Unexpected pieces to the senescence puzzle

Abstract

Although initially described as the end state of mortal cells after extended rounds of cell division in culture, it is now clear that cellular senescence induced by several different stimuli plays an important role in tumor suppression in vivo. Three recent studies by Acosta et al.; Kuilman et al.; and Wajapeyee et al. in Cell report that secreted proteins play an important role in enforcing the senescence response. These new findings identify unanticipated contributors to this cell state that suppresses tumor development.

Cellular senescence is characterized by an irreversible arrest in proliferation and distinctive morphological alterations. Although originally described in human fibroblasts at the end of their replicative potential, it is now clear that senescence can be induced both in vitro and in vivo by a pleiotropic group of stimuli that include oncogene activation, telomere dysfunction, and agents that damage DNA or alter chromatin structure (Campisi and d'Adda di Fagagna, 2007). These observations support the view that similar to apoptosis, senescence is a programmed cellular response that can be triggered by a variety of stresses. Abundant recent work indicates that senescence serves as an important mechanism of tumor suppression that is activated in benign lesions (Braig et al., 2005; Chen et al., 2005; Collado et al., 2005; Courtois-Cox et al., 2006; Michaloglou et al., 2005). Senescence is also implicated in the loss of regenerative potential in aging tissues. The loss of proliferative potential in senescent cells contributes to diminished tissue function (reviewed in Campisi and d'Adda di Fagagna, 2007). Moreover, because senescent cells remain metabolically active, it has been proposed that these cells may also play a role in tissue aging by influencing the neighboring tissue microenvironment either positively or negatively. Thus, senescence may serve as a key mechanistic link between aging and cancer.

At the molecular level the p53 and retinoblastoma (RB) tumor suppressor pathways serve as critical cell-cycle checkpoints that mediate both replicative and oncogene-induced senescence (OIS). Although recent work has begun to identify the molecules that enforce senescence in response to different signals, we still have only a rudimentary understanding of the components that comprise and execute the senescence program. Indeed, no truly validated molecular markers exist to identify senescent cells. In this context, three groups (Acosta et al., 2008; Kuilman et al., 2008; Wajapeyee et al., 2008) use unbiased genomic approaches to identify specific secreted proteins play an essential role in OIS. These unexpected findings add another level of complexity to our understanding of senescence and its role in aging and cancer.

Both Wajapeyee et al. (2008) and Acosta et al. (200) used loss-of-function screens in cultured diploid human fibroblasts to identify genes required to induce senescence. Wajapeyee et al. made use of an oncogenic allele of the serine/threonine kinase BRAF (BRAF^V600E) that contains a mutation known to occur with high frequency in malignant melanomas. After confirming that the expression of BRAF^V600E was sufficient to induce senescence and apoptosis, the authors introduced a short hairpin RNA (shRNA) library targeting human genes into the BRAF^V600E-expressing cells. Using this approach they identified 17 genes whose suppression permitted the cells to proliferate in the presence of BRAF^V600E. Of these genes, one encoded the insulin-like growth factor binding protein 7 (IGFBP7), a secreted protein previously implicated in senescence in breast cancer cells (Wilson et al., 2002). IGFBP7 is a low-affinity insulin-like growth factor (IGF) binding protein (Burger et al., 2005). Although it is unknown whether IGFBP7 mediates its effects via binding and suppressing IGF proteins, it is notable that IGFs have been long suspected to promote tumorigenesis. Indeed, therapeutic strategies focused on inhibiting the IGF signaling pathway are currently in development.

Wajapeyee and colleagues showed that BRAF^V600E upregulated IGFBP7 expression at least in part through stimulating the binding of the transcription factor activator protein 1 (AP-1) to the IGFBP7 promoter. Consistent with the hypothesis that IGFBP7 suppresses Raf-driven tumor development in vivo, melanomas that harbored BRAF^V600E mutations exhibited hypermethylation of a CpG island in the IGFBP7 promoter relative to either benign lesions (nevi) that harbored BRAF^V600E or melanomas that lacked the BRAF^V600E oncogene. Interestingly, cultured melanocytes expressing BRAF^V600E were sensitive to secreted IGFBP7 in the culture media, which was sufficient to induce apoptosis. Taken together, these observations implicate IGFBP7 as an essential component of a autocrine/paracrine feedback loop that constrains BRAF^V600E-stimulated cell proliferation.

Acosta et al. used a similar approach to search for genes that extended the lifespan of IMR-90 cultured human fibroblasts (Acosta et al., 2008) and found the chemokine receptor CXCR2 to be required for replicative senescence and oncogene-induced senescence. Furthermore, overexpression of CXCR2 or the CXCR2 ligands interleukin 8 (IL-8) and GROα/Gro-1 led to a proliferative arrest that was dependent upon the expression of p53. Notably, expression of oncogenic RAS was previously shown to enhance Gro-1 expression and induce senescence of murine stromal fibroblasts, underscoring the importance of this chemokine secretory response in senescence (Yang et al., 2006). Indeed, these CXCR2 ligands were also found to be upregulated in cells rendered senescent by mitogen activated kinase kinase (MEK) activation. Using small molecule inhibitors, Acosta et al. further demonstrated that blocking the p38 and NF-κB pathways suppressed the observed induction of IL-8 expression by MEK. Finally, they showed that CXCR2 is upregulated in benign lesions such as chemically induced papillomas and prostate intraepithelial neoplastic lesions, suggesting that this pathway is upregulated in spontaneously arising premalignant lesions.

Kuilman et al. used whole genome transcriptional analyses of cell lines expressing BRAF^V600E in their search for senescence factors (Kuilman et al., 2008). Similar to what has been previously reported for replicative senescence (reviewed in Campisi and d'Adda di Fagagna, 2007), they found that of the many transcriptional changes observed in senescent cells, a significant number of the upregulated genes are involved in chemokine and cytokine signaling. In particular, Kuilman et al. found that IL-6 was reproducibly upregulated in BRAF^V600E-expressing cells and verified that suppression of IL-6 or its cognate receptor was sufficient to allow these cells to re-enter the cell cycle and proliferate, thereby bypassing OIS. However, overexpression of IL-6 alone was not sufficient to induce senescence, suggesting that IL-6 acts in concert with other factors to initiate senescence. Indeed, the suppression of IL-8 resulted in similar biological effects as the suppression of IL-6. Kuilman et al. further showed that the levels of the transcription factors C/EBP (CCAAT-enhancer-binding proteins) increased in OIS. They determined by chromatin immunoprecipitation that C/EBP was present at both the IL-6 and IL-8 promoters, suggesting a likely mechanism for the upregulation of these cytokines in senescence. Moreover, they found that IL-6 and C/EBP function in a positive feedback loop, and that both are required to amplify and sustain the inflammatory network and senescence. Importantly, similar to the findings of Acosta et al. (2008), Kuilman and colleagues showed that the expression of IL-8 correlated with a non-proliferative phenotype in colonic adenomas, supporting the possibility that these cytokines function in vivo to promote cell senescence in benign human tumors.

Although each of these three studies identified different genes involved in the senescence response, the common theme that emerges is that secreted proteins play a critical role in the induction and maintenance of some forms of OIS. Although surprising, these observations provide new insights into our understanding of OIS. So, how do these findings interface with our current mechanistic understanding of OIS and human tumor development?

Connections with known mediators of senescence

At least three cellular responses to oncogenic stress have been proposed to mediate OIS:, heterochromatin formation, activation of the DNA damage response (DDR), and negative feedback signaling within the senescence pathways (reviewed in Campisi and d'Adda di Fagagna, 2007). Importantly, these mechanisms of OIS are not mutually exclusive and all of these processeses are intimately connected to the activity of the RB and p53 pathways. The derepression of the INK4/ARF locus is one mechanism known to contribute to the activation of RB and p53 and has been shown to play a critical role in initiating and maintaining the senescence response; however, several lines of evidence suggest that additional cooperating signals exist. Although the details differ in these studies, the secreted proteins featured herein appear to provide such signals and converge with each of these proposed senescence models. A unified model integrating these new findings is depicted in Figure 1.

Figure 1. Secreted proteins and oncogene-induced senescence.

Expression of an oncogenic allele of BRAF (BRAF^V600E) in human cells induces senescence. Three recent genetic screens using cells expressing BRAF^V600E identified the secreted proteins (green) insulin-like growth factor binding protein 7 (IGFBP7), interleukin 8 (IL-8), and interleukin 6 (IL-6) as essential mediators of BRAF^V600E-dependent senescence. These unexpected observations provide new insights into the overlapping yet distinct mechanisms of the p53 and RB tumor suppressor pathway activation, which induces the senescence response. IGFBP7 expression is mediated by activating protein 1 (AP-1. IGFBP7 inhibits the MEK mitogen activated kinase pathway via the upregulation of the Raf inhibitory protein (RKIP) but promotes p16 expression. The transcription factor (blue) C/EBP (CCAAT-enhancer-binding protein) promotes IL-6 expression. IL-6 activity is mediated by its receptor IL-6R. The chemokine receptor CXCR or its ligands IL-8 and GRO-α promote p53-dependent senescence. IL-8 upregulation is dependent upon the transcripton factors C/EBP and NFκB.

Heterochromatin formation

Although no universal marker of senescence has been described, many senescent cells undergo dramatic changes in chromatin structure and form easily visualized senescence-associated heterochromatic foci (SAHFs) (Narita et al., 2003). The formation of SAHFs requires the recruitment of RB and heterochromatin proteins (Adams, 2007; Narita et al., 2003). This provides a plausible mechanism for the stable suppression E2F target genes and other genes necessary to enforce an irreversible senescent state. One RB pathway regulator, p16^INK4a, has been shown to be essential for OIS in vivo and in vitro. However, others have noted that p16^INK4a is not expressed in every senescent cell within BRAF^V600E-expressing melanocytic nevi, supporting the notion that other signals contribute to the activation of RB in some settings (Michaloglou et al., 2005). One such signal may be the cyclin kinase inhibitor p15^INK4b, which was identified as a marker of Ras-induced senescence (Collado et al., 2005). The work of Kuilman and colleagues (2008) provides an interesting link between p15^INK4b, heterochromatin and the IL-6/C/EBP pathway. Specifically, inactivation of IL6 or C/EBP not only prevented the proliferative arrest triggered by oncogenic BRAF^V600E but also inhibited p15^INK4b expression and SAHF formation. Notably, C/EBP has previously been shown to bind and activate the p15^INK4b promoter (Gomis et al., 2006), providing a mechanistic explanation for these findings. Taken together these observations suggest that IL-6 and C/EBP coordinately function to upregulate p15^INK4b which in turn contributes to RB activation, SAHF formation, and OIS.

DNA damage

OIS can be mediated by DNA damage associated with replicative stress or the accumulation of reactive oxygen species (ROS) (Bartkova et al., 2006; Di Micco et al., 2006; Mallette et al., 2007). Following their identification of CXCR as a factor involved in senescence, Acosta et al. directly explored a role for chemokine signaling in the DNA damage response. Specifically, they showed that inactivation of CXCR2 inhibited the DNA damage response and abrogated the proliferative arrest of irradiated fibroblasts. In addition, CXCR2-induced senescence and SAHF formation depended on functional p53 (in murine cells) or p53 and RB (in human cells). Further work is necessary to decipher the effects of CXCR2 ablation on the DNA damage response and p53 expression in response to oncogenic stress. However, these observations do suggest that excessive CXCR signaling during OIS could possibly allow for increased ROS accumulation, thus promoting the DNA damage response and sustained activation of p53.

Negative feedback signaling

Wajapeyee et al. suggest that IGFBP7 mediates its anti-proliferative effects via the induction of a negative feedback signaling response. Specifically they showed that conditioned media from IGFBP7-expressing cells or recombinant IGFBP7 purified from baculovirus-infected insect cells potently suppressed the activation of the mitogen activated kinase (MAPK) ERK. Moreover, they demonstrated that in melanoma cells this suppressive effect was largely mediated by the up-regulation of the Raf Inhibitory Protein (RKIP), a protein that disrupts the interaction between upstream activators of ERK, Raf and MEK. Importantly, the anti-proliferative effects of IGFBP7 in melanoma cells were suppressed by the ectopic expression of an activated ERK2 or MEK allele. Thus, Wajapeyee et al. concluded that the suppression of cell proliferation conferred by IGFBP7 is mediated, in part, by a negative feedback signal that emanates from the initial oncogenic insult. These findings are conceptually similar to another report demonstrating that a negative feedback signaling network involving multiple negative regulatory genes is activated by an oncogenic allele of RAF during OIS (Courtois-Cox et al., 2006). Given that other studies have demonstrated that perturbation of components of the Ras/phosphatidylinositol 3-kinase (PI3K) signaling pathways also contribute to OIS (Chen et al., 2005; Courtois-Cox et al., 2006), perhaps negative feedback suppression of both the MEK/ERK and PI3K pathways cooperate to mediate tumor suppression. This model is supported by the observation that both the ERK and PI3K pathways are inactive in some benign human tumors (Courtois-Cox et al., 2006).

Mediating senescence by threshold

The process of OIS appears to be dependent on a rapidly growing list of genes. How then is it possible that the inactivation of any one gene can block senescence? There are two non-mutually exclusive possibilities. First, although it is clear that p53 and RB regulated senescence can be induced by many stimuli, the activation of the p53 or RB cell-cycle checkpoint under normal cellular contexts fails to induce a senescence response. Therefore, it is likely that a combination of activating signal amplitude and duration above a certain threshold is necessary to trigger an irreversible proliferative arrest. For example, although p16^INK4a is certainly an important regulator of OIS, it is possible that other cyclin kinase inhibitors such as p15^INK4b reinforce and amplify this signal. Similarly, the DDR pathway serine/threonine kinase ATM and the p53 regulator ARF may collaborate in response to certain oncogenic insults to potentiate p53 activation (Bartkova et al., 2006; Di Micco et al., 2006; Mallette et al., 2007).

A second possibility is that the activation of pathways linked to senescence triggers a positive feedback loop necessary to enforce the senescence response. Indeed, Kuilman et al. demonstrated that cytokines IL-6 and C/EBP form this type of feedback loop and observed that in the absence of either of these two genes, the entire network of signaling pathways that regulate the inflammatory response is disrupted.

Although such thresholds and feedback loops in this system are as of yet incompletely defined, these models could explain how multiple different signals may cooperate to induce the senescence response as dependent upon the context. If the threshold response model is correct, the level and strength of oncogenic signals should be carefully considered when interpreting studies that focus on OIS. In addition, while OIS is generally utilized to describe senescence triggered by any oncogenic insult, it is likely that specific oncogenes may be attenuated by different inhibitory signals. Consistent with this notion, Wajapeyee et al showed that melanomas expressing a mutant RAS allele were much less sensitive to the inhibitory effects of IGFBP7 relative to melanomas harboring the BRAF^V600E allele.

Regardless of the details of how senescence may be initiated, these new findings have important implications for our understanding of tumor progression. Inactivation of any one of the proteins identified in the three screens in a specific setting may allow premalignant cells to bypass OIS and thereby contribute towards the development of a malignant lesion. Thus, identifications of signals that promote senescence in a given tumor type may provide new therapeutic targets for cancer. The observation that mice harboring human tumor xenografts treated with recombinant IGFBP7 show tumor growth suppression lends support for this possibility (Wajapeyee et al., 2008).

While the genes identified in these studies clearly act to promote senescence in an autocrine manner, because the gene products are secreted factors, it is also necessary to consider the consequences of upregulating these genes in situ. In particular, it has been reported that senescent cells can induce their own clearance (Xue et al., 2007). Alternatively, senescent cells have also been observed to elicit effects on the tumor stroma that serve to enhance tumor growth (Krtolica et al., 2001). Thus, depending on the context, these secreted proteins may also serve to inhibit or enhance cancer initiation and progression in addition to enforcing OIS in a cell autonomous manner. Thus, despite the recent rapid progress in deciphering the molecular regulators and components of senescence, it is clear that we have only begun to unravel the complex regulation of cell senescence and its role in tumor suppression.