|Home | About | Journals | Submit | Contact Us | Français|
Cellular senescence is an irreversible proliferation arrest of primary cells and an important tumor suppression process. Senescence is often characterized by domains of facultative heterochromatin, called Senescence-Associated Heterochromatin Foci (SAHF), which repress expression of proliferation-promoting genes. Formation of SAHF is driven by a complex of histone chaperones, HIRA and ASF1a, and depends upon prior localization of HIRA to PML nuclear bodies. However, how the SAHF assembly pathway is activated in senescent cells is not known. Here we show that expression of the canonical Wnt2 ligand and downstream canonical Wnt-signals are repressed in senescent human cells. Repression of Wnt2 occurs early in senescence and independent of the pRB and p53 tumor suppressor proteins, and drives relocalization of HIRA to PML bodies, formation of SAHF and senescence, likely through GSK3β-mediated phosphorylation of HIRA. These results have major implications for our understanding of both Wnt-signaling and senescence in tissue homeostasis and cancer progression.
Cell senescence is an irreversible proliferation-arrest that is triggered by activated oncogenes and, consequently, is an important tumor suppression process in vivo (Braig et al., 2005; Campisi, 2005; Chen et al., 2005; Collado et al., 2005; Courtois-Cox et al., 2006; Dimri et al., 1995; Michaloglou et al., 2005; Serrano et al., 1997). Senescence is also caused by shortened telomeres that result from repeated rounds of cell division, and cellular stresses and inadequate growth conditions (Campisi, 2005; Ramirez et al., 2001). Consequently, senescence in vivo is thought to contribute to tissue aging, through exhaustion of renewable tissue stem cell populations (Campisi, 2005; Janzen et al., 2006; Krishnamurthy et al., 2006; Molofsky et al., 2006).
One molecular characteristic of senescence in many human cell types is formation of specialized domains of facultative heterochromatin, called Senescence Associated Heterochromatin Foci (SAHF) (Denoyelle et al., 2006; Narita et al., 2003; Zhang et al., 2005). SAHF result from condensation of individual chromosomes into isolated heterochromatic domains (Funayama et al., 2006; Zhang et al., 2007a). SAHF repress expression of proliferation-promoting genes, thereby contributing to senescence-associated cell cycle arrest. SAHF contain several molecular indicators of transcriptionally silent heterochromatin, including heterochromatin proteins 1 (HP1α, β and γ) and histone variant macroH2A. In addition, SAHF contain increased amounts of HMGA proteins (Funayama et al., 2006; Narita et al., 2006). Indicative of their physiological link to senescence, tissue aging and tumor suppression, SAHF have been reported in skin of aging primates (Herbig et al., 2006) and inactivation of HMGA proteins abrogates senescence and facilitates cell transformation and tumor formation (Narita et al., 2006).
Two chromatin regulators, HIRA and ASF1a, drive formation of SAHF in human cells (Zhang et al., 2005). HIRA and ASF1a are the human orthologs of proteins that create transcriptionally silent heterochromatin in yeast, flies and plants (Moshkin et al., 2002; Phelps-Durr et al., 2005; Sharp et al., 2001; Sherwood et al., 1993; Singer et al., 1998). HIRA is a histone chaperone that specifically deposits the histone replacement variant H3.3 in nucleosomes (Loppin et al., 2005; Tagami et al., 2004; van der Heijden et al., 2007). Likewise, yeast Asf1p has histone deposition activity (Tyler et al., 1999). Consistent with their overlapping properties, yeast Asf1p and Hir proteins physically interact and this interaction is necessary for telomeric silencing (Daganzo et al., 2003). Likewise, formation of SAHF in human cells depends upon a trimeric HIRA, ASF1a and histone H3 complex (Tagami et al., 2004; Tang et al., 2006; Zhang et al., 2007a; Zhang et al., 2005), most likely due to the ability of this complex to facilitate nucleosome assembly and increased nucleosome density.
One of the earliest cytological indicators of impending senescence in human cells is recruitment of HIRA into 20–30 distinct nuclear foci, about 0.1–1μM in diameter. These foci appear before other senescent phenotypes, such as cell cycle exit, SAHF, a large flat morphology and SA β-gal activity (Zhang et al., 2005). Underscoring the importance of HIRA’s regulation for physiological senescence and tissue aging, Sedivy and coworkers demonstrated a striking correlation between the level of expression of HIRA in dermal fibroblasts and the age of the donor primates (Jeyapalan et al., 2007). The nuclear foci into which HIRA is recruited in senescent human cells are PML (acute ProMyelocytic Leukemia) nuclear bodies (Zhang et al., 2005), subnuclear organelles enriched in PML and many other proteins (Salomoni and Pandolfi, 2002). In human cells, localization of HIRA to PML bodies appears to be essential for formation of SAHF (Ye et al., 2007) (see Discussion). Significantly, HIRA translocates into PML bodies at the same time as HP1 proteins, which ultimately exit PML bodies and are stably incorporated into SAHF (Zhang et al., 2005). At a molecular level, PML bodies are thought to serve as sites of assembly of macromolecular regulatory complexes and/or protein modification (Salomoni and Pandolfi, 2002). Thus, it seems likely that PML bodies are a molecular “staging ground” where HIRA-containing complexes are assembled or modified prior to their translocation to chromatin and formation of SAHF. We have set out to understand the mechanism by which HIRA localization is controlled.
The pRB and p53 tumor suppressor pathways are master effectors of senescence (Campisi, 2005). Formation of SAHF depends on a multi-step cascade of events, comprised of pRB and p53-dependent and independent steps (Narita et al., 2003; Ye et al., 2007; Zhang et al., 2007b). Downstream of HIRA’s recruitment to PML bodies, the HIRA/ASF1a pathway cooperates with the pRB and p53 pathways to create SAHF. However, HIRA’s initial localization to PML bodies is independent of pRB and p53. Consequently, we sought out a trigger of HIRA’s relocalization that is independent of the status of pRB and p53.
Wnt signaling frequently maintains proliferation of tissue stem cells, by stimulation of cell division and inhibition of differentiation and apoptosis (Reya and Clevers, 2005). The proliferative effects of Wnt are mediated, in large part, through the canonical Wnt pathway. Extracellular Wnt proteins bind to their cognate transmembrane receptors, members of the Frizzled family. This results in disruption of a cytoplasmic protein complex containing the kinase, GSK3β, and its substrate, β-catenin. This, in turn, prevents phosphorylation and degradation of the transcription factor β-catenin, allowing it accumulate in soluble form in the cytoplasm and translocate to the nucleus to drive expression of proliferative genes, such as c-myc and cyclin D1. Underscoring its role as a promoter of cell proliferation, the Wnt signaling pathway is frequently activated in human cancers (Reya and Clevers, 2005).
Since senescence is thought to limit proliferative lifespan of many human stem cell populations, whereas Wnt signaling tends to maintain proliferation of those populations, we hypothesized that senescence and Wnt signaling ought to be coordinated with each other. Here we report that pRB and p53-independent repression of Wnt signaling early in senescence is a signal for HIRA’s localization to PML bodies, SAHF formation and senescence. Consequently, senescence of primary human cells is modulated by activity of this key extracellular signaling system, with major implications for normal tissue homeostasis, as well as pathological conditions, such as cancer and aging.
Primary human WI38 fibroblasts express mRNAs coding for several Wnt family proteins and their Frizzled receptors (Supplementary Figure 1 and Figure 1A). To test whether onset of senescence is coordinated with Wnt signaling, we asked whether expression of any of these components is altered in senescent cells. We compared their expression in growing WI38 cells (population doubling (PD) #31) and cells made senescent by extended growth in culture (PD#58) or ectopic expression of an activated Ras oncogene. There was no change in expression in any of the Frizzled mRNAs, Wnt5a or Wnt5b (Figure 1A and data not shown). Expression of Wnt9a increased in Ras-infected cells, but was unchanged in cells induced to senesce through extended culture, suggesting that expression of this Wnt is not directly linked to senescence (data not shown). Significantly, compared to growing cells, there was a dramatic decrease in abundance of Wnt2 mRNA in senescent cells, regardless of whether senescence was induced by extended growth in culture or activated Ras (Figure 1A).
Wnt2 regulates the canonical Wnt signaling pathway (Gazit et al., 1999; Karasawa et al., 2002). Therefore, we examined other downstream markers of activity of this pathway; namely, the GSK3β kinase and abundance of free, soluble transcriptionally active β-catenin. Activity of GSK3β antagonizes Wnt signaling and its activity towards β-catenin is largely regulated through changes in protein : protein interactions in multimeric protein complexes. In senescent cells, we reproducibly detected an increase in immunofluorescent-stained and immunoprecipitable GSK3β (Figure 1B and C). However, we did not reproducibly detect a comparable increase in total cellular GSK3β, based on direct western blotting of whole cell lysates (data not shown). In senescent cells, we also detected more GSK3β kinase activity in IP-kinase assays, using either autophosphorylation or phosphorylation of a peptide substrate as a readout (Figure 1D and E). These changes in GSK3β immunoreactivity, without a large change in total GSK3β protein, are indicative of remodeling of a GSK3β containing multimeric protein complex in senescent cells. In the absence of canonical Wnt ligand, GSK3β phosphorylates β-catenin, triggering its degradation (Reya and Clevers, 2005; Young et al., 1998). Consistent with increased GSK3β kinase activity towards its substrates in senescent cells, abundance of the transcriptionally active “soluble” pool of β-catenin was decreased in senescent cells (Figure 1F). Previous reports also implicated GSK3β in senescence (Kortlever et al., 2006; Zmijewski and Jope, 2004). We observed similar changes whether senescence was induced by an activated Ras oncogene or extended growth in culture. In sum, by several different measures, activity of the canonical Wnt pathway is repressed in senescent human fibroblasts.
Previously, we showed that initial activation of the HIRA/ASF1a-mediated SAHF-assembly pathway, reflected by HIRA’s localization to PML bodies, occurs early during senescence and is independent of pRB and p53 activity (Ye et al., 2007; Zhang et al., 2005). To assess whether repression of Wnt-signaling and relocalization of HIRA might be mechanistically linked to each other, we analyzed the kinetics of Wnt2 repression in senescent cells and asked whether repression of Wnt2 is independent of pRB and p53. During extended cell culture-induced senescence, Wnt2 mRNA declined early in the program, slightly ahead of HIRA’s localization to PML bodies (Figure 2A), and during oncogene-induced senescence, it occurred between 2 and 4 days after infection (Figure 2B), well before final exit from the cell cycle (e.g. see Figure 7C–E and data not shown). Next, we asked whether repression of Wnt2 occurs in cells lacking functional pRB and p53 pathways. As expected, cells expressing SV40 T-antigen bypassed extended cell culture-induced senescence and did not form SAHF, but ultimately entered “crisis” (Figure 2D, E), demonstrating that SV40 T-antigen has inactivated pRB and p53 in these cells (Ahuja et al., 2005). However, like translocation of HIRA to PML bodies (Ye et al., 2007), repression of Wnt2 expression still occurred in cells expressing SV40 T-antigen (Figure 2C). In sum, repression of Wnt2 and relocalization of HIRA to PML bodies both occur early in the senescence program, and both occur independent of pRB and p53. This is consistent with these two events being mechanistically linked.
We hypothesized that repression of Wnt signaling promotes relocalization of HIRA to PML bodies. Specifically, we asked whether GSK3β activity is necessary and rate-limiting for localization of HIRA to PML bodies and formation of SAHF. To test whether GSK3β is necessary, cells were treated with 3 different inhibitors of GSK3β-LiCl, kenpaullone and a virus-encoded peptide inhibitor, GID (Cohen and Goedert, 2004; Hedgepeth et al., 1999). Regardless of whether senescence was induced by extended culture or activated Ras, each inhibitor blocked localization of HIRA to PML bodies and formation of SAHF (Figure 3A and B). To test whether GSK3β is rate-limiting for localization of HIRA to PML bodies and formation of SAHF, GSK3β wild type, a constitutively active mutant (GSK3βS9A) or an inactive mutant (GSK3βKD) were ectopically expressed in primary WI38 fibroblasts. Wild type GSK3β or GSK3βS9A induced relocalization of HIRA to PML bodies, formation of SAHF, senescence-associated cell cycle exit and SA β-gal activity. In contrast, there was no effect of GSK3βKD or a control retrovirus (Figure 3B and data not shown). If GSK3β drives formation of SAHF via HIRA’s localization to PML bodies, blocking the latter should prevent formation of SAHF. To test this, we utilized a dominant negative mutant, HIRA-C. HIRA-C does not bind to ASF1a and so cannot create SAHF (Tang et al., 2006; Zhang et al., 2005), but blocks translocation of endogenous HIRA to PML bodies and formation of SAHF (Ye et al., 2007). Myc-HIRA-C potently blocked formation of SAHF induced by GSK3βS9A, Figure 3C, D). Taken together, these results show that the Wnt2 -regulated kinase, GSK3β, is necessary and rate-limiting for localization of HIRA to PML bodies and formation of SAHF, and formation of SAHF by GSK3β occurs in a HIRA dependent manner.
HIRA contains 26 consensus GSK3β phosphorylation sites, of consensus S/TXXXpS/T, where S/T is the GSK3β phosphoacceptor site and pS/T is a “priming site” that is typically pre-phosphorylated by another kinase and enhances phosphorylation of S/T by GSK3β (Supplementary Figure 2) (Fiol et al., 1987). This suggests that phosphorylation of HIRA by GSK3β might be responsible for HIRA’s recruitment to PML bodies, and ultimately formation of SAHF. Consistent with this model, HIRA and GSK3β physically interacted in cells and in vitro (Figure 4A, B).
If HIRA is a physiological substrate of GSK3β in vivo, it should be phosphorylated on one or more sites in vivo that are phosphorylated by GSK3β in vitro. Purified recombinant GSK3β phosphorylated residues 421-729 of HIRA (GST-HIRA(421-729)) in vitro. However, conversion of all 13 consensus GSK3β phosphorylation sites in this fragment to non-phosphorylatable alanine residues (GST-HIRA(421-729)ΔP) largely abolished phosphorylation (Figure 4C and Supplementary Figure 2). A mutant of HIRA lacking S697, GST-HIRA(421-729)S697A, was a slightly less efficient substrate than GST-HIRA(421-729) (Figure 4D). However, a mutant in which all GSK3β consensus sites were mutated to alanine except S697, GST-HIRA(421-729)ΔP697S, was phosphorylated almost as well as GST-HIRA(421-729). Moreover, GSK3β also phosphorylated a synthetic phosphopeptide substrate encompassing S697 in vitro (Figure 4E). We conclude that GSK3β efficiently phosphorylates S697 in vitro, but can also phosphorylate other GSK3β consensus sites in HIRA, at least when S697 is unavailable.
To examine HIRA phosphorylation in vivo, HA-tagged HIRA(421-729) and HA-HIRA(421-729)ΔP were ectopically expressed in primary WI38 fibroblasts and the cells pulse labeled with 33P-orthophosphate. The wild type protein migrated as a doublet of 33P-labeled phosphoproteins in SDS-PAGE, and λ-phosphatase treatment confirmed that the slower mobility form depended upon phosphorylation (Figure 4F). HA-HIRA(421-729)ΔP incorporated less total 33P and migrated as a single band (Figure 4G), showing that HIRA is phosphorylated in vivo on sites that conform to the GSK3β consensus. To ask specifically whether serine 697 is phosphorylated in vivo, we raised a phosphospecific antibody to this site. This antibody detected ectopically expressed HA-HIRA(421-729), but not HA-HIRA(421-729)ΔP nor HA-HIRA(421-729)S697A (Figure 4H). The phosphospecific antibody reacted well with substitution mutants containing phosphomimetic aspartate or glutamate residues in place of S697 (namely HIRA(421-729)S697D and HIRA(421-729)S697E), indicating that this antibody reacts with the acidic phosphate rather than unphosphorylated S697 itself (Figure 4H). Phosphorylation of HIRA on serine 697 was inhibited by retrovirus-mediated expression of the GSK3β inhibitory GID peptide, demonstrating that phosphorylation of this site depends upon GSK3β activity in vivo (Figure 4I). In sum, our data indicate that HIRA is phosphorylated in vivo on S697, a site that is phosphorylated by GSK3β in vitro.
Since GSK3β drives HIRA into PML bodies and phosphorylates HIRA on S697, we asked whether serine 697 is required for translocation of HIRA to PML bodies, using an immunofluorescence assay. Compared to wild type HA-HIRA(421-729) (Zhang et al., 2005), both HA-HIRA(421-729)ΔP and HA-HIRA(421-729)S697A were completely defective for targeting to PML bodies, even though PML bodies were readily detectable (Figure 5A–C). The HIRA mutants were expressed and bound efficiently to endogenous cellular ASF1a, suggesting that the proteins are folded normally (Figure 5C). Other HIRA mutants, including a double alanine substitution of two other consensus GSK3β sites (HA-HIRA(421-729)S511A, S515A) and deletion of the ASF1a binding domain of HIRA (HA-HIRA(421-729)ΔB) (Zhang et al., 2005), did not affect localization to PML bodies (Figure 5A and data not shown). We conclude that Wnt signaling is downregulated in senescent cells and recruitment of HIRA to PML bodies depends on serine 697 of HIRA, most likely because of its phosphorylation by GSK3β, a key effector of the repressed Wnt signaling pathway.
If repression of Wnt2 expression is a signal for HIRA’s recruitment to PML bodies, formation of SAHF and onset of senescence, then premature repression of Wnt2 should accelerate the program. We used lentivirus-encoded shRNAs to knock down expression of Wnt2 in young, growing primary human WI38 fibroblasts. 4 out of 5 shRNAs to Wnt2 efficiently knocked down expression of the target. One shRNA, shWnt2–2, had only a slight effect (Figure 6A). There was a good correlation between knock down of Wnt2 and recruitment of HIRA to PML bodies, formation of SAHF judged by DAPI staining and expression of SA β-gal (Figure 6B–D). Most notably, the 4 shRNAs which knocked down Wnt2 induced HIRA relocalization and SAHF, whereas the largely inactive shWnt2–2 and the control virus had no effect. As observed previously (Zhang et al., 2005), an intermediate level of HIRA in PML bodies (shWnt2-1 and shWnt2–3) was accompanied by a low level of SAHF, consistent with recruitment of HIRA to PML bodies preceding formation of SAHF (Ye et al., 2007; Zhang et al., 2007a). SAHF induced by knock down of Wnt2 contained HP1 proteins and macroH2A (Figure 6E and data not shown). Knock down of Wnt2 was observed 4 days after virus infection, before onset of the senescence phenotypes, showing that Wnt2 is not repressed as an indirect consequence of senescence. Knock down of Wnt2 did not induce the large, flat senescent cell morphology (Figure 6D). This separation of senescence phenotypes has been previously observed in primary human melanocytes (Denoyelle et al., 2006), and suggests that the observed effects are not a non-specific consequence of cell stress, since this typically induces the large, flat phenotype (data not shown). Together, our data indicate that repression of Wnt2 is sufficient for HIRA’s recruitment to PML bodies and activation of the SAHF assembly pathway.
In light of these results, we asked whether an increase in the amount of extracellular canonical Wnt ligand delays senescence. We were unable to over-express Wnt2 in a form where it was secreted and active. Therefore, we tested another canonical Wnt ligand, Wnt3a (Willert et al., 2003), which is obtainable in a biologically active form in conditioned medium from an overexpressing cell line, or as purified recombinant protein. Strikingly, Wnt3a-conditioned medium or purified recombinant Wnt3a, delayed HIRA’s localization to PML bodies and formation of SAHF in cells passaged towards senescence (Figure 7A, B). Moreover, based on 5′-BrdU labeling, cell morphology and cell number, Wnt3a delayed cell cycle exit induced by activated Ras in WI38 fibroblasts and Retinal Pigment Epithelial (RPE) cells (Figure 7C–G and data not shown). Likewise, exogenous Wnt3a delayed onset of cell senescence caused by extended growth of WI38 cells in culture (Supplementary Figure 3). We conclude that an increased level of extracellular Wnt3a is able to delay oncogene- and extended cell culture-induced activation of the HIRA/ASF1a SAHF-assembly pathway and senescence in human fibroblasts and epithelial cells.
We report the following important findings. First, we define a molecular mechanism by which formation of SAHF is initiated in senescent cells, by down regulation of Wnt signaling. Second, we describe a function for Wnt signaling as a regulator of cell senescence, to add to its defined cellular roles as a regulator of cell division, differentiation and apoptosis.
We have used recruitment of HIRA to PML bodies as a cell biological read-out of activation of the HIRA/ASF1a SAHF assembly pathway. Several lines of evidence support this approach and indicate that recruitment of HIRA to PML bodies is an essential upstream step in the SAHF assembly process. First, localization of HIRA to PML bodies precedes formation of SAHF (Zhang et al., 2005). Second, in pre-senescent cells HIRA transiently colocalizes in PML bodies with HP1 proteins, prior to eventual incorporation of HP1 proteins in SAHF (Zhang et al., 2005). Third, ectopic expression of HIRA or its binding partner, ASF1a, drives premature formation of SAHF (Zhang et al., 2005). Fourth, RNAi-mediated knock down of ASF1a blocks formation of SAHF (Zhang et al., 2005). Fifth, disruption of PML bodies with a PML-RARα fusion protein blocks formation of SAHF (Ye et al., 2007). Sixth, a HIRA dominant negative mutant, which blocks recruitment of endogenous HIRA to PML bodies, also blocks formation of SAHF (Ye et al., 2007). Consequently, we have proposed that HIRA-containing complexes are assembled or modified in PML bodies, prior to catalyzing formation of SAHF elsewhere in the nucleus (Adams, 2007; Ye et al., 2007; Zhang et al., 2007a).
Here we show that a major signal for recruitment of HIRA to PML bodies and activation of SAHF assembly is down regulation of Wnt signaling in senescent cells. Several lines of evidence support this conclusion. First, repression of Wnt2 and localization of HIRA to PML bodies both occur early during senescence and independent of pRB and p53 activity, strongly suggesting that the two events are mechanistically linked. Second, GSK3β, a kinase whose activity towards its substrates is antagonized by Wnt ligands, is activated in senescent cells and appears to phosphorylate HIRA on a specific residue (S697) that is required for localization of HIRA to PML bodies. Third, inhibition of Wnt2-dependent signaling, by RNAi-mediated knock down of Wnt2 or ectopic expression of GSK3β, drives premature localization of HIRA to PML bodies and formation of SAHF. Fourth, forced activation of Wnt signaling, by exogenous Wnt3a or inhibition of GSK3β, delays recruitment of HIRA to PML bodies and formation of SAHF. These data show that down regulation of canonical Wnt signaling activity is a necessary and sufficient signal for efficient relocalization of HIRA to PML bodies and formation of SAHF.
Typically, in human cells, Wnt signaling maintains proliferation of stem cells by inhibiting cell differentiation and apoptosis and stimulating cell division (Reya and Clevers, 2005). Here we have identified an additional mechanism by which Wnt signaling can promote cell proliferation, by inhibiting senescence of primary fibroblasts and epithelial cells. We present three lines of evidence to show that Wnt signaling regulates senescence in fibroblasts and epithelial cells. First, Wnt2-dependent signaling is down regulated in senescent cells. Second, premature inhibition of Wnt signaling induces a premature senescence. Third, forced activation of the canonical Wnt pathway delays senescence.
These findings raise additional important issues. First, how is Wnt2 repressed? Recently, DNA damage signaling pathways were shown to act as upstream activators of the pRB and p53 pathways, in response to both replicative senescence and oncogene-induced senescence (Bartkova et al., 2006; Di Micco et al., 2006; Herbig et al., 2004; Mallette et al., 2007). Whether DNA damage signaling is responsible for repression of Wnt2 remains to be tested. Second, what is the relative contribution of different Wnt signaling pathways to delay of senescence by Wnt? A key effector of canonical Wnt is the transcription factor β-catenin (Reya and Clevers, 2005). One transcriptional target of β-catenin is c-myc, and decreased expression of c-myc triggers cell senescence (Guney et al., 2006). However, as reported previously, we found that increased β-catenin activity alone does not delay senescence, but instead induces cell cycle exit (Damalas et al., 2001)(Supplementary Figure 4). We showed previously that inactivation of HIRA/ASF1a-mediated SAHF formation delays onset of senescence (Zhang et al., 2005). On balance, it seems likely that canonical Wnt signals delay senescence through the coordinated and combined effect of multiple pathways, including activation of β-catenin, expression of c-myc and inactivation of the HIRA/ASF1a SAHF-assembly pathway.
Regardless of these issues, the demonstration that Wnt signaling is a regulator of cell senescence has important implications, because senescence is, perhaps, a source of tissue aging and a known tumor suppression process in vivo (Braig et al., 2005; Campisi, 2005; Chen et al., 2005; Collado et al., 2005; Courtois-Cox et al., 2006; Dimri et al., 1995; Janzen et al., 2006; Krishnamurthy et al., 2006; Michaloglou et al., 2005; Molofsky et al., 2006; Serrano et al., 1997). Consistent with Wnt signaling being an antagonist of senescence and tissue aging, this pathway contributes to regeneration of muscle and delays signs of aging in bone (Bennett et al., 2005; Polesskaya et al., 2003). Consistent with Wnt signaling being an antagonist of senescence and tumor suppression, elevated Wnt signaling is a driving force in many cancers (Reya and Clevers, 2005). Wnt2 itself is overexpressed in several cancers (Clement et al., 2006; Dale et al., 1996; Vider et al., 1996; Yoshida et al., 1994). During the earliest stages of tumorigenesis, elevated Wnt signaling might delay oncogene-induced senescence long enough to allow acquisition of additional key genetic and epigenetic alterations that further promote neoplastic progression. Although Wnt ligands appear to delay, and not prevent, oncogene-induced senescence, the rounds of DNA synthesis that occur in the presence of activated oncogenes are known to be error prone (Bartkova et al., 2006; Di Micco et al., 2006; Mallette et al., 2007). Interestingly, Baylin and coworkers have suggested that activation of Wnt signaling by epigenetic silencing of the secreted Wnt inhibitor sFRP might provide a selective advantage to very early colorectal Aberrant Crypt Foci (ACF) lesions that have frequently acquired activated Ras oncogenes but typically lack mutations in the APC tumor suppressor (Suzuki et al., 2004). Conceivably, activation of Wnt signaling may delay oncogene-induced senescence in this context, thereby promoting expansion of the lesion and, in some cases, facilitating progression along a tumorigenic path.
To our knowledge, canonical Wnt is only the second extracellular ligand to modulate senescence. Recently, Bernards and coworkers showed that the secreted extracellular factor, Plasminogen Activator Inhibitor-1 (PAI-1), contributes to senescence in mouse and human fibroblasts (Kortlever et al., 2006). Previously published findings are consistent with the idea that Wnt is a physiological antagonist of senescence in vivo. First, senescence and Wnt signaling antagonistically impact proliferation of the same cell populations, most notably epithelial cells and fibroblasts (Campisi, 2005; Reya and Clevers, 2005)(this work). In vivo Wnt signaling maintains proliferation of neuronal stem cells and hematopoietic stem cells (Chenn and Walsh, 2002; Reya et al., 2003), two populations that may be prone to senescence in vivo (Geiger and Van Zant, 2002; Palmer et al., 2001). Second, as discussed above, Wnt signaling and senescence appear to be antagonistic in terms of their effects on cancer progression and, perhaps, tissue aging. Thus, Wnt signaling and cell senescence appear to be antagonistic at the cellular, tissue and organismal levels.
In this manuscript we have investigated this antagonism at a molecular and cellular level. In the process, we have shown that down regulation of Wnt signaling activity is a signal for formation of SAHF in senescent cells. This defines a function for Wnt signaling as a regulator of cell senescence. Finally, we have shown that cell senescence is regulated by canonical Wnt signals. In view of the near-ubiquitous role of canonical Wnt ligands as regulators of tissue homeostasis, this discovery has major implications for our understanding of both senescence and Wnt signaling in tissue development and homeostasis, and aging and cancer.
For additional methods, see Supplementary Info.
WI38 cells were cultured according to the ATCC, in 2% (approximating physiological, Figures 6 and 7a–e, Sup. Fig. 3) or 21% (ambient, Figures 1, ,2,2, ,3,3, ,4,4, ,5,5, ,7b,7b, Sup. Fig. 1 and and4)4) oxygen as indicated. Throughout, similar results were obtained in 2% and 21% oxygen. Experiments were performed on WI38 cells between PD #32 and 36, unless otherwise stated. Wnt3a-expressing and parental L cells were obtained from ATCC (CRL-2647 and CRL-2648). Conditioned medium was collected after 4 days and 7 days of culture and applied to WI38 or RPE cells as a 5-fold dilution.
“Soluble” transcriptionally active β-catenin was extracted from cells in a buffer containing 10mM Tris-HCl (pH7.5), 0.05%(v/v) NP40, 10 mM NaCl, 3 mM MgCl2, 1mM EDTA and protease and phosphatase inhibitors.
We thank Peter Klein, Maria Barna, Bill Hahn and Gary Nolan for reagents; Peter Klein for advice; Jonathon Chernoff and Davide Ruggero for reading the manuscript. This work was supported by grants from the NIH (CA104429, GM062281) and DOD (DAMD17-02-1-0726) to PDA and AFAR fellowship to RZ. PDA is a Leukemia and Lymphoma Society Scholar.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.