Small molecule compounds that induce cellular senescence

1Cellular senescence is a stable arrest of the cell cycle and is characterized by complex phenotypic changes. It was first described in studies of human fibroblasts that ceased proliferation following an extended cultivation (Hayflick & Moorhead, 1961; Hayflick, 1965). Discovered by Hayflick and Moorhead, senescence in normal human cells was shown to depend on telomere dysfunction originating mainly from replication‐associated telomere shortening (Harley et al., 1990; Allsopp, 1996; Bodnar et al., 1998). This type of senescence is also known as replicative senescence and is the prototypical cellular senescence state. Other forms of senescence (i.e., not linked to proliferation‐dependent telomere shortening) include a variety of prematurely developed cellular senescence phenotypes, similar but not identical to replicative senescence. Many proliferative cell types can undergo so‐called stress‐induced premature senescence (SIPS) upon exposure to subcytotoxic stresses (UV, γ‐irradiation, H2O2, hyperoxia, etc.) (Toussaint et al., 2000, 2002). Oncogene‐induced senescence (OIS) represents another complex senescence phenotype that depends on activation and/or overexpression of oncogenes (Serrano et al., 1997; Bianchi‐Smiraglia & Nikiforov, 2012). The mechanism of OIS involves DNA damage that may be a result of DNA hyper‐replication (Di Micco et al., 2006), replication fork reversal (Neelsen et al., 2013), depletion of nucleotide pools (Mannava et al., 2013), and/or increased levels of reactive oxygen species (ROS) (Lee et al., 1999). Conceptually and mechanistically, OIS is closely related to tumor‐suppressor loss‐induced senescence (Chen et al., 2005; Di Mitri & Alimonti, 2016). Cell‐to‐cell fusion‐induced senescence can also be considered a premature senescence subtype (Chuprin et al., 2013; Burton & Faragher, 2015). The distinctive phenotypic changes typical of various types of cellular senescence are cell enlargement and flattening, senescence‐associated β‐galactosidase activity (SA‐β‐gal), formation of senescence‐associated heterochromatin foci (SAHF), persistent DNA damage response (DDR), and senescence‐associated secretory phenotype (SASP). However, these and several other facultative features of cellular senescence that manifest in each particular case of cell cycle arrest greatly depend on the senescence‐inducing stimulus and the cell type (Campisi, 2013; Salama et al., 2014).

2The contribution of cellular senescence to organismal aging is a question of ongoing research (van Deursen, 2014). However, strong evidence for this connection has been reported recently. Specifically, it was shown that clearance of age‐accumulated p16INK4A‐positive senescent cells in mice could extend their healthy lifespan (Baker et al., 2011, 2016). Several chemical compounds that specifically target senescent cells have been identified in the last 2 years (so‐called senolytic drugs) (Xu et al., 2015b; Zhu et al., 2015a,b). It was shown that clearance of senescent cells by such drugs may alleviate age‐related vasomotor dysfunction and frailty, enhance adipogenesis, rejuvenate haematopoietic stem cells after total‐body irradiation, and, generally, extend lifespan (Xu et al., 2015a; Zhu et al., 2015b; Roos et al., 2016). Furthermore, these studies confirm the known pathological impact of cellular senescence, exemplified by cellular dysfunction, impairment of tissue regeneration, detrimental effects on tissue microenvironment, etc. (Burton & Krizhanovsky, 2014). It is evident that along with its detrimental effects, cellular senescence has clearly defined beneficial physiological functions. For instance, it has been shown recently that cellular senescence plays a role in the differentiation of megakaryocytes (Besancenot et al., 2010), the maturation of the placenta (Chuprin et al., 2013), the restriction of fibrosis (Krizhanovsky et al., 2008; Jun & Lau, 2010; Zhu et al., 2013), tissue repair (Demaria et al., 2014), and embryonic development (Nacher et al., 2006; Munoz‐Espin et al., 2013; Storer et al., 2013). The role of cellular senescence in cancer prevention is well documented (Burton & Krizhanovsky, 2014; Munoz‐Espin & Serrano, 2014).

3It is generally agreed in the field that the most important features of cellular senescence are SASP and resistance to apoptosis (Munoz‐Espin & Serrano, 2014; Burton & Faragher, 2015). SASP stimulates immune system‐dependent elimination of unwanted precancerous cells or specific embryonic cells that undergo senescence. Notably, cellular senescence may serve as an alternative to apoptosis in embryonic development as well as in cancer prevention (Childs et al., 2014). It has been shown that failure to undergo senescence triggers apoptosis in a compensatory manner to eliminate transient structures during development (Munoz‐Espin et al., 2013; Storer et al., 2013). Therefore, it may be reasonable to consider some of the cellular senescence states (e.g., SIPS), along with apoptosis, autophagy, necrosis, etc., in terms of the cell stress response rather than aging. However, it is unclear whether or not the cellular senescence that is widely implicated in normal aging, chronic diseases, tumor suppression, tumorigenesis, cell differentiation, and embryogenesis represents a single physiological cellular state.

4To date, dozens of stress‐induced cellular senescence phenotypes have been reported. These senescence states may differ substantially from each other, as well as from replicative senescence, through the presence of specific senescence features. Additionally, it has been reported that some stress‐induced senescence states can be overcome, thus challenging the dogma that cellular senescence is an irreversible form of growth arrest (Romanov et al., 2001; Beausejour et al., 2003). Such caveats can lead to confusion regarding the terminology of stress‐induced cellular senescence states; it is not clear whether senescence‐like growth arrest (and variations thereof) resembles ‘true’ cellular senescence. The indispensable characteristics of this ‘true’ cellular senescence are also elusive. It can be argued that SASP (arising along with morphological changes and SA‐β‐gal) may be the most important physiologically relevant feature of cellular senescence; however, SASP has not been studied in most cases of stress‐induced senescence. Here, we attempt to catalog virtually all of the cellular senescence‐like states that can be induced by low molecular weight compounds (Table 1). We summarize the biological markers, molecular pathways involved in senescence establishment, and specific traits of cellular senescence states induced by small compounds, as well as the treatment conditions used. In total, we analyzed more than 50 chemical inducers of cellular senescence and senescence‐like states. These chemical compounds can be functionally classified into eight groups: (1) DNA replication stress inducers (aphidicolin, hydroxyurea, thymidine, bromodeoxyuridine, difluorodeoxycytidine, cyclopentenyl cytosine); (2) DNA‐damaging agents, including (2a) DNA topoisomerase inhibitors (doxorubicin, etoposide, daunorubicin, mitoxantrone, camptothecin), (2b) DNA cross‐linkers (cisplatin, mitomycin C, busulfan, cyclophosphamide, diaziquone), and (2c) drugs with complex effects (actinomycin D, bleomycin, temozolomide); (3) epigenetic modifiers that inhibit DNA methyltransferases (5‐aza‐2′‐deoxycytidine), histone deacetylases (sodium butyrate, trichostatin A, MS‐275, SAHA, LBH589, phenylbutyric acid, valproic acid), histone acetyltransferases (curcumin, C646), and histone methyltransferases (BRD4770); (4) inhibitors of telomerase activity (SYUIQ‐5, BMVC4, pyridostatin, compound 115405, perylene and indole derivatives, harmine, BIBR1532, azidothymidine); (5) inhibitors of cyclin‐dependent kinases (palbociclib, roscovitine, ribociclib); (6) activators of p53 (nutlin 3a, FL118); (7) activators of protein kinase C (TPA/PMA, PEP005, PEP008); and (8) reactive oxygen species (ROS) inducers (hydrogen peroxide, tert‐butyl hydroperoxide, phenyl‐2‐pyridyl ketoxime, phenylaminonaphthoquinones, paraquat).

5The table highlights the fact that cancer cells can undergo cellular senescence in vitro just as well as their normal nontransformed counterparts. It is apparent that there is no senescence marker or pathway unique to normal or cancer cells. In most cases, increased SA‐β‐gal, morphological changes, and persistent DDR foci were recorded. SAHF were found in only a few cases (aphidicolin, etoposide, palbociclib, and epigenetic modifiers). SASP was also noted only in some cases; however, this is likely because SASP is not commonly analyzed as a senescence biomarker. Apparently, an implicit consensus was established that the demonstration of SA‐β‐gal, morphological changes, and persistent DDR is sufficient to document a cellular senescence‐like state. It is notable that authors designated these phenotypes as a state of premature senescence or senescence‐like cell cycle arrest, regardless of the set of biomarkers observed in each case.

6Extremely prolonged drug exposure (from hours to days) was typically required to induce cellular senescence, as is evidenced by the table. In marginal situations, as in the case of aphidicolin‐induced cell cycle arrest, the full set of senescence biomarkers (SA‐β‐gal, cell enlargement, SAHF, and DDR foci) was maintained, while the drug was present in the culture medium and lost upon drug removal (Maya‐Mendoza et al., 2014). The requirement for prolonged incubation time was found for all groups of chemical compounds analyzed; however, the mechanism of senescence development appeared to differ among these groups. Whereas replication stress inducers, different DNA‐damaging agents, and telomerase inhibitors likely generate a persistent DDR following prolonged introduction of a small number of DNA lesions or telomere uncapping, long‐term incubation with epigenetic modifiers likely causes transcriptional activation of repressed loci (particularly INK4A, which encodes p16 CDK inhibitor). This hypothesis is supported by the fact that, in contrast to DNA damage‐induced cellular senescence, which depends on p21 CDK inhibitor, epigenetically induced senescence is mostly dependent on p16. This characterizes epigenetically induced senescence as ‘causeless’—epigenetic modifiers directly activate molecular pathways maintaining the cellular senescence state without generating any cell stress. In this regard, senescence induced by epigenetic modifiers can resemble developmentally programmed or organismal aging‐associated cellular senescence, while replication stress‐ and DNA damage‐induced senescence are examples of stress‐induced premature senescence states.

7It follows from the table that replication stress‐ and DNA damage‐induced cellular senescence mostly depend on the p53‐p21 pathway. The same is basically true for cellular senescence induced by physical stressors such as ionizing radiation (IR) and ultraviolet (UV) (Latonen et al., 2001; Suzuki et al., 2006). It is well known that IR as well as UV can stimulate senescence in a variety of normal and cancer cell lines (Chainiaux et al., 2002; Meng et al., 2003; Jones et al., 2005; Jee et al., 2009). Mechanistically, this type of cellular senescence mostly depends on DNA damage induced by these stressors; this links IR and UV to chemical DNA‐damaging agents. Moreover, IR and UV, along with most of the DNA‐damaging agents presented in the table, induce apoptosis rather than senescence when used at higher doses. These observations further emphasize the relationship between apoptosis and senescence. Accordingly, these cell stress response pathways may operate either as alternatives or as supplement to each other. While prominent (but short term) DNA damage induces apoptosis, prolonged mild DNA damage activates cellular senescence. The p53 transcription factor emerges as a master regulator controlling these cell fate decisions (Purvis et al., 2012).

8This work was supported by a Russian Science Foundation [grant number 14‐24‐00022].

9None declared.