Cellular Senescence as a Cause of Aging: from Wishful Thinking to Case Closed

Today's open access review covers what is known of cellular senescence as a cause of aging, and is a very readable example of the type. It is always pleasant to find a well-written paper that can serve as an introduction for people outside the scientific community, those with an interest in the topic but only a little knowledge of the relevant biology. If you have friends who fit that description, and who are not all that familiar with the science, then you might send this over as a more gentle introduction than some of the other reviews of cellular senescence published in recent years. In particular, you might point out the middle section from which I borrowed the title for this post.

Senescent cells are those that have entered an altered state in which replication is shut down, and a range of signals and other molecules are secreted. These provoke inflammation, attract immune cells, remodel the nearby extracellular matrix, and increase the likelihood of nearby cells also becoming senescent. Senescence has several forms, occurring in response to cell damage, a toxic environment, radiation, or in the vast majority of cases as the end state of a somatic cell that has reached the Hayflick limit on cell divisions. Most species have a two-tier hierarchy of cells: a small number of stem cells that can replicate indefinitely, and the limited somatic cells that make up the vast majority of any tissue. Stem cells produce a supply of somatic cells to make up those lost to the Hayflick limit. In such a system something like senescence has to exist if somatic cells are in fact to be limited in the number of times they can replicate. Why does this two-tier system exist? Probably because it is the most accessible way to suppress the risk of cancer sufficiently well for higher animal life to evolve at all: if any more of our cells were normally capable of unlimited replication, and thus easily subverted by cancerous mutations, then our lineage could not survive over evolutionary time.

Beyond the necessity of being a full stop at the end of a somatic cell's life span, cellular senescence appears to have evolved other uses along the way. Reuse is very common in biology. Thus cellular senescence acts to set limits to growth in embryonic development, coordinates with the immune system in wound healing, and acts to suppress cancer, at least when the number of senescent cells is still low, by shutting down replication in cells that are most at risk of becoming cancerous. Senescent cells so far appear to be best as short-lived entities that self-destruct or are destroyed by the immune system quite quickly after they appear. The contribution of senescent cells to aging is produced by the tiny minority of such cells that somehow linger instead. The signals they secrete, used in the short term to carry out their evolved tasks, become destructive when issued over the long-term, and in ever increasing volume. The solution to this problem is likely very simple: destroy these cells, clearing up the remnant population that natural processes fail to eliminate. Doing so will reverse this one portion of the aging process.

Senescence in the aging process

From its initial discovery, it was postulated that senescence, on some level, was linked with organismal ageing. Modern forms of this hypothesis propose that senescent cells are produced gradually throughout life. These then begin to accumulate in mitotic tissues and act as causal agents of the ageing process through the disruption of tissue function. This conceptual model carries three underlying assumptions: firstly that senescent cells are present in vivo, secondly that they accumulate with age, and finally that an accumulation of senescent cells can have a negative impact. Each is worthy of examination.

A steady production of senescent cells is quite plausible if the kinetics whereby populations of normal cells become senescent in vitro are assumed to be similar in vivo. Many early reports evaluated findings with the mistaken underlying assumption that cell cultures become senescent because all of the cells divide synchronously for a fixed number of times and then stop. In fact, it has been known since the early 1970s that each time a cell goes through the cell cycle (i) it has a finite chance of entering the senescent state and (ii) this chance increases with each subsequent division. Thus, senescent cells appear early on if a cell population is required to divide. Indirect demonstrations that senescent cells occur in vivo, accumulate with ageing, and do so at reduced rates in organisms where ageing is slowed (for example, by dietary restriction) were occasionally published from the 1970s onwards. However, they were technically difficult to perform and correspondingly hard to interpret.

Senescence triggers changes in gene expression. A central component of this shift is the secretion of biologically active proteins (for example, growth factors, proteases, and cytokines) that have potent autocrine and paracrine activities, a process termed the senescence-associated secretory phenotype (SASP). This results in cells that overproduce a wide variety of pro-inflammatory cytokines, typically through the induction of nuclear factor kappa B (NF-κB) and matrix-degrading proteins such as collagenase. Other radical phenotypic changes, such as calcification, have also been shown to occur in some cell types with the onset of replicative senescence. The individual components of the SASP vary from tissue to tissue and, within a given cell type, can differ depending upon the stimulus used to induce senescence (for example, in fibroblasts rendered senescent by oncogene activation compared with telomere attrition or mitochondrial dysfunction). Such studies demonstrated that senescent cells could, at least potentially, produce significant and diverse degenerative pathology.

However, the observation that something can produce pathology does not mean that it must produce pathology, and a historic weakness of the cell senescence literature was that the in vivo studies essential to testing the causal relationship between ageing and cellular senescence (induced by any mechanism) were lacking. However, the production of transgenic mouse models in which it was possible to eliminate senescent cells has finally made such tests experimentally feasible. Initial studies demonstrated first that senescent cells appeared to play a causal role in a variety of age-associated pathologies in the BubR1 mutant mouse and subsequently that either life-long removal of senescent cells or their clearance late in life significantly attenuated the development of such pathologies in these progeroid animals. This clearly demonstrated that senescent cells can have significant, deleterious effects in vivo. Interestingly, the removal of senescent cells in this system was not associated with increased lifespan (an observation that demonstrated that it is possible to achieve classic 'compression of morbidity' by deleting senescent cells). However, on more conventional genetic backgrounds, attenuated age-related organ deterioration was accompanied by increases in lifespan of the order of 25%.

A justifiable claim can be made to consider these studies 'landmarks' in the field in that (i) they demonstrate a causal relationship between senescent cells and 'ageing' and (ii) the same mechanism can cause changes associated with 'ageing' as well as those associated with 'age-related disease'. These results have unusually profound philosophical implications for a scientific paper and challenge a fundamental ontological distinction that has been drawn for almost two thousand years between 'natural' ageing and 'unnatural' disease.

Whether an enhanced emphasis on basic human studies is a useful parallel-track approach to the pioneering work now taking place in rodent models or an essential next step is a matter of perspective. Many fundamental mechanisms of ageing are conserved between species, but there are often important species-specific differences. Those inclined to stress the cross-species similarities will be inclined to deprioritise human studies and vice versa. Regardless of the species of origin, the extent of variation in the phenotype of senescent cells derived from the same tissue in different individuals is not well characterised. It would be surprising if important intra-individual variation did not exist within the general population as well as 'outliers' (for example, centenarians and those with accelerated ageing diseases such as Werner's syndrome). Although data on differential SASP profiles in response to a senescence stimulus are beginning to enter the literature, they remain fragmentary concerning the senescent cell phenotype in different tissues.

Despite these gaps, progress is being made towards the development of 'senolytic' drugs that can destroy senescent cells - with the goal of duplicating the effects of the transgenic mouse models first in normal animals and eventually in human patients. Initial results seem promising.