Reya et al., 2001). The self-renewal function of Bmi1 in adult stem cells relies largely on the silencing of one of its targets, the locus encoding the p16INK4a and ARF tumor suppressors (Molofsky et al., 2006). Deletion of p16INK4a and/or ARF partially rescues the self-renewal defects observed in various stem cell populations from Bmi1-null mice. Nevertheless, as described by the authors, this rescue is incomplete and thus other major Bmi1 regulated genes must exist. Candidates for additional Bmi1 targets in the context of self renewal are the Hox (homeobox) genes. A subset of Hox genes has been implicated in mammalian brain development, and several of them are highly expressed in neurospheres formed in vitro from cultured neural stem cells of the subventricular zone of


Adapted from Takaaki Konuma et al. Develop. Growth Differ. (2010) 52, 505–516

Table 1. Principcal components of the Polycomb Group Complexes and their hematopoietic defects in mutant mice.

Insights Into Stem Cell Aging 253

Bmi-null mice (Molofsky et al., 2006). More recently, Bmi1 and Ring1A were shown to play essential roles in H2A ubiquitylation and Hox gene silencing. Knockout of Bmi1 results in significant loss of H2A ubiquitylation and an upregulation of HoxC13 expression, whereas Ezh2-mediated H3-K27 methylation is not affected (Cao et al., 2005). Similar findings have been described for the HoxC5 gene. However, considering that PcG proteins modify the chromatin of large sets of genes (Kirmizis et al., 2004), a great number of additional targets are likely to exist. For instance, both PRC2 and Bmi1 have recently been shown to play roles in the repression of E-cadherin expression (Yang et al., 2010). Interestingly, PcG genes have been shown to have a tumor suppressive function. In *Drosophila*, PcG proteins repress JAK/STAT and Notch signaling activity, whose activation drives disc cell overproliferation (Classen et al., 2009). Specifically, the *Drosophila* complex Psc (posterior sex combs), which includes Bmi1, and Suz12 (suppressor of zeste 12) play a tumor suppressive role mediated by Wnt repression in follicle stem cells(Li et al., 2010). In mammals, Eed (embryonic ectoderm development protein) displays tumor suppressive activity in the mouse hematopoietic system (Richie et al., 2002). Thus, PcG genes have been suggested to behave either as proto-oncogenes or as tumor suppressors depending on the tissue, cell context,

Bmi1 is regulated by Sonic Hedgehog, providing a direct connection between PcG and a major stem cell-specific pathway (Leung et al., 2004). Furthermore, activation of either Hedgehog or Notch signaling has been shown to increase Bmi1 expression, whereas siRNA knockdown of Bmi1 abrogates the effects of Hedgehog or Notch signaling on sphere formation, a functional readout of stemness. Thus the effects of Hedgehog and Notch signaling on stem cell self-renewal appear to be largely dependent on Bmi1. A complex regulation of Bmi1 is suggested by the fact that distinct Bmi1 regulators have been found in different types of cancer, for example Twist1 in head and neck squamous cell carcinoma and the Zeb1 (zinc finger E-box binding homeobox 1) – miR-200 pathway in pancreatic cancers (Wellner et al., 2009). Furthermore, a single PcG function can be regulated by multiple factors, for example Snail1 regulates E-cadherin silencing by PRC2, whereas the action of

In summary, PcG proteins, in particular Bmi1, are essential for self-renewal and proliferative potential, which are crucial for the maintenance of stemness, acting as a critical failsafe mechanism against loss of stem cells in response to senescence signals. In turn, Bmi1 must be finely-regulated to prevent uncontrolled replicative expansion and tumor induction. Despite the importance of PcG proteins, we are only beginning to unravel how these master regulators are themselves regulated to achieve an appropriate balance between ensuring

The tumor suppressors p16INK4a and ARF. Cell-cycle regulators such as the *INK4/ARF* locus appear to play an important role in the reaction of adult stem cells to stress and aging. The *INK4/ARF* locus plays a central role in tumor suppression, reflected in its inactivation in almost 50% of human cancers (Sharpless, 2005). Indeed, this locus is regarded as one of the most important anti-oncogenic defenses of the mammalian genome, comparable in importance only to p53. The remarkable feature of the *INK4/ARF* locus is that it encodes three tumor suppressors in a genomic segment of about 50 kb: *p16INK4a*, its related family member *p15INK4b*, and *ARF* (called *p19ARF* in mice and *p14ARF* in humans). The actions of p16INK4a, p15INK4b and ARF are well understood. Both p16INK4a and p15INK4b inhibit the kinase

developmental stage and gene dosage.

PRC1 on this target is regulated by Twist1 (Yang et al., 2010).

stem cell longevity and preventing tumorigenesis.

Adapted from Takaaki Konuma et al. Develop. Growth Differ. (2010) 52, 505–516

defects in mutant mice.

Table 1. Principcal components of the Polycomb Group Complexes and their hematopoietic

Bmi-null mice (Molofsky et al., 2006). More recently, Bmi1 and Ring1A were shown to play essential roles in H2A ubiquitylation and Hox gene silencing. Knockout of Bmi1 results in significant loss of H2A ubiquitylation and an upregulation of HoxC13 expression, whereas Ezh2-mediated H3-K27 methylation is not affected (Cao et al., 2005). Similar findings have been described for the HoxC5 gene. However, considering that PcG proteins modify the chromatin of large sets of genes (Kirmizis et al., 2004), a great number of additional targets are likely to exist. For instance, both PRC2 and Bmi1 have recently been shown to play roles in the repression of E-cadherin expression (Yang et al., 2010). Interestingly, PcG genes have been shown to have a tumor suppressive function. In *Drosophila*, PcG proteins repress JAK/STAT and Notch signaling activity, whose activation drives disc cell overproliferation (Classen et al., 2009). Specifically, the *Drosophila* complex Psc (posterior sex combs), which includes Bmi1, and Suz12 (suppressor of zeste 12) play a tumor suppressive role mediated by Wnt repression in follicle stem cells(Li et al., 2010). In mammals, Eed (embryonic ectoderm development protein) displays tumor suppressive activity in the mouse hematopoietic system (Richie et al., 2002). Thus, PcG genes have been suggested to behave either as proto-oncogenes or as tumor suppressors depending on the tissue, cell context, developmental stage and gene dosage.

Bmi1 is regulated by Sonic Hedgehog, providing a direct connection between PcG and a major stem cell-specific pathway (Leung et al., 2004). Furthermore, activation of either Hedgehog or Notch signaling has been shown to increase Bmi1 expression, whereas siRNA knockdown of Bmi1 abrogates the effects of Hedgehog or Notch signaling on sphere formation, a functional readout of stemness. Thus the effects of Hedgehog and Notch signaling on stem cell self-renewal appear to be largely dependent on Bmi1. A complex regulation of Bmi1 is suggested by the fact that distinct Bmi1 regulators have been found in different types of cancer, for example Twist1 in head and neck squamous cell carcinoma and the Zeb1 (zinc finger E-box binding homeobox 1) – miR-200 pathway in pancreatic cancers (Wellner et al., 2009). Furthermore, a single PcG function can be regulated by multiple factors, for example Snail1 regulates E-cadherin silencing by PRC2, whereas the action of PRC1 on this target is regulated by Twist1 (Yang et al., 2010).

In summary, PcG proteins, in particular Bmi1, are essential for self-renewal and proliferative potential, which are crucial for the maintenance of stemness, acting as a critical failsafe mechanism against loss of stem cells in response to senescence signals. In turn, Bmi1 must be finely-regulated to prevent uncontrolled replicative expansion and tumor induction. Despite the importance of PcG proteins, we are only beginning to unravel how these master regulators are themselves regulated to achieve an appropriate balance between ensuring stem cell longevity and preventing tumorigenesis.

The tumor suppressors p16INK4a and ARF. Cell-cycle regulators such as the *INK4/ARF* locus appear to play an important role in the reaction of adult stem cells to stress and aging. The *INK4/ARF* locus plays a central role in tumor suppression, reflected in its inactivation in almost 50% of human cancers (Sharpless, 2005). Indeed, this locus is regarded as one of the most important anti-oncogenic defenses of the mammalian genome, comparable in importance only to p53. The remarkable feature of the *INK4/ARF* locus is that it encodes three tumor suppressors in a genomic segment of about 50 kb: *p16INK4a*, its related family member *p15INK4b*, and *ARF* (called *p19ARF* in mice and *p14ARF* in humans). The actions of p16INK4a, p15INK4b and ARF are well understood. Both p16INK4a and p15INK4b inhibit the kinase

Insights Into Stem Cell Aging 255

and though the effects of p53 and *INK4/ARF* locus expression in aging are context and dosage dependent, these results suggest that under physiological aging (labeled by moderate increase of still regulated p53 activity), the damaged cells are eliminated by either triggering their self-destruction (by apoptosis) or by pulling them out of the proliferative pool (by inducing senescence). In contrast, by massive DNA damage, the presence of uncontrolled activity of p53 results in excessive elimination of cells by p53 that exhausts the

The *INK4/ARF* locus and age-associated phenotypes. p16INK4a and ARF may also be broadly important to diseases of aging beyond their function in stem cells. Specifically, three research consortia that undertook genome-wide association studies across large, carefully annotated patient samples have reported an association between single nucleotide polymorphisms (SNPs) near to *INK4a/ARF* locus and frailty (Melzer et al., 2007), atherosclerotic heart disease (ASHD)(Helgadottir et al., 2007) (McPherson et al., 2007), and type-2 diabetes (Saxena et al., 2007; Zeggini et al., 2008) in large human cohorts. However, few of the associated SNPs near the locus, and associated with these phenotypes, are not in linkage disequilibrium with each other, which suggests that more than one polymorphism near the locus influences these aging phenotypes. Therefore, although these studies do not pinpoint specific polymorphisms that affect the risks of age-related diseases, there are only four genes in the vicinity of the mapped polymorphisms: p16INK4a, ARF, p15INK4b, and ANRIL (a noncoding RNA). More relative data suggest specific links: p16INK4a expression increases with age in pancreatic β cells, and p16INK4a deficiency increases β-cell regenerative capacity(Krishnamurthy et al., 2006), providing a mechanism by which polymorphisms that affect p16INK4a expression or activity might affect risk for type-2 diabetes. It remains unclear whether these polymorphisms influence the risk of frailty and heart disease through their effects on tissue regenerative capacity or by mechanisms that are completely independent of stem/progenitor cells. Nevertheless, in light of the murine genetic studies that link *INK4a/ARF* locus and stem cell function, proteins encoded by the locus are the strongest candidates to mediate the effects of these polymorphisms on the incidence of these common

The regenerative capacity of many stem cells declines functionally with age and, this decline triggers in part many age-related symptoms, and the development of certain diseases. Recent evidences have demonstrated that certain tumor suppressors, like *p16INK4a*, also suppresses the proliferation of stem or progenitor cells in the bone marrow, pancreas and brain. Thus, p16INK4a seems to balance equilibrium reducing cancer incidence, which promotes longevity, but also decreasing stem cell self-renewal and proliferation, compromising tissue regeneration and repair, which are likely to reduce longevity. These observations allow us to suggest the provocative but unproved hypothesis that mammalian aging results in part from the beneficial efforts of tumor suppressor proteins to interdict cancer. In this stage, characterization of how stem cells age, such as **the characterization of reliable biomarkers,** deregulated signaling pathways, loss of self-renewal or acquisition of defects in differentiation of stem cells, will contribute to understand the age-associated pathophysiological decline. Likewise, it is also essential to figure out the cellular and molecular components of stem cell niches, how the niche changes during aging, and

capacity of tissue regeneration leading to premature aging.

diseases that are associated with aging.

**5. Conclusions** 

activity of CDK4/6-cycD complexes, thus contributing to the maintenance of the active, growth-suppressive form of the retinoblastoma (Rb) family of proteins. ARF contributes to the stability of p53 by inhibiting the p53-degrading activity of MDM2. Through the activation of Rb and p53, the *INK4/ARF* locus is able to induce cell senescence and cell death (Gil and Peters, 2006; Lowe and Sherr, 2003). These tumor suppressors have taken on additional importance given recent evidence that at least one product of the locus, p16INK4a, also contributes to the decline in the replication potential of self-renewing cells during the aging of stem cells. The expression of p16INK4a is relatively low in the HSCs of young mice, but is upregulated with age or in response to cellular stresses (Janzen et al., 2006). Although the number of immunophenotypic HSCs increases with age in wild-type animals, HSC functionality is impaired. In particular, the HSC compartment of old animals is more rapidly exhausted by serial transplantation than that of young animals. In contrast, aging has the opposite effect on p16INK4a-/- HSCs, with p16INK4a-/- HSCs from old animals substantially outperforming young p16INK4a-/- HSCs in serial transplantation assays (Janzen et al., 2006). In fact, old p16INK4a-/- HSCs perform as well as young wild-type HSCs in this assay. Thus p16INK4a compromises HSC functionality in older mice. Similar results were obtained in studies of p16INK4a-/- neuronal stem cells and pancreatic islets (Krishnamurthy et al., 2006; Molofsky et al., 2006), revealing a general role for p16INK4a in the regulation of stem cell and progenitor cell aging. Therefore, on one face of this coin, p16INK4a acts as a potent tumor suppressor that promotes longevity by suppressing the development of cancer, while on the flipside, the increase of p16INK4a levels with age impairs the proliferation of stem or progenitor cells, ultimately reducing longevity. Thus, p16INK4a seems to balance an equilibrium reducing cancer incidence, but also contributing to aging by decreasing stem cell self-renewal and proliferation. These observations suggest the provocative but as yet unproven notion that mammalian aging results in part from the beneficial effects of tumor suppressor proteins (Figure 2).

The transcription factor p53. Besides p16INK4a tumor suppressor, p53 is also a tumor suppressor that influences stem cell self-renewal, tissue regenerative capacity, age-related disease, and cancer, which activity is lost in nearly half of all human cancers(Toledo and Wahl, 2006). The p53 protein is normally inactive, due in part to its rapid degradation by the specific ubiquitin ligase Mdm2. A multitude of stresses converge on p53 through complex, and partially understood, signaling pathways that stabilize and modify p53. The analysis of the effect of p53 in aging has revealed a dual role that seems to depend on the intensity of p53 activity. Overexpression of short isoforms of p53 in mice have greater protection against tumor development than wild-type mice, while at the same time they show signs of premature aging (Maier et al., 2004; Tyner et al., 2002). However, mouse models of increased wild-type p53 activity do not present premature aging. In particular, bacterial artificial chromosome transgenic mice that bear a third copy of the p53 locus show a decreased cancer incidence but normal longevity and normal onset of aging phenotypes (Garcia-Cao et al., 2002; Matheu et al., 2007; Matheu et al., 2004). An additional mouse model, the super-INK4a/ARF mice, with an extra copy of the entire *INK4a/ARF* locus (being ARF an activator of p53), show a significantly reduced incidence of cancer, although the mice aged normally (Matheu et al., 2004). To investigate whether the concomitant expression of both tumor suppressors had a synergistic effect, mice that bear a third copy of the p53 locus and a third copy of the *INK4/ARF* locus show increased longevity and delayed aging in a manner that cannot be explained by their reduced incidence of cancer (Matheu et al., 2007). Therefore,

activity of CDK4/6-cycD complexes, thus contributing to the maintenance of the active, growth-suppressive form of the retinoblastoma (Rb) family of proteins. ARF contributes to the stability of p53 by inhibiting the p53-degrading activity of MDM2. Through the activation of Rb and p53, the *INK4/ARF* locus is able to induce cell senescence and cell death (Gil and Peters, 2006; Lowe and Sherr, 2003). These tumor suppressors have taken on additional importance given recent evidence that at least one product of the locus, p16INK4a, also contributes to the decline in the replication potential of self-renewing cells during the aging of stem cells. The expression of p16INK4a is relatively low in the HSCs of young mice, but is upregulated with age or in response to cellular stresses (Janzen et al., 2006). Although the number of immunophenotypic HSCs increases with age in wild-type animals, HSC functionality is impaired. In particular, the HSC compartment of old animals is more rapidly exhausted by serial transplantation than that of young animals. In contrast, aging has the opposite effect on p16INK4a-/- HSCs, with p16INK4a-/- HSCs from old animals substantially outperforming young p16INK4a-/- HSCs in serial transplantation assays (Janzen et al., 2006). In fact, old p16INK4a-/- HSCs perform as well as young wild-type HSCs in this assay. Thus p16INK4a compromises HSC functionality in older mice. Similar results were obtained in studies of p16INK4a-/- neuronal stem cells and pancreatic islets (Krishnamurthy et al., 2006; Molofsky et al., 2006), revealing a general role for p16INK4a in the regulation of stem cell and progenitor cell aging. Therefore, on one face of this coin, p16INK4a acts as a potent tumor suppressor that promotes longevity by suppressing the development of cancer, while on the flipside, the increase of p16INK4a levels with age impairs the proliferation of stem or progenitor cells, ultimately reducing longevity. Thus, p16INK4a seems to balance an equilibrium reducing cancer incidence, but also contributing to aging by decreasing stem cell self-renewal and proliferation. These observations suggest the provocative but as yet unproven notion that mammalian aging results in part from the beneficial effects of tumor

The transcription factor p53. Besides p16INK4a tumor suppressor, p53 is also a tumor suppressor that influences stem cell self-renewal, tissue regenerative capacity, age-related disease, and cancer, which activity is lost in nearly half of all human cancers(Toledo and Wahl, 2006). The p53 protein is normally inactive, due in part to its rapid degradation by the specific ubiquitin ligase Mdm2. A multitude of stresses converge on p53 through complex, and partially understood, signaling pathways that stabilize and modify p53. The analysis of the effect of p53 in aging has revealed a dual role that seems to depend on the intensity of p53 activity. Overexpression of short isoforms of p53 in mice have greater protection against tumor development than wild-type mice, while at the same time they show signs of premature aging (Maier et al., 2004; Tyner et al., 2002). However, mouse models of increased wild-type p53 activity do not present premature aging. In particular, bacterial artificial chromosome transgenic mice that bear a third copy of the p53 locus show a decreased cancer incidence but normal longevity and normal onset of aging phenotypes (Garcia-Cao et al., 2002; Matheu et al., 2007; Matheu et al., 2004). An additional mouse model, the super-INK4a/ARF mice, with an extra copy of the entire *INK4a/ARF* locus (being ARF an activator of p53), show a significantly reduced incidence of cancer, although the mice aged normally (Matheu et al., 2004). To investigate whether the concomitant expression of both tumor suppressors had a synergistic effect, mice that bear a third copy of the p53 locus and a third copy of the *INK4/ARF* locus show increased longevity and delayed aging in a manner that cannot be explained by their reduced incidence of cancer (Matheu et al., 2007). Therefore,

suppressor proteins (Figure 2).

and though the effects of p53 and *INK4/ARF* locus expression in aging are context and dosage dependent, these results suggest that under physiological aging (labeled by moderate increase of still regulated p53 activity), the damaged cells are eliminated by either triggering their self-destruction (by apoptosis) or by pulling them out of the proliferative pool (by inducing senescence). In contrast, by massive DNA damage, the presence of uncontrolled activity of p53 results in excessive elimination of cells by p53 that exhausts the capacity of tissue regeneration leading to premature aging.

The *INK4/ARF* locus and age-associated phenotypes. p16INK4a and ARF may also be broadly important to diseases of aging beyond their function in stem cells. Specifically, three research consortia that undertook genome-wide association studies across large, carefully annotated patient samples have reported an association between single nucleotide polymorphisms (SNPs) near to *INK4a/ARF* locus and frailty (Melzer et al., 2007), atherosclerotic heart disease (ASHD)(Helgadottir et al., 2007) (McPherson et al., 2007), and type-2 diabetes (Saxena et al., 2007; Zeggini et al., 2008) in large human cohorts. However, few of the associated SNPs near the locus, and associated with these phenotypes, are not in linkage disequilibrium with each other, which suggests that more than one polymorphism near the locus influences these aging phenotypes. Therefore, although these studies do not pinpoint specific polymorphisms that affect the risks of age-related diseases, there are only four genes in the vicinity of the mapped polymorphisms: p16INK4a, ARF, p15INK4b, and ANRIL (a noncoding RNA). More relative data suggest specific links: p16INK4a expression increases with age in pancreatic β cells, and p16INK4a deficiency increases β-cell regenerative capacity(Krishnamurthy et al., 2006), providing a mechanism by which polymorphisms that affect p16INK4a expression or activity might affect risk for type-2 diabetes. It remains unclear whether these polymorphisms influence the risk of frailty and heart disease through their effects on tissue regenerative capacity or by mechanisms that are completely independent of stem/progenitor cells. Nevertheless, in light of the murine genetic studies that link *INK4a/ARF* locus and stem cell function, proteins encoded by the locus are the strongest candidates to mediate the effects of these polymorphisms on the incidence of these common diseases that are associated with aging.
