Senolytics and cell senescence: historical and evolutionary perspectives

Abstract

Senolytics are a new class of anti-aging drugs developed to selectively kill ‘senescent’ cells that are considered harmful in normal aging. More than 20 drug trials are ongoing with diverse ‘senolytic cocktails’. This commentary on recent reviews of senolytics gives a historical context of mammalian cell senescence that enabled these new drugs. While cell senescence is considered harmful to aging tissues, many studies show its essential role in some regenerative and developmental processes for which senolytic drugs may interfere. Longer-term studies of side effects are needed before senolytics are considered for general clinical practice. The wide occurrence of cell senescence in eukaryotes, yeast to fish to humans, and suggests an ancient eukaryotic process that evolved multiple phenotypes.

INTRODUCTION

Two recent reviews of senolytic drugs and cell senescence (CS) from the University of Tokyo [1] and from Mayo Clinic [2] insightfully described the multiple clinical targets and diverse molecular mechanisms of senolytics. For new readers of the burgeoning field of clinical CS, I briefly examine the variability of CS, as known from its early in vitro history, and how senolytic drugs increased rodent lifespan. Yet, CS is essential for some regenerative processes. Finally, I sketch an evolutionary perspective that CS is widely found in animals, humans to jelly fish, and suggest that CS is an ancient eukaryotic process with multiple evolved phenotypes.

HISTORY

The modern era of CS began in 1961 when Leonard Hayflick and Paul Moorhead showed the limited capacity for serial propagation of diploid fetal lung fibroblasts, which yielded postmitotic cells (Phase III) that could not be further subcultured [3, 4]. These benchmark studies challenged received wisdom that primary cell cultures could be propagated indefinitely; most indefinitely propagating cell cultures are heteroploid mutants. These findings intrigued me in choosing a problem in aging for my PhD research. My interest soon waned for in vitro cell models of aging because of the widely varying number of population doublings (PD) before CS. Moreover, adult-derived cultures ranged 2-fold in PD without progressive decline from the donor ages 26–87 (Table VI) in [3]. These variations, coupled with lengthy cultures, dissuaded me from in vitro CS and drew me to my career choice of mice as a model to study aging for physiological gene regulation.

The Hayflick diploid human fibroblast cultures (WI-38) immediately benefited public health as host cells for growing the pure viruses urgently needed for polio and other rampant infections. The Hayflick CS model also fostered the new field of biogerontology in the 1970’s. Findings on lung fibroblasts were soon extended to skin and other organs. Hayflick sometimes stated that Phase III cells died [5]. However, If carefully maintained, non-replicating ‘Phase III’ cultures can persist for nearly 2 years [5, 6], and perhaps even longer. Moreover. Phase III cells proved to be highly resistant to apoptosis [6, 7]. Holiday showed that cultures from the same explant differ a million-fold in growth potential [8]. Because of this extreme heterogeneity, the extent to which CS cells change with human aging remains undefined [9]. This heterogeneity across adult ages was also shown in the initial data [3].

EXPANDING ROLES OF CS

The role of CS in aging diseases was expanded with Sager’s 1991 proposal for CS as an anti-tumor mechanism [10]. In 2008, Campisi made the key finding that CS secrete diverse inflammatory proteins that impact neighboring cells [11], widely known as the ‘SASP’ (senescence-associated secretory phenotype) [2]. β-Galactosidase was among the first CS markers in fibroblasts and increases with age in many tissues. SASP responses include proinflammatory cytokines, chemokines, and oxidized lipids that differ cell-by-cell in the same tissue. No universal CS biochemical marker is recognized [12]. Depending on the context, SASP from CS can benefit or harm neighboring cells.

As both reviews noted [1, 2], DNA damage is among the most reproducible inducers of CS, but it is unclear how this applies to CS of embryos and adult wound healing. Stochastic DNA damage from telomere shortening and mitochondrial DNA mutations is widely associated with cell-cycle arrest [12–14]. The stochasticity of DNA damage may underlie the hugely variable proliferative capacity of primary fibroblasts, as noted above. Future single-cell DNA analysis may reveal cell-cycle checkpoint thresholds for levels of deleted mtDNA and telomere DNA and their interactions. Adjacent neurons from older brains differ in single-point mutations and deletions [15]. Somatic mutations body-wide are being analyzed in the NIH program Somatic Mosaicism Across Human Tissues (SMaHT).

The ever-expanding mechanisms for CS have not yet yielded a unifying theory of mechanisms, which limits clinical understanding and design of senolytics. Single-cell RNA (scRNA) has recently identified cell-type-specific senescent gene networks (SnG) for SASP responses and cell-cycle regulators [16]. Further scRNA analysis of CS in a broader range of species may show gene regulatory networks that operate by modular logic of cis-regulatory elements in embryonic development in specific phyletic lineages [17, 18].

Senolytic drugs

Twenty phase 1–2 clinical trials are testing a wide range of senolytic drugs [1, 2]. The extraordinary range of targeted conditions includes Alzheimer’s disease, osteoarthritis, and COVID-19 infections. The diverse senolytic drugs range from the chemotherapeutic diastinib, a flavonoid, to the diet-supplement quercetin. While clinical outcomes are pending, animal models show impressive response to the ‘senolytic cocktail’ of diastinib + quercetin. After 8 weeks of treatment, 2-year-old mice had log-fold fewer CS in their hippocampus, while spatial memory improved by 30% [19]. However, the presence of CS in neurons may be related to accumulating mutations as found in the aging human brain [16]. CS induction can benefit organ repairs, as discussed in both reviews [1, 2]. For example, partial hepatectomy of young mice rapidly induced CS; while the senolytic ABT263 decreased CS by 90%, regeneration was slowed more, by 50% [20].

Mouse lifespans are increased by some senolytics. Analysis of four experimental studies showed that median survival was increased more than maximum longevity, differing 2-fold between studies [21]. By Gompertz’s analysis for the actuarial aging rate, senolytics accelerated mortality rates. This finding is unexpected because caloric restriction slows mortality acceleration and increases the maximum lifespan in many mouse genotypes [22]. Kowald and Kirkwood are laudably cautious [21]: ‘While small pilot studies of senolytics show some initial success, the gold-standard of double-blinded placebo trials is a long road ahead … Unless such trials show target engagement, minimal short- and long-term adverse events, and clinical utility, these interventions should not be prescribed in routine clinical practice’.

THE COMPARATIVE BIOLOGY OF CS

Finally, I briefly note the wide occurrence of CS across eukaryotes, shown for four species of different phyla, and the absence of CS in two other species. The African killifish Nothobranchius furzeri is among the shortest-lived vertebrates; brain CS arises by its middle-age at 18 weeks [23]. Senolytic treatment by dasatinib and quercetin for 6 days decreased CS by 35% while increasing the number of neuroprogenitor cells. Similarly, neuroprogenitor cells in obese mice were increased by treatment with the senolytic AP20187 (AP) [24].

Invertebrates also show diverse roles for CS. The jellyfish relative Hydractinia symbiolongicarpus can regenerate its head within 3 days of amputation [25]. Regeneration is mediated by the migration of ‘i-cells’, which express CS-related paralogues including β-galactosidase and cell-cycle regulators. Head regeneration was blocked by the senolytic navitoclax (ABT-263). Authors Salinas-Saavedra et al. [25] suggested that CS is an ancient mechanism that transiently induces stress signals in neighboring tissues for regeneration. Recall from above that senolytics impaired liver regeneration.

Aging yeast have replicative senescence without telomere shortening [26]. Telomeres are protected in yeast, as in mammals, by TERRA, a non-coding RNA, transcribed from telomeric DNA, which can base-pair with telomere repeat sequences and bind telomer-protective proteins [27, 28]. TERRA is expressed in normal somatic cells and meiotic germ cells and is a recent therapeutic target for cancer [29].

Two familiar invertebrate models of aging, nematode and fly, appear to lack CS in wildtypes. Adult Caenorhabditis elegans has cell division, and mtDNA mutations have not been detected during aging [29]. However, Drosophila melanogaster which has double the lifespan of C. elegans, does accumulate mtDNA damage during aging [30]. Transgenic and neurotoxic fly models for Parkinson’s disease responded to quercetin with increased lifespan and geotactic balance [31].

The age increase of mtDNA mutations may be considered as ‘private mechanisms of aging’ [32], which anticipates further CS variations in long-lived species. Mammalian CS is strongly associated with DNA damage and may be cancer protective [10, 11, 33]. Telomere stabilization by the non-coding RNA, TERRA, found from yeast to humans, may have evolved in early eukaryote innovation in stabilization of aging genomes. On the other hand, DNA damage seems implausible as a CS inducer during regeneration of mouse liver and cnidarian heads, which require transient induction of CS proteins. This mechanistic diversity suggests that CS evolved more than once for multiple functions during metazoan evolution. A deeper evolutionary theory for CS is relevant to senolytic design, as well as life history theory.

AUTHOR CONTRIBUTIONS

Caleb Finch (Conceptualization [Lead], Writing—original draft [Lead], Writing—review & editing [Lead])