Senescence: updated evidence

Summary

1. What is senescence?
2. Why is senescence harmful?
3. Types of senescence and the link with telomeres
4. A brief history of senescence research: seminal papers
5. Pathways leading to senescence and markers of senescence
6. Senescence controversies

7. Does senescence contribute to aging? (indirect data)
8. Does senescence increase with aging?
9. Human data suggesting that senescence contributes to aging
10. Comparative data across species
11. Comparative data within species
12. Comparative data inlong-lived mouse models
13. Age-related diseases and senescence
14. Misc data – the SASP, inflammation, plausibility and indirect evidence for causality

15. Intervening (direct data)
16. Types of interventions
17. A list of senolytic interventions
18. A word of caution about mouse healthspan studies
19. Key studies: improved healthspan
20. Key studies: improved lifespan
21. Human data and ongoing trials

22. Models of accelerated senescence
23. Disease associated with high senescent cell burden
24. Senescent cell transplantation


25. Dose response studies

26. Summary, outlook, open questions

Housekeeping note:
hopefully most of the formatting errors are fixed now. I think most of them were introduced because blogspot (just like office, gmail, etc) insist to copy fully formatted text instead of supporting plain text paste per default. Once you accidentally paste formatted text it becomes hard to remove the formatting via the editor interface.

Summary
The holy grail of aging research, if you will, is a one time-treatment that can rejuvenate an aging organism. No other kind of treatment could be realistically and efficiently applied to humans. Compared to mice humans are giant, lazy slobs that don't take their medication. That is the huge problem. Unfortunately, most interventions that can extend lifespan in mice do so most efficiently when started from birth (e.g. growth hormone dwarfism). There are few interventions that lead to rejuvenation of older animals (e.g. rapamycin). Even fewer interventions allow true rejuvenation with just a short treatment or intermittent treatments. The removal of senescent cells is shaping up to be the latter which is why it is worth critically discussing the idea.

Here I would like to briefly review the evidence for the senescence theory of aging using Koch- or Braford-Hill-like criteria. I've been reading the literature through the lens of Koch & Hill for a while because there is no other way to make sense of it. We're lucky because others have already reviewed this topic from a similar angle (Yanai & Fraifeld 2018) and I just need to provide an update and, when appropriate, criticism of their approach. For example, I would like to point out that like many others Yanai & Fraifeld put too much faith in models of age-related disease and models of accelerated aging. In contrast they barely consider the concept of temporality (4) and of biological gradient (5).

Braford-Hill suggested to evaluate a correlation (or in this case hypothesis) by considering several factors like (1) strength of association, (2) consistency, (3) specificity, (4) temporality, (5) biological gradient, (6) plausibility, (7) coherence, (8) experiment, and (9) analogy.

Senescence increases with aging and this is supported by "longitudinal" data (moderate to strong evidence). This is true in many different studies and in various species. Nevertheless the data remains heterogeneous and plagued by biases. In reference to Bradford-Hill we can note that the data is generally strong (1) and more often than not consistent (2).

Slow-aging mice have reduced levels of senescence (moderate to strong evidence). While this is generally true, studies are extremely heterogeneous. Summarizing the data and going back to Bradford-Hill we can also see a biological gradient (5), for example, when an increasing degree of caloric restriction (CR) leads to an increasing suppression of senescence associated inflammation.
Fast-aging or so called progeroid mice have increased levels of senescence and removal of senescent cells improves the aging-like phenotype (strong evidence - but weak model).

All the above models have one thing in common. They fail to establish temporality (4) because - outside of experimental studies with their own shortcomings - we do not know if senescence precedes or follows aging and manipulations of lifespan. This is still a critical gap in the literature.

Senescence is associated with age-related diseases, especially CVD and metabolic disease (moderate evidence - but weak model). When we slow aging we do expect a slowing of most diseases, so per definition most anti-aging treatment would violate the concept of specificity (3). Even considering this limitation the senescence data is all over the place. The causes of senescence are too many and not well defined; the correlations between senescence and health outcomes are too numerous; and senescence is too dynamic and modulated by too many treatments; there are too many conflicting markers of senescence. Nothing is "specific" in any way, shape or form. This could be also described as a lack of coherence (7) with many outstanding questions and contradictions, although, none strong enough to sink the hypothesis.

Comparative studies across species have not studied senescence using modern-day markers. Indirect evidence supports the senescence theory (overall weak evidence). In some sense this can be described as using analogy (9) and extrapolation to make claims, as proposed by Bradford-Hill.

Senolytic therapy extends lifespan and healthspan in mice (moderate to strong evidence, good model). However, the improvement in lifespan is often small. Similarly, perhaps a bit more consistently, healthspan is improved, but this does not apply to each and every aging pathology, or even to the majority of pathologies.



What is senescence?
Recent studies have found that the removal of so called senescent cells can extend the lifespan and healthspan of mice.

Understandably the concept of senescence has generated a lot of (well-deserved) hype and a bit of confusion. Many people may be confused by what exactly senescence is? It has to be distinguished from aging per se, also sometimes called senescence. It is also not to be confused with so called T-cell senescence, any other type of immunosenescence, or even erythrocyte senescence. Although related, senescence studied in tissues is not the same as proliferative senescence in cell culture.

Senescence is an irreversible non-proliferative state different from quiescence, more relevant to dividing cells. In reality it is a bit more complex and we do see senescence of non-dividing, postmitotic cells (Jurk et al. 2012) although often less pronounced. The basic idea is simple. Cells that have suffered DNA damage that fails to be repaired, have to choose between continued proliferation, senescence or apoptosis. If cells with unrepaired damage do not proliferate, this may protect from cancer without jeopardizing tissue integrity as excessive cell death would. Over time this mechanism, however, leads to the accumulation of many dysfunctional cells that also secrete inflammatory proteins (i.e. the senescence associated secretory phenotype; SASP). This is consistent with evolutionary theories of aging and antagonistic pleiotropy, the idea that some pathways are beneficial until middle age while being harmful for older organisms.

Senescence is considered anti-tumorigenic and senescent cells are involved in wound healing and possibly other physiologic processes like insulin secretion (Helman et al. 2016). Therefore the removal of theses cells poses a risk. Despite great advances the field has been moving very slowly towards human studies. The first proper human studies were only published in 2019 (Hickson et al. 2019). This raises the question if these delays are justified due to the inherent risks? It remains my opinion that phase I studies in humans should have gone ahead much sooner. Why are we allowing people to risk their lives by working dangerous jobs between gold mines and war zones, but we don't allow them to risk their lives to advance science? We would easily find thousands of volunteers that would willingly take the risk to test senolytics.

Why is senescence harmful?
Senescent cells are rare. Therefore, before the senescence theory of aging could generate the current success, we needed to develop some plausible framework explaining how a few cells can impair tissue functioning. While we have some good theories now, none of these are proven, although it might be fair to say they moved from the hypothesis stage to the theory stage. The strongest one is the SASP theory.

  • without understanding the mechanisms in detail we know that the secreted inflammatory cytokines (the SASP) are associated with adverse health outcomes, especially cancer and CVD. This data will be discussed under "Misc data – the SASP..."
  • "senescence-stem lock model" in which the chronic secretion of pro-inflammatory factors by these senescent cells keeps neighboring cells in a permanent stem-like state and thereby prevents proper tissue renewal.
  • inflammatory or other cytokines promote fibrosis of nearby cells (cave: sometimes the opposite depending on the tissue)
  • inflammatory or other cytokines promote invasion of immune cells which, at least for atherosclerosis, contributes to disease pathology
  • inflammatory or other cytokines, especially IL-6 and IL-1β, can promote insulin resistance in adjacent fat cells
  • Asking: why are we assuming the conclusion that senescent cells are rare if our markers are trash and we haven't thoroughly screened all human tissue types? Maybe there are tissues that do show high enough levels of senescence to matter through bulk effects, look e.g. at human skin in Idda et al. (2020).
Types of senescence and the link with telomeres
The types of senescence can be divided by the cause or trigger for senescence. There is for example proliferative, oncogene-,  ROS-, DNA damage- or chemotherapy-induced senescence. We also know that senescent cells can induce senescence in neighboring cells ("paracrine senescence", da Silva et al 2019). The list of senescence triggers is endless and almost too broad, making you question if senescence is particularly specific to aging. For example there is also mitochondrial dysfunction induced senescence (Wiley et al. 2016; MiDAS) or senesence induced by epigenetic poisons.

Proliferative senescence deserves its own discussion because this idea is so misunderstood. Cells that grow in a petri dish will eventually stop dividing, either due to telomere attrition or due to other reasons. This is associated with morphologic and molecular changes and is called replicative senescence. This type of senescence was discovered by Leonard Hayflick in the 1960s and it has very little relevance for senescence in living organisms. While it could be one cause of in vivo senescence, presently we think that most senesence is caused by DNA damage (or other events). In particular, it is thought that cells in vivo do not divide often enough to show telomere attrition, but length-independent telomere damage has also been described as a cause of senescence (Anderson et al. 2019).

Therefore it is difficult to interpret telomere and ex vivo studies as support for the senescence theory of aging (i.e. the idea that senescence is important to many aspects of aging). So with a grain of salt I will note here that telomere attrition has been linked to bird and species maximum lifespan (Tricola et al. 2018, Dantzer and Fletcher 2015). Studies of ex vivo senescence will be discussed later on (in the history and comparative data chapters).




A brief history of senescence research: seminal papers
In the early to mid 1960s Hayflick discoverd replicative senescence and showed that human embryonic fibroblasts found in the lung have a higher proliferative potential than those from adults. The idea that cells show senescence over time was born. Several ostensibly rigorous studies supported the corollary that cells isolated from older organisms would senesce faster. Much of that early work is reviewed by Stanulis-Praeger (1987) and Martin (1977). However, later on the data was contradicted by Cristofalo et al. (1998), which may be the reason why people shied away from doing any more comparative studies of ex vivo senescence. Briefly put, the authors think that prior studies were biased by the inclusion of fetal cells and cadaver derived cells.

Dimri et al., 1995: Provided the first strong evidence that senescent cells accumulate in human skin (beta-galactosidase staining). Understandably, the concept was still quite controversial (Severino et al. 2000; Cristofalo et al. 2004; de Magalhaes 2004; Lim 2006) and in fact it took quite long for the data to be replicated. There was Zindy et al. (1997) who found elevated p16 expression in mouse tissues, Nielsen et al. (1999) using p16 as a marker in human tissues and Pendergrass et al. (1999) and Mishima et al. (1999) using beta-gal staining in rhesus macaques while at the same time Severino et al. (2000) stated they couldn't replicate the Dimri data.

The big breakthrough in the 90s was the use of in situ assays that allowed us to circumvent the need for cell isolation and growth in culture, which is very stressful and unphysiologic. It also allowed us to ignore the ugly finding by Cristofalo et al. Then a flurry of mostly supportive studies followed that is reviewed by Yanai & Fraifeld (2018) and others.

Of note, Sedelnikova et al. 2004 are perhaps the first to use markers of unrepaired DNA damage (γ-H2AX foci) as a correlate of in situ senescence showing that these increase in 5 tissues of aging mice.

At the same time the van Deursen group (Baker et al. 2004) showed that senescent, often p16- and beta-gal positive, cells accumulate in a progeroid mouse model (BubR1 hypomorphic mice). Later they showed that crossing the progeroid mouse with a strain that does not express a key driver of senescence, i.e. p16Ink4a, extended lifespan (Baker et al. 2008). Since these mice are born without p16, it was impossible to exclude an interaction between the genotype and mouse development and, more importantly, this study design did not suggest any path to clinical development. Therefore it proved essential to show that induced ablation of p16-positive cells had the same effect. This the authors achieved using the INK-ATTAC system that allows to express an inducible caspase from the p16 promoter (the famous "suicide gene" approach; Baker et al 2011). Now the remaining question was whether we could kill senescent, p16-positive cells using a drug?

The long break between Baker et al. (2011) and follow-up studies is a bit surprising and I have no explanation. While it always takes a long time to do follow-up research, some of the blame also goes to society because we do not value research enough. Everything would be faster if we had human pilot trials on a voluntary basis and more research groups working in parallel.

At the same time as the van Deursen work was being fleshed out, the Campisi group solved another key problem. For a long time senescence skeptics had argued that these cells are too rare to be harmful and it was perhaps Coppé et al. (2008) from her group who first suggested that senescent cells can secrete harmful cytokines which are detrimental to tissue functioning.

The prize for the study that first measured health benefits after senescent cell ablation goes to Xu et al. 2015, because they provided reasonable evidence that the INK-ATTAC system blunts age-related lipodystrophy in situ. To some extent, Sousa-Victor et al. 2014 also qualifies as they used  adenoviral shRNA to silence p16INK4a in muscle of aging mice and found improved satellite cell activation afterwards. However, their system feels kind of contrived to me.

The year 2015 is very productive indeed. Following the above studies, two canonical treatment concepts are discovered that year, targeting the secretory phenotype (SASP) and senolytics. Zhu et al. (2015) publish the famous dasatinib/quercetin paper in Aging Cell detailing the senolytic effects of this combination and Xu et al. (2015) provide data on the jak1/2 inhibitor ruxolitinib to mitigate the SASP which was published in Elife and PNAS. (I just noticed that both of these approaches are from the Kirkland group.)

Soon thereafter the van Deursen group pushes back vying for the title of best translational paper, with Baker et al. (2016) providing data that INK-ATTAC mediated p16Ink4a+ cell ablation extends lifespan. It takes the Kirkland group a couple more years to show the same using their much more attractive drug induced model. Finally, however, Xu et al. (2018) confirms that dasatinib/quercetin (D+Q) extend mouse lifespan. On to human studies?

Well, yes, albeit slowly. Three important studies in patients were published in 2019 with Hickson et al. (2019) key among them, showing that dasatinib/quercetin (D+Q) reduce senescent cells in skin and adipose tissue and SASP in plasma of patients with diabetic kidney disease. Again we can see that the time lag is very large. Perhaps studies could have been started earlier if we had the right regulatory environment.

Pathways leading to senescence and markers of senescence
From the introduction you would think we have ironed out the definition of senescence, but don't be mistaken. We have a rough idea that senescence is associated with things like growth arrest, p16, p21 expression and SASP proteins. However, whether these are the key pathways is still unclear. Further complicating the picture, often there is a discrepancy between in vivo and in vitro markers, either due to absence of evidence (not studied) or due to evidence of absence (studied and failed to replicate between both models). We know, for example, quite a bit about morphological senescence-associated changes in vitro but I don't recall many studies that tried to measure this in vivo. For further reading on this topic, I would recommend two reviews (Sharpless & Sherr 2015; Hernandez-Segura et al. 2018).

A meta-analysis in 2017 revealed (Hernandez-Segura 2017):
“The selected datasets covered 6 different fibroblast strains (BJ, IMR90, HFF, MRC5, WI38, and HCA-2) and 3 different senescence-inducing stimuli (replicative senescence [RS], oncogene-induced senescence [OIS], and ionizing radiation-induced senescence [IRIS]) and were generated by 5 independent laboratories”
Remarkably, none of the classical senescence markers (CDKN2A, CDKN1A, LMNB1, and members of the SASP) were among [our] hits, since either they were not differentially expressed in all the cell types, or they were shared with quiescence”
From what I can tell this continues to be true on the transcriptomic level (Casella et al. 2019):
“these findings suggest that [our] 68 transcripts may serve as superior RNA biomarkers of senescence rather than the classical p16 and p21 mRNAs”

Similarly, recent data on the secretory phenotype measured by a proteomic approach in three different in vitro models (Basisty et al. 2020) suggests unprecedented heterogeneity. As expected the data was also different between fibroblast and epithelial models. Keeping that in mind we can briefly discuss the pathways leading to senescence as long as we understand the data to be preliminary.

DNA damage response
After a lesion is identified a decisions is made between repair, senescence and apoptosis. A pathway going from DNA damage to ATM, H2AX histone phosphorylation (phosphorylated at Ser139 it is also called γH2AX), activated p53 and p21 expression is well described (d'Adda di Fagagna et al. 2008). As far as mechanisms are concerned, it is possible that transient activation of p53 promotes expression from promoters with high p53 affinity (e.g. p21) and only prolonged activation promotes expression from low affinity promoters found in pro-apoptotic genes (Roos et al. 2016).

53BP1 is another one of the vast number of proteins recruited to DNA double strand breaks that aids in repair. 53BP1 serves as an adaptor for other repair proteins and promotes NHEJ repair (non homologous end joining). H2AX phosphorylation is an early step for both NHEJ and HR. Both of these serve as markers for (unrepaired?) lesions and we could probably use many other DNA repair proteins as markers except they're technically difficult to detect.

Unsurprisingly several proteins and pathways of the DNA damage response are used as a marker of senescence. For example, gamma-H2AX and 53BP1 foci are used as senescnece marker and increase in the skin of aging baboons (Herbig et al. 2006). When yH2AX colocalizes with telomeres it is called a telomere induced focus (TIF).

p53 pathway and FOXO4
The p53 protein is activated at least during the early stages of the DNA damage response. During senescence p53 is thought to be sequestered in the nucleus by Foxo4 within special chromatin regions called DNA-SCARS (Baar et al. 2017) probably limiting apoptosis.

Immune surveillance
The idea that senescent cells are cleared by the immune system makes biologic sense because many of the secreted SASP components are able to attract immune cells. Although, clearance of senescent cells is well described during tumorigenesis or tissue damage via NK-cells, T-cells or macrophages, we know very little about surveillance during normal aging. It does seem to occur, though. Ovadya et al. (2018) showed that Prf1-/- mice which “which lack functional cytotoxic (T, NK or NKT) lymphocytes” accumulate senescent cells.

The accompanying editorial asks the right questions: “Would it be possible to use immuno-oncological drugs such as immune checkpoint blockers or other non-specific immunostimulants for boosting anti-senescence immunosurveillance [without killing the patient]?”

Lysosomal changes
It seems that senescent cells are a bit sluggish, perhaps with reduced overall protein turnover. This leads to an accumulation of lipofuscin which can be measured by a biotinylated Sudan Black B (SBB) analog (GL13) - among other assays (Georgakopoulou et al. 2013).
The famous beta-galactosidase assay also works in the same vein. Basically it measures the lysosmal (enzyme) content, because beta-galactosidase is a lysosomal enzyme (GLB1) and lysosomal enzymes are usually co-expressed and regulated together. Activity of this enzyme is measured using X-Gal as an artificial substrate and at pH 6, both rather unphysiologic. See Muñoz-Espín et al. (2018) for some creative and more physiologic ways of imaging beta-galactosidase overexpressing cells.
What this means for the activities of other lysosomal hydrolases is not clear although we do have limited in vitro evidence suggesting upregulation of other enzymes as well (Knaś et al. 2012). 
Since senescent cells accumulate undigested material in their lysosomes it is possible that the autophagic-lysosomal system is compromised. Either way, there is some evidence of general expansion of the lysosomal compartment as measured by lysotracker and LAMP1 (as per Hernandez-Segura et al. 2018).



This figure from Paradis et al. 2001 shows a typical study measuring beta-gal in humans tissues. The other methods are rarely used to study the longitudinal accumulation of senescent cells. In this study the authors show that some people do not have any hepatic senescence even at old age (RS negative group). However, there is a tendency for people with abundant senescent cells in the liver to be older (RS positive group).

CDK inhibitors
p21 is considered to be an early phase senescence marker that is transactivated by p53 (that knowledge goes back to 1990s! Dulić et al. 1994 among others).

p16(Ink4a) is a key player in senescence and one of the most common markers. This protein inhibits the kinase activity of Cyclin D dependent kinases (CDK4 and CDK6) and the progression through S phase. Normally prior to S phase entry these CDKs phosphorylate and inhibit pRB. Whereas during senescence pRB will be active and recruit chromatin modifiers to silence further expression of cell cycle proteins (review: Tonnessen-Murray et al. 2017).

I never quite figured out how p16 is regulated and we probably don't know much, but epigenetic regulation seems to be key here (Hernandez-Segura et al. 2018). It is also activated by the Ras-MAPK pathway during oncogene induced senescence and many other activators existThe p16 protein might be involved in long term maintenance of senescence and form a negative feedback loop with p53 (Mirzayans et al. 2012). This makes sense because while p53 is important to trigger senescence through the DDR, eventually it has to be inhibited to prevent cell death.

The p16 protein is encoded by the CDKN2A locus which also encodes other proteins, Arf and Ink4b. Often the primer design is a problem for p16 because these other proteins have a similar sequence. Antibodies for this protein are (used to be?) poor yet people publish western blot based studies on it. One workaround using transgenic mice is to introduce a marker protein under the p16 protein that you can detect more easily. Furthermore, p16 is dispensable for senescence. Nevertheless it is the most common marker used and we know that it increases with aging in several human tissues (Tuttle et al. 2019).

Nuclear changes and chromatin remodelling
We've already discussed DNA damage associated foci so let's focus on other nuclear changes. Senescence-associated heterochromatin foci (SAHF) are DAPI-dense spots seen in the microscope and often associated with senescence in human cells, although, not mouse cells. As all known markers, they are dispensable and variable. Crucially, they're not seen in tissue which is essential for a good marker (Kosar et al. 2011).

I don't want to talk too much about chromatin remodelling because I am no expert on this and I simply don't think we understand how this contributes to senescence. The review by Hernandez-Segura et al. (2018) has a bit more information on this (so does Parry & Narita 2016). Let me just give on example. Presumably the senescent state has to be stabilized in the long term by chromatin remodeling. Cell cycle genes are thought to localize to the SAHFs where they are silenced. Furthermore, a decrease of HMGB2 and HMGB1 is observed with senescence and this appears to have rather complicated effects. In fact, reduced HMGB2 lead to the redistribution of SASP loci to SAHFs and their silencing suggesting that perhaps senescent cells strive to reduce SASP secretion (Guerrero and Gil 2016). It is possible that HMGB1 ends up being secreted (Basisty et al. 2020) acting as a alarmin that recruits immune cells, which is good for a while, but eventually may promote pathologic tissue remodeling.

Another phenomenon that is observed is called senescence-associated distention of satellites (SADS) which may lead to aberrant expression of so called satellite DNA that is normally silenced. It is thought that some of these satellites, called LINE-1 retrotransposons may activate the cGAS/STING pathway and the SASP because they lead to the production of cDNA in the cytosol via their reverse transcriptase activity (Loo et al. 2020, De Cecco et al. 2019). This fits with our views on aging because we have quite a bit of evidence for the reactivation of once silent retrotransposons as animals get older (De Cecco et al. 2013).

Yet another nuclear change is called DNA-SCARS, i.e. DNA segments with chromatin alterations reinforcing senescence. These structures contain H2AX, p53 and CHK2. They seem to be a form of persistent 53BP1 foci that normally promote DNA repair in the early stages of DNA damage. One key difference is that DNA-SCARS are longer lived and partially co-localize with the PML protein in nuclear bodies and serve to stabilize senescence independently of DNA damage. Importantly, they are observed in vivo in irradiated mice (Rodier et al. 2010). While regular markers of DNA damage repair are known to accumulate with aging (H2AX+53BP1, Galbiati et al. 2017), I cannot find any paper looking at DNA-SCARS during aging specifically.

There is also evidence for blebbing, distortion and leakiness of the nuclear envelope. The downregulation of lamin B1 that is observed in vitro, in mice and in human skin is thought to contribute to this (Freund et al. 2012, Dreesen et al. 2013). Cytoplasmic chromatin fragments (CCFs) are observed in senescent cells (Ivanov et al. 2013). The authors also observed downregulation of lamin B1 and histones in vivo in human nevi that are thought to contain abundant senescent cells. This is quite intriguing because distortion of the nucleus and nuclear lamina is also a prominent problem seen in Hutchinson-Gilford-Progeria. There are also interesting parallels between CCFs and micronuclei, the latter is a typical marker of genomic instability.

Finally, the cGAS-STING pathway acts as a sensor for this cytoplasmic DNA that then triggers inflammation (Dou et al. 2017) and at least in mice this was confirmed by STING knock-out. While I can't quite find evidence of CCFs in tissue, this pathway nonetheless seems rather promising with quite a bit of in vivo support.

Senescence controversies
Several authors have suggested that non-specific expression of so called senescence markers hampers interpretation of the data (Childs et al. 2016, Bennett and Clarke 2016, Hernandez-Segura et al. 2018). While this isn't wrong, it's also not that simple. In the meantime we have a lot of different senolytics that target different pathways and most of them are beneficial for age-related diseases, even though the studies choose to use very heterogeneous health outcomes. Another counter argument is to just say, who cares? Maybe senolytics also kill activated macrophages, or maybe they do completely different, but either way this ends up being beneficial and makes biologic sense in the context of inflamm-aging theories. From a clinical point of view it doesn't matter how a drug works.

I'd like to give an example for the conflicting evidence about selectivity. In the study of the FOXO4-DRI peptide (Baar et al. 2017) the authors found that dasatinib+quercetin (D+Q) is not selective for the killing of senescent IMR90 cells. This could mean two things, one, that D+Q is also promiscuous in vivo and we can't take the data seriously, or equally likely, that the IMR90 model isn't faithfully replicating in vivo senescence. Either way this is problematic, but again we come back to the idea of multiple studies with different treatments arriving at the same conclusion.

Something that worries me personally is that senescent cells are a bit too "dynamic" (Rossman et al. 2017; Ogrodnik et al. 2017, also this paper), being affected by many treatments, not just de facto senolytics, in very short time frames. Intuitively, reversing age-related damage shouldn't be that easy and reversal should be harder to achieve than prevention. For example, we know that CR has smaller benefits when it is initiated in old animals. In fact, the benefits of CR on lifespan are cumulative but the benefits on senescent cells are acute (Ogrodnik et al. 2017), because 3 months and lifelong CR have the same effects on senescent cell burden. This suggests that senescent cells are not the key pathology that is slowed by CR and not a contributor to lifespan. The review by Ogrodnik et al. (2019) makes more or less the same point and is worth a read. Just to give one other - very limited! - example, this time in humans. Intuitively, senescence should not depend strongly on general health because the rate of aging also doesn't, yet observational studies do suggest such a dependence (Rossman et al. 2017). The situation in mice may be more consistent with the working hypothesis, as p16 is induced quite specifically by genotoxic stress rather than a high fat diet (Sorrentino et al. 2014).

Another interesting issue is that most papers claim to ablate p16 positive cells and these senescent cells are known to accumulate with aging. Given this, we would expect p16 overexpression to be harmful and loss of p16 to be beneficial. Neither is true (loss: Serrano et al. 1996, gain: Matheu et al. 2009; McHugh and Gil 2018). The reason for this paradox is presumably the tumor suppressive function of p16. You want p16 to be active in order to halt early stage cancers, but at the same time you'd prefer not to accumulation too many p16 positive cells.

Does senescence contribute to aging?
Here we discuss studies that aim to establish causality without the use of interventions, this includes observational studies, uncontrolled or open label trials, comparative biogerontology and genetics. These studies are the cornerstone of biogerontology because interventional studies, even in mice, are very expensive and time consuming.

As a rule, we would expect senescence to increase with aging and hence old individuals should show more senescence. Age-related diseases should be linked with senescence as well. Whereas healthy people should have less senescence, this may include centenarians or their offspring, vegans, athletes, or people on a particularly healthy diet. Within a healthy individual senescence should correlate with objective and subjective markers of aging and predict long term health.

It is absolutely crucial to test this in humans, even if we later screen mice for therapies, because we must know that we're aiming for the right target!

None of this can fully prove causality, though. For example, mice are short-lived and may not show the same type of aging as humans hence the data from "long-lived" mice might not be as meaningful as we hoped. On the other hand, we can look at exceedingly long-lived species and compare them. This may then include species that age even slower than humans, solving the issue we have with the fast-aging mouse model. However, some mechanisms of aging will only occur in certain related species (private vs public mechanisms). Personally, I think private mechanisms of aging or anti-aging pathways are rather rare (although I just recently stumbled on one possible example, Dziegelewska et a. 2016 for hydrogen sulfide signaling).

In summary, the data shows that senescence is strongly correlated with chronologic age, suggesting it is important to age-related pathology. Some key issues remain, however, like the paucity of longitudinal studies testing whether senescence predicts health outcomes (in any species), publication bias, no true centenarian-offspring studies, no strong cross-species comparative studies, etc.

Does senescence increase with aging?
First of all, as a remainder. These studies estimate the longitudinal change in senescence, across the lifespan, but technically most of them use a cross-sectional design. Longitudinal design would be following an organism(s) at multiple points, while in the cross-sectional study we look at organisms of different ages, but we only have a single data point for each of them.

Senescence increases with aging and this is well documented in aging humans, primates, mice and rats. This data is reviewed in Yanai & Fraifeld (2018). What is more, Tuttle et al. 2019 have performed a meta-analysis of the available human data and highlight several issues that I also noticed with the literature. They found that the literature suffers from publication bias, probably because most studies are small and positive studies are published more often. Nonetheless they find the data to be supportive of the senescence hypothesis. A complementary approach is to look at RNAseq data across the lifespan. In a reanalysis of GTEx data, a large consortium that curates human RNAseq data, Hudgins et al. (2018) found that p16 and p21 were increased quite consistently with aging across nine tissues, but none of the the SASP components were consistently elevated. As a side note, based on this and other findings some people have speculated that the increase of SASP markers in the blood may be driven by secretion from a particular tissue or group of cells (Wiley et al. 2017).


(Figure take from Tuttle et al. 2019)

Interestingly primates show extensive skin senescence (Herbig et al. 2006) while mice barely show any (Yousefzadeh et al. 2020) again highlighting the problems with the mouse models. In another study mice showed aging-related increases in SASP mRNAs but human tissues did not (Hudgins et al. 2018), although, p16 was elevated as already noted. Still taken together most tissues show senescence in most tested species, e.g. as measured by expression of p16 or the beta-galactosidase assay.

At this stage it makes more sense to ask which tissues do NOT show senescence rather than the opposite. Multiple reports suggest muscle doesn’t show senescence  in both humans and primates (Idda 2020, Herbig 2006/2007), although, progenitor cells in muscle may show senescence. Perhaps there is also little senescence in the human lung (Idda et al. 2020).
To give another example, in 
Tuttle et al. (2019) the data is very weak for the gut, blood and thymus (i.e. not significant and low effect size). Whereas they identified a positive study for lung which does not agree with the data by Idda et al. (2020) and is probably due to different methods employed by the studies. Additionally, it is still a bit unclear, at least to me, if immune cells show bona fide cellular senescence or if it they express senescence markers as part of their normal physiology (Biran et al. 2017, Hall et al. 2016).

Hudgins et al. (2018) found 3 out of 9 tissues had non-significant changes in p16 mRNA and this was the thyroid, heart and oddly enough the skin based on RNAseq data of tissue homogenates. So as we can see very different methods may arrive at varying results. It also matters whether we look at tissue homogenate or individual cells.

Human data suggesting that senescence contributes to aging
Fontana et al. (2018) found that graded CR reduced several sensecence markers in the colon of mice, i.e. p16, p21, MMP9 and others measured by qPCR. Additionally, they compared people from the CRON cohort, self-imposed caloric restriction in humans, with a western diet group and found evidence of reduced senescence (n=12 for CRON and WD).

The study by Lattanzi et al. (2014) suggests reduced levels of beta-galactosidase staining in fibroblasts isolated from centenarians as compared to controls. Unfortunately, the study does not include so called centenarian offspring controls and thus may be subject to various forms of survivorship bias (unhealthy individuals per definition rarely reach old age).

The following papers are perhaps a bit more convincing. Waaijer et al. (2016) found that the number of cells that were P16Ink4a positive in the skin correlated with perceived aging, facial wrinkles and elastic fibre morphology. At least two genome wide association studies (GWAS) have implicated the CDKN2A/B locus (encoding among other proteins also p16) in healthspan regulation (Jeck et al. 2012, Fortney et al. 2015).

Although not all that impressive the following ones are still noteworthy. Liu et al. 2009 found that expression of P16Ink4a (n=170) in T-cells was associated with aging, smoking, physical inactivity and plasma IL6. Perhaps a bit stronger, Justice et al. 2018 found that P16Ink4a positive cells in biopsied adipose tissue were associated with physical function of older, overweight women and their post hoc analysis showed that there was less senescence after 5 months of resistance training, albeit, with no control group. However, we have to remember that p16 expression could be a marker of poor health and not aging. Consider what the authors themselves note: up to 30-fold greater senescent cell accumulation is observed in visceral adipose tissue from obese compared with non-obese adults


There are probably a couple more such studies that I am missing, but again, I do not think they are particularly powerful. For example, Sørensen et al. 2016, found that p16ink4a expression in salivary glands is associated with cognitive decline in middle aged men (not longitudinal, n=181). Lawrence et al. 2018 measured two senescence markers, HMGB2 and p16INK4A in plasma and showed that they correlated with age, albeit very weakly. Subjects with a low mobility or mental health score also had higher HMGB2 levels. It is a bit unclear what the cellular source of plasma p16INK4A is.


Let us also remember that most of these studies are cross-sectional. None of them measure senescence and uses this knowledge to predict future health outcomes. Evidently we need rigorous prospective studies, but the invasive nature of measuring senescence in biopsies is quite limiting in this regard.

Comparative data across species
Few studies directly tested the modern senescence theory of aging. Early studies worked under the assumption that proliferation leads to senescence, hence cells from longer-lived species should be able to proliferate longer before becoming senescent. These studies are quite mixed.

The work by Attaallah et al. (2019) is the only comparative study I could find that does not focus on ex vivo proliferation, but it is still relying on cells grown in vitro. Across 6 species the ability to induce senescence in vitro is positively correlated with species longevity. Senescence was induced by neocarzinostatin (NCS) and measured by the beta-gal assay. This is one of those results that is difficult to explain, because in situ the acquisition of senescence appears to be slower and not faster in long-lived species. Besides the obvious caveat that in vitro may not apply to in vivo, we can consider possible explanations. Perhaps even though long-lived species quickly induce senescence, they also quickly clear senescent cells e.g. by immune cells, eventual cell death or even reversal of the phenotype. Only when the organism ages does this balancing act fail leading to accumulation.

Rohme et al. 1981 found a correlation between species lifespan (MLS) and proliferative potential that looks a bit too perfect across their 8 species. A mere 25 years later, Lorenzini et al. (2005) contradict this finding in 11 species, suggesting that the relationship is fully explained by bodymass. We can see that this branch of research isn't going anywhere.

However, there are superior approaches that we have not yet explored. What we would like to test is how quickly senescent cells increase in situ; this could be done if we look at the rate of increases in beta-galactosidase in tissues, for example. Just eyeballing the different studies it would be obvious - or at least plausible - that mice (Wang et al. 2009) accumulate senescent cells faster than primates (Herbig et al. 2006), but what about other species?
If we wish to continue studying isolated fibroblasts due to their simplicity, we should focus on different types of senescence, esp. DNA-damage induced and oncogene-induced senescence (as in (Attaallah et al. 2019) .

If we seek even more indirect evidence, we can look towards DNA damage and telomeres. Current studies indicate that DNA double strand breaks (DSBs) are central to aging and the ability to repair those is correlated with species lifespan (Tian et al. 2019, Lorenzini et al. 2009) but this does not apply to other types of DNA damage (BER and NER; Tian et al. 2019, Page & Stuart 2012). Incidentally DSBs also induce senescence.

Although, several studies support the idea that telomeres shorten more slowly in species that are longer-lived (Tricola et al. 2018, Dantzer and Fletcher 2015), this is still at best only indirect evidence on the topic of senescence. Especially because there is some controversy about the causality of this hypothesis (Arai et al. 2015, Simons et al. 2015) and also because telomere length shows only small associations with human pathology, that may be biased because pre-existing disease causes shorter telomeres. For example, Araújo Carvalho et al. 2019 write that: “The current available evidence suggests that telomere length may be not a meaningful biomarker for frailty” and a large mendelian randomization study suggests that marginal benefits on CVD are offset by cancer mortality and that "Telomere lengthening may offer little gain in later-life health status and face increasing cancer risks" (Kuo et al. 2019).


Figure note: My goal is to clearly distinguish between strong and weak evidence. All the "pathways" in the above graph can be considered weak and indirect evidence in favour of the senescence theory. Nonetheless they are worth discussing. If for no other reason so that we can better criticize the published literature! Comparative studies of telomere dysfunction, DNA damage and replicative senescence are discussed in this chapter. Chemotherapy will be briefly discussed in the "Models of accelerated senescence". The inflammation data supporting the senescence theory is discussed in "Misc data – the SASP..."

Comparative data within species
This paradigm was entirely forgotten by Yanai & Fraifeld (2018) in their review of the evidence, yet it has been used to great effect in the GH/IGF-1 field (Miller et al. 2002, Yuan et al. 2009). In a longitudinal study design, where we are able to follow animals, we would expect animals with higher senescence at age X, to also have a higher mortality or disease risk at age X+n. Although this study design is best suited for minimally invasive sampling (e.g. blood draw to measure hormones; body-weight measurements) I see no reason why this would not be feasible using skin biopsies.

Comparative data in long-lived mouse models
Let’s review the mouse findings using the GENtervention database as a guide (Tyshkovskiy et al. 2019). The Gladyshev group combined transcriptomic data for all the mouse models of extended longevity that they could find (n=15). Their database includes validated models whose lifespan was confirmed by several groups (e.g. Ames dwarfs), single study results (e.g. FGF-21 transgenics) and models with modest and sex-specific increases in lifespan (e.g. Acarbose, 17alpha estradiol, protandim, S6K1 deletion). Nevertheless, we would expect to see a pattern if there is something real going on here.

So how is the expression of senescence-associated genes affected based on their transcriptomic data? Let us use the following markers for senescence for a quick and dirty analysis: p16, p21, LMNB1 and some SASP components IL-6, IL-1A, IL-8, CXCL1, CXCL5. Only the underlined ones are available in their database, possibly because the other targets are not expressed at a high enough level in the liver. Interestingly, none of these markers are reduced in long-lived mice, which may be explained by the heterogeneity of their dataset. Since this database includes data from both young and old animals this could mask a true age-related improvement in senescence. Given these limitations let’s not take this issue too seriously and wait for age-corrected data to be published.

However, we can still take a look at the primary data for each of those models that was published. Below I provide an example graph from the GENtervention database; overlaid on the graph I have marked models in green that support the senescence hypothesis, in red models that do not show decelerated senescence despite an extended lifespan and with NA models that have no available data that I could find during my quick review.

Figure note: Surprisingly the SASP-associated chemokine CXCL1 tends to be elevated in long-lived mouse models. Data for other genes is not shown but also doesn't look very convincing. Green=favors senescence hypothesis, Red=does not favor, Source: GENtervention database.

Overall data from long-lived mice supports the senescence hypothesis but there are some ups and downs. A recent study (Mau et al. 2020) found that Acarbose, 17-a-estradiol and rapamycin do not decrease adipose tissue mRNA markers of senescence. CR in contrast was quite effective reducing p16, p21 and p53 mRNA levels in both male and female mice. Other studies in contrast do show reduced senescence after rapamycin treatment (Wang et al. 2017, Herranz et al. 2015, Kucheryavenko et al. 2019, etc). This may be explained by tissue specific effects but in at least one other paper rapmaycin did have a positive impact on fat tissue p16 and some SASP cytokines (Wang et al. 2017).

The data on CR (Ogrodnik et al. 2017) and the various growth hormone dwarfs (Stout et al. 2014is rather convincing even though the methods to quantify senescence and the target tissues vary widely between studies. In contrast I would call the MR data preliminary because it is supported by only one or two studies from the same lab (Wang et al. 2019). The biggest challenge is the data from mice with reduced Myc expression. The authors clearly expected to see a change in senescence. They were quite thorough and tested beta-gal positive cells, p16 and p21 expression in liver, lung, heart, muscle and spleen, completely dismantling their working hypothesis (Hofman et al. 2015). This baffling because as the authors note "Increased expression of the MYC protein strongly promotes cell proliferation and has been documented as a frequent event in a wide variety of human cancers" which may be expected to promote senescence. We will have to see how this develops further. In addition, I am quite keen to see more work on the FGF21 transgenic mouse which is among the most promising models.

Age-related diseases and senescence
I do not want to dwell on this topic, because I don't think it is very productive. Yanai & Fraifeld (2018) provide a good overview. The main issue is that there are dozens of age-related diseases and all of them are multifactorial with hundreds of causes. For most of them our mouse models are inadequate, which is one of the reasons why so many drugs fail in the clinic. We do not understand the main causes of these diseases in humans and even if we do, we often fail to model them in animals.

Atherosclerosis seems to be a promising target, because a strong inflammation link is well documented after the CANTOS trial and because senescence is (probably) pro-inflammatory. In humans SA-β-Gal-positive cells are often found adjacent to atherosclerotic plaques and in mouse models of atherosclerosis removal of p16-positive cells reduces plaque burden. However, as others have noted "macrophages can substantially contribute to SAβG staining, and might be killed via suicide genes expressed from p16ink4a promoters" (Childs et al. 2016, Bennett and Clarke 2016). Furthermore, even Childs et al. who should be defending their own study freely admit that "proliferative senescence arrest may initially restrict atherosclerotic lesion development" (Childs, Li and van Deursen 2018). Finally, since naturally-aging mice do not develop atherosclerosis, we can't use the countless senolytic studies in normal mice to address this issue.

Diabetes and metabolic dysfunction are other topics that are somewhat controversial. On the one hand, for DM II we can see accelerated fat cell senescence in humans (Minamino et al., 2009) and senescent cell ablation improved insulin sensitivity in a mouse model of obesity (Palmer et al. 2019). Similarly, in normally aging mice the JAK1/2 inhibitor ruxolitinib reduced SASP secretion and improved insulin sensitivity (Xu et al. 2015). On the other hand, insulin and glucose levels were unchanged in normally aging mice after senescent cell ablation via a p16 suicide gene (Baker et al. 2016) and many senolytic studies fail to report these outcomes. What is more, we know that in mice p16Ink4a promotes insulin secretion from beta-cells (Helman et al. 2016) so ablation of p16 could have mixed effects on glucose homeostasis.

I hope this makes it clear why we can't take disease associations too seriously. These studies are of interest to researchers who want to plan follow-up work, however, they can't address whether senescence is important to aging.

Misc data – the SASP, inflammation, plausibility and indirect evidence for causality
If we stipulate that senescence -> inflammation -> mortality, we can ask 1/ what are the inflammatory mediators that are relevant for the SASP in humans and 2/ are these associated with mortality? Do we have any evidence for causality?

Let us ignore for a moment the fact that pro- and anti-inflammatory cytokines and pathways have very tissue dependent effects, some of which are positive, and try to hone in on the so called pro-inflammatory cytokines.

Regarding 1:
Let’s remember that the SASP and transcriptional phenotype of senescence is extremely heterogenous and still controversial. It is still possible that the SASP is a laboratory artifact and that the in vivo secretome looks totally different than predicted by early studies (see the Basisty et al. (2020) work and the chapter on markers). Nevertheless a selection of studies points to the following inflammatory mediators:

MMP9, IL1A and maybe CXCL1 and IL8 because they are candidate SASP genes with lower expression in the colon of CRON volunteers (Fontana et al. 2018). Another study supports the following factors at least in so far as they were lowered in diabetic kidney disease patients who were treated with dasatinib + quercetin: Plasma IL-1α, −2, −6, and − 9 and Matrix Metalloproteinases (MMP)-2, −9, and −12 were significantly lower 11 days after than before the 3 days of D + Q administration (Hickson et al. 2019)

Regarding 2:
The strongest data links CRP as a pan-inflammation marker with a range of adverse health outcomes, and especially CVD and some cancers. Evidently the theory was bolstered by the CANTOS trial but it is a far cry from proven. Out of the senescence associated cytokines, IL-6, is perhaps most strongly linked with aging. In a meta-analysis IL-6 was associated with CVD and all-cause mortality of the elderly (Li et al. 2017) and also with frailty (Soysal et al. 2016). In multi-marker studies IL-6 or IL-6 containing inflammation scores often rank at the top performing better than other biomarkers of aging and predicting lifespan (Varadhan et al. 2013, Arai et al. 2015). Similarly, we see promising signals in different types of genetic studies (GWAS in centenarians: Zeng et al. 2016, Mendelian randomization: Rosa et al. 2019).

Among some of the more recent senescence markers identified in vitro, the mitokine GDF15 stands out because it is rather strongly linked with all-cause mortality and frailty (see Basisty et al. 2020 for a brief discussion of this and other markers). However, I would caution against taking this seriously because we do not know if senescent cell ablation actually reduces GDF15.

As we can see, the SASP concept itself does not yet provide convincing evidence for the senescence theory. However, the data is rather promising.

Intervening
Here we will discuss studies that tried to reduce the senescent cell burden. These studies, while important, are not necessarily superior to the "indirect" studies we discussed before. Any intervention that successfully targets aging may be expected to reduce senescence, often through indirect pathways. Let's say senescence is just a side-effect of DNA damage and tumorigenesis with no causal role in aging. Then any intervention that reduces DNA damage, for example CR, would also decrease senescence while extending lifespan, giving the wrong impression that a decrease in senescence promoted lifespan. As a result we have to give more weight to studies that use *selective* anti-senescent therapies and consider the totality of the evidence from all study types. Also for this reason, I consider studies of dwarf mice and CR to be indirect studies (discussed above).

Types of interventions
Let us first review how we can intervene. As mentioned before, there are pleitropic anti-aging interventions, for example dwarfism or CR that may affect senescence. Regarding more specific interventions, the field started by studying the genetic loss of p16 (Baker et al. 2008). This is not good enough, because p16 may affect growing and developing mice. So the induced ablation of p16 was invented (Baker et al. 2011). Again that's not good enough, as we can't engineer humans to have a mutant p16! Therefore drugs were the next stage. Groups developed or repurposed drugs that affect the inflammatory secretion of the SASP, e.g. the JAK1/2 inhibitor ruxolitinib (Xu et al. 2015) and rapamycin (Herranz et al. 2015). While others screened for, and found, drugs that can kill senescent cells and called them senolytics. The combination of dasatinib+quercetin is the prototype here (Zhu et al. 2015). The distinction between specific and pleitropic intervention is a bit arbitrary. Most of these "specific" drugs have many effects. Personally I will consider them to be "specific" if they were discovered first as senotherapeutics and not incidentally discovered to affect senescence through post-hoc analysis (e.g. CR, rapamycin, etc).

Senolytics and other intermittent therapies are the preferred option because a short-term treatment can have sizable benefits. This type of treatment is the best, if not only, avenue to develop universal and cost-effective anti-aging therapies for humans.

A list of senolytic interventions
What follows is a (most likely) incomplete list of senolytics that show in vivo efficacy. By now there may be many unpublished screening efforts that are being developed commercially and remain under lock until they are more clinically advanced.

List based on Kim & Kim (2019), Kobbe (2019) and Kirkland (2017):
  • Two drugs targeting the Bcl-2 pro-survival pathway:
    ABT263 aka Navitoclax (Chang et al. 2016, Zhu et al. 2016, in naturally aging mice)
    ABT737 (Yosef et al. 2016, in vivo induced senescence in mice)
  • 17-DMAG (Fuhrmann-Stroissnigg et al. 2017 in accelerated aging mice; HSP90 pathway)
    Currently in testing by the NIA ITP programe.
  • FOXO4-DRI peptide (Baar et al. 2017, in naturally aging mice; p53)
  • Dasatinib/quercetin (Zhu et al. 2015, in naturally aging mice; several pathways)
  • Cardiac glycosides (Triana-Martínez et al. 2019; in vivo induced senescence in mice; membrane depolarization of senescent cells) 
  • Fisetin (Yousefzadeh et al. 2018 in naturally aging mice; mechanism unclear)
    Intermittent treatment currently in testing by the NIA ITP programe.
With some limitations based on the assays used:
  • galacto-oligosaccharide encapsulated cytotoxic drugs (Muñoz-Espín et al. 2018, in vivo induced senescence in mice)
A word of caution about mouse healthspan studies
Healthspan studies in mice have to be taken with a grain of salt. Not only are mice very different from humans but studies show a high risk of bias, because they are easy to carry out. This means that publication bias and unintentional p-hacking to confirm a seminal findings are bound to happen. Because disease are multifactorial we can find disease association with ineffective drugs like curcumin or reservatrol but this does not mean that they affect aging at all. Most published studies of disease models are positive, whereas most published studies of lifespan tests are negative. Since most drugs fail in the clinic and we assume disease to be complex this suggests that our preferred model should have a high failure rate to reflect this complexity and weed out weak candidates.

A couple of strategies help to address this problem. When we evaluate mouse studies we have to pay attention to studies with multiple outcomes and consistent effects between studies. We also have to look for proof of principle studies that help guide future work and effective treatment strategies, for example, short-term treatments are always more interesting than long-term treatments.

Of course senescence plays a role in many age-related diseases, so do mitochondria, lysosomes, so does mTOR, AMPK, TFEB, NFKB, sirtuins and even ROS! – but we need evidence that such pathways have a considerable impact on aging and are tractable.

The story of resveratrol and curcumin should be the perfect warning. While some described resveratrol as a SASP inhibitor or a “senomorphic” (Kobbe 2019) it failed to extend lifespan in the NIA ITP study. Similarly, curcumin is known to be a senolytic in vitro (Yousefzadeh 2018) and also fails to extend lifespan in the NIA ITP study. Both of these compounds are rather unselective with low bioavailability as well. Let's not forget that quercetin is also thought to have low bioavailability and yet it works better than the other compounds. This should teach us that “plausible” and preliminary data must be taken with a giant grain of salt. Maybe they don’t work; maybe they lack bioavailability – there are millions of reasons why preliminary “healthspan” data may not translate into more rigorous studies.

Key studies: improved healthspan

Fisetin; see improved lifespan
INK-ATTAC; see improved lifespan

Although not a senolytic, a temporary treatment with the JAK inhibitor ruxolitinib improved physical function, systemic and fat tissue inflammation after 2 months (Xu et al. 2015), most likely by inhibiting the SASP. While promising we have to remember that 2 months for a mouse are akin to 6 years for a human.

Baar et al. cell 2017 (Campisi group) used a FOXO4 peptide that destabilizes the FOXO4-p53 interaction and thereby presumably induces cell death specifically in senescent cells. To quantify ablation of p16 positive cells the authors put a renilla luciferase (RLUC) protein under the p16 promoter. This allows them to measure sensecence in a light-generating assay and not worry about technical problems like p16 antibodies and primers. After the treatment multiple markers of healthspan improved, among those frailty, fur density and renal function. It appears the treatment was started in mice around ~104wks of age.

Key studies: improved lifespan
Lifespan studies are more stringent than healthspan studies, therefore the preferred evidence to a biogerontologist.

The plant polyphenol fisetin, reduced beta-galactosidase staining, p16 and p21 expression in several mouse tissues, improved pathologic lesion scores in brain and kidney and extended lifespan (Yousefzadeh et al. 2018). Although, acute administration is enough to eliminate senescent cells, the lifespan part of the study used continuous administration. Of note is that intermittent fisetin is currently being studied in the NIA ITP – the most rigorous mouse aging study in the world – to answer whether these findings can be replicated.
Unfortunately, the Yousefzadeh et al. study cannot be considered a definite replication of Baker et al. 2016 for two reasons. First, the short lifespan of the control mice and second the pleiotropic effects of fisetin.

Xu et al. 2018 found intermittent treatment with dasatinib+quercetin, D+Q, every two weeks, beginning at the age of ~700days extended mouse lifespan and improved physical function like e.g. grip strength. Overall I am quite happy with this study because the lifespan of the controls is impressive with a median lifespan around 940 days. The caveat here is that the extension in maximum lifespan is smaller than the change in mean lifespan. Given this, I find it a bit unfair the way they characterize the study of Baker et al. 2016 as showing "an increase in median rather than maximum survival". Additionally, the prevalence of age-related diseases + tumor burden was unchanged at autopsy, probably because the treated mice lived longer and were older.

Baker et al. 2016 used the INK-ATTAC model to ablate p16 positive senescent cells finding that both healthspan and lifespan were improved. The authors start treatment at 12mo and continued for 6 months. Regarding age-related diseases tumour latency was increased and “spontaneous activity and exploratory behaviour” were improved. However, there was no difference in “motor coordination and balance, memory, exercise ability, and muscle strength”. In contrast, glomerulosclerosis and heart (cardiomyocyte) atrophy were improved.

Of course the study is great, but it remains hampered by the lifespan of the controls. Even looking at the post-hoc data the authors provide to claim acceptable lifespans. The mean LS of male controls is unacceptable as they admit themselves. However, the females fare only slightly better. Six out of nine other studies showed better survival and three of them by around 100 days or 20%. That means that even the female treatment group in this study (LS+17%) doesn't beat the lifespan of control mice in other studies.

Human data and ongoing trials
As already mentioned, three important human studies were published in 2019 with Hickson et al. (2019) key among them, showing that dasatinib/quercetin (D+Q) reduce senescent cells in skin and adipose tissue and SASP in plasma of patients with diabetic kidney disease (n=9). In this study subjects were given 100mg Dasatinib and 1000mg Quercetin for 3 days. While promising, we have to remember that this study is a far cry from the primary prevention trials that we need to run in order to prove the senescence hypothesis. Since the participants had diabetic kidney disease and were around 68 years old we have to assume that they were quite sick to begin with. This study also doesn't include a control group making interpretation difficult. It is possible that D+Q only works in the elderly or that the authors measured regression to the mean. Nevertheless the paper is a good read and in the discussion the authors give a brief update about ongoing trials.

The two other trials published this year are weaker so I won't discuss them in any detail. Justice et al. 2019 showed that physical function in idiopathic pulmonary fibrosis (IPF), a lung fibrotic disease linked to senescence, was improved after D+Q but effects on blood SASP were mixed (n=14). Then Martyanov et al. 2019 provided evidence for reduced SASP in the skin of systemic sclerosis patients treated with just quercetin.

Okay, I said three, so the fourth study is running out of competition, because it is not a study of oral senolytics. Nevertheless it is quite impressive. Chung et al. (2019) report a placebo controlled trial, they studied dermal application of rapamycin at a dose of 10 μM dissolved in DMSO. Congratulations to the PIs Lorenzini and Sell for getting this study off the ground! The authors showed that p16INK4A in the skin was down by IHC staining after 6 to 8 months of treatment. Photoaging-related dyspigmentation and wrinkling of the hands also improved significantly. 

What about future studies? Out of all the ongoing trials I could find only one qualifies as an actual anti-aging trial because it includes normal people and not a highly selected and sick group.  "AFFIRM-LITE: A Phase 2 Randomized, Placebo-Controlled Study of Alleviation by Fisetin of Frailty, Inflammation, and Related Measures in Older Adults". Paez-Ribes et al. (2019) and Hickson et al. (2019) review several ongoing studies for those who are interested to read more.

Self-experimentation
Given the success in animal studies combined with the lack of human studies to recruit volunteers who want to pioneer senolytics, many people choose to self-experiment with these drugs in dangerous ways. This is the flip-side of "protecting" patients by not rushing into clinical trials! As researchers our job would have been to protect these people by offering them a safe venue to self-experiment, but we failed. Let me share a couple anecdotes and suggestions for safe use. Reason from fightaging.org has written an excellent, although slightly outdated, post on this topic. While his suggestions for quality control and harm reduction are spot on, I do think he is wrong about the selection of senolytics. Given that the ultimate goal of these treatments is healthspan extension, and I consider mouse lifespan and preliminary human data to be the gold standard here, I would say that the only drugs even worth considering at this stage are fisetin and dasatinib + quercetin (D+Q). Very soon we should also have lifespan data for (chronic administration of) alvespimycin/17-DMAG.

Also the best estimate for a correct dose need not be derived from animal studies anymore. We have two ongoing trials one with "Fisetin [at] 20mg/kg/day, orally for 2 consecutive days" and another with 100mg Dasatinib and 1000mg Quercetin for 3 days (Hickson et al. 2019).

So does it work?

One user, for example, reports improved balance and reduced fatigue after taking India-sourced Dasatinib+Quercetin at 200mg D & 2000mg Q (which seems rather high). We can make a reasonable case for a certain kind of self-experimentation. Warn your doctor about your use/abuse of research drugs and keep track of your health before and after use. In that sense, this user is doing reasonably well, because they're tracking their progress. Also remember that dasatinib is a dangerous drug not to be taken lightly even with intermittent dosing. On the other hand, we should not be hysteric about the risks, which generally manifest during chronic treatment of leukemias that lasts for weeks, months or years and even so they're manageable: "[some leukemic patients can] anticipate being on a TKI [like dasatinib] for many years...results indicated that dasatinib treatment at a dose of 100 mg once daily demonstrated durable efficacy and a tolerable long‐term safety profile."

Let us contrast this with another user. At the same time as some are experimenting with D+Q that is backed up by many animal studies and a couple of pilot human trials, others are trying self application of azithromycin based on in vitro studies that used ridiculously high doses of 100uM! Basic science literacy would show that this is not particularly prudent, because in vitro studies are weak and this one in particular used a dose that is way too high. So azithromycin blood concentrations are reported to peak around 0.4mg/L. With a Mw of 749 mg/mmol; at 400ug/L in the blood and 749ug/umol this translates to blood levels of ~0.5uM (same ballpark as reported by: Vandooren et al. 2017). I mean if you already self-experiment, and there is nothing wrong with that, as people can take whatever risks they want; why not minimize the risk by using a protocol with better risk/reward?

So then perhaps using fisetin for self-experimentation is better? In this case the user, for example, reports improved skin texture and no beneficial changes in blood work and another user reports reduced back pain. Well, I think you get the pattern, don't you? The reported changes are often very subjective and hard to quantify. It will be interesting to follow them up in more rigorous ways. If you want to read more, the two forums most involved with self-experimentation are probably longecity and age-reversal.

Models of accelerated senescence
At last I would like to discuss the models of premature aging ("progeroid mice"), which are often hailed as a great invention, if not the holy grail of aging research. Of course, they're useful. However, there are many misunderstandings about progeroid models. Many of these models show severe pathology and age-related "damage" that is not seen during mouse and human aging, for example the mutator PolG mouse.

Either way, showing that a pathology X is seen in an accelerated aging model and normal aging, while amelioration of pathology X extends lifespan in the progeroid model is an important finding. However, it is neither necessary nor sufficient to prove that pathology X is involved in normal aging. Such a finding is akin to playing a scratchcard. You reveal one digit after another and all of them seem to match. While you're about to scratch out the last two digits, you're not going to call your mother and tell her that you're a millionaire just yet. For better or worse, luck plays a role in the way promising scientific findings translate to the clinic and more robust systems.

To be entirely clear here, none of the progeroid models actually show accelerated aging. This is impossible to achieve. However, they may faithfully replicate some key aspects of aging. Perhaps it would be better to talk about segmentally accelerated aging here. Nevertheless, there is no doubt that a progeroid mouse may replicate aspects of mouse aging, just like C. elegans may replicate aspects of mouse aging.

With that in mind, let's quickly discuss the findings here. It was noted quite a while ago that mice with reduced BubR1 levels show premature "aging" and this is associated with aneuploidy and renal senescence as judged by β-galactosidase staining (Baker et al. 2004). These mice show muscle weakness and have a hunchback (lordokyphosis) so they do look quite "old". Even earlier, there were suggestions that progeroid fibroblasts from mice and humans arrest earlier when grown ex vivo (as per Yanai & Fraifeld 2018). As it turns out, the BubR1 mice also show elevated p16Ink4a expression in some tissues.

Based on this knowledge, the van Deursen lab crossed BubR1 hypomorphic mice with p16Ink4a null mice showing great improvements in lifespan (Baker et al. 2008). Three years later they repeated this experiment using a genetic system that allows ablation of p16Ink4a positive cells in adult mice (Baker et al. 2011; INK-ATTAC; p16 was fused to a drug-inducible caspase 8 protein). They showed that this is beneficial similarly to inborn loss of p16 and this paved the way for senescent cell ablation.

Other progeroid models
Ercc1 hypomorphic mice that mimic the human XFE progeroid syndrome have elevated levels of senescence and removal of these cells improves their healthspan. The same is true for a mouse model of human trichothiodystrophy. As for the the SOD1 knock-out mouse, it does have elevated levels of p16 and p21 but no one has tested sen cell ablation yet (Yousefzadeh et al. 2019). Despite the success of senescent cell ablation we haven't tested, or I couldn't find the data, if senescence is increased in human progerias. I think we do see beta-galactosidase accumulation with Hutchinson-Gilford progeria but I can't find data on other progerias (Foo et al. 2019).

Senescent cell transplantation
A special case of mice with accelerated senescence is the targeted transplantation of senescent cells. Given the right controls, this model appears slightly more powerful than the "accelerated aging" models. In the latter we have the following chain of events: disease X -> senescent cell increase -> pathologic outcome -- it is obviously more elegant to omit X.

Xu et al. (2018) found that transplantation of senescent preadipocytes into 17mo mice reduced physical function in these mice and shortened their lifespan, without shifts in disease burden, and transplantation of control cells had no such effect. Another study transplanting fewer senescent cells, confirmed some of these findings (da Silva et al 2019). Senescent cells transplanted into muscle or skin induced neighboring cells to express senescent markers but interestingly there was no impact on fibre area in muscle and no impact on skin thinning.

Chemotherapy
While it may be true that chemotherapy induces senescence and that we see "early aging phenotypes" in cancer survivors this is again only circumstantial evidence in favor of the senescence theory (see Naylor et al. 2012). If you think about it, chemotherapy messes up so many things that there could be a million reasons for the ill health of cancer survivors.

Dose response studies
Assuming that senescence is causally involved in aging, the more cells you ablate the larger should be the health benefit. Similarly in models with extended lifespan, the degree of the lifespan extension should correlate with the number of senescent. The former hasn't be done to my knowledge and for the latter we only have a couple of studies.

Let us start with models of graded lifespan extension. I am aware of two commonly used "models". We have a rather basic model of graded GH expression, which involves the use of GH-deficient dwarfs, controls and bovine growth hormone (bGH) overexpressing mice or sometimes injected bGH. What this model lacks in sophistication it makes up in simplicity. Stout et al. 2014 from the Kirkland group found that 18mo Snell and GHR -/- mice have less beta-gal positive cells in several fat depots and GH injection directly increases senescence. Similarly, bGH-overexpressing mice also had higher levels of beta-gal positive cells. Interestingly the changes on the transcript level of p16, p21, IL-6 are somewhat inconsistent in contrast to the beta-gal data. Again highlighting a problem in the senescence field: heterogeneous markers, heterogeneous outcomes.


In another study the authors (Fontana et al. 2018) measured gene expression of senescence markers (p16, p21) and of SASP markers (IL1a, MMP9 and CXCL1) in the colon of mice after 90 days of graded calorie restriction (CR). CR tends to improve these markers. The medial colon shows good evidence of a graded dose-response and we also see a bit of that in the distal colon. The proximal colon barely responds to CR at all. This type of research also highlights the problem with mouse studies. What to make of the result if only 2 out of 3 tested tissues show an effect? Indeed without the inclusion of a graded protocol this study would be much less impressive.

Summary, outlook, open questions
The future of senolytic treatments has never looked so bright. We may truly hope that this concept will give rise to life-extending therapies in humans. Despite all the promise and effort, however, we have to understand that beautiful theories often fail in the clinic. After decades of human trials, where is the drug or nutraceutical that can prevent cancer or cardiovascular disease in healthy middle aged adults? Most of our favourite treatments, backed by promising preliminary data, have faltered in larger trials. These were, for example, antioxidants in the late 90s and early 2000s. Later on fish oil and vitamin D.


We should embrace self-experimentation because it is not going away. We direly need someone to follow-up and evaluate these people, who are taking senolytics from the black market, perhaps in a similar study design to the CRON cohort studied by Luigi Fontana and others. Unfortunately, given the heterogeneity of these drugs and their users, this would be tremendously hard. What is the control group? How do we know people were taking the drugs correctly and if the delivered chemicals even contained the advertised drug?

Another way to accelerate progress would be to involve pet owners again. Matt Kaeberlein has pioneered the idea of testing life extending drugs in dogs. The coming decade could be the time for this protocol to shine. With the advent of smartphones, we should try to take it one step further and devise ultra-low-cost app-based study designs that work purely via standard mail and smartphone apps. Most relevant outcomes in pets can be judged without a veterinarian even if we can't do any advanced health checks and measurements. For example, users should have no problem to self-report their pet's quality of life through pictures and videos; mortality can be measured in the same way; available data from normal vet visits can be reported by taking a picture, etc.

While the senescence theory is promising, the coming decade is also the right time to carry out rigorous studies to kill a beautiful hypothesis before it turns into an immortal zombie lie, as did antioxidants for lifespan and healthspan extension. Now is the time for translational work and more rigorous replication studies and papers addressing publication bias and shortcomings of the published data. For example, I hope we can fund a study that tests dozens of tissues with different senescence markers in primates or humans. We're getting more and more replication studies, like this one, largely in support of the senescence theory (Idda et al. 2020) but they also keep uncovering inconsistencies. For example the lung senescence data from mice does not replicate in humans; which considering the current situation is rather depressing.

What happens after removal of senescent cells? No one knows the long term consequences or even the short term consequences in more detail, but we can expect that a healthier niche for new cells would open up. I am sure researchers are already thinking about combined ablation + regeneration approaches, as well they should, e.g. Brooks and Robbins (2018): "Overcoming the possible adverse effects conferred by extensive depletion of dysfunctional stem cells could be overcome through transplantation of healthy young stem cells or the molecules in which they secret (see below)"

We're lucky in a way because the skin is among the organs with the highest burden of senescent cells. Seminal work has shown that the topical treatment with rapamycin is beneficial and now is the time to test senolytics for the skin. This is much safer and cheaper than running trials with oral drugs and would act as a proof of concept that senescent cells can be removed in the key target population (i.e. middle aged adults, or young elderly, who may be unwilling to undergo genuine phase I-style oral dosing studies).


References (selection)
Helman, A., Klochendler, A., Azazmeh, N., Gabai, Y., Horwitz, E., Anzi, S., ... & Fixler, Y. (2016). p16 Ink4a-induced senescence of pancreatic beta cells enhances insulin secretion. Nature medicine, 22(4), 412.

Hickson, L. J., Prata, L. G. L., Bobart, S. A., Evans, T. K., Giorgadze, N., Hashmi, S. K., ... & Kellogg, T. A. (2019). Senolytics decrease senescent cells in humans: Preliminary report from a clinical trial of Dasatinib plus Quercetin in individuals with diabetic kidney disease. EBioMedicine, 47, 446-456.

Anderson, R., Lagnado, A., Maggiorani, D., Walaszczyk, A., Dookun, E., Chapman, J., ... & Proctor, C. (2019). Length‐independent telomere damage drives post‐mitotic cardiomyocyte senescence. The EMBO journal, 38(5).

Tricola, G. M., Simons, M. J., Atema, E., Boughton, R. K., Brown, J. L., Dearborn, D. C., ... & Juola, F. A. (2018). The rate of telomere loss is related to maximum lifespan in birds. Philosophical Transactions of the Royal Society B: Biological Sciences, 373(1741), 20160445.

Cristofalo, V. J., Allen, R. G., Pignolo, R. J., Martin, B. G., & Beck, J. C. (1998). Relationship between donor age and the replicative lifespan of human cells in culture: a reevaluation. Proceedings of the National Academy of Sciences, 95(18), 10614-10619.

Dimri, G. P., Lee, X., Basile, G., Acosta, M., Scott, G., Roskelley, C., ... & Pereira-Smith, O. (1995). A biomarker that identifies senescent human cells in culture and in aging skin in vivo. Proceedings of the National Academy of Sciences, 92(20), 9363-9367.

Cristofalo, V. J., Lorenzini, A., Allen, R. G., Torres, C., & Tresini, M. (2004). Replicative senescence: a critical review. Mechanisms of ageing and development, 125(10-11), 827-848.

Zindy, F., Quelle, D. E., Roussel, M. F., & Sherr, C. J. (1997). Expression of the p16 INK4a tumor suppressor versus other INK4 family members during mouse development and aging. Oncogene, 15(2), 203-211.

Severino, J., Allen, R. G., Balin, S., Balin, A., & Cristofalo, V. J. (2000). Is β-galactosidase staining a marker of senescence in vitro and in vivo?. Experimental cell research, 257(1), 162-171.

Mishima, K., Handa, J. T., Aotaki-Keen, A., Lutty, G. A., Morse, L. S., & Hjelmeland, L. M. (1999). Senescence-associated beta-galactosidase histochemistry for the primate eye. Investigative ophthalmology & visual science, 40(7), 1590-1593.

Pendergrass, W. R., Lane, M. A., Bodkin, N. L., Hansen, B. C., Ingram, D. K., Roth, G. S., ... & Wolf, N. S. (1999). Cellular proliferation potential during aging and caloric restriction in rhesus monkeys (Macaca mulatta). Journal of cellular physiology, 180(1), 123-130.

Sedelnikova, O. A., Horikawa, I., Zimonjic, D. B., Popescu, N. C., Bonner, W. M., & Barrett, J. C. (2004). Senescing human cells and ageing mice accumulate DNA lesions with unrepairable double-strand breaks. Nature cell biology, 6(2), 168-170.

Baker, D. J., Jeganathan, K. B., Cameron, J. D., Thompson, M., Juneja, S., Kopecka, A., ... & Van Deursen, J. M. (2004). BubR1 insufficiency causes early onset of aging-associated phenotypes and infertility in mice. Nature genetics, 36(7), 744-749.

Baker, D. J., Perez-Terzic, C., Jin, F., Pitel, K. S., Niederländer, N. J., Jeganathan, K., ... & Eberhardt, N. L. (2008). Opposing roles for p16 Ink4a and p19 Arf in senescence and ageing caused by BubR1 insufficiency. Nature cell biology, 10(7), 825-836.

Coppé, J. P., Patil, C. K., Rodier, F., Sun, Y. U., Muñoz, D. P., Goldstein, J., ... & Campisi, J. (2008). Senescence-associated secretory phenotypes reveal cell-nonautonomous functions of oncogenic RAS and the p53 tumor suppressor. PLoS biology, 6(12).

Xu, M., Palmer, A. K., Ding, H., Weivoda, M. M., Pirtskhalava, T., White, T. A., ... & Jensen, M. D. (2015). Targeting senescent cells enhances adipogenesis and metabolic function in old age. elife, 4, e12997.

Sousa-Victor, P., Gutarra, S., García-Prat, L., Rodriguez-Ubreva, J., Ortet, L., Ruiz-Bonilla, V., ... & Perdiguero, E. (2014). Geriatric muscle stem cells switch reversible quiescence into senescence. Nature, 506(7488), 316-321.

Zhu, Y., Tchkonia, T., Pirtskhalava, T., Gower, A. C., Ding, H., Giorgadze, N., ... & O'Hara, S. P. (2015). The Achilles’ heel of senescent cells: from transcriptome to senolytic drugs. Aging cell, 14(4), 644-658.

Xu, M., Tchkonia, T., Ding, H., Ogrodnik, M., Lubbers, E. R., Pirtskhalava, T., ... & Giorgadze, N. (2015). JAK inhibition alleviates the cellular senescence-associated secretory phenotype and frailty in old age. Proceedings of the National Academy of Sciences, 112(46), E6301-E6310.

Baker, D. J., Childs, B. G., Durik, M., Wijers, M. E., Sieben, C. J., Zhong, J., ... & Khazaie, K. (2016). Naturally occurring p16 Ink4a-positive cells shorten healthy lifespan. Nature, 530(7589), 184-189.

Xu, M., Pirtskhalava, T., Farr, J. N., Weigand, B. M., Palmer, A. K., Weivoda, M. M., ... & Onken, J. L. (2018). Senolytics improve physical function and increase lifespan in old age. Nature medicine, 24(8), 1246-1256.

Hickson, L. J., Prata, L. G. L., Bobart, S. A., Evans, T. K., Giorgadze, N., Hashmi, S. K., ... & Kellogg, T. A. (2019). Senolytics decrease senescent cells in humans: Preliminary report from a clinical trial of Dasatinib plus Quercetin in individuals with diabetic kidney disease. EBioMedicine, 47, 446-456.

Sharpless, N. E., & Sherr, C. J. (2015). Forging a signature of in vivo senescence. Nature Reviews Cancer, 15(7), 397-408.

Hernandez-Segura, A., Nehme, J., & Demaria, M. (2018). Hallmarks of cellular senescence. Trends in cell biology, 28(6), 436-453.

Hernandez-Segura, A., de Jong, T. V., Melov, S., Guryev, V., Campisi, J., & Demaria, M. (2017). Unmasking transcriptional heterogeneity in senescent cells. Current Biology, 27(17), 2652-2660.

Casella, G., Munk, R., Kim, K. M., Piao, Y., De, S., Abdelmohsen, K., & Gorospe, M. (2019). Transcriptome signature of cellular senescence. Nucleic acids research, 47(14), 7294-7305.

Basisty, N., Kale, A., Jeon, O. H., Kuehnemann, C., Payne, T., Rao, C., ... & Campisi, J. (2020). A proteomic atlas of senescence-associated secretomes for aging biomarker development. PLoS biology, 18(1), e3000599.

Roos, W. P., Thomas, A. D., & Kaina, B. (2016). DNA damage and the balance between survival and death in cancer biology. Nature Reviews Cancer, 16(1), 20.

Herbig, U., Ferreira, M., Condel, L., Carey, D., & Sedivy, J. M. (2006). Cellular senescence in aging primates. Science, 311(5765), 1257-1257.

Baar, M. P., Brandt, R. M., Putavet, D. A., Klein, J. D., Derks, K. W., Bourgeois, B. R., ... & van der Pluijm, I. (2017). Targeted apoptosis of senescent cells restores tissue homeostasis in response to chemotoxicity and aging. Cell, 169(1), 132-147.

Muñoz‐Espín, D., Rovira, M., Galiana, I., Giménez, C., Lozano‐Torres, B., Paez‐Ribes, M., ... & Garaulet, G. (2018). A versatile drug delivery system targeting senescent cells. EMBO molecular medicine, 10(9).

Paradis, V., Youssef, N., Dargère, D., Bâ, N., Bonvoust, F., Deschatrette, J., & Bedossa, P. (2001). Replicative senescence in normal liver, chronic hepatitis C, and hepatocellular carcinomas. Human pathology, 32(3), 327-332.

Tonnessen-Murray, C. A., Lozano, G., & Jackson, J. G. (2017). The regulation of cellular functions by the p53 protein: cellular senescence. Cold Spring Harbor perspectives in medicine, 7(2), a026112.

Baker, D. J., Wijshake, T., Tchkonia, T., LeBrasseur, N. K., Childs, B. G., Van De Sluis, B., ... & van Deursen, J. M. (2011). Clearance of p16 Ink4a-positive senescent cells delays ageing-associated disorders. Nature, 479(7372), 232-236.

Mirzayans, R., Andrais, B., Hansen, G., & Murray, D. (2012). Role of p 1 6 I N K 4 A in Replicative Senescence and DNA Damage-Induced Premature Senescence in p53-Deficient Human Cells. Biochemistry research international, 2012.

Tuttle, C. S., Waaijer, M. E., Slee‐Valentijn, M. S., Stijnen, T., Westendorp, R., & Maier, A. B. (2020). Cellular senescence and chronological age in various human tissues: A systematic review and meta‐analysis. Aging cell, 19(2), e13083.

Kosar, M., Bartkova, J., Hubackova, S., Hodny, Z., Lukas, J., & Bartek, J. (2011). Senescence-associated heterochromatin foci are dispensable for cellular senescence, occur in a cell type-and insult-dependent manner and follow expression of p16ink4a. Cell cycle, 10(3), 457-468.

Guerrero, A., & Gil, J. (2016). HMGB2 holds the key to the senescence-associated secretory phenotype. J Cell Biol, 215(3), 297-299.

Loo, T. M., Miyata, K., Tanaka, Y., & Takahashi, A. (2020). Cellular senescence and senescence‐associated secretory phenotype via the cGAS‐STING signaling pathway in cancer. Cancer Science, 111(2), 304.

De Cecco, M., Ito, T., Petrashen, A. P., Elias, A. E., Skvir, N. J., Criscione, S. W., ... & Le, O. (2019). L1 drives IFN in senescent cells and promotes age-associated inflammation. Nature, 566(7742), 73-78.

De Cecco, M., Criscione, S. W., Peterson, A. L., Neretti, N., Sedivy, J. M., & Kreiling, J. A. (2013). Transposable elements become active and mobile in the genomes of aging mammalian somatic tissues. Aging (Albany NY), 5(12), 867.

Rodier, F., Muñoz, D. P., Teachenor, R., Chu, V., Le, O., Bhaumik, D., ... & Davalos, A. R. (2011). DNA-SCARS: distinct nuclear structures that sustain damage-induced senescence growth arrest and inflammatory cytokine secretion. Journal of cell science, 124(1), 68-81.

Freund, A., Laberge, R. M., Demaria, M., & Campisi, J. (2012). Lamin B1 loss is a senescence-associated biomarker. Molecular biology of the cell, 23(11), 2066-2075.

Dreesen, O., Chojnowski, A., Ong, P. F., Zhao, T. Y., Common, J. E., Lunny, D., ... & Colman, A. (2013). Lamin B1 fluctuations have differential effects on cellular proliferation and senescence. Journal of Cell Biology, 200(5), 605-617.

Ivanov, A., Pawlikowski, J., Manoharan, I., van Tuyn, J., Nelson, D. M., Rai, T. S., ... & Wu, H. (2013). Lysosome-mediated processing of chromatin in senescence. Journal of Cell Biology, 202(1), 129-143.

Dou, Z., Ghosh, K., Vizioli, M. G., Zhu, J., Sen, P., Wangensteen, K. J., ... & Capell, B. C. (2017). Cytoplasmic chromatin triggers inflammation in senescence and cancer. Nature, 550(7676), 402-406.

Childs, B. G., Baker, D. J., Wijshake, T., Conover, C. A., Campisi, J., & Van Deursen, J. M. (2016). Senescent intimal foam cells are deleterious at all stages of atherosclerosis. Science, 354(6311), 472-477.

Bennett, M. R., & Clarke, M. C. (2017). Killing the old: cell senescence in atherosclerosis. Nature Reviews Cardiology, 14(1), 8-9.

Ogrodnik, M., Miwa, S., Tchkonia, T., Tiniakos, D., Wilson, C. L., Lahat, A., ... & Grellscheid, S. N. (2017). Cellular senescence drives age-dependent hepatic steatosis. Nature communications, 8(1), 1-12.

Rossman, M. J., Kaplon, R. E., Hill, S. D., McNamara, M. N., Santos-Parker, J. R., Pierce, G. L., ... & Donato, A. J. (2017). Endothelial cell senescence with aging in healthy humans: prevention by habitual exercise and relation to vascular endothelial function. American journal of physiology-Heart and circulatory physiology, 313(5), H890-H895.

Matheu, A., Maraver, A., Collado, M., Garcia‐Cao, I., Cañamero, M., Borras, C., ... & Serrano, M. (2009). Anti‐aging activity of the Ink4/Arf locus. Aging cell, 8(2), 152-161.

Serrano, M., Lee, H. W., Chin, L., Cordon-Cardo, C., Beach, D., & DePinho, R. A. (1996). Role of the INK4a locus in tumor suppression and cell mortality. Cell, 85(1), 27-37.

Hudgins, A. D., Tazearslan, C., Tare, A., Zhu, Y., Huffman, D., & Suh, Y. (2018). Age-and tissue-specific expression of senescence biomarkers in mice. Frontiers in genetics, 9, 59.

Yousefzadeh, M. J., Zhao, J., Bukata, C., Wade, E. A., McGowan, S. J., Angelini, L. A., ... & Kato, J. I. (2020). Tissue specificity of senescent cell accumulation during physiologic and accelerated aging of mice. Aging Cell, 19(3), e13094.

Wiley, C. D., Flynn, J. M., Morrissey, C., Lebofsky, R., Shuga, J., Dong, X., ... & Campisi, J. (2017). Analysis of individual cells identifies cell‐to‐cell variability following induction of cellular senescence. Aging cell, 16(5), 1043-1050.

Fontana, L., Mitchell, S. E., Wang, B., Tosti, V., van Vliet, T., Veronese, N., ... & Demaria, M. (2018). The effects of graded caloric restriction: XII. Comparison of mouse to human impact on cellular senescence in the colon. Aging cell, 17(3), e12746.

Waaijer, M. E., Gunn, D. A., Adams, P. D., Pawlikowski, J. S., Griffiths, C. E., Van Heemst, D., ... & Maier, A. B. (2016). P16INK4a positive cells in human skin are indicative of local elastic fiber morphology, facial wrinkling, and perceived age. Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences, 71(8), 1022-1028.

Jeck, W. R., Siebold, A. P., & Sharpless, N. E. (2012). a meta‐analysis of GWAS and age‐associated diseases. Aging cell, 11(5), 727-731.

Fortney, K., Dobriban, E., Garagnani, P., Pirazzini, C., Monti, D., Mari, D., ... & Kim, S. K. (2015). Genome-wide scan informed by age-related disease identifies loci for exceptional human longevity. PLoS genetics, 11(12), e1005728.

Liu, Y., Sanoff, H. K., Cho, H., Burd, C. E., Torrice, C., Ibrahim, J. G., ... & Sharpless, N. E. (2009). Expression of p16INK4a in peripheral blood T‐cells is a biomarker of human aging. Aging cell, 8(4), 439-448.

Justice, J. N., Gregory, H., Tchkonia, T., LeBrasseur, N. K., Kirkland, J. L., Kritchevsky, S. B., & Nicklas, B. J. (2018). Cellular senescence biomarker p16INK4a+ cell burden in thigh adipose is associated with poor physical function in older women. The Journals of Gerontology: Series A, 73(7), 939-945.

Sørensen, C. E., Tritsaris, K., Reibel, J., Lauritzen, M., Mortensen, E. L., Osler, M., & Pedersen, A. M. L. (2016). Elevated p16ink4a expression in human labial salivary glands as a potential correlate of cognitive aging in late midlife. PloS one, 11(3).

Lawrence, I., Bene, M., Nacarelli, T., Azar, A., Cohen, J. Z., Torres, C., ... & Sell, C. (2018). Correlations between age, functional status, and the senescence-associated proteins HMGB2 and p16 INK4a. Geroscience, 40(2), 193-199.

Röhme, D. (1981). Evidence for a relationship between longevity of mammalian species and life spans of normal fibroblasts in vitro and erythrocytes in vivo. Proceedings of the National Academy of Sciences, 78(8), 5009-5013.

Lorenzini, A., Tresini, M., Austad, S. N., & Cristofalo, V. J. (2005). Cellular replicative capacity correlates primarily with species body mass not longevity. Mechanisms of ageing and development, 126(10), 1130-1133.

Wang, C., Jurk, D., Maddick, M., Nelson, G., Martin‐Ruiz, C., & Von Zglinicki, T. (2009). DNA damage response and cellular senescence in tissues of aging mice. Aging cell, 8(3), 311-323.

Tian, X., Firsanov, D., Zhang, Z., Cheng, Y., Luo, L., Tombline, G., ... & Goldfarb, A. (2019). SIRT6 is responsible for more efficient DNA double-strand break repair in long-lived species. Cell, 177(3), 622-638.

Page, M. M., & Stuart, J. A. (2012). Activities of DNA base excision repair enzymes in liver and brain correlate with body mass, but not lifespan. Age, 34(5), 1195-1209.

Lorenzini, A., Johnson, F. B., Oliver, A., Tresini, M., Smith, J. S., Hdeib, M., ... & Stamato, T. D. (2009). Significant correlation of species longevity with DNA double strand break recognition but not with telomere length. Mechanisms of ageing and development, 130(11-12), 784-792.

Tricola, G. M., Simons, M. J., Atema, E., Boughton, R. K., Brown, J. L., Dearborn, D. C., ... & Juola, F. A. (2018). The rate of telomere loss is related to maximum lifespan in birds. Philosophical Transactions of the Royal Society B: Biological Sciences, 373(1741), 20160445.

Arai, Y., Martin-Ruiz, C. M., Takayama, M., Abe, Y., Takebayashi, T., Koyasu, S., ... & von Zglinicki, T. (2015). Inflammation, but not telomere length, predicts successful ageing at extreme old age: a longitudinal study of semi-supercentenarians. EBioMedicine, 2(10), 1549-1558.

Miller, R. A., Harper, J. M., Galecki, A., & Burke, D. T. (2002). Big mice die young: early life body weight predicts longevity in genetically heterogeneous mice. Aging cell, 1(1), 22-29.

Yuan R, Tsaih SW, Petkova SB, Marin de Evsikova C, Xing S, Marion MA, Bogue MA, Mills KD, Peters LL, Bult CJ, Rosen CJ, Sundberg JP, Harrison DE, Churchill GA, Paigen B: Aging in inbred strains of mice: study design and interim report on median lifespans and circulating IGF1 levels. Aging Cell 2009;8:277–287.
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2768517/

Tyshkovskiy, A., Bozaykut, P., Borodinova, A. A., Gerashchenko, M. V., Ables, G. P., Garratt, M., ... & Gladyshev, V. N. (2019). Identification and application of gene expression signatures associated with lifespan extension. Cell metabolism, 30(3), 573-593.

Mau, T., O’Brien, M., Ghosh, A. K., Miller, R. A., & Yung, R. (2020). Life-span Extension Drug Interventions Affect Adipose Tissue Inflammation in Aging. The Journals of Gerontology: Series A, 75(1), 89-98.

Herranz, N., Gallage, S., Mellone, M., Wuestefeld, T., Klotz, S., Hanley, C. J., ... & Georgilis, A. (2015). mTOR regulates MAPKAPK2 translation to control the senescence-associated secretory phenotype. Nature cell biology, 17(9), 1205-1217.

Wang, R., Yu, Z., Sunchu, B., Shoaf, J., Dang, I., Zhao, S., ... & Löhr, C. V. (2017). Rapamycin inhibits the secretory phenotype of senescent cells by a Nrf2‐independent mechanism. Aging cell, 16(3), 564-574.

Ogrodnik, M., Miwa, S., Tchkonia, T., Tiniakos, D., Wilson, C. L., Lahat, A., ... & Grellscheid, S. N. (2017). Cellular senescence drives age-dependent hepatic steatosis. Nature communications, 8(1), 1-12.

Stout, M. B., Tchkonia, T., Pirtskhalava, T., Palmer, A. K., List, E. O., Berryman, D. E., ... & Oberg, A. L. (2014). Growth hormone action predicts age-related white adipose tissue dysfunction and senescent cell burden in mice. Aging (Albany NY), 6(7), 575.

Wang, S. Y., Wang, W. J., Liu, J. Q., Song, Y. H., Li, P., Sun, X. F., ... & Chen, X. M. (2019). Methionine restriction delays senescence and suppresses the senescence-associated secretory phenotype in the kidney through endogenous hydrogen sulfide. Cell Cycle, 18(14), 1573-1587.

Hofmann, J. W., Zhao, X., De Cecco, M., Peterson, A. L., Pagliaroli, L., Manivannan, J., ... & Li, X. (2015). Reduced expression of MYC increases longevity and enhances healthspan. Cell, 160(3), 477-488.

Bennett, M. R., & Clarke, M. C. (2017). Killing the old: cell senescence in atherosclerosis. Nature Reviews Cardiology, 14(1), 8-9.

Minamino, T., Orimo, M., Shimizu, I., Kunieda, T., Yokoyama, M., Ito, T., ... & Ishikawa, F. (2009). A crucial role for adipose tissue p53 in the regulation of insulin resistance. Nature medicine, 15(9), 1082.

Palmer, A. K., Xu, M., Zhu, Y., Pirtskhalava, T., Weivoda, M. M., Hachfeld, C. M., ... & Johnson, K. O. (2019). Targeting senescent cells alleviates obesity‐induced metabolic dysfunction. Aging cell, 18(3), e12950.

Helman, A., Klochendler, A., Azazmeh, N., Gabai, Y., Horwitz, E., Anzi, S., ... & Fixler, Y. (2016). p16 Ink4a-induced senescence of pancreatic beta cells enhances insulin secretion. Nature medicine, 22(4), 412.

Li, H., Liu, W., & Xie, J. (2017). Circulating interleukin-6 levels and cardiovascular and all-cause mortality in the elderly population: a meta-analysis. Archives of gerontology and geriatrics, 73, 257-262.

Varadhan, R., Yao, W., Matteini, A., Beamer, B. A., Xue, Q. L., Yang, H., ... & Fallin, M. D. (2014). Simple biologically informed inflammatory index of two serum cytokines predicts 10 year all-cause mortality in older adults. Journals of Gerontology Series A: Biomedical Sciences and Medical Sciences, 69(2), 165-173.

Zeng, Y., Feng, Q., Gu, D., & Vaupel, J. W. (2017). Demographics, phenotypic health characteristics and genetic analysis of centenarians in China. Mechanisms of ageing and development, 165, 86-97.

Rosa, M., Chignon, A., Li, Z., Boulanger, M. C., Arsenault, B. J., Bossé, Y., ... & Mathieu, P. (2019). A Mendelian randomization study of IL6 signaling in cardiovascular diseases, immune-related disorders and longevity. NPJ genomic medicine, 4(1), 1-10.

Senotherapeutics: emerging strategy for healthy aging and age-related disease.
Kim EC, Kim JR.
BMB Rep. 2019 Jan;52(1):47-55. Review.

Chang, J., Wang, Y., Shao, L., Laberge, R. M., Demaria, M., Campisi, J., ... & Luo, Y. (2016). Clearance of senescent cells by ABT263 rejuvenates aged hematopoietic stem cells in mice. Nature medicine, 22(1), 78.

Yosef, R., Pilpel, N., Tokarsky-Amiel, R., Biran, A., Ovadya, Y., Cohen, S., ... & Ben-Porath, I. (2016). Directed elimination of senescent cells by inhibition of BCL-W and BCL-XL. Nature communications, 7(1), 1-11.

Fuhrmann-Stroissnigg, H., Ling, Y. Y., Zhao, J., McGowan, S. J., Zhu, Y., Brooks, R. W., ... & Corbo, L. (2017). Identification of HSP90 inhibitors as a novel class of senolytics. Nature communications, 8(1), 1-14.

Triana-Martínez, F., Picallos-Rabina, P., Da Silva-Álvarez, S., Pietrocola, F., Llanos, S., Rodilla, V., ... & Hernández-González, F. (2019). Identification and characterization of Cardiac Glycosides as senolytic compounds. Nature communications, 10(1), 1-12.

Yousefzadeh, M. J., Zhu, Y., McGowan, S. J., Angelini, L., Fuhrmann-Stroissnigg, H., Xu, M., ... & McGuckian, C. (2018). Fisetin is a senotherapeutic that extends health and lifespan. EBioMedicine, 36, 18-28.

Justice, J. N., Nambiar, A. M., Tchkonia, T., LeBrasseur, N. K., Pascual, R., Hashmi, S. K., ... & Kirkland, J. L. (2019). Senolytics in idiopathic pulmonary fibrosis: results from a first-in-human, open-label, pilot study. EBioMedicine, 40, 554-563.

Martyanov, V., Whitfield, M. L., & Varga, J. (2019). Senescence signature in skin biopsies from systemic sclerosis patients treated with senolytic therapy: potential predictor of clinical response?. Arthritis & Rheumatology, 71(10), 1766-1767.

Chung, C. L., Lawrence, I., Hoffman, M., Elgindi, D., Nadhan, K., Potnis, M., ... & Sell, C. (2019). Topical rapamycin reduces markers of senescence and aging in human skin: an exploratory, prospective, randomized trial. GeroScience, 41(6), 861-869.

Paez‐Ribes, M., González‐Gualda, E., Doherty, G. J., & Muñoz‐Espín, D. (2019). Targeting senescent cells in translational medicine. EMBO molecular medicine, 11(12).

da Silva, P. F., Ogrodnik, M., Kucheryavenko, O., Glibert, J., Miwa, S., Cameron, K., ... & von Zglinicki, T. (2019). The bystander effect contributes to the accumulation of senescent cells in vivo. Aging Cell, 18(1), e12848.

Naylor, R. M., Baker, D. J., & Van Deursen, J. M. (2013). Senescent cells: a novel therapeutic target for aging and age‐related diseases. Clinical Pharmacology & Therapeutics, 93(1), 105-116.

Brooks, R. W., & Robbins, P. D. (2018). Treating age-related diseases with somatic stem cells. In Exosomes, Stem Cells and MicroRNA (pp. 29-45). Springer, Cham.

Yousefzadeh, M. J., Melos, K. I., Angelini, L., Burd, C. E., Robbins, P. D., & Niedernhofer, L. J. (2019). Mouse Models of Accelerated Cellular Senescence. In Cellular Senescence (pp. 203-230). Humana Press, New York, NY.

Hall, B. M., Balan, V., Gleiberman, A. S., Strom, E., Krasnov, P., Virtuoso, L. P., ... & Leonova, K. (2016). Aging of mice is associated with p16 (Ink4a)-and β-galactosidase-positive macrophage accumulation that can be induced in young mice by senescent cells. Aging (Albany NY), 8(7), 1294.

Baker, D. J., Perez-Terzic, C., Jin, F., Pitel, K. S., Niederländer, N. J., Jeganathan, K., ... & Eberhardt, N. L. (2008). Opposing roles for p16 Ink4a and p19 Arf in senescence and ageing caused by BubR1 insufficiency. Nature cell biology, 10(7), 825-836.

Kommentare