Cellular senescence stops the growth of cells. This process, first glimpsed in cell culture, is now confirmed by in vivo evidence as a vital mechanism that constrains the malignant progression of many tumours.
Societies have traditionally taken three approaches to handling recidivist criminals: exile, execution and lifetime imprisonment. It seems that human cells use similar strategies to prevent rogue cells harbouring dangerous mutations from turning into fully fledged cancers. Epithelial tissue, such as that lining the airways and intestines, continuously renews and sloughs off, thereby sentencing some pre-cancerous cells to extra-corporeal exile. There is also a cellular version of the death penalty — apoptosis, a well-established anticancer mechanism. And in this issue, four groups1,2,3,4 report striking in vivo evidence that the body can subject potential cancer cells to the equivalent of a life-sentence: cellular senescence.
Senescence is a specific form of stable growth arrest provoked by diverse stresses, including the enforced expression of cancer-promoting genes in cultured cells. This ‘oncogene-induced senescence’ (OIS)5 is linked to known cancer pathways in cultured cells, notably the ARF–p53 and p16INK4a–RB pathways (Fig. 1). But whether OIS is an authentic anticancer process in vivo, or simply an artefact of enforced oncogene expression in cells experiencing culture shock6, has been controversial.
This issue is settled by the new papers1,2,3,4, which show that OIS occurs in vivo in several diverse precancerous tissues from both human and mouse. In addition, the work identifies much-needed markers of senescence, and further delineates the molecular underpinnings of this key tumour-suppressing process. A compelling feature of these studies is the consistency of OIS in response to a variety of cancer-causing mutations in different human tumour types and mouse-model systems. At the same time, the reports reveal that the molecular circuitry of OIS may be wired differently among tumour types.
Michaloglou et al. (page 720)1 worked with cultures of human melanocytes (pigmented skin cells) and nevi (skin moles, the benign precursors of malignant melanoma). They found that nevi harbouring mutations of the BRAF protein (mutations that are frequently found in melanomas) have robust expression of senescence markers and do not seem to proliferate. In melanoma cells, however, senescence is extinguished and proliferation accelerated.
Curiously, the tumour suppressor p16INK4a — a known activator of senescence that is deleted in melanoma cells — showed spotty expression in nevi, and experimental depletion of p16INK4a failed to increase BRAF-induced senescence in melanocyte cultures. Mutated BRAF in melanocytes also failed to induce the ARF and p53 tumour suppressors, two proteins integral to the activation of senescence in many systems. These results expose serious gaps in our understanding of the genes and pathways that function to constrain the transformation of nevi into lethal melanomas.
Exploring the evolution of prostate cancer, Chen et al. (page 725)2 discovered senescence in early-stage prostate abnormalities in humans and in mice engineered to sustain prostate-specific deletion of the PTEN tumour-suppressor gene. However, in contrast to the situation in melanocytes, prostate OIS is dependent on p53, and co-deletion of PTEN and p53 cancelled senescence, promoting full-blown prostate cancer. Parallel studies using mouse models to dissect the role of the Ras oncogene in the lung and pancreas3 and in lymphoid cells4 reinforced similar principles. So, although previous work has established that the role of p53 as a tumour suppressor depends on its ability to mediate apoptosis, these papers emphasize that p53 can also mediate senescence in primary tumours.
Collado et al. (page 642)3 address a crucial need for better in vivo markers of OIS. So far, the gold standard has been the detection of an enzymatic activity associated with senescence (that of SA-β-gal)7. Although SA-β-gal has been used successfully to analyse human and mouse samples, this marker is not molecularly well-defined and demonstrates background activity in certain organs. Collado et al. employed an ingenious microarray screen to identify a small set of genes, the expression of which correlates strongly with senescence induced by the ERK protein. (ERK mediates the effects of certain cancer-causing mutations.) The correlations with gene expression are not seen when ERK is induced in the absence of senescence. These markers of OIS include protein-encoding genes and at least three RNA-encoding genes that are relevant to mouse tumour models of different tissues. These markers might predict OIS in precancerous abnormalities in humans.
Braig et al. (page 660)4 provide a penetrating biochemical view of senescence. Their experiments were guided by the observation of unusual foci of tightly packed DNA in senescent cells8. These foci possessed features of a form of silenced DNA called heterochromatin, and were characterized by specific methylation of some of the histone proteins around which the DNA is packed. The Suv39h1 protein is known to methylate histones, and binds physically to the RB tumour suppressor9,10. So Suv39h1 seemed a reasonable candidate for a mediator of the senescence-promoting functions of RB.
Accordingly, Suv39h1 activity was shown to be required for Ras-induced OIS in lymphocyte cells. And in Suv39h1-deficient lymphomas, RB was not able to promote senescence as it normally does. These data suggest that Suv39h1 functions in concert with RB to alter DNA packaging in a manner that is required for senescence.
The findings of Braig et al. emphasize that OIS is an active process requiring specific molecular events that are sometimes perturbed, with malignant consequences. The authors reasoned that OIS would be inhibited by blocking histone deacetylases (HDACs) and DNA methyltransferases (DNMTs) because the activities of these enzymes seem to potentiate the ability of Suv39h1 to induce the formation of heterochromatin. Indeed, they found that lymphoma progression in a mouse model is accelerated by treatment with inhibitors of HDACs and DNMTs.
These findings are of added significance given the emerging application of ‘epigenetic’ therapies — targeted, for instance, at HDACs and DNMTs — in cancer and chronic diseases such as sickle-cell anaemia. The work raises the spectre that such therapies would provoke adverse consequences if they inhibited the senescence-mediated suppression of would-be cancer cells. Nonetheless, DNMT inhibitors have shown remarkable clinical activity in the bone-marrow disease myelodysplasia, and HDAC inhibitors have exhibited promising anticancer profiles in early trials.
Several issues regarding senescence remain to be addressed. Does senescence really mean life imprisonment for a precancerous cell, or is parole possible — that is, can senescence be reversed under some conditions? For example, could senescence in benign nevi be quelled, leading to melanoma, by further exposure to sunlight? The preliminary data here are mixed: some forms of senescence, such as that mediated by p53 in mouse cells11,12, can be reversed by inactivation of p53 and/or RB, whereas other types, for example that mediated by p16INK4a in human cells12, cannot. Moreover, what eventually happens to senescent cells in vivo? Do they accumulate with age or are they culled through some unknown mechanism? The fact that certain markers of senescence mount up with age suggests that senescent cells do accumulate to some extent.
Perhaps most importantly, these reports provide additional evidence that senescence is molecularly heterogeneous, requiring different pathways in different cell types, and in response to different oncogenic insults. A more precise molecular understanding of OIS is clearly needed. Just as apoptotic mechanisms are being used for therapeutic benefit, so a greater understanding of the stimuli that induce and enforce OIS will allow oncologists to exploit this crucial tumour-suppressor mechanism in cancer prevention and treatment.
Michaloglou, C. et al. Nature 436, 720–724 (2005).
Chen, Z. et al. Nature 436, 725–730 (2005).
Collado, M. et al. Nature 436, 642 (2005).
Braig, M. et al. Nature 436, 660–665 (2005).
Serrano, M., Lin, A. W., McCurrach, M. E., Beach, D. & Lowe, S. W. Cell 88, 593–602 (1997).
Sherr, C. J. & DePinho, R. A. Cell 102, 407–410 (2000).
Dimri, G. P. et al. Proc. Natl Acad. Sci. USA 92, 9363–9367 (1995).
Narita, M. et al. Cell 113, 703–716 (2003).
Nielsen, S. J. et al. Nature 412, 561–565 (2001).
Vandel, L. et al. Mol. Cell. Biol. 21, 6484–6494 (2001).
Sage, J., Miller, A. L., Perez-Mancera, P. A., Wysocki, J. M. & Jacks, T. Nature 424, 223–228 (2003).
Beausejour, C. M. et al. EMBO J. 22, 4212–4222 (2003).
About this article
Journal of Nephrology (2020)
The American Journal of Pathology (2019)
The evolution into personalized therapies in pancreatic ductal adenocarcinoma: challenges and opportunities
Expert Review of Anticancer Therapy (2018)
There are only four basic modes of cell death, although there are many ad-hoc variants adapted to different situations
Cell & Bioscience (2018)
JCI Insight (2018)