Abstract
Cellular senescence is an irreversible form of cell cycle arrest that can be induced by persistent DNA damage, and is well known to function as an important tumour suppression mechanism. Cellular senescence is detected in aged organisms; thus, it is also recognized as a hallmark of organismal ageing. Unlike apoptotic cells, senescent cells can survive for long periods of time. Recently, it has been shown that the late stage of senescent cells are capable of expressing a variety of secreted proteins such as cytokines, chemokines and proteases, and this condition is now known as senescence-associated secretory phenotype (SASP). These secreted factors are involved in myriad of physiological functions including tissue repair and clearance of damaged cells. Alternatively, these factors may promote detrimental effects, such as chronic inflammation or cancer progression, should the SASP persist. Recent scientific advances have indicated that innate immune responses, particularly involving the cGAS–STING pathway, trigger SASP induction. Therefore, developing a strategy to regulate SASP may provide scientific insights for the management of age-associated diseases and the implementation of healthy ageing in the future.
Cellular senescence is an irreversible cell proliferation arrest that can be induced by a variety of oncogenic stimuli, such as persistent DNA damage. Therefore, cellular senescence is believed to function as an important tumour suppression mechanism (1, 2). Cellular senescence is also recognized as a hallmark of organismal ageing, since senescent cells are known to accumulate with age (3–5). Recent studies have shown that the late stage of senescent cells express a variety of secreted proteins including inflammatory cytokines, chemokines and growth factors, which is a phenomenon called senescence-associated secretory phenotype (SASP) (6, 7). Cellular senescence is, therefore, thought to be involved in several age-related inflammatory diseases (2). Moreover recently, cytosolic DNA fragments generated in senescent cells have been shown to stimulate the cyclic GMP–AMP synthase (cGAS)–stimulator of interferon genes (STING) pathway, a recently discovered component of the innate DNA sensor machinery (8–12). Interferon regulatory factor 3 becomes phosphorylated by cGAS–STING stimulation, which thereby activates type 1 interferon (IFN) signalling, leading to the expression of a series of senescence-associated secretory proteins in SASP, a distinct characteristic of late stage senescent cells (13). In this review I introduce recent findings on the mechanisms responsible for the induction of SASP mediated by cGAS–STING and other innate immune responses in vivo by gut microbial components.
Cellular Senescence is Induced by DNA Damage-mediated Signals
Human somatic cells are capable of undergoing only a finite number of cell divisions. This phenomenon of cellular senescence is identified when cell proliferation becomes irreversibly arrested in vitro by repeated passaging (14). Since the senescent cells can no longer divide even with the proliferation-driving forces such as high percentage of serum in culture, this type of irreversible cell proliferation arrest is called ‘replicative senescence’. Replicative senescence is thought to be induced by telomere shortening, which has been attributed to incomplete replication at the chromosomal ends where telomeric repetitive sequences reside. The shortening of terminal telomeric repeats occurs with each cell division during chromosomal DNA replication. Telomeres eventually shorten progressively with successive cell division, and the resulting telomere erosion promotes DNA damage responsive signals, thereby causing irreversible cell proliferation arrest (14, 15). Conversely, cancer cells can proliferate infinitely without undergoing senescence cell cycle arrest. This is thought to be due to the high telomerase activity. Alternatively, some cancer cells have telomeres that contain end-to-end fusion, which prevents telomere erosion (7, 15, 16).
However, it has become apparent that cellular senescence can be rapidly induced without telomere shortening when cells are exposed to persistent DNA damage-inducing stresses such as oxidative stress and excessive replication stress caused by activation of oncogenes (17–21). This suggests that cellular senescence functions as an important, fail-safe tumour suppression mechanism. Cellular senescence is, therefore, an irreversible cell proliferation arrest that is induced by strong cell cycle checkpoint mechanisms when cells are exposed to persistent DNA damage (7, 8). Moreover, p16 and p21 cyclin dependent kinase inhibitors, play crucial roles in inducing cellular senescence (22). The signals induced by persistent DNA damage stabilize the transcription factor, p53, and induce expression of its target gene, p21. In the presence of persistent DNA damage signal, expression of p16 is also induced by the ETS family of transcription factor through a pathway that is independent of p53 (23). Specifically, p16 and p21 have been shown to cooperatively activate the retinoblastoma protein, the active dephosphorylated form of which is known to function as a strong cell cycle stopper, which in turn facilitates cellular senescence (24, 25). Moreover, p53 and p16 are known to be inactivated or dysfunctional in many types of human cancers (26, 27), suggesting that the pathways responsible for induction of cellular senescence are crucial for the suppression of human cancer.
Senescence-associated Secretory Phenotype
Moreover, recent studies have shown that senescent cells express a myriad of secretory proteins, and this is a phenomenon called SASP (6, 7, 28, 29). These cells secrete a wide range of proteins; however, a large portion of them appear to be regulated by NF-ĸB target genes, which encode a series of inflammatory cytokines, chemokines, growth factors, extracellular matrix metalloproteinases and cyclooxygenase (6, 7, 28, 29). It has been shown that the factors secreted from senescent cells in the state of SASP (so-called SASP factors) induce senescence of surrounding cells in a paracrine manner, and of themselves in an autocrine manner, thereby autonomously enhancing cellular senescence (30, 31). It has also been suggested that specific SASP factors, including chemokines, recruit immune cells to clear the senescent cells, which is called ‘senescence-surveillance’ (32, 33). This kind of SASP as a senescence-surveillance system is suggested to function as a mechanism of senescent (precancerous) cell (34) depletion. However, in some cases, SASP in cancer-associated fibroblasts (CAFs) in the tumour-microenvironment of advanced tumour tissues, could function to promote cancer progression (29, 35, 36). Therefore, it is important to consider which cells exhibit SASP in each biological context (Fig. 1). Indeed, we showed that SASP occurs in hepatic stellate cells (HSCs, myofibroblasts in the hepatic stroma) in advanced liver tumour tissues, and the factors released by these senescent HSCs suppressed anti-tumour immunity and contributed to the progression of obesity-associated liver cancer (35, 36). Our findings suggest that, in the advanced stages of cancer, SASP in stromal cells may confer adverse and deleterious effects in the tumour microenvironment (29).
The effect of SASP is alternative depending on each biological context. The SASP as senescence-surveillance plays a role as a mechanism of depleting senescent (precancerous) cells (left). However, in some reported cases, SASP in CAFs in the tumour-microenvironment of advanced tumour tissues could promote cancer progression (right). Therefore, it is important to consider which cells exhibit SASP in each biological context.
In addition, the study performed by Campisi et al. demonstrated that platelet-derived growth factor-AA, described as one of the SASP factors, was generated in skin fibroblasts during skin tissue repair. This SASP temporarily appeared subcutaneously during the tissue repair process (37). The senescent fibroblasts in the skin in wound healing have been suggested to secrete inflammatory cytokines and chemokines, which facilitate chemotaxis of immune cells, thereby contributing to the removal of dead cells in the injured tissue (3, 38). Simultaneously, senescent fibroblasts produce growth factors, which are also categorized as SASP factors, to promote the proliferation of skin precursor cells needed for the regeneration of skin tissues. Once tissue repair is complete, senescent cells undergoing SASP are cleared by immune cells. Similarly, in a study involving liver injury, HSCs were shown to produce SASP factors, contributing liver tissue regeneration, and then were cleared away by immune cells to suppress the excessive collagen production (39). Thus, the effects of SASP, to be beneficial or deleterious, seem dependent on whether senescent cells are promptly cleared away. If senescent cells survive for a long period of time, adverse side effects may result, such as chronic inflammation and the formation of a cancer promoting microenvironment (36).
SASP is Induced by Innate Immune Responses via the DNA Sensor, cGAS–STING
As mentioned above, persistent DNA damage triggers cellular senescence. Moreover, the positive feedback mechanisms further increase the production of reactive oxygen species in the senescent cells and stabilize DNA damage-induced cellular senescence (40). Therefore, DNA damage becomes more accumulated as cellular senescence persists. Recent reports have shown that the DNA fragments and micronuclei generated during cellular senescence could act as ligands for cGAS–STING, a cytosolic DNA sensor, stimulating senescent cells to activate type 1 IFN-associated cytokine production (9–12, 33, 41). Originally, cGAS was identified as an innate immune receptor that recognizes viral and bacterial DNA (42). cGAS, a cyclic GMP–AMP synthase, is known to catalyses the synthesis of cyclic dinucleotides composed of GMP and AMP (43, 44). These cyclic dinucleotides are then recognized by STING, a protein localized in the endoplasmic reticulum. Activated STING stimulates the type I IFN signalling pathway.
Specific mechanisms for cGAS stimulation in senescent cells have recently been emerged. The expression of lamin B1, an intermediate filament meshwork that lies beneath the inner nuclear membrane, decreases in the late stages of cellular senescence causing breakdown of the nuclear envelope. This breakdown and the resulting cytosolic releasing chromatin fragments stimulate cGAS, thereby activating the type I IFN signalling pathway (11). Furthermore, DNA damage-associated micronuclei increased in senescent cells are reported to be recognized by cGAS and induce stimulation of the cGAS–STING pathway (9, 10, 33). The micronuclei and chromosomal fragments are thought to be generated due to inhibition of cytokinesis in the senescent cells (10, 12, 40). These DNA fragments remain in the cytoplasm where they may act as ligands for innate immune receptors, including DNA sensors such as cGAS, thereby initiating inflammatory signals. However, cells attempt to control this inflammatory response by enclosing the small cytoplasmic DNA fragments within exosomes and excreting them from the cell (41). When exosome secretion is inhibited, DNA fragments are shown to be further accumulated in the cytoplasm, and provoke excessive activation of DNA sensors that induce further innate immune responses, primarily through the type 1 IFN pathway, suggesting that the secretion of exosome enclosing DNA fragments prevents the accumulation of cytoplasmic DNA, contributing to the maintenance of cellular homeostasis (41).
Very recently, it has been reported that cytosolic complementary DNA (cDNA) produced by a reverse transcription from retrotransposons such as long interspersed nuclear elements 1 (LINE-1), induces SASP through the cGAS–STING pathway (45). Transposable elements (TEs) including transposons and retrotransposons, which are made up of repetitive DNA sequences, have the capacity to insert into new genomic locations. The transposon DNA sequence can be digested and jump to a new location in the genome. Alternatively, the retrotransposons are transcribed to mRNA, and then reverse transcribed to cDNA, which is able to transfer to new locations in the genome. Particularly, the retrotransposon, LINE-1, is known to exhibit a high transfer activity. If such TEs are inserted into a gene encoding region in the genome, unrepairable DNA breaks may occur. Therefore, TEs are known to be normally silenced by small RNAs (46, 47). In the report, cDNA fragments, from which LINE-1-derived mRNA was reverse-transcribed, were found to be significantly increased in the cytoplasm of senescent cells. The LINE-1 cDNA was reported to activate the cGAS–STING pathway and contributed to the induction of SASP (45). It has also been shown that an increase in cytoplasmic cDNA derived from the retrotransposons cause chronic inflammation by over-production of IFN in aged individuals. Moreover, since inhibition of retrotransposon reverse transcriptase (RT) activity has been shown to reduce inflammation, retrotransposon RT inhibitors could be a promising candidate for age-associated inflammation (45). Furthermore, a recent study using SIRT6 deficient mice are known to express high levels of active LINE-1 and LINE-1-derived cDNA, accompanying chronic inflammation and premature ageing. This report also demonstrated that the administration of a retrotransposon RT inhibitor alleviated the chronic inflammation and prolonged lifespans of the SIRT6 knockout mice (48). These results suggest that a persistent inflammatory signal, induced by retrotransposon-derived cDNA, was responsible, at least in part, for the premature ageing (48). Together these studies demonstrate that undegraded small DNA fragments in the cytoplasm trigger activation of the cGAS–STING pathway, thereby inducing persistent SASP in senescent cells.
Given that the abnormally accumulated cytoplasmic DNA fragments in senescent cells could trigger innate immune responses, how they are normally scavenged? Our colleagues in Eiji Hara’s group found that DNase2 and TREX1 (DNase3) were responsible for degrading cytoplasmic DNA, yet they were both suppressed in senescent cells due to inactivation of E2F transcription factor, thereby causing the accumulation of DNA fragments in the cytoplasm (12). This accumulation of cytosolic DNA fragments triggers the cGAS–STING DNA sensor to stimulate IFN-β and concomitantly activate the transcription factor, NF-ĸB, which has been identified as a mechanism for SASP induction (Fig. 2). In addition, this mechanism has also been shown to play an important role in SASP induction within HSCs in vivo, thereby promoting development of obesity-associated hepatocellular carcinoma (HCC) (12).
Downregulation of DNases promotes the accumulation of cytosolic DNA fragments and SASP induction. DNase2 and TREX1 (DNase3) which are responsible for degrading cytoplasmic DNA, are both suppressed in senescent cells due to inactivation of E2F transcription factor, thereby causing the accumulation of DNA fragments in the cytoplasm. The accumulated cytosolic DNA fragments bind to cGAS, which in turn create cGAMP that triggers type 1 IFN signal activation through STING activation. This is now thought to be one of the important mechanisms of SASP.
Gut Microbial Components Trigger Induction of SASP
Given that SASP may be triggered by innate immune responses, other extrinsic ligands for innate immune receptors could be associated with the induction of SASP. Recently, we found that mice receiving a high-fat diet (HFD) that had been treated with 7,12-dimethylbenzathracene as neonates, developed HCC with very high frequency (35). When we investigated the histology of HCC, we found that HSCs underwent cellular senescence, and showed SASP, thereby creating a tumour promoting microenvironment (35, 36). To elucidate the mechanism responsible for causing senescence in HSCs, metabolome analysis was performed. Results showed that deoxycholic acid (DCA), which circulates between the liver and intestine via the enterohepatic circulation, was identified as an inducer of senescence and SASP in HSCs within the liver tumour region in obese mice.
Interestingly, mice treated with difructose anhydride III, an oligosaccharide that suppresses the proliferation of DCA-producing bacteria, or with ursodeoxycholic acid, which promotes the excretion of bile acids, showed reduced levels of serum DCA. Furthermore, the incidence of liver cancer and of senescence phenotypes in HSCs was significantly reduced in these mice. Conversely, when DCA was administered orally in combination with antibiotics to reduce the gut microbiota, the development of liver cancer, which was suppressed by antibiotic treatment, was markedly restored by the administration of DCA. This was also accompanied by the appearance of cellular senescence and a SASP in HSCs within the tumour region, suggesting that DCA is indeed a risk of HFD-associated HCC.
However, mice that were administered DCA with normal diet, did not develop HCC following 30 weeks observation, the time point when HFD-fed mice developed HCC. Rather these normal diet plus DCA fed mice required several months for full SASP induction and HCC development to occur (36). It is thus proposed that additional factors are required for the promotion of liver cancer. We thought that signalling from lipoteichoic acid (LTA) and TLR2 receptor that recognizes LTA may also act as promoters of liver cancer, since the number of Gram-positive gut microbiota was increased dramatically within HFD-fed mice. Accordingly, due to the presence of leaky gut, LTA was also highly abundant in the liver tumour region. Since relatively high levels of DCA and LTA were both thought to be present in the tumour microenvironment, we examined the effect of DCA and LTA on HSCs. Interestingly, DCA and LTA were found to synergistically upregulate the expression of many SASP factors and COX-2 in the DCA-induced senescent HSCs. More interestingly, the expression of TLR2 was significantly upregulated following treatment of DCA and LTA. Thus, the concurrent presence of DCA and LTA could activate the positive feedback loop for LTA–TLR2 signalling, thereby causing further activation of the pathway through TLR2 (36).
With the high expression of COX-2, prostaglandins (PGs), along with other COX-2-mediated metabolites of arachidonic acid, were found to be significantly over-produced in the tumour area. Importantly, we observed that the over-produced PGE2 suppressed anti-tumour immunity via the EP4 receptor on immune cells (36), which is upregulated in the tumour, and may thus promote progression of HFD-induced liver cancer in our mouse model. Conversely, administration of an EP4 antagonist caused activation of anti-tumour immunity, thereby attenuating tumour formation (Fig. 3). Furthermore, high expression of COX-2 and over-production of PGE2 were observed in some cases of human non-alcoholic steatohepatitis-associated liver cancer characterized by a high levels of accumulated lipids in the tumour region, suggesting that a similar mechanism observed in mice, may be involved in humans as well (36).
SASP induction by gut microbial component promote obesity-associated HCC progression. Obesity-induced Gram-positive gut microbiota produce DCA in the intestine, and enterohepatic circulation of DCA promotes the cellular senescence of the HSCs. In addition to DCA, LTA, a cell wall component of the Gram-positive bacteria is also transferred to the liver, and provokes the TLR2-mediated signalling to express COX2 and SASP factors. The produced PGs particularly PGE2 together with SASP factors could promote obesity-associated HCC. DCA, deoxycholic acid; COX-2, cyclooxygenase 2; LTA, lipoteichoic acid. This figure was modified from Figure 7B of our original paper, reference No. 36.
Conclusions and Future Directions
As mentioned above, the SASP, which is promoted by DNA fragment-driven stimulation of the cGAS–STING pathway, was recently shown to be induced in late stages of senescent cells, promoting deleterious side effects such as chronic inflammation and cancer, if the senescent cells should be allowed to persist. In addition to DNA fragments in the cytoplasm, we also suggested that the abundance of LTA, a cell wall component in Gram-positive gut microbiota, which was increased in HFD-induced obesity, was transported to the liver where it triggered SASP in HSCs expressing TLR2, an innate receptor of LTA (36). Moreover, it has been increasingly recognized that ‘senolysis’, the removal of senescent cells, is effective in preventing age-related diseases because this also inhibits the development of SASP (49, 50). Clarifying the specific molecular mechanisms involved in SASP induction, and developing strategies for regulating SASP, will provide important insights for alleviation of ageing-associated inflammatory diseases and cancer to realize healthy longevity.
Funding
This work was supported in part by JSPS KAKENHI (19H04002 to N.O.) and AMED, Japan Agency for Medical Research and Development, Japan (JP19gm1010009 and 19ck0106260 to N.O.).
Conflict of Interest
None declared.
