Senescent cells, senolytics and tissue repair: the devil may be in the dosing

Abstract

There is tremendous interest in the development of drugs that target senescent cells (‘senolytic’ drugs) to treat a range of age-related morbidities. However, studies in mice that demonstrate impaired tissue repair following clearance of senescent cells raise questions about the potential risks of senolytic therapies. Closer examination of the available studies reveals the hopeful possibility of a ‘therapeutic window’ in which these risks can be minimized.

There is increasing evidence that senescent cells accumulate across tissues with age and that clearance of these cells leads to an improvement in multiple age-associated morbidities and extends lifespan in mice¹. This has led to intense interest in the development of senolytic compounds to ameliorate human age-related disorders, including frailty, osteoporosis, cardiometabolic disorders, neurodegenerative diseases and others². Indeed, on the basis of the now-extensive mouse data, early-phase clinical trials have been initiated using a number of candidate drugs². Although preliminary, some — but not all — studies have shown a signal for possible efficacy in humans in terms of clearance of senescent cells in patients with diabetic kidney disease³ or improvement in measures of physical function in patients with idiopathic pulmonary fibrosis⁴. It remains to be determined, however, whether the potential efficacy of senolytic drugs in humans is truly due to the clearance of senescent cells or to other, off-target effects of these drugs. Regardless, as enthusiasm for targeting senescent cells for therapeutic benefit grows, there are calls for caution — particularly based on the potential physiological role of senescent cells (or at least ‘injury’ cells that express features of cellular senescence, including increased p16^INK4a (also known as CDKN2A) and p21^CIP1 (also known as CDKN1A)) in tissue repair⁵.

Senescent cell clearance impairs skin and lung injury repair

Evidence for a role for senescent cells in tissue repair was provided by Demaria et al.⁶ in a skin injury model in mice. These investigators developed and validated the p16-3MR mouse model, in which the p16^Ink4a (also known as Cdkn2a) promoter — which is upregulated in senescent cells — drives production of a fusion protein that includes a truncated thymidine kinase from herpes simplex virus 1 (ref.⁶). In this model, the administration of ganciclovir selectively kills senescent cells. The key finding of this study was that the administration of ganciclovir to p16-3MR mice transiently delayed skin wound healing, although the wound did heal fully by day 12 (similar to vehicle-treated mice). Further, these investigators demonstrated that p16^Ink4a-expressing cells in the wound were positive principally for markers of fibroblasts and endothelial cells⁶. In mechanistic studies, they identified PDGF-A as a key early senescence-associated secretory phenotype (SASP) factor that was critical for optimal skin wound healing.

In a more recent study, Reyes et al.⁷ developed highly sensitive p16^Ink4a-reporter mice that express multiple copies of eGFP driven by the endogenous p16^Ink4a promoter, and found a marked increase in p16^Ink4a-expressing cells following lung injury induced by naphthalene. Using single-cell RNA sequencing, they showed that (similar to the skin injury model⁶) GFP-positive cells were enriched for Pdgfra expression and consisted of either fibroblasts or inflammatory monocytes and monocyte-derived interstitial macrophages. They then treated the mice with the senolytic combination of dasatinib and quercetin following lung injury, and found that this not only markedly reduced the number of GFP-positive cells but also reduced the appearance of regenerative epithelial stem cells, consistent with an impaired response to injury following senescent cell clearance.

The findings of both studies — along with previous evidence that demonstrates important roles for senescent (or at least p16^Ink4a-expressing) cells in limiting liver fibrosis⁸, as well as in limb regeneration of salamanders⁹ — serve to dampen enthusiasm for targeting senescent cells. That said, an important caveat here is that cells that transiently express p16^Ink4a following injury may not be truly ‘senescent’ as we currently understand canonical, growth-arrested senescent cells associated with aging, and this issue requires further study. In addition, further studies are needed to define whether interventions that clear p16^Ink4a-expressing cells reduce only senescent cells or also affect other, nonsenescent cells that express p16^Ink4a. Regardless, the question remains as to whether senolytic drugs are essentially doomed if patients being treated with these drugs cannot respond appropriately to a range of tissue injuries.

Senescent cell clearance facilitates bone fracture repair

In contrast to the above findings in skin and lung injury, three separate studies in bone injury (fracture) have demonstrated that clearing senescent cells leads to the opposite outcome — namely, an acceleration of fracture repair. Using telomere-associated foci (TAF) as a marker for senescent cells¹⁰, we found that TAF-positive cells peaked in the fracture callus at day 14 after fracture, but had largely returned to baseline by day 28 (ref.¹¹). In further studies, we demonstrated that treatment with dasatinib and quercetin accelerated the time course, increased the callus bone volume and, importantly, increased the biomechanical strength of the healed fracture¹¹. These findings were replicated by a further study from our group using genetic clearance of senescent cells (D. Saul et al., unpublished). Here we used p21-ATTAC mice, in which administration of the synthetic compound AP20187 selectively clears p21^Cip1 (also known as Cdkn2a)-positive senescent cells, and also demonstrated acceleration of fracture healing following senescent-cell clearance. Findings from our group have further been validated by an independent study by Liu et al.¹², who also found that treatment with dasatinib and quercetin enhanced fracture healing in mice.

Importance of dosing for reconciling these findings

The resolution of the apparently discrepant findings from skin and lung versus bone fracture has important implications, not only for our understanding of the underlying biology of senescent cells but (as noted above) has tremendous implications for the future of senolytic drugs. One possibility is that tissue repair in skin and lung is fundamentally different from bone repair. Arguing against this, however, is a previous study by Schafer et al.¹³ that shows that in a bleomycin-induced lung injury model, clearing senescent cells with dasatinib and quercetin in wild-type mice or with AP20187 in p16-INK-ATTAC mice improved — rather than impaired — pulmonary function and physical health. These findings in lung injury are clearly different from the more recent findings of Reyes et al.⁷ and are consistent with the bone fracture data, arguing that lung and bone may not in fact have fundamental biological differences in terms of the role of senescent cells in tissue repair.

The summary in Table 1 perhaps provides an alternate explanation for the apparently discrepant findings in lung and skin versus bone, along with the lung study of Schafer et al.¹³. Careful examination of the dosing regimens of senolytic drugs reveals that in the two studies^6,7 in which clearing senescent cells impaired tissue healing, the drugs were essentially administered continuously. By contrast, in the three fracture studies as well as the lung study of Schafer et al.¹³, the senolytic drugs were administered intermittently. Thus, the appearance of senescent cells during injury is probably important for tissue repair, and elimination of these cells by continuous dosing of a senolytic regimen does, as anticipated, impair healing. Given that during acute injury these senescent cells appear rapidly (as demonstrated by our TAF data in fracure¹¹), the hypothesis based on the collective findings summarized in Table 1 is that modestly reducing — but not eliminating — the burden of these cells (as would be the case with intermittent dosing of a senolytic drug) may essentially ‘tamp down’ the exuberant injury response and actually facilitate — or at the least, not impair — tissue repair. This hypothesis, which is summarized in Fig. 1, clearly needs to be directly tested using injury models that compare intermittent versus continuous dosing of senolytic drugs and examine effects on p16^Ink4a and p21^Cip1 expression, SASP factors and tissue healing. In addition, direct measurements of drug concentrations with formal dose–response curves would more rigorously establish possible therapeutic windows for senolytic interventions.

Table 1 |.

Summary of available studies evaluating the use of senolytic drugs in tissue repair

Study	Injury site	Mouse model	Dosing regimen	Effect on repair
Demaria et al.6	Skin wound	p16-3MR	Ganciclovir continuously on days 1–6	Transient delay
Reyes et al.7	Lung (naphthalene injury)	INKBRITE	D + Q continuously for 2 weeks	Impaired epithelial stem cell regeneration
Saul et al.11	Bone fracture	Wild type	D + Q, once weekly for 5 weeks	Accelerated fracture repair
Liu et al.12	Bone fracture	Wild type	D + Q on days 1, 3, 5 and 7; no dose then to day 28	Accelerated fracture repair
D. Saul et al., unpub1isheda	Bone fracture	p21-ATTAC	AP20187 twice a week for 5 weeks	Accelerated fracture repair
Schafer et al.13	Lung (bleomycin injury)	Wild type and p16-INK-ATTAC	Brief treatment with D + Q (wild type, 3 days) or AP20187 (p16-INK-ATTAC, 6 days) 2 weeks after injury; outcomes assessed at 4 weeks after injury	Improved pulmonary function

The Demaria et al.⁶ and Reyes et al.⁷ studies demonstrate the adverse effects of clearing senescent cells on the repair response. D, dasatinib; Q, quercetin.

^aCurrently unpublished study that replicated results from ref.¹¹ using genetic clearance of senescent cells.

Fig. 1 |. Schematic of postulated effects of continuous versus intermittent senolytic treatment on tissue repair.

Tissue injury triggers the appears of p16^Ink4a- and p21^Cip1 (p16^Ink4a/p16^Cip1)-expressing ‘injury’ cells that express multiple features of senescent cells, including a SASP. The relationship of these cells to canonical, growth-arrested senescent cells associated with aging remains unclear and requires further study (depicted by question mark). Continuous senolytic treatment, leading to elimination of these p16^Ink4a- and p21^Cip1-expressing injury cells, results in impaired tissue repair. By contrast, intermittent senolytic treatment, as would generally be used clinically, reduces the burden of these cells and leads to improved tissue repair (or at least to no adverse effects). Although consistent with existing data, this hypothesis clearly needs to be directly tested by treating various types of injuries in preclinical models with continuous versus intermittent senolytic drugs.

Adding further support to the hypothesis of differing effects of continuous versus intermittent clearance of senescent cells on the response to injury are previous findings by Ritschka et al.¹⁴, who found that primary mouse keratinocytes transiently exposed to the SASP exhibited increased expression of stem cell markers and regenerative capacity in vivo. However, prolonged exposure to the SASP caused a subsequent cell-intrinsic senescence arrest to counter the continued regenerative stimuli. These findings indicate that there may well be a ‘sweet spot’ for optimal senescent-cell numbers during tissue repair. Too few cells, as in the case of continuous senolytic administration, would impair tissue repair. Too many senescent cells, as in the untreated mice with fracture, may actually be less than optimal for most efficient repair. But intermittent senolytic treatment, by reducing — but not eliminating — senescent cells may well be hitting this sweet spot and enhancing fracture (and perhaps other tissue) repair.

The future of senolytic drugs

Senolytic therapies for age-associated morbidities hold great promise, but also potential risks. It is important to note that senolytic therapy has generally been thought to be best administered intermittently, rather than continuously, as previously summarized¹. Thus, the continuous presence of senolytic drugs is probably not required for the drugs to be effective, as senescent cells — at least in the context of aging — take weeks or months to develop and acquire a SASP. As such, a brief disruption of senescent-cell prosurvival pathways is adequate to kill senescent cells. This was directly demonstrated in our studies evaluating dasatinib and quercetin in preventing age-related bone loss in mice, in which once-monthly administration of these drugs was effective¹⁵.

On the basis of these considerations, additional preclinical studies in mice and other nonhuman species are needed to further evaluate whether the intermittent elimination of senescent cells does not impair (or may even improve) repair across diverse tissues. In addition, studies are needed to evaluate whether the timing of senolytic administration (that is, before, during or following injury) has differing effects on tissue repair. Moreover, as human trials with intermittent senolytic administration progress, monitoring participants for any evidence of impaired responses to injury is critical. An additional point here is that although intermittent administration of senolytic drugs probably will not impair tissue repair, senomorphic drugs (for example, JAK inhibitors) — which inhibit the secretion of the SASP and thus need to be administered continuously¹ — may well mimic the effects of continuous senolytic administration and have adverse effects on tissue repair.

In summary, as with any intervention, senolytic drug are likely to have risks and benefits. In terms of tissue repair, the available data would indicate that there may be an adequate therapeutic window using intermittent administration of these drugs to justify their continued preclinical and clinical development.