Senescence-Associated MicroRNAs

Abstract

Senescent cells arise as a consequence of cellular damage and can have either a detrimental or advantageous impact on tissues and organs depending on the specific cell type and metabolic state. As senescent cells accumulate in tissues with advancing age, they have been implicated in many age-related declines and diseases. The major facets of senescence include two pathways responsible for establishing and maintaining a senescence program, p53/CDKN1A(p21) and CDKN2A(p16)/RB, as well as the senescence-associated secretory phenotype. Numerous MicroRNAs influence senescence by modulating the abundance of key senescence regulatory proteins, generally by lowering the stability and/or translation of mRNAs that encode such factors. Accordingly, understanding the molecular mechanisms by which MicroRNAs influence senescence will enable diagnostic and therapeutic opportunities directed at senescent cells. Here, we review senescence-associated (SA)-MicroRNAs and discuss their implications in senescence-relevant pathologies.

1. INTRODUCTION

1.1. Senescence

Senescence was first described by Leonard Hayflick 50 years ago as a state in which normal cells divide for a finite number of times before they cease proliferation indefinitely (Hayflick, 1965; Hayflick and Moorhead, 1961), although they remain viable and metabolically active (Gey and Seeger, 2013). Senescence has been studied extensively using primary cells cultured ex vivo until they stop dividing; this process is known as replicative senescence (Hayflick, 1992). Additionally, senescence can be achieved by exposure of primary cells to a range of sublethal damaging agents; this process is known as premature senescence. Knowledge from these two processes has been recently complemented by studies on senescence in vivo and its impact in physiology and pathology.

1.1.1. Replicative and Premature Senescence

Replicative senescence is achieved in primary cells isolated from tissues and placed in culture. Over time, their population doubling time (PDL) gradually increases until they are terminally arrested and cease to proliferate. At the same time, the telomeres that protect chromosome ends and permit DNA polymerase to complete replication become progressively shorter and eventually trigger a DNA damage response (DDR) resulting in growth cessation and senescence (Kuilman et al., 2010). The short, unprotected telomeres activate several downstream effectors including the kinases ataxia telangiectasia mutated (ATM) and ATM-related (ATR), the checkpoint kinases CHK1 and CHK2, and the tumor suppressor and transcription factor p53 (Ben-Porath and Weinberg, 2005; Herbig et al., 2004). In turn, p53 transcriptionally induces expression of the cyclin-dependent kinase (CDK) inhibitor CDKN1A/p21 and the alternate open reading frame protein ARF, enabling the p53/p21 axis that inhibits replication (Ben-Porath and Weinberg, 2005; Campisi, 2005). Replicative senescence is further established by the CDKN2A(p16)/RB axis, comprising of the retinoblastoma protein (RB) activated through the CDK inhibitor p16 (CDKN2A/INK4A) and the related proteins p15/INK4B, p18/INK4C, and p19/INK4D (Ben-Porath and Weinberg, 2005; Campisi, 2005).

Premature senescence, also known as stress-inducible senescence, can be rapidly achieved by exposing cells to stresses such as radiation, oxidants, toxins, chemotherapy, oncoprotein activation, or through tumor-suppressor inactivation. These harmful stimuli trigger senescence by activating stress signals including DDR; the ensuing growth arrest by CDK inhibitors and heterochromatin changes occur without apparent loss of telomere function (Kuilman et al., 2010). Premature senescence can be triggered by activation of oncoproteins such as KRAS(V12) and BRAF(V600E), and also by inactivation of the tumor suppressors phosphatase and tensin homologue (PTEN), von Hipple-Lindau (VHL), or neurofibromatosis (NF) 1 (Gorospe and Abdelmohsen, 2011). While ectopic expression of telomerase can restore cell proliferation during replicative senescence, it does not prevent premature senescence (Wei et al., 1999). The stresses that activate senescence also utilize the p53/p21 and p16/RB pathways, as described in a recent review (Loaiza and Demaria, 2016).

1.1.2. In Vivo Senescence

There are many models of replicative and premature senescence, mainly involving cultured primary cells. However, these senescence models remain limited, as they may not fully recapitulate the physiological consequences of senescence. Increasing numbers of studies have provided evidence that senescence occurs in tissues and organs in vivo and can be detected in physiologic and pathologic situations, with both detrimental and protective consequences.

In tissues from humans, mice, and monkeys, senescent cells have been detected showing the enhanced presence of DDR proteins and senescence-associated β-galactosidase activity, as explained later (Cristofalo et al., 2004; Dimri et al., 1995; Herbig et al., 2006). Senescent cells were found to accumulate with age in renewable tissues like epithelia, stroma, and the hematopoietic system (Choi et al., 2000; Dimri et al., 1995; Jeyapalan et al., 2007; Krishnamurthty et al., 2004; Krizhanovsky et al., 2008). Stem cells in muscle from aged mice express p16 as they switch from reversible quiescence to irreversible senescence, suggesting that the maintenance of quiescence depends upon the effective suppression of senescence pathways in vivo (Sousa-Victor et al., 2014). The reliance on p16 for the quiescence-to-senescence transition may explain, at least in part, the elevated proliferation of progenitors of neurons and islet cells in p16-deficient mice, which show improved neuronal and pancreatic function compared with WT mice (Krishnamurthy et al., 2006; Molofsky et al., 2006). Importantly, mice in which p16-expressing cells are selectively eliminated via caspase activation show delayed appearance of age-related conditions, including cataracts, sarcopenia, and loss of adipose tissue (Baker et al., 2011).

In physiologic contexts, senescence is essential for morphogenesis during development. It also plays a pivotal role during wound repair, as the wound-induced senescent fibroblasts and endothelial cells secrete platelet-derived growth factor AA (PDGF-AA) which accelerates myofibroblast differentiation, wound healing, and tissue repair (Demaria et al., 2014). Similarly, following liver damage, hepatic stellate cells acquire senescence, thereby increasing degradation of the extracellular matrix, preventing the formation of fibrotic scars, and preserving liver function (Krizhanovsky et al., 2008).

On the other hand, senescence has also been detected and implicated in various pathologies, prominently cancer. Interestingly, the same oncogenes that can trigger senescence (e.g., KRAS(V12), BRAF (V600E)) can also trigger tumorigenesis if cells are unable to mount a senescent response (Braig et al., 2005; Collado et al., 2005; Dankort et al., 2009; Michaloglou et al., 2005). The complexity between senescence and cancer is further illustrated by the senescence-associated secretory phenotype (SASP), as this phenotype can favor or suppress tumorigenesis. Favoring tumorigenesis is the secretion by senescent cells of angiogenic factors (e.g., VEGF) that promote local angiogenesis, and matrix metalloproteinases (MMPs) which make tissues permeable and facilitate both migration and metastasis (Freund et al., 2010; Ohtani et al., 2012). Suppressing tumorigenesis, particularly at certain tumor stages, are SASP factors that promote chemotaxis leading to the recognition and clearance of cancer cells by the immune system (Acosta et al., 2008; Kuilman et al., 2008; Massague, 2008).

Senescent cells can also exacerbate other pathologic situations. They have been implicated in age-related diseases such as atherosclerosis, diabetes, neurodegeneration, and cardiovascular illnesses (Childs et al., 2015; Chinta et al., 2015). For example, accumulation of senescent microglia with aging is linked to neurodegenerative diseases such as Parkinson’s and Alzheimer’s diseases (Luo et al., 2010), while the impaired vascularization in diabetes is linked to the enhanced senescence of endothelial progenitor cells (Rosso et al., 2006). Finally, senescence of skeletal muscle and precursor cells is linked to muscle aging and the related disease sarcopenia (Navarro et al., 2001). Collectively, there is rising recognition that in vivo senescence impacts human physiology and pathology.

1.1.3. Senescence Markers

As cells approach senescence, they undergo specific morphological and molecular changes. Morphologically, senescent cells become flat and enlarged, and accumulate stress granules and vacuoles (Campisi, 1997; Goldstein, 1990). Senescent cells normally display increased activity of the acidic senescence-associated β-galactosidase (SA-β-gal), which is widely used as a marker to identify senescent cells in tissues and cell cultures (Dimri et al., 1995). Senescent cells display significant changes in their nuclear morphology including large nuclei, irregular nuclear envelope, and condensed and disordered chromosomes (Freund et al., 2012; Mehta et al., 2007; Zhang et al., 2007).

The senescence phenotype can also be confirmed by other methods. High levels of ROS are widely used as markers of senescence; as ROS can induce DNA damage. DDR proteins including TP53, p21, γ-H2AX, p16, as well as senescence-associated heterochromatin foci (SAHF) are also major markers of senescence (Narita et al., 2003; Zhang et al., 2005). The localization of HMGB1 can indicate senescence as it enters the extracellular space to stimulate cytokine production through the Toll-like receptor (TLR) 4 pathway (Davalos et al., 2013). Senescent cells can also be identified by the SASP, a phenotype that includes heightened secretion of major chemokines (e.g., IL8, GROA/B/G, MCP2, and HCC4), cytokines (IL6, IL1A, IL1B, and IL7), MMPs (including MMP1/3/10, TIMP1/2), and additional factors (e.g., VEGF, EGF, IGFBP2, NGF, and FGF7) that influence both the immediate cellular microenvironment and distant tissues and organs (Campisi, 2005; Coppe et al., 2006; Kuilman and Peeper, 2009). Together, senescent cells display many specific changes that can serve as markers, although it is important to mention that no single marker is conclusive and thus assessment of multiple markers is needed to identify senescent cells.

1.2. MicroRNAs

MicroRNAs (MiRNAs) are small noncoding RNA molecules (~22 nucleotides long) that generally influence gene expression by binding to subsets of mRNAs with which they share partial complementarity, and suppressing the stability and/or translation of such mRNAs. They are transcribed as long hairpin structures, primary (Pri)-miRNAs, which are processed in the nucleus by Drosha/DGCR8 into precursor (Pre)-miRNAs that are exported to the cytoplasm and further processed by DICER1 into mature single-stranded MiRNAs. Mature MiRNAs are key components of the RNA-inducible silencing complex (RISC) that direct RISC to target mRNAs (Bartel, 2009; Fabian et al., 2010; Kim et al., 2009; Newman and Hammond, 2010). Through their interaction with a wide spectra of mRNAs encoding proteins with diverse functions, MiRNAs have emerged as potent posttranscriptional regulators of several cellular processes, including cell survival, death, division, differentiation, and senescence (Abdelmohsen and Gorospe, 2015; Bushati and Cohen, 2007; Catalanotto et al., 2016; Kloosterman and Plasterk, 2006; Lingor, 2010; Yao, 2016). Like senescence, MiRNAs have been implicated in several disease conditions such as cancer, cardiovascular disease, and neurodegeneration (Saeidimehr et al., 2016; Shah et al., 2016).

The activity of the p53/p21 and p16/RB senescence pathways, as well as the SASP, is governed by the relative levels of the constituents and modulators of these pathways. Among other mechanisms, the levels of the protein components of these signaling systems are strongly regulated by MiRNAs, as discussed in Sections 2–4.

2. SA-MicroRNAs IN THE P53/P21 PATHWAY

In addition to the finding that suppression of global MiRNA biogenesis induces p53/p21 signaling and senescence, individual MiRNAs have been identified that control the expression levels of protein components of this pathway (Jones and Lal, 2012; Mudhasani et al., 2008). Specifically, we will focus on the SA-MicroRNAs that regulate expression of p53, p21, MDM2, SIRT1, MYC, TERT, and MCD1 (Table 1).

Table 1.

SA-MicroRNAs in the p53/p21 Pathway

MiRNA	Target	References
MiR-125b, MiR-504, MiR-25, MiR-30d	p53	Kumar et al. (2011)
MiR-192, MiR-194, MiR-215, MiR-605	MDM2	Jones and Lal (2012)
MiR-34a, MiR-22, MiR-138, MiR-181a, MiR-217	SIRT1	Jazbutyte et al. (2013), Menghini et al. (2009), Ming et al. (2016), and Yamakuchi (2012)
MiR-34a	FoxM1 and MYC	Cui et al. (2016) and Xu et al. (2015)
MiR-29c-3p	CNOT6	Shang et al. (2016)
MiR-22	CDK6, SIRT1, SP1	Xu et al. (2011)
MiR-195	TERT, SIRT1	Okada et al. (2016)
MiR-22	AKT3	Zheng and Xu (2014)
MiR-200c	ZEB1, BMI1	Magenta et al. (2011)
MiR-181a, MiR-181b, MiR-138	SIRT1	Rivetti di Val Cervo et al. (2012) and Zhou et al. (2016)
MiR-130b, MiR-302a, MiR-302b, MiR302c, MiR-302d, MiR-515-3p	p21	Borgdorff et al. (2010)
MiR-106b	p21	Ivanovska et al. (2008) and Li et al. (2011)
MiR-20a	p21	Sokolova et al. (2015)
MiR-17	p21	Gibcus et al. (2011)
MiR-208	p21	Zhang et al. (2011)
MiR-663	p21	Yi et al. (2012)
MiR-519	DUT1, EXO1, RPA2, POLE4	Abdelmohsen et al. (2012) and Marasa et al. (2010)

2.1. p53

MiRNAs MiR-125b, MiR-504, MiR-25, and MiR-30d directly bind p53 mRNA, suppress p53 protein levels, and inhibit senescence (Kumar et al., 2011). For example, lowering MiR-125b levels increased p53 levels and enabled senescence, while elevating MiR-125b levels suppressed p53 abundance and induced apoptosis in human lung fibroblasts responding to stress and developmental cues (Le et al., 2009) (Fig. 1).

Fig. 1.

SA-MicroRNAs implicated in cellular senescence. Schematic representation of the MicroRNAs that promote or inhibit senescence in three major domains of senescence: the p53/p21 pathway, the p16/RB pathway, and the trait SASP. MicroRNAs that promote senescence are in yellow boxes, MicroRNAs that suppress senescence are in white boxes.

2.2. p21

Several MiRNAs directly suppress p21 expression and thus influence cell senescence and proliferation. For instance, senescence-inhibitory MiRNAs including MiR-130b, MiR-302a, MiR-302b, MiR302c, MiR-302d, and MiR-515-3p attenuate RAS(G12V)-induced senescence of human mammary epithelial cells by preventing the RAS(G12V)-induced upregulation of p21 levels (Borgdorff et al., 2010). MiR-106b reduced p21-mediated cell cycle arrest in prostate cancer cells exhibiting radiation resistance, likely through direct targeting of MiR-106b to the p21 mRNA (Ivanovska et al., 2008; Li et al., 2011). Similarly, MiR-20a was shown to target the 3′UTR of p21 mRNA and blocked p21-mediated growth arrest in TGF-β-responsive colon carcinoma (Sokolova et al., 2015). MiR-208 promoted insulin-induced vascular smooth muscle cell proliferation by down-regulating p21 (Zhang et al., 2011), and MiR-663 suppressed p21 expression and promoted the proliferation and tumorigenesis of nasopharyngeal carcinoma cells (Yi et al., 2012).

Other MiRNAs indirectly enhance p21 expression leading to cell cycle arrest and senescence. For instance, MiR-519a, MiR-519b, and MiR-519c trigger p21-induced senescence by modulating multiple processes including DDR, calcium influx, and autophagy. The impact of MiR-519 on senescence was observed in HeLa cells and WI-38 human diploid fibroblasts (HDFs) (Abdelmohsen et al., 2012; Marasa et al., 2010). The cytoplasmic deadenylase CNOT6 is a component of the deadenylase complex, which contributes to the prevention of cell death and senescence (Mittal et al., 2011). MiR-29c-3p promoted senescence through the p53/p21 and p16/RB pathways by reducing CNOT6 expression levels in human mesenchymal stem cells (Shang et al., 2016).

2.3. MDM2

The murine double minute clone 2 (MDM2) is an oncogene that suppresses the expression of p53 by functioning as a ubiquitin ligase which promotes p53 ubiquitination and degradation by the 26S proteasome (Momand et al., 2000; Piette et al., 1997). Accordingly, the repression of MDM2 expression by MiR-192, MiR-194, MiR-215, and MiR-605 enhanced p53 levels and senescence (Jones and Lal, 2012).

2.4. SIRT1

The silent mating type information regulation 2 homologue 1 (SIRT1) is a deacetylase that regulates metabolic activity in response to cellular stress. SIRT1-mediated deacetylation of p53 suppressed its transcriptional induction of p21 expression (Brooks and Gu, 2009). MiR-34a, MiR-22, MiR-138, MiR-181a, and MiR-217 decrease SIRT1 expression, thereby inducing p53 levels and cell senescence (Jazbutyte et al., 2013; Menghini et al., 2009; Ming et al., 2016; Xu et al., 2011; Yamakuchi, 2012). Interestingly, p53 amplified this effect by transcriptionally inducing the expression of MiR-34a, MiR-34b, and mir-34c, which triggered senescence in various cell systems (Bai et al., 2011; Cui et al., 2016; Kumamoto et al., 2008; Tazawa et al., 2007; Xu et al., 2015).

2.5. MYC

The oncoprotein MYC influences DNA replication, cell cycle progression, and apoptosis. MYC suppresses p21 expression levels, thereby modulating genotoxic and apoptotic responses (Seoane et al., 2002). Several MiRNAs, including MiR-34a, MiR-429, MiR-33b, MiR-126, MiR-136, MiR-145, MiR-449c, and let-7 were found to target the MYC mRNA, although it is not known if they all induce senescence (Deng and Sui, 2013). MiR-34a further impacts senescence via MYC-modulated telomerase activity in human hepatocellular carcinoma (Cui et al., 2016; Xu et al., 2015). Senescence of WI-38 and BJ-1 HDFs triggered by metformin treatment led to increased MiR-429 levels (Cufi et al., 2012), although it is not yet known if MiR-429 targets MYC mRNA in this paradigm.

2.6. TERT

The reverse transcriptase enzyme telomerase (TERT) is essential for maintaining telomere length and thus prevents senescence (Shawi and Autexier, 2008). MiR-195 induced senescence in old mesenchymal stem cells (OMSCs) by directly reducing the levels of TERT and SIRT1. Conversely, inhibiting MiR-195 enhanced the expression of TERT and SIRT1, reduced p53 levels, and thereby delayed senescence (Okada et al., 2016). MiR-1207-5p and MiR-1266 suppressed cell growth and gastric cancer cell invasion by targeting TERT, but it remains to be tested if they induce cell senescence (Chen et al., 2014).

2.7. MDC1

The mediator of DNA damage checkpoint 1 (MDC1) regulates p53 activity and in turn the DDR, DNA repair, cell survival, and cell death (Nakanishi et al., 2007). Accordingly, a rise in MiR-22, which directly represses MDC1 production, hindered the DDR machinery, and induced premature senescence (Coster and Goldberg, 2010; Lee et al., 2015).

3. SA-MicroRNAs ASSOCIATED WITH THE P16/RB PATHWAY

Several MiRNAs have been implicated in p16-mediated senescence. While some MiRNAs influence p16 expression directly through binding to the mRNA, other MiRNAs act upstream to influence p16 levels. Here, we will focus on SA-MiRNAs that affect key components of this pathway, including p16, PRC2, MKK4, and RB (Table 2).

Table 2.

SA-MicroRNAs in the p16/RB Pathway

MicroRNA	Target	References
MiR-24s	p16	Lal et al. (2008) and Philipot et al. (2014)
MiR-26b	EED; EZH2	Overhoff et al. (2014)
MiR-181a	CBX7	Overhoff et al. (2014)
MiR-210	EED, EZH2, suz12	Overhoff et al. (2014)
MiR-424	EED, suz12	Overhoff et al. (2014)
MiR-138	EZH2	Liang et al. (2013)
MiR15b, MiR-24, MiR-25, MiR-141	MKK4	Marasa et al. (2009)
let-7	p66SHC	Xu et al. (2014)
MiR-141	BMI1	Dimri et al. (2013)
MiR-9	CBX7	O’Loghlen et al. (2015)
MiR-29, MiR-30	B-Myb	Martinez et al. (2011)
MiR-29c-3p	CNOT6	Shang et al. (2016)
MiR-203	E2F3A, E2F3B, ZBP89	Noguchi et al. (2012)
MiR-205	E2F1	Dar et al. (2011)
MiR-449a	CCND1, E2F1, E2F3	Noonan et al. (2010), Mao et al. (2016), and Ren et al. (2014)

3.1. p16

The SA-MiRNA MiR-24 binds the 3′UTR of p16 mRNA and suppresses its translation in WI-38 HDFs and in HeLa cells (Lal et al., 2008). While MiR-24 was not found to inhibit senescence, its levels are inversely correlated with the levels of p16 in osteoarthritis-associated senescence (Philipot et al., 2014). In addition, MiR-24 promoted cell proliferation by targeting p16 and the CDK inhibitor p27 in various human cancer cell lines (Giglio et al., 2013). Other MiRNAs suppress p16 expression by targeting p16 mRNA include MiR-300, MiR-514, MiR-663, and MiR-141, although their effect on senescence remains to be studied (Al-Khalaf et al., 2013; Kabir et al., 2015).

3.2. PRC2

p16 is encoded by the INK4A-ARF locus, which is transcriptionally suppressed by the Polycomb-repressive complex 2 (PRC2). Polycomb group (PcG) proteins were found to disassociate from INK4A-ARF locus in senescent cells, allowing p16 levels to increase via increased transcription (Bracken et al., 2007). MiRNAs such as MiR-26b, MiR-181a, MiR-210, and MiR-424 enhanced p16 expression by directly suppressing PRC2 factors like the PcG proteins CBX7, embryonic ectoderm development (EED), enhancer of zeste homologue 2 (EZH2), and suppressor of zeste 12 homologue (Suz12) (Overhoff et al., 2014). Additionally, in human renal cell carcinoma, MiR-138 induced p16 and cell senescence by targeting EZH2 (Liang et al., 2013).

3.3. MKK4

The mitogen-activated protein (MAP) kinase kinase 4 (MKK4) phosphorylates and thereby activates other MAP kinases in response to stress. High levels of MKK4 in senescent cells decreased cell proliferation and triggered senescence (Cunningham et al., 2010). MKK4 mRNA is a target of four MiRNAs (MiR-15b, MiR-24, MiR-25, MiR-141) that jointly suppress MKK4 expression. Downregulation of MKK4 by these MiRNAs decreased p16 levels and delayed senescence, while their coordinated inhibition elevated MKK4 and p16 levels and induced senescence in WI-38 HDFs (Marasa et al., 2009).

3.4. RB

Activation of the RB pathway increased the levels of MiR-29 and MiR-30, which associated with the 3′UTR of the MYBL2 (B-Myb) mRNA, repressed MYBL2 expression and triggered senescence (Martinez et al., 2011; Zhou et al., 2016). While DU-145 prostate cancer cells with mutant RB are resistant to MiR-449a-induced senescence, overexpression of wild-type RB induces senescence by restoring the function of MiR-449a (Noonan et al., 2010). Other studies point to additional MiR-449a targets in the RB pathway; for example, in prostate cancer cells, MiR-449a suppressed E2F1 and induced senescence following ionizing radiation (Mao et al., 2016), and in human lung cancer cells, MiR-449a lowered E2F3 abundance and thus induced senescence (Ren et al., 2014). Ectopic expression of MiR-203 in melanoma cells induced hypophosphorylated RB and elevated other senescence markers by suppressing the expression of E2F3a, E2F3b, and ZBP-89 (Noguchi et al., 2012).

3.5. Other SA-MiRNAs Regulating the 16/RB Pathway

In HDFs, let-7 delayed senescence by reducing the translation of the adaptor protein p66SHC, a key regulator of lifespan in mammals (Xu et al., 2014). In addition, MiR-141 induced p16 levels and consequently senescence by suppressing the production of BMI1 in HDFs (Dimri et al., 2013). In IMR-90 and WI-38 HDFs, MiR-9 was found to promote p16 expression by lowering the abundance of CBX7 (O’Loghlen et al., 2015). Finally, overexpression of MiR-205 in melanoma cells suppressed proliferation and induced markers of senescence including p16 (Dar et al., 2011).

4. SA-MicroRNAs IN SASP

The senescence-associated production and secretion of cytokines, growth factors, and proteases constitutes the SASP (Coppe et al., 2008). This phenotype is governed by transcription factors, RBPs and ncRNAs, including both MiRNAs and lncRNAs. In this section, we focus on MiRNAs that regulate SASP, in particular interleukins (ILs), MMPs, and tumor necrosis factor alpha (TNF) (Table 3).

Table 3.

SA-MicroRNAs in SASP

MicroRNA	Target	References
MiR-146a/b	IRAK1, TLR8	Bhaumik et al. (2009) and Gysler et al. (2016)
MiR-143	MMP13	Osaki et al. (2011)
MiR-335	PTEN	Kabir et al. (2016)
MiR-183	ITGB1	Li et al. (2010)
MiR-9	IL6	Zhang et al. (2016a)
MiR-222	MMP1	Liu et al. (2009)
MiR-125b	TNF, MMP13	Tili et al. (2007) and Xu et al. (2012)
MiR-152	MMP3	Zheng et al. (2013)
MiR-187	TNF, IL6	Rossato et al. (2012)

4.1. Interleukins

The production and secretion of ILs increases in senescent cells and creates a proinflammatory environment that promotes tumor progression (Coppe et al., 2010). In primary human fibroblasts, MiR-146a and MiR-146b directly suppressed the expression of interleukin 1 receptor-associated kinase 1 (IRAK1), a key regulator of the IL1A (IL-1α) receptor signaling, leading to inhibition of IL6 and IL8 production. Interestingly, high levels of secreted cytokines upregulated MiR-146a/b expression, which in turn prevented excessive SASP activity (Bhaumik et al., 2009). The tumor-suppressor MiR-9 inhibits IL6 expression in cervical adenocarcinoma (Zhang et al., 2016c), and MiR-187, induced by IL10, similarly suppressed the production of IL6 and TNF (Rossato et al., 2012). The circulating MiRNA MiR-21 is a ligand for TLR8 in immune cells, and via this alternative mechanism, it stimulates IL6 and TNF production (Fabbri et al., 2012). MiR-199a suppresses the production of IκB kinase-β (IKBKB), which is required for NF-κB activation, thereby lowering IL6 and IL8 production (Chen et al., 2008).

4.2. Matrix Metalloproteinases

The trait SASP includes the production of numerous MMPs, responsible for degrading the extracellular matrix and cleaving cell surface receptors and ligands. As senescent cells accumulate, the rising MMP activity promotes tissue damage through chronic inflammation and extracellular matrix remodeling caused by SASP (Verma and Hansch, 2007). MMP levels are regulated by different MiRNAs. MiR-222 directly suppresses MMP1 expression in oral tongue squamous cell carcinoma, and further suppresses the production of superoxide dismutase 2 (SOD2), which would otherwise induce MMP1. In addition, MiR-222 indirectly lowers MMP13 levels by downregulating the histone deacetylase HDAC4 (Liu et al., 2009; Song et al., 2015). MiR-152 targets the 3′UTR of MMP3 mRNA, lowering MMP3 expression, and significantly reduces the invasiveness of glioma cells (Zheng et al., 2013). MiR-143 targets the MMP13 3′UTR and suppresses the expression of MMP13. In lung osteosarcoma, the robust down-regulation of MiR-143 leads to a strong increase of MMP13 levels and cancer cell invasiveness (Osaki et al., 2011). The lower levels of MiR-125b in cutaneous squamous cell carcinoma (cSCC) compared with healthy skin allows higher production of MMP13 and accordingly promotes invasion of cSCC (Xu et al., 2012). Finally, MiR-181a-5p directly suppresses MMP14 expression in breast cancer cells (Roth and Cao, 2015).

4.3. TNF

The proinflammatory cytokine TNF is also regulated by MiRNAs. As mentioned earlier, the circulating MiR-21 was capable of binding and activating TLR8 in immune cells, thereby stimulating TNF and IL6 secretion (Fabbri et al., 2012). By contrast, in primary human monocytes, IL10 induced MiR-187, which in turn suppressed TNF and IL6 (Rossato et al., 2012), while in RAW 264.7 macrophages, LPS suppressed MiR-125b, which targets TNF mRNA, and thus enabled TNF production (Tili et al., 2007).

5. SA-MicroRNAs IN PATHOLOGY

As discussed earlier, there is a complex functional connection between senescence and several age-related pathologies. In this section, we discuss some of the major SA-MiRNAs implicated in disease processes.

5.1. Impact on Pathology by SA-MicroRNAs in the p53/p21 Pathway

Many MiRNAs having an impact on the p53/p21 pathway are implicated in diverse pathologies. For example, MiR-181a and MiR-30c synergistically regulate the p53/p21 pathway in diabetes-induced cardiac hypertrophy (Raut et al., 2016). The MiR-34a-mediated reduction of SIRT1, a suppressor of p53 activity, was implicated in tumorigenesis of the colon, lung, and other tissues (Kim et al., 2016; Lai et al., 2012; Misso et al., 2014; Rupaimoole and Slack, 2016; Wang et al., 2013). The Kruppel-like zinc finger protein ZNF224 transcriptionally increased MiR-663a, an MiRNA that lowers p53 and p21 production, and thus ZNF224 induced cell growth and resistance to apoptosis in human breast ductal carcinoma cells (Cho et al., 2016). MiR-621 suppressed production of the p53 inhibitor FBXO11, elevating p53 activity and sensitizing breast cancer cells to apoptosis (Xue et al., 2016), while MiR-24 increased metastasis and invasion of hepatocellular carcinoma cells by lowering p53 levels (Chen et al., 2016b). Collectively, these findings suggest that the network of MiRNAs affecting p53 function can influence pathologies, particularly cancer (Krell et al., 2013).

Recent studies suggest that the MiRNAs affecting p21 expression are particularly tightly involved in cancer. For instance, MiR-95-3p promoted tumorigenesis by targeting and reducing p21 levels in hepatocellular carcinoma (Ye et al., 2016), while inhibition of p21 by MiR-146b enhanced the proliferation of anaplastic thyroid cancer cells (Zhang et al., 2016a). The radiation-induced MiR-208a increased human lung cancer cell proliferation by lowering p21, and both MiR-106b and 208 enhanced radioresistance by lowering p21 expression (Tang et al., 2016; Zheng et al., 2015). In nasopharyngeal carcinoma, MiR-17-5p and MiR-663 promoted cell proliferation and tumorigenesis by lowering p21 abundance (Chen et al., 2016a; Yi et al., 2012), while in esophageal carcinoma, MiR-31 suppressed tumorigenesis in a p21-dependent manner (Ning et al., 2014). The MiRNA cluster MiR-17-92 regulated stem cell potential in MLL leukemia by reducing p21 expression (Wong et al., 2010), while upregulation of p21 by MiR-6734 induced cell cycle arrest and apoptosis in colon cancer cells (Kang et al., 2016). The impact of some of these MiRNAs on senescence remains to be investigated.

5.2. Impact on Pathology by SA-MicroRNAs in the p16/RB Pathway

MiR-24 inhibits p16 expression (Lal et al., 2008) and promoted cell growth, proliferation, and metastasis in gastric cancer cells, nonsmall cell lung cancer, and breast cancer, although a role for p16 in these processes await confirmation (Xu et al., 2013; Zhang et al., 2016b; Zhao et al., 2015). In another disease model, osteoarthritis, MiR-24 was identified as a negative regulator of p16 (Philipot et al., 2014). A rise in p16 levels following the loss of MiR-877-3p suppressed proliferation and tumorigenicity of bladder cancer by arresting cells in the G1 phase (Li et al., 2016b).

5.3. Impact on Pathology by SASP-Associated SA-MicroRNAs

Strong connections between SASP and human pathology are rapidly emerging. Among some prominent examples, lowering MiR-142-3p in macrophages contributed to increasing IL6 levels in aged mice and was proposed to enhance inflammation (Liu et al., 2016), and MiR-218-mediated suppression of the leucine-rich repeat-containing G protein-coupled receptor 4 (LGR4) was proposed to prevent the IL6-induced proliferation and invasion of prostate cancer cells (Li et al., 2016a). In addition, MiR-33a suppressed IL6-induced tumor metastasis of gallbladder cancer by inhibiting the expression of Twist, a transcription factor involved in epithelial-to-mesenchymal transition (Margetts, 2012; Zhang et al., 2016d). Interestingly, p16 suppressed tumorigenesis in breast stromal fibroblasts by repressing the expression and secretion of IL6. This effect was mediated by MiR-146b-5p, which bound to the 3’UTR of IL6 mRNA and suppressed IL6 production (Al-Ansari and Aboussekhra, 2015). In patients with myasthenia gravis, low levels of MiR-181c were found to correlate with high serum levels of IL7, encoded by the MiR-181-target IL7 mRNA (Zhang et al., 2016e). MiR-100-3p and MiR-877-3p suppressed production of IL8 and IL1B, respectively, in mesangial cells from IgA nephropathy patients (Liang et al., 2016). In a number of cultured neuroblastoma cells (SK-N-AS) and glioma cells (U251 and T98G), as well as in patients with glioma, MiR-93-5p suppressed the expression of IL8 and VEGF (Fabbri et al., 2015, 2016). The MiRNA that suppresses IRAK1 expression, MiR-146a, inhibited cell growth, lowered cell migration, and induced apoptosis in nonsmall cell lung cancer (Chen et al., 2013), although the role of IRAK1 in this phenotype was not examined. These reports highlight the extensive functional crosstalk among senescence pathways, inflammation, and disease by MiRNAs that modulate SASP.

The network of MiRNAs affecting senescence-associated MMPs has also been connected to cancer. MiR-143 and MiR-125b suppressed MMP13 expression and inhibited the growth of breast, ovarian, gastric, bladder, and other cancers (Han et al., 2013; Luo et al., 2015; Wang et al., 2016; Wu et al., 2015; Zhai et al., 2016), although the contribution of MMP13 to these effects remains to be confirmed. In human osteosarcoma, MiR-143 lowered the expression of plasminogen activator inhibitor-1 (PAI-1, SERPINE1); the finding that PAI-1 promoted the production and secretion of MMP13, and thereby enhanced invasion and metastasis, is in keeping with the tumor-suppressor function of MiR-143 (Hirahata et al., 2016). Furthermore, as MiR-143 is a circulating MiRNA, it may serve as a biomarker for bladder cancer diagnosis (Motawi et al., 2016). Other MiRNAs like MiR-15b and MiR-152 reduced glioma cell invasion and angiogenesis by suppressing MMP3 and NRP2 (Zheng et al., 2013). MiR-181a-5p suppressed metastases by lowering MMP14 levels in breast cancer cells (Roth and Cao, 2015). It is worth noting that MiR-181a, by lowering SIRT1 and PRC2, was also implicated in activating p53 and inducing p16 levels, respectively, as explained earlier, suggesting that these pathways are coordinately regulated by MiRNAs.

6. CONCLUDING REMARKS AND PERSPECTIVES

The impact of senescent cells on tissue homeostasis is complex and evolves over time, as senescent cells accumulate in the human body with age. Senescent cells are beneficial in some instances, such as during wound healing, muscle regeneration, suppression of pancreatic and liver fibrosis, and prevention of tumorigenesis in young organisms. However, senescent cells can be detrimental in other instances, particularly in the elderly, as they contribute to the development of age-related declines and diseases such as cancer, diabetes, cardiovascular disorders, diminished immunity, and neurodegeneration. Thus, there is rising interest in understanding the molecular regulators of senescence so that we can develop therapeutic approaches that promote its beneficial actions and ameliorate its harmful effects. To help achieve this larger objective, we have reviewed the MiRNAs that promote or suppress senescence.

We discussed SA-MiRNAs and their impact on cellular senescence by focusing on MiRNAs that enhance or diminish the SASP as well as the main senescence pathways, p53/p21 and p16/RB (Tables 1–3). Some SA-MiRNAs function as direct regulators, for instance, MiR-25 and MiR-30d suppressed senescence by directly targeting the p53 mRNA, while others function as indirect regulators, as shown for MiR-34a and MiR-22, which induced senescence by downregulating SIRT1 levels, which in turn enhanced p53 function, or for MiR-138, which targeted EZH2 leading to transcriptional upregulation of p16 expression. Other SA-MiRNAs, such as MiR-146a and MiR-146b, can modulate SASP, although there is increasing evidence that SA-MiRNAs can coordinately influence multiple senescent traits (Fig. 1).

Importantly, the majority of SA-MiRNAs have been investigated in cell culture models, and so it is important to assess if they change with age and whether they have an impact upon aging and age-related diseases. The development of mouse models of senescence and MiRNA function is starting to enable the systematic analysis of SA-MiRNA function in vivo, although much work is still needed. Understanding the impact of SA-MiRNAs on aging- and age-related diseases will also illuminate their diagnostic and therapeutic potential. For example, SA-MiRNAs in bodily fluids such as blood, urine, and saliva could serve as diagnostic or prognostic markers, perhaps helping to identify disease processes influenced by senescence.

Targeting MiRNAs by gain- or loss-of-function for therapeutic purposes is also an attractive proposition. Despite progress in this area, important challenges remain with the delivery of therapeutic MiRNA-relevant reagents with high specificity and efficacy (Chen et al., 2015). The gain-of-function approach using MiRNA mimics is widely used; however, the outcome should be cautiously interpreted as they may lead to nonspecific changes in gene expression (Jin et al., 2015). The loss-of-function approach using antisense oligos has often been used to inhibit MiRNA activity, but they can also target other sequences with few mismatches resulting in nonspecific effects. To overcome the above limitations, gene knockout would be an ideal choice for MiRNA deletion for functional analysis. Recently, genome-editing technologies based on zinc finger nucleases (ZFNs), transcription activator-like effector nucleases (TALENs), and CRISPR-Cas9-RNA-guided endonucleases (RGENs) have been widely used to knockout genomic loci of interest (Cai and Yang, 2014). Expansion of these technologies to study MiRNAs is emerging (Chang et al., 2016; Kim et al., 2013). Thus, using such new techniques to target MiRNAs will provide an excellent opportunity to further elucidate the role of SA-MiRNAs in vitro and in vivo.

Understanding the full spectrum of transcriptional and posttranscriptional regulators of SA-MiRNAs will further help with accurate targeting of the senescent phenotype. The studies discussed here highlight the efforts underway to exploit senescence regulatory factors toward favorable clinical outcomes in cancer and other diseases influenced by senescent cells.