Senescence Elicits Stemness: A Surprising Mechanism for Cancer Relapse

Abstract

Cellular senescence is traditionally viewed as a permanent form of cell cycle arrest that restrains tumorigenesis. In a recent study in Nature, however, Milanovic et al. (2018) challenge this conventional view, showing that senescence can counterintuitively promote cancer sternness and tumor aggressiveness. This finding suggests that attacking senescence can be exploited in cancer therapy.

A promising aspect of current therapeutic approaches to aging is the notion that addressing underlying causes of aging would ameliorate a broad array of age-associated diseases. Age is a major risk factor for cancer, and thus, advances to forestall aging may impede cancer. Certain characteristics of aging occur in cellular senescence. Commonly known as the Hayflick limit, senescence is a stable form of cell cycle arrest that is implicated in cancer and in aging (He and Sharpless, 2017). On the one hand, senescence is a potent tumor-suppressive mechanism. This is mediated by the cell cycle arrest program and the senescence-associated secretory phenotype (SASP) that induces immuno-surveillance of pre-malignant cells. On the other hand, however, senescent cells accumulate in aged tissues, blocking tissue renewal and contributing to chronic inflammation associated with age-related diseases, including cancer. Hence, selective manipulation of senescence is currently an important biomedical objective. In a recent study by Milanovic et al. (2018) in Nature, a novel feature of senescence is revealed: senescence unexpectedly is associated with sternness and with the potential to develop highly aggressive tumors (Figure 1). This new finding broadens our understanding of the biology of senescence, has significant implications in cancer therapy, and suggests yet another reason for why manipulating senescence would be beneficial to health. Using an established therapy-induced senescence lymphoma model, Milanovic et al. (2018) compared gene expression profiles in senescence-competent and -incompetent cells and found that an adult tissue stem cell signature is strongly enriched in senescent cells. They found that this holds true in multiple models of senescence, including replica-tive senescence and stress-induced senescence, in both mouse and human cells. Hence, senescence-associated sternness (termed “SAS” by Milanovic et al., 2018) is a new common feature of senescence in addition to the established cell cycle arrest and SASP programs.

Figure 1. Therapy-Induced Senescence Can Promote Cancer Sternness and Cancer Aggressiveness.

In response to cancer therapy, such as chemotherapy or radiation therapy, tumors regress by cell death (apoptosis or necrosis) and senescence. Senescence is traditionally viewed as an endpoint of cancer therapy, as senescent cells are non-dividing. However, Milanovic et al. (2018) report that senescent cells can acquire features of sternness, partly through activation of Wnt. This can give rise to tumor-initiating cells (e.g., cancer stem cells), which can eventually lead to cancer relapse and metastasis.

Does SAS associate with functional consequences? Using the lymphoma model combined with genetic disruption of key mediators of senescence, including p53, Milanovic et al. (2018) generated senescence-released cells that reverted senescence. Compared to “never senescent” counterparts, the senescent-released cells displayed significantly elevated tumor-initiating potential as measured by cell proliferation, colony formation assay, and tumor formation in mice. In this acquisition of sternness—a fundamental feature of cancer stem cells (Reya et al., 2001)—the senescent-released cells unexpectedly resemble cancer stem cells. Milanovic et al. (2018) further extended their studies to leukemia models, including T-ALL and AML, and in both cases, therapy-induced senescent-released cells showed markers of sternness and elevated tumor-initiating potential, indicative of cancer stem cells. These results collectively suggest that SAS may be a general mechanism leading to the generation of highly aggressive and malignant tumor-initiating stem cells in response to cancer therapy.

The SASP program has recently been shown to promote reprogramming and sternness in several in vitro and in vivo models (Cahu et al., 2012; Mosteiro et al., 2016; Ritschka et al., 2017). Using inhibition of NFκB, a key transcription factor for SASP genes, Milanovic et al. (2018) conclude that SAS can arise even when ablating a functional SASP program in the lymphoma model. Thus, the roles of SASP in senescence and tissue remodeling are context specific depending on distinct biological models and pathological stressors. It appears that, while SAS can be a cell-intrinsic program as shown by Milanovic et al. (2018), the SASP can mediate additional features of sternness and tissue plasticity in the senescence microenvironment, as reported by others.

How does the cell-intrinsic program drive the SAS? One mechanism is through activation of Wnt signaling that is essential for stem cells (Reya et al., 2003). Activation of the Wnt signature in senescence is identified by Milanovic et al. (2018) and others (Pawlikowski et al., 2013). Notably, inhibition of Wnt blunts SAS and the associated tumor-initiating potential. In addition to Wnt, additional mechanisms are likely to be involved in SAS. One such mechanism could be global reorganization of the epigenome in senescence, including alterations of both active euchromatic and repressive heterochromatic regions (Sen et al.,2016). Indeed, epigenome reorganization is considered a central event for the creation of induced pluripotent stem cells (Maherali et al., 2007). Furthermore, senescent cells must overcome the cell cycle arrest and the shortened telomere in order to convert to functional stem cells. The mechanisms that initiate this SAS program represent an interesting topic for future investigation. Understanding the “choice” of senescent cells between cell cycle arrest and becoming stem cells will be key to unraveling mysteries of the senescence program and will provide critical new knowledge in tumor suppression and progression.

While senescent-released cells in this study were in large part generated by genetic ablation of critical senescence mediators, Milanovic et al. (2018) did notice a rare and spontaneously occurring senescent-released population in the lymphoma model. While it remains unclear whether this population acquires sternness or tumor-initiating potential, this observation raises an important general question regarding senescence evasion beyond the scope of this study: do senescent-evaded cells arise by bypassing initiation of senescence or by release after establishment of senescence? The SA-β-gal-positive and EdU-positive nature of these cells, shown by Milanovic et al. (2018), supports derivation from senescent-released cells. However, in many other oncogene-induced cancer models in which senescence is involved, it is yet unknown whether the tumor-initiating cells are generated from failure to initiate senescence or by senescence release. If it is the latter case, do the malignant cells acquire sternness? Answering this ques tion may require a new mouse model in which senescent cells can be traced from initiation to release. Importantly, Milanovic et al. (2018) showed that longitudinal studies of lymphoma patients before and after chemotherapy revealed activation of Wnt in relapsed samples, consistent with acquirement of sternness. Although it is unclear whether these relapsed samples were generated from senescence release, the current study paves the way for future investigation of SAS in tumorigenesis and cancer therapy.

While the cell cycle arrest and SASP programs of senescence have important biological significance, the physiological function of SAS remains unknown. Milanovic et al. (2018) postulate that SAS may serve an alternative role as an actual rescue mechanism in development and homeostasis to counter imminent cell loss in tissues. While this is speculative as a pathway to replenish tissue via revitalization of senescent cells, indirect supportive evidence is that senescence can be involved in wound healing and embryonic development (He and Sharpless, 2017). An alternative theory could be that SAS is solely a byproduct of cancer therapy and is not associated with physiological functions per se. Although future studies are needed to address a potential physiological function of SAS, Milanovic et al. (2018) clearly show that, in the context of cancer therapy, SAS has a detrimental potential to generate highly aggressive tumor-initiating cells that contribute to tumor relapse. Finally, it is relevant that several agents that kill senescent cells (termed “senolytics”) are in clinical trials. This current study provides a further rationale for investigating senolytics in cancer therapy and provokes an exciting new avenue to study senescence and sternness in broader scenarios of health and diseases.