1. Introduction
Cellular senescence is a complex stress response where cells undergo an irreversible cell cycle arrest in response to DNA damage and mitochondrial dysfunction. Although no exclusive and unique markers of senescence exist, senescent cells can be identified by a persistent DNA damage, expression of cyclin dependent kinase inhibitors p16 and p21, resistance to apoptotic stimuli, increased lysosomal content, ER stress, accumulation of mitochondria, and alterations in compositions of the plasma and nuclear membranes [1]. In addition, senescent cells secrete various factors including pro-inflammatory cytokines, growth factors, chemokines and matrix metalloproteinases (MMP's), collectively named the senescence-associated secretory phenotype (SASP) [1]. Senescent cells are found in many age-associated disorders, where their detrimental effects are influenced by the SASP. Clearing naturally occurring senescent cells delays the onset of several of these diseases and increases median lifespan in mice [2]. Accumulating evidence suggests that modulating the SASP can also improve healthspan and reduce frailty [3]. Owing to these exciting findings, compounds which selectively eliminate senescent cells (senolytics) or modulate the SASP (senomorphics) are under current investigation for their effectiveness to alleviate age-associated disorders [4].
Resistance to apoptotic stimuli occurs through upregulation of several members of the Bcl-2 family of anti-apoptotic proteins, namely Bcl-xL and Bcl-w. ABT-263 and ABT-737 are two confirmed senolytic agents which sensitize senescent cells to undergo intrinsic apoptosis via inhibition of these proteins. In vivo these compounds have been demonstrated to eliminate senescent cells induced by ionizing radiation or chemotherapeutic agents, as well as naturally occurring senescent cells in aged mice, in various tissues such as the lung, skin and hematopoietic system [[5], [6], [7]]. ABT-263 has also been shown to kill senescent cells in atherosclerotic lesions, reducing the overall disease burden [8]. More recently, ABT-263 has been shown to reduce senescence markers in the cortex and hippocampus and attenuate tau phosphorylation in a tau-dependent neurodegenerative mouse model [9].
Other compounds with senolytic activity include Dasatanib (D), an inhibitor of multiple tyrosine kinases already used in cancer treatment, and a natural flavanol called Quercetin (Q). When combined (D + Q), these drugs rescued cardiovascular function and frailty in aged mice, extended health span in progeroid ERCC1-/Δ mice and ameliorated obesity associated loss of neurogenesis and anxiety-like behavior [[10], [11], [12]]. However, the mechanism as to how D + Q induce cell deathspecifically in senescent cells is currently unknown.
HSP90 inhibitors such as 17-DMAG have been demonstrated to eliminate senescent cells through downregulation of the anti-apoptotic PI3K-AKT pathway and, similarly to D + Q treatment, to extend healthspan in ERCC1-/Δ mice [13].
Recent studies have also shown that molecules which modulate p53 activity and/or localization can act as senolytics. FOXO4 is required for maintaining viability in senescent cells, but a peptide which interferes with the binding of this protein to p53, resulted in p53 nuclear exclusion and apoptosis in senescent cells. This peptide improved healthspan in progeroid XPDTTD/TTD and naturally aged mice, and rescued chemotoxicity in doxorubicin-treated mice [14]. Another compound, termed UBX0101, functions as a p53/MDM2 inhibitor and has been demonstrated to kill senescent cells in osteoarthritic lesions in mice and decrease overall disease phenotypes [15].
The NF-kB (nuclear factor-kB) transcription factor acts as a master regulator of SASP factor expression [16]. The NF-kB pathway during senescence is modulated by the mammalian target of rapamycin (mTOR) [17] and mitogen-activated protein kinase (MAPK) cascades [18]. mTOR stimulates IL-1α translation, which leads to activation of IL1R signaling in an autocrine manner to trigger NF-ĸB mediated transcription of pro-inflammatory SASP factors [17]. mTOR also regulates translation of MK2, which in turn induces downstream phosphorylation of ZFP36L1. This results in a suppression in the protein's ability to degrade transcripts of various SASP factors during senescence [19]. Interfering with the mTOR pathway could therefore also be an approach to ameliorate detrimental effects of the SASP and improve mammalian healthspan. Rapamycin is a specific suppressor of mTORC1 and its administration has been shown to alleviate senescence-associated inflammation and increase lifespan in mice [17,20]. Specific MAPK inhibitors SB203580, UR13756 and BIRB 796 significantly inhibited SASP expression in senescent cells and decreased their effect in promoting breast cancer cell invasion in vitro [18,21]. Inhibition of MK2 was shown to have similar effects in dampening the SASP [21]. Glucocorticoids have also been discovered to inhibit selected components of the SASP through blockage of the IL1α-NF-kB signaling axis [22].
Accumulating evidence suggests that caloric restriction (CR) is also an effective method to diminish the SASP in vivo and promote health- and lifespan in a wide variety of organisms [23]. Markers of senescence including p16 and p21 are downregulated in colons from mouse and humans which underwent CR [24]. Therefore it is likely that CR inhibits the natural occurrence of senescent cells although the mechanism is currently unknown. In addition, drugs that are believed to be ‘caloric restriction mimetics’ such as resveratrol and metformin were shown to have similar effects on the SASP. Metformin administrations in mice resulted in an increase in both health- and lifespan, presumably through activation of the Nrf2/ARE antioxidant pathway [25]. This would possibly result in an overall reduction in oxidative damage and decrease in both NF-Kb activation and chronic inflammation [25]. Metformin also directly inhibited NF-Kb signaling by preventing phosphorylation of IKK/B and nuclear translocation of RELA and RELB [26]. The natural flavonoid resveratol increased lifespan in worms and mice presumably via activation of SIRT1, a mammalian sirtuin that normally suppresses the SASP epigenetically [27]. Kaempferol and apigenin are also natural flavonoids which inhibit the SASP but function by inhibiting NF-Kb activity [28].
Although senotherapies appear promising as a therapeutic intervention to treat age-associated disorders, there are some limitations with their use due to intrinsic and extrinsic toxicities. ABT-263 and ABT-737 can cause thrombocytopenia or neutropenia owing to the importance of Bcl-2 anti-apoptotic proteins in platelet and neutrophil survival [4]. Rapamycin is an immunosuppressant as mTOR suppression can inhibit T and B cell activation and proliferation [29]. Importantly, senescent cells have also been reported to play beneficial roles in tissue repair and remodeling, which further complicates the development of senotherapies with minimal adverse effects (Fig. 1).
Fig. 1. Beneficial effects of senescent cells during tissue repair and potential detrimental effects of senotherapies.
2. Regulation of tissue repair, reprogramming and regeneration by cellular senescence
In recent years, beneficial roles for senescent cells in promoting wound healing and limiting excessive fibrosis in response to tissue damage have been described (Fig. 1). At the site of a wound, senescent cells transiently appear in a coordinated manner to promote fibroblast to myofibroblast differentiation through secretion of platelet derived growth factor AA (PDGF-AA). Myofibroblasts drive wound contraction and promote optimal repair. Clearing senescent cells in engineered mouse models (p16-3MR and INK-ATTAC) resulted in a delay in wound closure [30,31]. In humans, fibroblast cultures from venous ulcers and pressure ulcer beds show an increase in senescence markers such as SA-β-gal positivity and shorter replicative lifespan, compared to normal skin fibroblasts [32,33]. Together, these data suggest that dysregulation of the senescent response during wound healing has the potential to alter wound healing kinetics in both humans and mice.
A fibrotic response is initiated during wound healing where excessive extracellular matrix (ECM) components are deposited at the site of injury. Although this typically results in scar formation, in extreme cases it can lead to organ failure. Liver fibrosiscan be induced in mice through CCl4 administration, where hepatic stellate cells(HSCs) are activated into senescent HSCs [34]. The matricellular protein CCN1 is a key senescence regulator in these cells [35,36]. CCN1 activates Ras-related C3 botulinium toxin substrate 1 (RAC1), leading to downstream generation of reactive oxygen species (ROS) via NADPH oxidase 1 (NOX1) activation. ROS activity leads to upregulation of p16 via ERK and p38 MAPK, and p53 via a DNA damage response. Senescent HSCs transiently accumulate along fibrotic scars where they secrete ECM degrading enzymes such as matrix metalloproteinases (MMPs) to prevent further scaring. CCL4 treated livers from p53-/-;INK4A/ARF-/- mice showed significant decreased numbers of senescent cells yet severe cirrhosis, confirming an essential beneficial role for senescent cells [34].
CCN1 has also been reported to limit fibrosis in other tissue contexts by regulating senescence via RAC1-NOX1. CCN1 induces fibroblast senescence in cutaneous wounds in mice. These cells also secrete antifibrotic SASP factors to prevent excess collagen deposition in the skin [36]. Interestingly clearing senescent cells in cutaneous wounds in the p16-3MR mouse model also resulted in excess skin fibrosis, most likely due to the resultant absence of proteases and MMPs [30]. Exogenous expression of CCN1 in infarcted hearts in mice resolved fibrosis and rescued heart function, suggesting also an essential role for restricting myocardial fibrosis [37].
A breakthrough in regenerative medicine has been the discovery that forced expression of only four factors (now called Yamanaka factors) was sufficient to reprogram differentiated cells into a pluripotent state [38]. Cellular senescence has been reported to positively influence this process in vivo by creating a favorable microenvironment for reprogramming to occur. Genetically engineered mice that express the four Yamanaka factors upon doxycycline administration (i4F) display NANOG positive cells and teratomas in several tissues, indicative of successful reprogramming [39]. Remarkably, this process does not occur in INK4A/ARFheterozygous mice, suggesting an intact INK4A/ARF locus is required [40]. Further work has demonstrated that INK4A, rather than ARF, is specifically required for in vivo reprogramming in mice [41]. SA-β-gal positive cells were found in close proximity to NANOG positive cells in wild-type i4F mice. These senescent cells promoted reprogramming through secretion of IL6, which functions in a paracrine manner to activate PIM1 and induce cellular plasticity in reprogrammed cells [40]. Senescent cells also favor reprogramming of tissues which are normally less susceptible to the process. Inducing tissue damage in skeletal muscle in i4F mice gave rise to senescent muscle cells. Doxycycline treatment post injury resulted in generation of pluripotent muscle stem cells. This did not occur in uninjured mice owing to the absence of a reprogrammable niche [42]. These findings may help generate new approaches in regenerative medicine for difficult to reprogram tissues.
As well as mediating transcription of pro-inflammatory SASP components, NF-kB has also been recently discovered to regulate transcription of stemness genes in senescent cells, providing further evidence of their pro-regenerative capabilities. Engraftment of newborn keratinocytes transiently exposed to the SASP from HRASV12-expressing keratinocytes resulted in an increased number of hair folliclesand hair growth in nude mice. Interestingly, when cells were exposed to the SASP for prolonged periods they undergo cell-cycle arrest and paracrine induced senescence, possibly to prevent tumorigenesis in response to persistent regenerative stimuli [43].
In contrast to mammals, salamanders have the ability to regenerate a variety of complex body parts including limbs. Senescent cells appear at the regenerating site in a defined spatial-temporal manner. Interference with macrophage mediated clearance of senescent cells, before or during limb amputation, led to complete loss of regenerative capacity. Therefore macrophage-mediated clearance of senescent cells is a critical step for proper limb regeneration in salamanders, although it is not yet known why [44].
Senescence has recently been discovered to be involved in embryonic development. SA-β-gal+ cells are visible in the apical ectodermal ridge (AER), neural roof plate, mesonephros and endolymphatic sac where they contribute to ensure correct patterning and limb formation [45,46]. Interestingly, embryonic senescence is p21-dependent but p16-and p53-independent. p21 is instead activated via TGF-β/SMAD and PI3K/FOXO pathways. Embryos that are p21-deficient only showed mild patterning and limb formation defects, suggesting compensatory pathways such as apoptosis exist.
3. Possible toxicities of senotherapies
3.1. Senolytics
Cellular senescence limits liver fibrosis by initially halting the proliferation of HSCs to prevent further tissue damage, and by also increasing expression of fibrolytic factors to facilitate repair [34]. Therefore there is a risk that patients with unresolved liver fibrosis develop cirrhosis if senolytics are administered systemically. In a similar manner, senolytics could lead to heart failure in patients with unresolved myocardial fibrosis. As cellular senescence also plays beneficial roles in limiting skin fibrosis [36] and promoting cutaneous wound healing [30], there is a risk that senolytic administrations could lead to ineffective tissue repair and exacerbated fibrosis, which should especially be considered if patients have undergone surgery. Local senolytic administrations could be considered to bypass these possible toxicities if possible. For example, UBX0101 can be delivered via intra-articular injection to clear senescent cells in osteoarthritic lesions, and this administration is currently under evaluation in a Phase I clinical trial [15]. Drug-eluting stents which release senolytic molecules could be considered for atherosclerosis treatment. As senescent cells are present in eyes of glaucoma patients, intraocular administrationscould be utilized to eliminate them [47].
Secreted HSP90 is present throughout the time course of acute and diabetic wound healing in the extracellular space where it promotes dermal fibroblast and keratinocyte migration and recruitment [[48], [49], [50]]. Inhibition of secreted HSP90 activity completely blocks dermal fibroblasts migration and slowed down wound healing kinetics in mice [48,50]. Intriguingly, application of recombinant HSP90 or a 115aa sized part of HSP90alpha to the wound accelerates wound closurekinetics in mice [49,50]. Mechanistically, HSP90 secretion is promoted by Hif-1alpha and TNF alpha. Hif-1alpha is induced by low oxygen supply after wounding [50] and can induce TNF-alpha in renal cancers [51]. Whether Hif-1alpha and TNF-alpha are causally linked in wound healing is unclear. These data indicate that the elimination of senescent cells by HSP90 inhibitors might lead to impaired wound healing capacity. Topical treatment of peptides that resemble part of the HSP90 protein may be an effective treatment to avoid these negative effects.
Senescent cells have been shown to promote in vivo reprogramming in neighboring somatic cells [40]. Therefore senescent cells are likely to play important roles in future therapies which utilize techniques in regenerative medicine. Studies show remarkable advances in in vivo reprogramming in mammals. For example, pancreatic exocrine cells could be converted into functional B-cells (care of diabetes) [52], cardiac fibroblasts into cardiomyocytes (treatment after heart infarcts) [53] and astrocytes into proliferative neuroblasts (in neurodegenerative diseases) [54] through in vivo reprogramming. Interestingly, some of these approaches are less efficient in vitro, possible owing to a lack of senescent cells found in vivo. It has already been demonstrated that ABT-263 prevented reprogramming in doxycycline fed i4F mice [40,42]. In possible future settings, patients who undergo regenerative medicine should avoid senolytic administrations in order for tissue regeneration to be effective.
Depletion of senescent cells during embryonic development in mice leads to mild patterning and limb formation defects [45,46]. Therefore, senotherapies should be avoided, similar to most drugs, during pregnancies to prevent possible developmental abnormalities and miscarriages.
3.2. Senomorphics
Rapamycin is an approved immunosuppressant drug to prevent acute graft rejection, but complications involving wound healing are associated with its use [55]. Additionally, accumulating evidence suggests that mTOR promotes wound healing kinetics in rodents possibly through the SASP [[56], [57], [58]]. VEGF (Vascular Endothelial Growth Factor) is a SASP factor that transiently populate cutaneous wounds and is needed for angiogenesis in wound healing [30,59]. In rats, rapamycin treatment delayed wound healing kinetics and reduced the number of VEGF expressing cells compared to placebo treated control rats after incision of the skin [56,58]. Rapamycin treatment significantly lowered the number of ECM producing myofibroblasts leading to decreased tension strength of wounds after injury. Epithelial cell specific disruption of PTEN or Tsc1, key suppressors of the PI3K/Akt/mTOR signaling axis, accelerated wound closure [57]. These data suggest that the use of rapamycin as a SASP modulator to alleviate age-related phenotypes may be detrimental for wound healing. Local stimulation of mTOR through targeting PTEN or Tsc1 may help to prevent this.
Similar to rapamycin, glucocorticoids also alleviate the SASP through IL1α-NF-kB inhibition [22]. The membranous glucocorticoid receptor inhibits keratinocyte migration and wound closure through activation of the Wnt-like phospholipase(PLC)/protein kinase C (PKC) signaling cascade [60,61]. Activation of this pathway leads to the induction of known biomarkers for non-healing wounds including β-catenin and c-myc. In addition, genetically modified mice that express a defective form of endogenous glucocorticoids showed remarkably enlarged granulation tissue after cutaneous wounding [62]. These data show the use of glucocorticoids should be avoided during tissue repair, limiting the potential use of glucocorticoids to alleviate age-related disorders.
Accumulating evidence suggest that the MAPK pathway is important for wound healing. Mitogen-activated protein kinase (MEK) kinase 1 (MEKK1) is expressed in cutaneous wounds and is required for ECM homeostasis through MMP expression, epithelial cell migration and wound epithelialization. Interference with MEKK1 during cutaneous wound healing significantly slowed down the wound healing rate [63]. Specific inhibition of MAPK by SB203580 and subsequent reduction of ERK-1activity led to decreased epithelial cell proliferation and migration, resulting in delayed corneal wound closure [64]. These data suggest that inhibition of the MAPK pathway to alleviate the SASP may impact on wound healing.
Systemic metformin administration in rats decreases wound closure rates through inhibition of keratinocyte proliferation [65]. Wounds of metformin treated rats showed significantly more redness and scarring, indicating inhibitory effects of metformin on tissue remodeling. Metformin is frequently used as an antidiabetic drug. Diabetic patients treated with metformin have foot ulcers with significantly bigger diameters. Contrarily, local administration of metformin in the wound beds of rats increased wound vascularization through activation of AMPK and subsequent expression of VEGF [66]. This suggests that metformin act on different phases of wound healing and the effect of metformin is dose-dependent. Thus, the systemic chronic intake of metformin to alleviate SASP mediated detrimental effects on health may have a negative impact on tissue repair and remodeling in the skin.
Caloric restriction in rodents show decreased wound healing rates, ablated synthesis of collagen and decreased number of endothelial cells and fibroblasts in the wound compared to refed control animals possibly due to impaired IGF-1 signaling [67,68]. This suggests that caloric restriction should be avoided during wound healing which limits the applicability of this intervention to alleviate aging. Whether caloric restriction influences wound healing through senescent cells remains to be investigated.
A major challenge in the senescence field is the lack of a uniform definition of senescence supported by specific molecular markers [1]. A combination of individually non-specific markers should be used to define senescence and to evaluate the efficacy of senotherapies [1,69]. Currently, the senescence-associated markers used for testing senotherapies are often variable and inconsistent, making cross-validations between different labs and the comparison among different compounds extremely challenging (Table 1).
Table 1. Efficacy and validation of senotherapies. Please check at source.
4. Concluding remarks
Accumulating evidence finds detrimental roles for senescent cells in many age-associated disorders. Senotherapies are attractive options to alleviate a number of these diseases at the same time, especially with an ever increasing aging population. However, as senescent cells have been shown to also influence beneficial physiological roles, care must be taken with possible side effects. These can include delayed wound healing post-surgery, cirrhosis in patients with liver fibrosis or developmental abnormalities in embryos (Fig. 1). No study to our knowledge has investigated whether senotherapies do negatively affect these outcomes, especially ones focused on the impact on senescent cells. Local senolytic/senomorphic administrations could counter these possible effects, although this may depend on disease context and on compound chemistry. Future studies addressing whether local administrations are effective at alleviating disease burden and not associated with toxic side effects should be performed.
Moreover, more studies should be performed to identify whether it is possible to discriminate between detrimental and beneficial senescence programs. This understanding could provide novel targets to develop therapeutic approaches aimed at interfering solely with deleterious senescent cells. Finally, head-to-head comparisons of senotherapies in different disease models should be done to provide insights in compound-specific toxicities. In the case of systemic application, the identification of delivering systems to preferentially target senescent cells and/or specific types of senescent cells might reduce off-target effects of senotherapies, an approach which has been recently shown to be possible [70].
