Cellular senescence: Molecular mechanisms and pathogenicity

3.1 DNA damage

Increasing evidence indicates that cellular stress, environmental stress, and chronic inflammation can induce cellular senescence through proinflammatory responses (Salminen, Kauppinen, & Kaarniranta, 2012). Oxidative stress (OS), in particular, contributes to DNA damage and telomere shortening (Passos, Saretzki, & von Zglinicki, 2007). For example, Nox4-dependent ROS production induced by treatment with 20% O₂ can trigger senescence of rat nucleus pulposus cells by activating p53-p21-Rb and p16-Rb pathways via extracellular signal-regulated kinase (ERK) signaling (Feng et al., 2017).

It has been found that DNA damage not only activates the type I interferon system and inflammation through DNA sensors, such as cyclic guanosine monophosphate (GMP)–adenosine monophosphate (AMP) synthase (cGAS), absent in melanoma 2 (AIM2), and interferon-inducible protein 16 (IFI16), but also leads to cellular senescence (Hartlova et al., 2015). DNA damage results in the accumulation of DNA fragments in the cytoplasm. The sensors that recognize DNA damage and prime cellular senescence are unknown. Recent studies have shown that cGAS is essential for cellular senescence to occur in response to spontaneous immortalization and DNA damaging agents (Glück et al., 2017; H. Yang, Wang, Ren, Chen, & Chen, 2017). The activation of cGAS stimulator of interferon genes (STING) induces the appearance of SASP (Glück et al., 2017). It is still unknown whether downstream effectors, TBK1 and IRF3, are involved in cGAS-induced senescence and the SASP (Figure 1).

Regulators of DNA damage play a role in maintaining cellular senescence. GATA4 was found to activate transcription factor NF-κB to initiate the SASP (C. Kang et al., 2015). DNA damage regulators, ataxia telangiectasia mutated (ATM) and ATM-Rad3 related (ATR), are required for the activation of GATA4, which is stabilized during cellular senescence (Kang et al., 2015; Figure 1). GATA4 is degraded by p62-mediated autophagy under normal conditions, but cellular senescence inhibits the degradation of GATA4 (C. Kang et al., 2015; Figure 1). Another study showed that CDK inhibitor p21 maintains the viability of DNA damage-induced senescent cells and inhibition of p21 can induce the apoptosis of senescent cells by activating ATM and NF-κB kinase (Yosef et al., 2017).

Long noncoding RNA (lncRNA) also plays a role in cellular senescence by regulating the DDR. For example, the lncRNA HOX transcript antisense RNA (HOTAIR) can induce cell senescence and influence the sensitivity of tumors to platinum (Özeş et al., 2016). HOTAIR expression is increased in recurrent platinum-resistant ovarian tumors, compared with primary ovarian tumors (Özeş et al., 2016). HOTAIR can induce the DDR after platinum treatment and activate NF-κB signaling by reducing the expression of NF-κB inhibitor Iκ-Bα (Özeş et al., 2016).

3.2 Oncogenic genes

Two decades ago, the oncogenic gene Ras was found to have the potential to induce cellular senescence by signaling for the accumulation of p53 and p16 (Serrano, Lin, McCurrach, Beach, & Lowe, 1997; Figure 1). Subsequently, Zhu, Woods, Mcmahon, and Bishop (1998) found that the oncogenic gene Raf also can regulate senescence through MAPK signaling (Figure 1). The molecular mechanism of Ras action and Ras-mediated p53 and p16 accumulation is still unknown. It has been known that Ras inactivates the BRCA1 DNA repair complex by separating BRCA1 from chromatin, which leads to the accumulation of DNA damage and arrest of the cell cycle (Z. Tu et al., 2011; Figure 1). In addition, BRG1 acts downstream of BRCA1 and is involved in SAHF formation and Ras-mediated senescence (Z. Tu, Zhuang, Yao, & Zhang, 2013; Figure 1). NORE1A, a member of the Ras-associated domain family, also plays a role in Ras-mediated cellular senescence by interacting with p53 and Rb (Donninger, Barnoud, & Clark, 2016; Figure 1). Besides, Ras induces the formation of the NORE1A and HIPK2 complex and enhances the interaction between HIPK2 and p53 (Figure 1). Furthermore, NORE1A promotes the prosenescence by promoting the acetylation of p53, and inhibiting the proapoptotic phosphorylation of p53 (Donninger et al., 2015).

4.1 Nuclear lamina

It is well established that telomere shortening due to cellular proliferation leads to senescence (Victorelli & Passos, 2017). A previous study showed that centromeres and telomeres are localized by nuclear lamina in senescent cells and telomeres are progressively aggregated during cellular senescence (Raz et al., 2008). In fact, telomere aggregation is associated with the changes in lamina spatial organization (Raz et al., 2008). Another study showed that there is a difference between the shape of nuclear lamina in healthy and senescent cells. As such, lamina morphology can be used as a phenotype for identification of senescent cells (Raz et al., 2008). As a result, some methods have been developed to quantify the intensity and curvature of nuclear lamina for classification of cell populations (Raz et al., 2008; Righolt, Hoff, Vermolen, Young, & Raz, 2011). Both lamin B1 and lamin B1 receptor (LBR) were found to be depleted during the tumor cell transition to senescence after gamma irradiation (Lukášová, Kovarˇík, Bacˇíková, Falk, & Kozubek, 2017). The reduction of lamin B1 and LBR leads to the release of heterochromatin from the lamina, resulting in changes in the chromatin architecture and gene expression (Lukášová et al., 2017). These results were is in line with those of a previous study, which showed that silencing lamin B1 can induce cellular senescence, which is dependent on activation of p53 and pRb (Shimi et al., 2011). However, another study showed that, although both lamin B1 overexpression and depletion can inhibit proliferation, only lamin B1 overexpression can induce cellular senescence, which can be hindered by inactivation of p53 and telomerase expression (Dreesen et al., 2013). The result of this study is consistent with an early study that found that OS increases the expression of lamin B1 by activating the p38 MAPK pathway which induces cellular senescence (Barascu et al., 2012). The molecular mechanisms by which lamin B1 silencing and overexpression induce cellular senescence are still elusive. Christopher J Hutchison et al. suggested that the difference in the lamin B1/lamin A ratio may lead to the change in nuclear lamina structure, which causes cellular senescence (Hutchison, 2012).

As some DNA viral infections can disrupt nuclear lamina, it would be interesting to study whether the viruses can induce cellular senescence because cell cycle arrest during cellular senescence is beneficial for viral replication (Cano-Monreal, Wylie, Cao, Tavis, & Morrison, 2009; C.-P. Lee et al., 2008; Sharma, Kamil, Coughlin, Reim, & Coen, 2014; W. Wei et al., 2016; X. Zhang et al., 2017).

4.2 Mitochondrial dysfunction

Increasing evidence has shown that mitochondria play an important role in cellular senescence. Mitochondrial DNA damage leads to mitochondrial dysfunction and increased ROS production (Passos et al., 2007). ROS produced by mitochondria may contribute to cellular senescence by inducing telomere dysfunction (Passos et al., 2007). Cigarette smoke (CS) can also induce senescence by damaging mitophagy and increasing the number of impaired mitochondria in the lung fibroblasts and small airway epithelial cells. CS interferes with the translocation of Parkin into mitochondria by inducing the accumulation of cytoplasmic p53 which interacts with Parkin. Mitochondria-specific antioxidants can delay cellular senescence (Ahmad et al., 2015). Cells lacking hFis1, a mitochondrial fission molecule, showed sustained mitochondrial elongation and a low rate of cell proliferation. In these cells, SA-β-gal activity is also elevated (S. Lee et al., 2007). Peroxisome-proliferator-activated receptor-γ coactivator 1α deficiency promotes vascular cellular senescence due to increased OS, reduced telomerase activity, and mitochondrial dysfunction (Xiong et al., 2013). Reduced expression of dynamin-related protein 1, a mitochondrial fission protein, in senescent endothelial cells (ECs) aggravates EC dysfunction due to impaired autophagic flux (Lin, Shen, Yan, & Gao, 2015).

Mitochondrial-dysfunction-associated senescence (MiDAS) triggers the SASP with the production of TNFα, IL-10, and CCL27. Lower NAD⁺/NADH ratios seen in MiDAS are caused by activation of the AMP-activated protein kinase (AMPK)-p53 pathway activation (Wiley et al., 2016). Caveolin-1 (Cav-1) deficiency also induces premature senescence by causing mitochondrial dysfunction. Cav-1 deficiency leads to decreased mitochondrial respiration and NAD/NADH ratios, reduced activity of oxidative phosphorylation complex, and inactivation of Sirtuin 2 (Sirt2). Cellular senescence caused by Cav-1 deficiency is dependent on p53-p21 pathway (D. M. Yu et al., 2017). Under OS, Cav-1 interacts with Sirt1 in the caveolar membrane and inhibits the activity of Sirt1. In agreement with this, a previous study found that ROS induces cellular senescence by increasing the acetylation of p53 and IL-6 production in wild type, but not Cav-1 knockout mouse embryonic fibroblasts (Volonte et al., 2015).

5.1 p53 signaling

The roles of p53 signaling in the cellular senescence have been studies for many years. Some proteins are found to promote p53-mediated cell senescence, such as Aurora B kinase (H.-J. Kim, Cho, Quan, & Kim, 2011), secretory phospholipase A(2) (H. J. Kim et al., 2009), and interferonγ (K. S. Kim, Kang, Seu, Baek, & Kim, 2009), whereas other proteins suppress p53-mediated cell senescence, such as Sirt2 (Langley et al., 2002), Hsp27 (O'Callaghan-Sunol, Gabai, & Sherman, 2007), and MAD2 (Lentini, Barra, Schillaci, & Di, 2012). p53 was found to act as a molecular switch regulating IGF-1-induced premature senescence (Tran et al., 2014). Short exposure of IGF-1 promotes cell proliferation, whereas prolonged exposure to IGF-1 induces cellular senescence (Tran et al., 2014). Prolonged exposure to IGF-1 inhibits Sirt1 deacetylase activity and increases p53 acetylation, resulting in p53 stabilization (Tran et al., 2014). Besides, Akt and p21 are required for the induction of cellular senescence downstream of p53 (Y. Y. Kim et al., 2017). Moreover, Foxp3 also accelerates senescence downstream of p53 signaling in fibroblasts and epithelial cancer cells. Foxp3 overexpression can increase the amount of p21 and ROS, which leads to the initiation of cellular senescence (J. E. Kim et al., 2017; Figure 2(a)).

5.2 p38 MAPK signaling

p38 signaling mediated by AMPK and transforming growth factor-beta-activated protein kinase (TAK) binding protein 1 drives the senescence of human T cells by inhibiting telomerase activity, T-cell proliferation, and expression of the T cell receptor (TCR) signalosome molecules (Lanna, Henson, Escors, & Akbar, 2014). CD45RA-expressing CD8⁺ T cells have elevated levels of both ROS and p38 MAPK. Inhibiting p38 MAPK signaling can induce mTOR-independent autophagy and enhance telomerase activity in the senescent CD8⁺ T cells (Henson et al., 2014; Henson, Macaulay, Riddell, Nunn, & Akbar, 2015). p38 MAPK activation also triggers the SASP, independent of the DDR (Freund, Patil, & Campisi, 2011). Blocking signaling by p38 and PD-1 can enhance cell proliferation (Henson et al., 2015). Similarly, p38 MAPK inhibitor was found to increase the number of human corneal ECs in vitro by suppressing cellular senescence, providing evidence for treatment of corneal endothelial dysfunction (Hongo, Okumura, Nakahara, Kay, & Koizumi, 2017). However, another study showed that AMPK activation can inhibit OS-induced senescence by restoring autophagic flux and NAD⁺ levels in the senescent fibroblast cells (Han et al., 2016; Figure 3). The difference in the results may be due to the different cell types used or the mechanisms of action of AMPK and p38 MAPK pathways.

5.3 NF-κB signaling

Many studies have correlated NF-κB signaling with cellular senescence. NF-κB signaling is implicated in indoxyl sulfate-induced cellular senescence via the ROS-NF-κB-p53 pathway (Shimizu et al., 2011). Indoxyl sulfate can induce the expression of NF-κB p65 and the phosphorylation of p65 at Ser-276. Inhibiting NF-κB signaling can suppress SA-β-gal activity as well as expression of p53, TGF-β, and α-smooth muscle actin (Shimizu et al., 2011). DNA damage can also activate NF-κB signaling, which stimulates the appearance of the SASP (Salminen et al., 2012). However, noncanonical NF-κB signaling was recently found to regulate the expression of the enhancer of zeste homolog 2 (EZH2) in melanoma and that inhibiting NF-κB signaling can reduce the expression of EZH2 and restore cellular senescence, resulting in the reduction of melanoma growth (De Donatis et al., 2016). EZH2 is directly involved in chromatin DNA methylation (Viré et al., 2006). Also, activation of EZH2 inhibits expression of p21, resulting in suppression of cellular senescence (Fan et al., 2011; Figure 4). More studies are still required to elucidate the potential downstream effectors of EZH2 during cellular senescence.

5.4 mTOR signaling

mTOR signaling involves cellular senescence through NF-κB, STAT3, and Akt signaling, as well as autophagy. mTOR inhibitor rapamycin was found to regulate cellular senescence by suppressing the translation of membrane-bound IL-1α which reduces the transcriptional activity of NF-κB. The authors proposed that rapamycin can ameliorate age-related diseases by inhibiting the senescence-related inflammation (Laberge et al., 2015). Besides, rapamycin can inhibit cellular senescence by upregulating the expression of Nrf2 and downregulating STAT3 pathway (Wang et al., 2017). Moreover, rapamycin can decelerate cellular senescence by restoring autophagy, as impairment of autophagy has been found to cause cellular senescence (Tai et al., 2016). In particular, autophagic flux is impaired during the development of OS-induced senescence. Rapamycin can restore the autophagic flux by restoring mitochondrial and lysosomal functions (Tai et al., 2016; Figure 5).

5.5 TGF-β signaling

Human pluripotent stem cells (hPSCs) have great potential for use in repairing ischemic tissue and treating of vascular disease. However, hPSCs display increased senescence under serum-free in vitro conditions, which affects vessel forming. TGF-β signaling is involved in the cellular senescence in hPSCs-derived ECs. Inhibiting TGF-β signaling can promote proliferation of hPSC-derived ECs and ameliorate cellular senescence (Bai, Gao, Hoyle, Cheng, & Wang, 2017). Mechanically, TGF-β may be involved in the formation of NF1/Smad complexes, which can repress the expression of adenine nucleotide translocate-2 (ANT2) by binding to the ANT2 promoter (Kretova et al., 2014). TGF-β-mediated suppression of ANT2 leads to increased ROS and DNA damage during the initiation of cellular senescence (Kretova et al., 2014). Integrin beta 3 (ITGB3) was found to regulate cellular senescence by activating TGF-β signaling in a paracrine and autocrine manner. In addition, polycomb protein CBX7 regulates the expression of ITGB3 during oncogene-induced senescence (Rapisarda et al., 2017). Furthermore, endogenous Notch1 may also mediate TGF-β-related cellular senescence (Kagawa et al., 2015; Figure 6).

5.6 Wnt signaling

The role of Wnt signaling in the cellular senescence is still not clear now. It has been reported that melanoma cells with high levels of Wnt5A expression demonstrate senescence-like features, such as increased expression of p21 and classic senescence-related molecules normally present under therapeutic stress induced by irradiation (Webster et al., 2015). However, the cells still have the potential to migrate and invade. Wnt5A knockdown reduces their invasion ability, indicating that Wnt5A drives melanoma invasion and promotes therapy resistance. The author proposed that the appearance of markers of senescence does not mean that the cells are truly senescent, but are exhibiting an adaptive stress response (Webster et al., 2015).

Studies on the molecular mechanisms of cellular senescence are just coming to light. Additional studies are needed to investigate the roles of other signaling pathways in the induction of cellular senescence, such as c-Jun N-terminal kinase (JNK)-signal transducer and activator of transcription 3 (STAT3), Wnt/β-catenin, epidermal growth factor receptor, toll-like receptor, cGAS-STING, and retinoic-acid-inducible gene I, and melanoma differentiation-associated antigen 5.

6.1 Cellular senescence is associated with chronic inflammation

Senescent cells can normally be recognized and cleared by the immune system (Sagiv & Krizhanovsky, 2013). However, senescent cells accumulate with age, and are involved in certain age-related diseases (Chang et al., 2016). It has been found that the SASP is required for the immune system to recognize senescent cells, which contributes to chronic inflammation and aging-associated diseases (Contrepois et al., 2017). Interestingly, p53 signaling seems to decrease with age, whereas NF-κB signaling is increased. It is proposed that p53 and NF-κB signaling pathways antagonize each other by a negative regulatory loop (Salminen & Kaarniranta, 2011). Also, p53 signaling initiates senescence, whereas NF-κB signaling induces the SASP (Salminen & Kaarniranta, 2011).

Moreover, DNA damage is implicated in senescence-associated inflammatory cytokine production (Rodier et al., 2009). DNA sensors, such as AIM2 and NLR family pyrin domain containing 3 (NLRP3), can recognize cytoplasmic DNA and drive the progression of inflammation. The significance of AIM2 and NLRP3 in senescence stills needs to be explored. Histone variant H2A.J was found to accumulate in senescent cells and can enhance the expression of inflammatory and immune response genes, including SASP-associated genes (Contrepois et al., 2017). It is unknown whether persistent DNA damage can enhance H2A.J accumulation and how H2A.J stimulates SASP-associated gene expression (Jurk et al., 2014).

7.1 Cellular senescence and obesity/diabetes

Increasing evidence supports the involvement of cellular senescence in metabolic disorders, such as obesity, insulin resistance, and diabetes. Senescent cells may contribute to obesity-associated inflammation. For example, the protein deleted in breast cancer-1 (DBC1) may regulate cellular senescence in obesity. Deletion of DBC1 can protect preadipocytes from cellular senescence and senescence-related inflammation (Escande et al., 2014). In addition, p38 MAPK signaling is associated with obesity and diabetes as it promotes tissue inflammation and injury. MAP kinase kinase (MKK)3-p38 MAPK signaling was found to be involved in the development of diabetic nephropathy. In diabetic MKK3-deficient mice, p38 MAPK signaling is downregulated, which is indicative of its protective role in the development of nephropathy (Lim et al., 2009). Besides, both high-mobility group protein AT-hook 2 (HMGA2) and p14, an inducer of senescence, are linked with the proliferation of stem and precursor cells of adipose tissue, cellular senescence, and an increased risk of type 2 diabetes (T2D). HMGA2 expression is higher, not only in obese patients, but also in the white adipose tissue of patients with T2D (Markowski et al., 2013). Increased expression of HMGA2 can lead to the increased expression of p14 and p53, followed by an increase in senescent cells (Markowski et al., 2013).

Cellular senescence is also involved in insulin resistance. Cheraghpour et al. studied the role of senescence in the insulin resistance and diabetes. They found that the expression of p53 is increased in fat and other peripheral tissues, whereas the expression of p16 is increased only in fat tissue. In addition, the serum levels of glucose, cholesterol, and triglycerides in the fat tissue of obese rats were found to be higher than in the controls (Cheraghpour, Shahbazi, Homayounfar, H, & Zand, 2015). The authors suggested that cellular senescence in fat tissue may be associated with the development of insulin resistance and age-related diseases (Cheraghpour et al., 2015).

Furthermore, suppressing senescence can increase glucose sensitivity following cold therapy in the elderly. Cold temperatures can be used to control diabesity by inducing progenitor cells to become beige adipocytes. However, the amount of cold-induced beige adipocytes decreases with age because of aging progenitor cells showing senescence-like phenotypes. Inhibiting senescence-associated pathways, like p38/MAPK-p16, can restore cold-induced beige adipocytes and subsequently increase glucose sensitivity (Berry et al., 2016).

7.2 Cellular senescence and liver disease

Senescence of hepatocytes is correlated with the stage of fibrosis (Aravinthan et al., 2013). Senescent cells are found in the livers of patients with NAFLD and cirrhotic, as well as the livers of mice fed a high-fat diet. An increase in the number of senescent cells can promote the development of NAFLD. The mitochondria in senescent cells cannot metabolize fatty acid normally, which results in the accumulation of hepatic fat and steatosis, and contributes to NAFLD. In addition, hepatocytes in patients with NAFLD express more p21 than hepatocytes of control patients (Ogrodnik et al., 2017).

NF-κB signaling plays a crucial role in a variety of liver diseases. For example, hepatic activation of IKK/NF-κB can induce liver fibrosis caused by macrophage mediated-chronic inflammation. Elimination of macrophages can reduce NF-κB- induced liver fibrosis (Sunami et al., 2012). NF-κB signaling is also implicated in trauma-associated damage to liver tissue. Inhibition of NF-κB signaling decreases the secretion of inflammatory cytokines, such as TNFα and IL-6, after liver trauma (W. Yang et al., 2012). A recent study implicates ERK, STAT3, and estrogen receptor α in hepatic steatosis (Choi et al., 2018). It is unknown whether these signaling pathways are involved in cellular senescence and the SASP which may play a role in the progression of liver disease.

7.3 Cellular senescence and lung disease

Senescent cells in the lung are associated with a variety of aging-associated lung diseases. Accumulation of senescent cells and the SASP in the lung of aged patients may lead to mild persistent inflammation, which results in tissue damage and increases the vulnerability to invasive pathogens. This increased vulnerability may be one reason why older people are susceptible to pneumonia (Yanagi et al., 2017). Besides, senescence of pulmonary artery smooth muscle cells (PA-SMCs) induced by telomerase shortening and OS contributes to the pulmonary hypertension (Noureddine et al., 2011). PA-SMCs can stimulate the proliferation and motility of normal PA-SMC cells and the production of paracrine factors (Noureddine et al., 2011). Osteopontin, an extracellular matrix protein secreted by PA-SMCs, was found to play a crucial role in the development of pulmonary hypertension. Osteopontin is highly expressed in the lungs of aged mice and patient with chronic pulmonary disease. Treatment of PA-SMCs from aged mice with an anti-osteopontin antibody can suppress the cell growth (Saker et al., 2016).

Senescence also contributes to the development of idiopathic pulmonary fibrosis (IPF). Alveolar type 2 (ATII) cells in the lungs of patients with IPF express high levels of p21, p16, and plasminogen activator inhibitor 1 (PAI-1). PAI-1 plays an important role in the activation of p53-p21-Rb pathway, ATII cell senescence, and bleomycin-induced fibrosis (Jiang et al., 2017). mTOR and ZEB1 may also be involved in the progression of pulmonary fibrosis. High levels of mTOR and ZEB1 expression are correlated with severe pulmonary fibrosis. STAT3 signaling also contributes to the progression of lung fibrosis. Inhibition of STAT3 reduces the expression of IL-6 and TGF-β-regulated gene expression, and decreases fibroblast-to-myofibroblast differentiation (Pedroza et al., 2015). Because pulmonary fibrosis is caused by chronic lung inflammation or environmental pollutants, it is possible that multiple cellular signaling pathways induce SASP that contributes to lung fibrosis (Park et al., 2014). Treatment with dasatinib and quercetin eliminates senescent cells to restore the pulmonary function and the physical health of aged mice. Senescent cell clearance also reduces the expression of proinflammatory factors, including Mcp1, Mmp12, and TGF-β, which coincides with the resolution of fibrosis (Schafer et al., 2017).

7.4 Cellular senescence and osteoarthritis

Aging and trauma are two etiological factors in the development of osteoarthritis (OA; Jeon et al., 2017). During aging, a subset of cells within the bone microenvironment become senescent (Farr et al., 2016). OS markedly accelerates telomere shortening and promotes senescence of chondrocytes (Brandl et al., 2011). Senescent cells have been shown to cause OA in the elderly, whereas the nonsenescent cells have little effect (Xu et al., 2017). Senescent cells are found in the articular cartilage and synovium after anterior cruciate ligament transaction (Jeon et al., 2017). Local clearance of senescent cells can ameliorate the development of posttraumatic osteoarthritis (Jeon et al., 2017). Xu et al. found that senescent cells transplanted into the knee region can cause symptoms of OA (Xu et al., 2017).

Sirt6/NF-κB signaling plays an important role in the regulation of senescence of chondrocytes (Duarte, 2015). Sirt6 overexpression can inhibit chondrocyte senescence and the NF-κB mediated inflammatory response, which limits the development of OA (Wu et al., 2015). In contrast, Sirt6 inhibition can lead to the increased expression of metalloproteinase 1 (MMP-1), MMP-13, and p16, exacerbated DNA damage and mitochondrial dysfunction, and subsequent premature senescence (Duarte, 2015). IL-6 was found to induce chondrocyte catabolism through STAT3 signaling. Importantly, blocking IL-6 or STAT3 can reduce destabilization of medial meniscus induced OA (Latourte et al., 2017). p38 MAPK is also involved in the senescence of chondrocytes. p38 inactivation can delay the onset of senescence, promote proliferation, and extend the life span of articular chondrocytes in vitro (S. Kang, Jung, Kim, & Shin, 2005).

7.5 Cellular senescence and tumorigenesis

Senescent tumor cells do not proliferate, and cellular senescence may be a strategy to inhibit tumor growth. For example, TGF-β plays an important role in suppressing the progression of head and neck squamous cell carcinoma (HNSCC) progression by inducing cellular senescence and inhibiting inflammation. Double knockout of TGF-β/pten in mice promotes cell proliferation, decreases apoptosis, and increases the expression of CCND1 in the basal layer of head and neck epithelia, whereas pten only knockout in mice has only a few lesions that progress to HNSCC (Bian et al., 2012). p38 MAPK also plays an important role in cellular senescence in some tumors. p38 MAPK/ROS pathway is involved in physalin A-induced arrest of the cell cycle at the G2/M phase. Physalin A, a Chinese herbal extract, is used to treat tumors and other diseases. Inhibition of p38 can reduce the production of ROS and the proportion of cells in the G2/M phase (N. Kang et al., 2016). It is uncertain whether physalin A can induce cellular senescence by activating the p38 MAPK/ROS pathway. A previous study showed that IL-4 not only induces cell cycle arrest in the G1 phase, but also cellular senescence mediated by STAT6 or p38 MAPK in renal cell carcinoma (H. D. Kim et al., 2013).

Despite the beneficial role of senescence in inhibiting tumor cell proliferation, SASP can promote tumorigenesis and confer therapy resistance. Cytokines or other proteins secreted from senescent cells can promote the proliferation and migration of tumor cells. For example, IL-6 secreted from senescent mesenchymal stem cells has been shown to enhance the proliferation and migration of breast cancer cells by activating of STAT3 in vitro and in vivo (Di et al., 2014). In addition, IL-6 can stimulate the epithelial-mesenchymal transition, promote the migration and proliferation, and inhibit apoptosis of human intrahepatic biliary epithelial cells (HIBECs), delaying of cellular senescence (Li et al., 2018). Moreover, chemokines, like CCL2, which is induced during cellular senescence, play an important role in shaping the tumor-promoting microenvironment (Eggert, Ji, Zender, Wang, & Greten, 2015). For instance, senescent hepatocytes can promote hepatocellular carcinoma cell growth by signaling for the accumulation of immunosuppressive CD11b⁺Gr1⁺ myeloid cells which can inhibit T-cell proliferation (Eggert et al., 2015). Importantly, CCL2 plays a crucial role in the accumulation of myeloid cells and accelerates tumor growth (Eggert et al., 2015). Unexpectedly, the expression of p53-dependent CCL2 in senescent tumor cells is needed to recruit natural killer (NK) cells to eliminate senescent tumor cells with induced expression of NKG2D (Iannello, Thompson, Ardolino, Lowe, & Raulet, 2013). Furthermore, senescent tumor cells have a high potential for invasion. For example, senescent tumor cells in papillary thyroid carcinoma (PTC) have high invasion potential through the SASP and are involved in lymph node metastasis. Mechanically, upregulated CXCL12/CXCR4 signaling plays a crucial role in the invasion of PTC cells by graded expression near the invasion border (Y. H. Kim et al., 2017).

It has been found that exosome-like small extracellular vesicles (sEVs) produced by senescent cells promote the tumor cell growth. sEVs-associated EphA2 can bind to the ephrin-A1 in tumor cells and enhance cell proliferation. Further study showed that ROS was found to be involved in sEV sorting of EphA2 through upregulating EphA2 phosphorylation (Takasugi et al., 2017). It is unknown whether senescence-associated sEVs have other contents, such as miRNA, DNA, and proteins, and whether the sEVs play a role in normal and pathological situations (Kadota et al., 2017; Takasugi et al., 2017). It should be noted that, in the human genome, there are ten EphA receptors and six EphB receptors, so it would be interesting to explore whether other ephrin/Eph forward or reverse signaling pathways are involved in cellular senescence (W. Wei, Wang, & Ji, 2017).

7.6 Cellular senescence and pathogen infection

It has been known that bacterial and viral infection can induce cellular senescence of host cells. By now, the roles of bacterial and viral infection in inducing of cellular senescence are still largely unknown.

Human cytomegalovirus (HCMV) infection causes cellular senescence, which might promote chronic inflammation (Wolf, Weinberger, & Grubeck-Loebenstein, 2012). HCMV induces gene expression related to the innate and adaptive immune responses in fibroblasts, some of which are also induced during cellular senescence, such as IL-6 (Wolf et al., 2012). Latent infection of HCMV is linked with age-associated immune dysfunction of T cells, which may contribute to immunosenescence, aging of the immune system, in the elderly (W. Tu & Rao, 2016). Besides, hepatitis B virus (HBV) infection can cause cellular senescence and may require HBcAg for the initiation of cellular senescence (Tachtatzis et al., 2015). In some chronically infected HBV patients, p21 expression is increased and correlates with liver fibrosis and telomere length is reduced. It should be noted that, although HBV can cause cellular senescence, biologically young hepatocytes are the preferred cells for viral replication (Tachtatzis et al., 2015). In addition, hepatitis C virus (HCV) monoinfection and human immunodeficiency virus (HIV)/HCV coinfection can accelerate T-cell immune senescence during early infection in injection drug users, which could contribute to premature morbidity and mortality (Grady, Nanlohy, &...Baarle, 2016). Human papilloma virus (HPV) E2 can induce senescence through the pRb- and p21-dependent pathways (Wells, 2000). The HPV E6 F47R mutant can also induce senescence, but by the p53-dependent pathway (Ristriani, Fournane, Orfanoudakis, Travé, & Masson, 2009). HPV E6/E7 protein can inactivate Notch1 and inhibit Notch1-mediated senescence, which promotes colony formation and xenograft tumor growth (Kagawa et al., 2015). Dengue virus (DENV) infection was found to induce cellular senescence in human umbilical vein ECs, but senescent cells do not support viral infection (Abubakar, Shu, Johari, & Wong, 2014). Another study showed that senescent cells can increase the monolayer permeability of ECs, suggesting that DENV infection may increase vascular permeability, which leads to the plasma leakage observed in dengue hemorrhagic fever and dengue shock syndrome (Krouwer, Hekking, Langelaarmakkinje, Reganklapisz, & Post, 2012).

Bacterial LPS is a potent inducer of inflammation. Interestingly, LPS is also implicated in the induction of cellular senescence. For example, LPS was found to induce cellular senescence in lung alveolar epithelial cells, in a hydrogen-peroxide-dependent manner (C. O. Kim et al., 2012). Another study showed that repeated LPS stimulation can lead to microglia senescence, which may result in failure to respond to detrimental stimuli and subsequent neurodegeneration (H.-M. Yu et al., 2012). In addition, bacterial toxins are also involved in cellular senescence. For example, the genotoxin colibactin, encoded by genomic island of Escherichia coli, can induce cellular senescence and promote the growth of tumor cells and colon cancer (Cougnoux et al., 2014; Secher, Sambalouaka, Oswald, & Nougayrède, 2013).

Infection by other microbes, such as Escherichia coli (Cuevas-Ramos et al., 2010), Helicobacter pylori (Toller et al., 2011), respiratory syncytial virus (Gibbs, Ornoff, Igo, & Imani, 2009), and transmissible gastroenteritis virus (Ding et al., 2013) can also cause DNA damage and cell cycle arrest. The ability of these pathogens to induce cellular senescence and the corresponding molecular mechanisms are worth studying in the future.

8.1 Inducing senescent cell apoptosis

Some peptides, chemicals, and small molecules have been developed to induce senescent cell apoptosis by activating apoptotic signaling or inhibiting the activity of antiapoptotic proteins. For example, FOXO4 is required to maintain cell viability by interacting with p53. A peptide that perturbs the FOXO4-p53 interaction can induce cell apoptosis of senescent cells, and restore tissue homeostasis (Baar et al., 2017). Agmatine, a chemical substance, inhibits neuronal senescence by activating p53 and inhibiting p21. Agmatine can also reduce SA-β-gal activity and the expression of IL-6, TNFα, and CCL2 in neuronal cells under high glucose condition (Song et al., 2016). In addition, inhibition of BCL-W and BCL-XL by siRNAs or the drug ABT-737 induces apoptosis of senescent cells in the epidermis and increases the proliferation of hair-follicle stem cells (Yosef et al., 2016). Similarly, ABT263, an inhibitor of BCL-2 and BCL-xL, can selectively eliminate stem cells in culture by inducing cell apoptosis, which is beneficial for rejuvenation of aged tissue stem cells (Chang et al., 2016).

8.2 Inhibiting cellular senescence by suppressing oxidative stress

Natural chemicals with antioxidant potential can suppress OS-induced senescence. For example, curcumin, a natural phenol with antioxidant and anti-inflammatory activities, can attenuate hydrogen peroxide (H₂O₂)-induced premature EC senescence by activating Sirt1 (Sun, Hu, Hu, Xu, & Jiang, 2015). Some molecules can delay senescence by directly suppressing ROS production. For example, nicotinamide, an amide derivative of vitamin B3, can attenuate senescence by reducing the level of ROS (Kwak, Ham, Kim, & Hwang, 2015). A recent study showed that 27-hydroxycholesterol (27HC) was found to promote the accumulation of ROS and then activate the IL-6/STAT3 pathway, whereas ROS scavenger N-acetylcysteine can inhibit ROS production and activation of the IL-6/STAT3 pathway. Notably, blocking ROS production or inhibiting the IL-6/STAT3 pathway can attenuate 27HC-induced cellular senescence (Liu et al., 2017).

8.3 Inhibiting cellular senescence-associated pathways

Given that several signaling pathways play important roles during senescence, it is possible to inhibit senescence by suppressing these pathways. A study found that therapy can induce tumor cell senescence and Bcl2-associated athanogene 3 (Bag3) is increased in this process (Pasillas et al., 2015). Moreover, Bag3 interacts with the major vault protein (MVP), which upregulates the ERK1/2 signaling, resulting in resistance to chemotherapy (Pasillas et al., 2015). Importantly, Bag3 or MVP knockdown weakens the activation of ERK1/2 and promotes apoptosis of therapy-induced senescent cells (Pasillas et al., 2015). A similar study found that inhibition of the MEK/ERK pathway can promote the elimination of Ras-transformed senescent cells as the cells cannot form the autophagosomes necessary for the clearance of damaged mitochondria, resulting in apoptosis (Kochetkova et al., 2017).

8.4 Immune regulation of cellular senescence

A variety of immune cells, including macrophages, NK cells, and T cells, can be recruited and activated by SASP, including chemokines, cytokines, adhesion molecules. For example, F4/80⁺ macrophages are involved in the elimination of senescent cells from the uterus of postpartum mice and regulation of the senescence-associated inflammatory microenvironment increases the success rate of pregnancy (Egashira et al., 2017). Moreover, NK cells also eliminate senescent tumor cells, in an NKG2D expression-dependent manner. p53 restoration can promote the secretion of various chemokines that can recruit NK cells (Iannello et al., 2013). Furthermore, type I interferons (IFNs) can amplify DNA-damage responses and promote senescence. It is known that IFNs induce senescence and suppress the development of melanoma (Katlinskaya et al., 2016). Interestingly, IFNs produced by senescent cells can increase the expression of MIC-A and ULBP2 ligands in the NK cells and promote the recognition and clearance of senescent cells by NK cells (Katlinskaya, Carbone, Yu, & Fuchs, 2015).

8.5 Using Oncolytic viruses to destroy senescent cells

Measles vaccine virus (MeV) has been used for lysis of therapy-induced senescent tumor cells. One study found that MeV can efficiently replicate and produce infectious progeny virus particles (Weiland et al., 2014). Another study found that MeV infection can induce cellular senescence in normal and tumor cells (Chuprin et al., 2013). Oncolytic adenovirus, in combination with doxorubicin and ifosfamide, can efficiently eliminate soft-tissue sarcomas cells (Siurala et al., 2015). These viruses could be used to lyse therapy-induced senescent tumor cells.

8.6 Exercise

Nutrient excess contributes significantly to increased expression of senescence-associated proteins, including p16, p53, p21 and increased activity of SA-β-gal (Schafer et al., 2016). Exercise can reduce the expression of the SASP-associated genes and prevent the accumulation of senescent cells caused by the fast-food diet (Schafer et al., 2016). Mechanically, exercise could reduce metabolic and replicative stresses in adipose tissue and limit the transition into a senescent state. Exercise could also trigger protective responses against cellular senescence in adipose tissue cells. Moreover, exercise may promote the clearance of senescent cells by the immune system (Schafer et al., 2016).