Multicellular life requires individual cells to cooperate in a way that benefits the organism. Cells that are uncooperative because they are damaged or dysfunctional, and that pose a threat, are either eliminated by cell death or undergo a usually irreversible growth arrest called senescence1. Senescent cells typically never divide (although there are some rare examples of cells exiting senescence and resuming division), but they can persist in tissues and contribute to ageing and cancer progression2,3. Writing in the Journal of Cell Biology, Tonnessen-Murray et al.4 reveal a deadly activity that underlies the persistence of senescent cells — they can eat their neighbours alive.
Cellular entry into senescence benefits an organism because it inhibits cancer development by preventing the division of cells that have accumulated extensive DNA damage or that express cancer-promoting genes called oncogenes2,5. Senescent cells are metabolically active6, and this is characterized by their secretion of proinflammatory molecules as part of a phenomenon termed the senescence-associated secretory phenotype (SASP) response2. Senescent cells can promote cancer progression and resistance to anticancer therapy in some contexts, as a result of the secretion, through SASP, of growth factors and immune-signalling molecules called cytokines2.
Chemotherapy that damages the DNA of cancer cells can result in their death or their entry into senescence. Tonnessen-Murray and colleagues investigated the effects of chemotherapy-driven senescence in breast cancer cells in mice treated with the chemotherapeutic drug doxorubicin. Under the microscope, they saw senescent cells eating and digesting entire neighbouring cells (Fig. 1). This striking observation was made in breast tumours formed of mixtures of transplanted cancer cells, which were engineered to express red or green fluorescent proteins. It can be difficult to observe a cell being internalized by another cell (a process termed engulfment) in cancer tissues. By growing tumours with mixtures of fluorescently labelled cells, the authors could clearly identify red- or green-labelled cells being taken up into neighbouring cells labelled by the other colour.
Engulfment also occurred at high rates for mouse and human breast cancer cells grown in vitro and treated with doxorubicin or another chemotherapeutic drug, paclitaxel. Ingestion peaked at 4–6 days after drug treatment, a time that correlated with the induction of senescence. The cells that were engulfed by the senescent cells were neighbouring senescent or non-senescent cancer cells. They showed no sign of being dead, and engulfment occurred even in the presence of a cell-death inhibitor molecule. This led the authors to conclude that the ingested cells were being eaten alive.
Ingested cells are broken down in a digestive organelle called the lysosome. Crucially, senescent cells that ate their neighbours survived longer in vitro than those that did not. This finding suggests that metabolic building blocks retrieved from the lysosomal digestion of neighbouring cells were being used by senescent cells to promote their survival.
This surprising finding that cell death supports the survival of senescent cells highlights the complexity of cell-death regulation in multicellular animals. Numerous mechanisms of cell death occur in animal tissues. These include forms of cell suicide, such as apoptosis, which leads to the fragmentation of individual cells, and regulated forms of necrotic cell death that induce cell rupture7. Some cell deaths are also carried out as ‘murders’7,8. These typically involve the presence of engulfing cells, and occur by at least two distinct mechanisms9.
One is a form of cell death called entosis, in which living cells that are destined to die invade a neighbouring cell and become engulfed10. Another mechanism is cellular cannibalism, in which living cells that will be ingested are targeted by a type of engulfment that resembles phagocytosis — the process typically used by immune-system cells such as macrophages to ingest and destroy dying cells9. Such cellular murders can support the survival of particular cells in a population that benefit from the metabolic banquet derived from ingesting and degrading whole cells11,12.
The authors examined the mechanism of senescence-associated engulfment and found that, although entosis could occur in the type of tumour cell studied, the engulfment of senescent cells did not involve the proteins required for entosis10. The authors analysed the gene-expression profile of cancer cells treated with chemotherapy drugs (most of these cells were senescent), and found that genes characteristic of phagocytosis were expressed. This gene expression peaked within a timeframe that correlated with the cellular engulfment. Senescent cells were also observed to engulf dead cells added in vitro, providing further evidence for the authors’ model that senescent cells engulf cells by phagocytosis.
Cell cannibalism in cancers has been reported previously9,12. However, Tonnessen-Murray et al. specifically identify an association between cannibalism and senescence, and show that this phenomenon might make a substantial contribution to the persistence of senescent cells in cancer tissues. The authors observed that cannibalism by senescent breast cancer cells occurs irrespective of whether or not the cell has functional p53, a notable tumour-suppressor protein that can control entry into senescence13. The authors tested chemotherapy-induced senescent cells of other types of cancer, including lung cancer and a bone cancer called osteosarcoma, and found that these cells also cannibalize neighbouring cells. Together, these findings suggest that cell cannibalism might be an activity that is broadly associated with the induction of senescence, rather than being linked to particular types of cancer or to the status of proteins such as p53. It will be important to investigate whether cannibalism is linked to senescence in other contexts, for example during tissue development when senescence can occur14,15, or in aged tissues that accumulate senescent cells3.
Entosis in cancer-cell populations can promote competition between individual cells in which ‘winner’ cells ingest and kill neighbouring ‘loser’ cells, removing them from the population16. Whether cells behave as winners or losers depends on certain cellular characteristics, for example differences in the tension of the internal cellular framework called the cytoskeleton16. It would be interesting to investigate whether senescent cells choose particular target cells to cannibalize in a competitive fashion. In cancers, complex mixtures of cells coexist in the tumour microenvironment, and this cellular composition changes over time or in response to anticancer therapy. The authors propose that cell cannibalism might affect cancer progression by supporting the SASP response. However, it is worth considering whether it might also contribute directly to cancer progression by removing particular cells from the tumour microenvironment. And if normal cells are found to be removed by senescent cells in aged tissues, this depletion might contribute directly to tissue degeneration.