INTRODUCTION

      The process of progressive organismal and cellular decline known as aging remains one of the most challenging problems in biology and medicine. Investigating and understanding the underlying mechanisms that promote aging is of paramount importance to elucidating interventions in agerelated diseases. Of the multiple observations that have been made in aging systems, one of the more mysterious has been the role of the transposable elements that have colonized the mammalian genome. While most of these elements lack the ability to replicate beyond their host cell, they are nonetheless still capable of replicating their DNA and pose a potential threat to the integrity of the host genome. One of the more successful retrotransposons is the class of long interspersed nuclear element 1 (L1), which is a ubiquitous feature of mammalian genomes, comprising approximately 20% of the genomic DNA in mice and humans (Lander et al., 2001; Waterston et al., 2002). These 6 kb, fully functional retrotransposons can replicate not only themselves, but also other retroelements that use L1-encoded proteins necessary for retrotransposition: ORF1, a nucleic acid chaperone, and ORF2, an endonuclease and reverse transcriptase (RT) (Dewannieux et al., 2003; Hancks and Kazazian, 2012; Richardson et al., 2015). As this ability to replicate and expand within a host genome requires DNA breakage and insertion, L1 activity has been linked to DNA damage and mutagenesis (Gasior et al., 2006; Gilbert et al., 2002; Iskow et al., 2010).

      While the preponderance of research on L1s has focused on their activity in the germline, recent evidence has suggested that L1 activity in somatic tissues contributes to a number of age-related diseases, such as neurodegeneration and cancer (Hancks and Kazazian, 2012; Iskow et al., 2010; Lee et al., 2012; Reilly et al., 2013). Given the potential harm that L1s can cause, host genomes have evolved a number of molecular mechanisms for silencing these parasitic elements (Crichton et al., 2014; Levin and Moran, 2011). Recent studies have reported, however, that these mechanisms become less efficient during the aging process, resulting in the derepression of L1s (De Cecco et al., 2013a; St Laurent et al., 2010; Van Meter et al., 2014). Much of this derepression appears to stem from redistribution and reorganization of the heterochromatin that normally constrains the activity of these elements and can lead to inflammation through the innate immune response (Ablasser et al., 2014; Oberdoerffer et al., 2008; Thomas et al., 2017; Van Meter et al., 2014). Despite the results that suggest that upregulation of L1 expression is a hallmark of aging, it is unclear to what extent, if any, L1s contribute to age-related pathologies and whether inhibiting L1 activity can delay these pathologies.

      Mice deficient in the mono-ADP-ribosylase/deacetylase protein SIRT6 develop a severe premature aging phenotype, characterized by a failure to thrive, intestinal sloughing, hypoglycemia, and a severely shortened lifespan (Mostoslavsky et al., 2006). Recently, SIRT6 has been demonstrated to be involved in silencing of L1 promoters (Van Meter et al., 2014). SIRT6 mono-ADP-ribosylates KAP1 and promotes its complex formation with HP1, thereby packaging the L1 DNA into transcriptionally silent heterochromatin (Van Meter et al., 2014). Remarkably, SIRT6 knockout (KO) mice show a strong activation of L1, suggesting a role for L1 misexpression in their age-related phenotypes (Mostoslavsky et al., 2006; Van Meter et al., 2014). Additionally, SIRT6 KO cells display excess genomic instability and DNA damage (Mostoslavsky et al., 2006). Considering the short lifespan of SIRT6-deficient mice, they provide a unique model to test whether inhibition of L1 activity can extend lifespan and alleviate the pathology of these mice.

      Given the multiple progeroid phenotypes and highly elevated L1 activity inherent to SIRT6 KO mice, we sought to use this system to address the role for L1s in age-related pathology. We used anti-RT drugs to treat SIRT6 KO cell lines and animals to inhibit L1 retrotranspositon and found that many of the pathologies in these animals were alleviated. L1-specific RNAi knockdown was able to recapitulate the results with the anti-RT drugs. We also discovered that SIRT6 KO-associated L1 activity triggers activation of the innate immune response via type I interferon production. Finally, we show that L1 DNA accumulates in the cytoplasm in SIRT6 KO tissues, triggering the cytoplasmic DNA sensor, cGAS, and initiating the innate immune response.

      Our data reveal that L1 activity directly contributes to the progeroid phenotypes of SIRT6 mice and correlates with observations in normal aging tissues and animals.

RESULTS

NRTI Treatment Abrogates L1 Activity in SIRT KO Cells and Tissues

      Nucleoside RT inhibitors (NRTIs) are a powerful class of clinical antiviral compounds used to treat HIV-1 infection via poisoning of RT enzymes by terminating chain elongation (Painter et al., 2004). This activity effectively blocks retroviruses and retroelements from completing genomic invasion. Several studies have reported that in addition to impeding the activity of viral polymerases, NRTIs such as 3TC and d4T are also potent inhibitors of L1 RT (Dai et al., 2011; Jones et al., 2008). In order to assess the efficacy of NRTIs in inhibiting L1 in the context of SIRT6 KO, we transfected wild-type (WT) and SIRT6 KO mouse embryonic fibroblasts (MEFs) with either human or mouse L1-enhanced green fluorescent protein (EGFP) reporter cassettes (Ostertag et al., 2000) to measure de novo retrotransposition events. In brief, successful retrotranspostion is detected when the GFP marker is retrotranscribed with the interrupting intron spliced out during mRNA processing. Successful events are measured as percent of GFP-positive cells. L1s were approximately three times more active in SIRT6 KO relative to WT cells (Figures 1A, 1B, and S1A). Both 3TC and d4T treatments abrogated L1 retrotransposition events in both WT and SIRT6 KO cells, demonstrating a robust antagonistic activity to the L1 life cycle (Figures 1A, 1B, and S1A). Additionally, we found that SIRT6 KO MEFs demonstrate a progressive accumulation of L1 DNA with each population doubling (PD) (Figure 1C). NRTI treatment of WT and SIRT6 KO fibroblasts over the course of 40 PDs effectively inhibited the expansion of L1 DNA copies in SIRT6 KO cells, demonstrating that NRTIs are sufficient for ameliorating L1 DNA accumulation (Figure 1C).

Inhibition of L1s Rescues Elevated DNA Damage in SIRT6 KO Cells

      DNA breaks induced by L1 ORF2 protein and de novo insertion events pose a threat to genomic stability. Overexpression of ORF2p is known to induce excessive DNA damage and can induce senescence, demonstrating the potential danger in endogenous misregulation of L1 activity (Gasior et al., 2006; Gilbert et al., 2002; Kines et al., 2014). In order to assess the effect of NRTI treatment on L1-mediated DNA damage, we treated mouse fibroblasts with NRTIs for 10 PDs. SIRT6 KO cells showed elevated double-strand breaks by both gH2AX (Figures 1D and S1B) and 53BP1 staining (Figures 1E and S1B) and by neutral comet assay (Figure 1F). This DNA damage was significantly ameliorated by NRTI treatment (Figures 1D–1F and S1B). Thus, NRTI treatment significantly reduces genomic instability linked to L1 RT in SIRT6 KO cells.

      NRTIs function as broad RT inhibitors. SIRT6 KO tissues demonstrate misregulation and increased expression of many retroelements, including SINEs and other retrotransposons (Van Meter et al., 2014), any of which could also contribute to SIRT6 KO phenotypes and would potentially be suppressed by NRTI treatment. Consistent with the previous report, we found several evolutionarily active L1 families to be elevated in MEF SIRT6 KO cells using several primer pairs targeting different regions (Figures S1C–S1E). To elucidate the direct role of L1 activity in SIRT6 KO biology, we generated two L1 RNAi vector systems to directly target L1s using conserved sequences between different L1 families, especially the more active families, including L1MdA_I, II, III, and L1MdTf (Hardies et al., 2000; Sookdeo et al., 2013). RNAi cassettes, or control vectors, were integrated into SIRT6 KO MEFs. Both RNAi systems demonstrated a significant reduction in L1 RNA abundance (Figure 1G). Immunostaining was conducted on these cell lines to assess DNA damage. Strikingly, both RNAi systems rescued the elevated gH2AX foci observed in the SIRT6 KO cells (Figure 1H). Thus, L1 activity alone in SIRT6 KO cells is the major contributor to the excessive DNA damage observed.

Elevated L1 Activity in SIRT6 KO Mouse Organs Is Suppressed by NRTI Treatment

      Based on the successful in vitro inhibition of L1 RT activity in SIRT6 KO cells and the significant amelioration of cellular DNA damage, we began administering 3TC and d4T to SIRT6 KO mice. Heterozygous SIRT6+/ mice were bred and pregnant animals were administered either 3TC or d4T in the drinking water immediately after mating. NRTIs continued to be administered in the drinking water throughout the postnatal period; in addition, the pups were given NRTIs orally once a day using a pipette, while the control pups were given water. As expected, in the homozygous SIRT6 KO control group, multiple tissues showed upregulation of L1 transcription compared to WT littermates (Figure 2A). This was complemented by observed increases in the LINE1 protein, ORF1p, in SIRT6 KO MEFs (Figure 2B). Total L1 DNA content (Figure 2D) was also highly elevated in these tissues. Remarkably, this increase in L1 DNA was suppressed by NRTI treatments (Figure 2D). Additionally, WT littermate cohorts treated with NRTIs also demonstrated significant decreases in L1 DNA content (Figure 2E). These data demonstrate that NRTIs can suppress L1 activity to a significant extent in these animals, regardless of the baseline level of L1 activity.

Cytoplasmic L1 DNA Is Enriched in SIRT6 KO Cells and Tissues

      It has previously been reported that conditions such as autoimmunity (Thomas et al., 2017; Stetson et al., 2008) are associated with accumulation of extrachromosomal L1 cDNA copies. Additionally, it has been reported that extranuclear DNA sensing is essential for cellular senescence (Dou et al., 2017; Li and Chen, 2018; Takahashi et al., 2018; Yang et al., 2017), which is elevated with SIRT6 deficiency and increases in aging mammals (Mao et al., 2012; Nagai et al., 2015). To test whether an increased L1 copy number is associated with the increase in the extrachromosomal L1 cDNA copies, immunofluorescence staining was performed using an anti-ssDNA antibody. Extranuclear ssDNA foci were consistently observed in SIRT6 KO cells, but not in WT cells (Figure 3A). Further, fluorescence in situ hybridization (FISH) staining using an L1 DNA-specific probe revealed multiple foci in SIRT6 KO cells, indicating that L1 DNA is present to a significant degree in the cytoplasm of these cells (Figures 3A and S1F). Cell fractionation and subsequent qPCR quantification demonstrated SIRT6 KO fibroblasts contained a 2-fold greater number of cytoplasmic L1 DNA than the WT cells, and this increase was completely rescued by the NRTIs (Figure 3B). Similarly, cytoplasmic L1 DNA content was also highly elevated in tissues that had demonstrated high L1 expression and total DNA content (Figures 3C–3F). These data indicate that much of the elevated L1 DNA observed in SIRT6-deficient tissues are not integrated copies and exist as extra-chromosomal cytoplasmic DNA.

L1 Activity Correlates with Type I Interferon Response and Is Rescued by NRTI Treatment

      L1 transposition intermediates such as L1 RT reverse-transcribed cytoplasmic cDNA can be recognized by cellular antiviral defense machineries triggering a type I interferon response (Volkman and Stetson, 2014). Indeed, we observed a significant increase in total L1 DNA copies, as well as cytoplasmic L1 cDNA, in SIRT6 KO cells and tissues (Figures 3B–3F), which could trigger a type I interferon response (McNab et al., 2015). In SIRT6 KO MEFs, the cytosolic DNA sensor, cGAS, exhibited significantly higher expression, which coincided with elevated expression of type I interferons (IFN-a and IFN-b1) (Figure 3G). To confirm that cGAS signaling was responsible for the observed interferon response, RNAi against cGAS was used. Using three separate short hairpin RNAs (shRNAs) with varying degrees of efficacy, cGAS knockdown correlated with suppression of both IFN-a and IFN-b1 (Figure 3H). Finally, MEF cells were crosslinked with UV radiation and cGAS was immunoprecipitated using two separate antibodies. Subsequent purification and analysis of the bound DNA revealed an  17- and  34-fold increase in the abundance of L1 DNA in SIRT6 KO cells compared to WT, depending on the antibody used (Figure 3I). Thus, cytosolic L1 DNA triggers type I interferon expression via the cGAS signaling pathway.

      Consistent with the results in MEFs, multiple tissues of SIRT6 KO mice showed a dramatic increase in type I interferon (IFN-a and IFN-b1) expression, which was completely rescued by NRTI treatment (Figures 4A and 4B). Additionally, both L1 RNAi systems completely rescued IFN-a and IFN-b1 expression, demonstrating similar trends to those observed in the NRTItreated animals and indicating that L1 activity is the cause of the innate immune response (Figures 4C and 4D). Further, suppression of basal levels of type I interferons by NRTI treatment was also observed in WT tissues (Figures S2A and S2B). Other inflammation markers were also elevated in SIRT6 KO mice, many of which were reversed by NRTI treatment (Figure S2C).

      Previously, it was reported that NRTIs possess an intrinsic anti-inflammatory activity independent of their anti-RT activity (Fowler et al., 2014), raising the possibility that the observed rescue of interferon response is independent of L1 inhibition. To address this possibility, we synthesized a 50 -O-methyl (meStav) version of d4T, which was previously reported to lack anti-RT activity due to the removal of the 50 -OH group but still retained anti-inflammatory activity (Fowler et al., 2014). SIRT6 KO MEFs treated with meStav at the same dosage as d4T (10 mM) failed to suppress IFN-a and IFN-b1 expression, but did display suppression of SASP factors associated with the NLRP3 inflammasome. Importantly, Fowler et al. observed inhibition of NLRP3 inflammasome at 100 mM. Here, we confirm their finding that meStav inhibits inflammasome even at a lower concentration; however, it does not inhibit the type I interferon response. This indicates that NRTI anti-RT activity is essential for suppression of the type I interferon response (Figures 4E– 4G and S2D).

NRTI Treatment Alleviates Progeroid Phenotypes and Extends Lifespan in SIRT6 KO Mice

      Consistent with previous studies (Mostoslavsky et al., 2006), the control (water treated) SIRT6 KO mice rapidly developed progeria and postnatal wasting with complete penetrance. Remarkably, the NRTI-treated mice presented as generally healthier, with shiny fur, improved body size, and less kyphosis (Figure 5A). While the control SIRT6 KO mice all died within 35 days, the NRTI-treated SIRT6 KO mice exhibited a more than 2-fold increase in their mean and maximum lifespans (Figure 5B). Additionally, we observed that NRTI-treated SIRT6 KO mice had improved body mass and delayed wasting compared to the controls (Figure 5C). Thus, NRTI treatment significantly improves SIRT6 KO lifespan and healthspan.

      One of the phenotypes of SIRT6 KO mice is hypoglycemia (Xiao et al., 2010; Zhong et al., 2010), which was not rescued by NRTI treatment (Figure S3A). Blood glucose levels continued to decline with age in NRTI-treated SIRT6 KO animals, similar to the control animals (Figure S3A). Previous reports have indicated that SIRT6 KO postnatal wasting can be partially rescued by providing animals with supplemental glucose (Xiao et al., 2010). To test the potential contribution of blood glucose to NRTI-mediated rescue, we provided SIRT6 KO animals with 10% glucose with and without NRTIs. Glucose supplementation alone did not significantly improve median SIRT6 KO lifespan, but yielded two (out of ten) ‘‘survivor’’ mice that outlived the untreated KO mice (Figure S3B). However, when glucose supplementation was combined with the NRTIs it abrogated the NRTI rescue (Figures S3C and S3D). While the mechanism of this effect was not clear, we hypothesize that the animals may have suffered from increased NRTI dosage due to increased consumption of glucose-containing water. The fact that hypoglycemia was not rescued by NRTIs, and the negative impact of glucose supplementation on the lifespan of NRTI-treated animals, indicates that NRTI-mediated rescue occurs independently of blood glucose changes.

      In addition to improved body weight and lifespan, NRTItreated animals showed dramatic improvements in mobility and behavior. Typically, SIRT6 KO animals demonstrate low mobility and unresponsiveness. However, NRTI-treated SIRT6 KO animals displayed a normal flight reflex from exposed locations (Figure 5D; Videos S1, S2, S3, and S4), foraging activity (Figure 5E), and greatly improved physical function on an inverted screen test (Figure 5F). Postnatal ophthalmological development is also stunted in SIRT6 KO mice, with pups unable to fully open their eyes. NRTI-treated animals showed significant improvements over control animals (Figures S4A–S4C). SIRT6 KO animals demonstrated reduced bone density (Figures 5G and S4D) and muscle mass and muscle fiber thickness (Figures 5H, 5I, and S4E), which were significantly improved by NRTI treatment. Additionally, SIRT6 KO animals showed deficiencies in hematopoietic (Figure S5) and intestinal (Figure S6A) stem cell compartments, which were significantly ameliorated by NRTI treatment. SIRT6 KO mice have also been reported to display severe lymphopenia (Mostoslavsky et al., 2006). In agreement with this, we observed elevated levels of apoptosis in thymus and spleen of the SIRT6 KO mice that were attenuated by the NRTI treatment (Figures 5J and 5K). Cumulatively, these data show that NRTI treatment drastically improves the healthspan of animals displaying elevated L1 activity, indicating an active role for L1s in these pathologies.

NRTI Treatment Reduces Apoptosis and Improves the Health of SIRT6 KO Intestines

      SIRT6 KO mice have been previously reported to suffer from a severe colitis phenotype (Mostoslavsky et al., 2006). Intestines of SIRT6 KO animals exhibited dramatically reduced thickness of the epithelial layer, atrophied villi, and pockets of inflammation, whereas the WT intestines did not (Figures 6A and 6B). Intestines of SIRT6 KO animals treated with either 3TC or d4T showed improved epithelial thickness and reduced inflammation (Figures 6B and 6D). In the small intestine, knockout mice displayed shortened villi (Figures 6A and 6E), decreased number of lamina propria plasma cells and lymphocytes, and pockets of neutrophilic acute inflammation dispersed throughout the lamina propria (Figure 6B). Additionally, apoptosis was signifi-cantly elevated in SIRT6 KO intestinal tissue and mucus-produc ing goblet cells were severely depleted (Figures 6C, 6F, and 6G). NRTI treatment partially rescued these phenotypes, increasing villi size, vastly reducing debris, ameliorating apoptosis, and suppressing inflammation (Figure 6). Restoration of tissue integrity in the intestine stands out in the context of extending SIRT6 KO lifespan, as the mice are characterized by a failure to thrive. Disruption of intestinal integrity is a well-characterized symptom of aging and perturbing intestinal permeability and barrier function is thought to contribute to age-related enteric diseases. No obvious histological differences were seen in the brains, kidneys, hearts, livers, and lungs of the different mouse groups (Figure S6B). Combined with the data described in Figure 5, these results suggest that some tissues (specifically, the intestines) are much more affected by LINE1 misregulation. Given the severity of the intestinal phenotypes in SIRT6 KO animals, we hypothesize that the improvements NRTI-treated mice display are greatly attributed to improved nutrient uptake facilitated by amelioration of the intestinal pathology.

L1 Misexpression and Sterile Inflammation in Aged Mice Coincide with Cytoplasmic L1 DNA

      Since sterile inflammation is a hallmark of aging (Lo ´ pez-Otı´n et al., 2013), we hypothesize that a type I interferon response mediated by age-related activation of L1 elements could play a causal role in this process. Consistent with this hypothesis, L1 expression and type I interferon expression were elevated in several tissues in the WT aged mice (Figures 7A–7C), suggesting that L1 activation may be contributing to sterile inflammation associated with normal aging. Consistent with the RNA expression, aged tissues also demonstrated elevated L1 ORF1p accumulation, indicating that L1s are biologically active (Figures 2B and 2C). WT aged mice also showed increased L1 cytoplasmic DNA (Figures 3C–3F). This result indicates that L1 activation is not limited to progeroid SIRT6 mice but is a hallmark of aging. Increase in L1 RNA and DNA in old tissues has been previously reported (De Cecco et al., 2013b).

      In order to assess if reverse-transcription inhibition impacts normal physiological aging, WT mice were treated with 2 mg/mL d4T from weaning age and tissue type I interferon expression was assayed at 55 weeks. We found that treatment of WT aged mice with d4T significantly reduced type I interferons in several tissues (Figures 7D and 7E), similar to what was observed in SIRT6 KO animals. Additionally, several tissues demonstrated reduced L1 DNA content (Figure 7F). D4T treatment also showed a trend toward reducing plasma concentration of multiple cytokines and chemokines, including those belonging to senescence-associated secretory phenotype (SASP), e.g., IL-6, IGFBP-3, IGFBB-6, CXCL1, CXCL4, TNF RI, CCL5, G-CSF, and MCP-1 as measured by antibody arrays in WT aged mice (Figure S7). Inducibility of these cytokines by poly(I:C), the inducer of the interferon response, remained the same with and without d4T treatment, suggesting that d4T does not affect interferon signaling pathway per se but rather reduces the levels of endogenous inducers such as cytoplasmic L1 cDNAs.

      To further assess the impact of L1 inhibition in aging WT animals, p16 expression was evaluated using control and d4Ttreated WT mice harboring a reporter where the p16 promoter is fused to the luciferase gene p16(LUC) (Burd et al., 2013). p16(LUC) mice were started on a regiment of 2 mg/mL d4T at 46 weeks and luciferase activity was measured at 46, 55, and 68 weeks of age. As expected, the luminescence of p16(LUC) mice increased with age. Regardless of a high degree of variability that is typical for this biomarker (Burd et al., 2013), significant differences between control and d4T-treated groups were observed in p16(LUC) female mice assessed on the 55th week of life (Figure 7I). A similar tendency was observed in males; however, they did not reach statistical significance within the group sizes used. Deaths from chronological aging occurred by the time of the next measurement of luminescence (68th week), resulting in drops of group sizes leading to increase in signal variability and reduction of statistical power.

      In order to address the effect of d4T treatment on the DNA methylation age (DNAm age), reduced representation bisulfite sequencing (RRBS) was performed on 12 blood samples of treated and untreated mice of both sexes applying the lifespan multi-tissue DNA methylation clock (Meer et al., 2018). The clock showed that the DNAm age of d4T-treated samples was lower (p = 0.046, two-tailed Mann-Whitney U test), indicating that d4T treatment reduced mouse methylation age (Figure 7J). Given these results, it is reasonable to suggest that the progressive loss of L1 silencing in aged systems significantly contributes to age-related sterile inflammation and other hallmarks of aging, including p16 expression and the methylation clock.

     To test whether SIRT6 can suppress L1 in the WT aged mice, we tested RNA from the livers of aged mice constitutively overexpressing SIRT6 (MOSES mice) (Kanfi et al., 2012). Remarkably, the age-related increase in L1 and type I interferon transcription were rescued in aged MOSES mice compared to their littermate controls (Figures 7A–7C), indicating that elevated SIRT6 activity prevents age-related L1 activation during normal aging. Taken together, these data indicate that L1 suppression not only impacts the progeroid phenotypes observed in SIRT6 KO animals, but similarly impacts normal aging pathologies.

DISCUSSION

      L1s are elevated and, in some cases, causative in pathologies, including inflammation, cancer, and neurodegenerative diseases. Here, we demonstrate an active role for these retrotransposons in the progeroid phenotypes associated with SIRT6 deficiency in mice. In this study, we demonstrate that inhibiting L1 RT activity not only alleviates the cellular and physiological dysfunctions of SIRT6 KO mice, but also extends the lifespan of these animals and correlates with observations inWT aged animals. Specifically, we show that L1 activity results in increased cytoplasmic L1 DNA, inducing a type I interferon response through cGAS cytosolic DNA sensing and promoting pathological inflammation. These results, combined with similar observations in aged WT mice, strongly support a detrimental role for L1s in the process of normal aging. Moreover, this study indicates that L1 activity contributes to the pathology of SIRT6-deficient mice.

      It should be noted that NRTIs are not specific to specific RTs. There remains the possibility that other retroelements, such as endogenous retroviruses, may contribute to aging pathologies. However, we were able to strongly recapitulate the results in cell culture experiments using NRTIs by using RNAi targeting several conserved regions on L1 families, indicating that L1 activity is the major contributor to the observed pathologies. Indeed, with L1s constituting the most abundant class of retroelements, it is reasonable to surmise that they make the major contribution to the RT-related type I interferon response.

      Interestingly, we found that WT aged mouse tissues display elevated L1 expression and induction of type I interferons, similar to the progeroid SIRT6 KO. These observations complement those reported for the higher incidence of DNA damage in aged tissues, similar to that seen in SIRT6-deficient models. Taken together, these results suggest that activation of L1 elements contributes to both age-related genomic instability and sterile inflammation associated with aging. The progressive activation of L1s with age may occur due to loss of silencing (De Cecco et al., 2013a; Villeponteau, 1997) and redistribution of chromatin modifiers such as SIRT1 and SIRT6 (Oberdoerffer et al., 2008; Van Meter et al., 2014).

      NRTI treatment was able to greatly improve the health and cellular pathologies in SIRT6 KO mice; however, it did not fully rescue the shortened lifespan. This indicates that there are other factors outside of L1 activity impacting these pathologies such as misregulation of glucose homeostasis (Mostoslavsky et al., 2006; Zhong et al., 2010). Furthermore, NRTIs rescue the accumulation of cytoplasmic L1s and type I interferon response but do not target L1 transcription, which may also be contributing to the burden these abundant retroelements impose on the cell.

      NRTIs themselves can pose adverse health effects, as patients treated chronically with NRTIs have demonstrated hepaptoxicity and other pathologies due to inhibition of mitochondrial DNA polymerase gamma (Montessori et al., 2003; Wu et al., 2017). Consistent with these reports, we did not observe an improvement in inflammation markers in the liver tissue, save for type I interferons. Additionally, NRTIs are known to inhibit telomerase and are believed to contribute to observations of accelerated aging in HIV patients (Hukezalie et al., 2012; Leeansyah et al., 2013). In mice, these drugs may have deleterious effects due to inhibition of telomerase, either by contributing to telomere erosion or by interfering with other functions of telomerase. While NRTI may prove an effective treatment to conditions predicated by L1 activity, it would not be an ideal solution to combat normal physiological aging. As such, the development of more targeted, anti-L1 interventions is now a necessary extension of these and other recent findings linking aberrant L1 activation to human disease. Such interventions may include specific non-NRTI inhibitors of L1 RT that do not inhibit other cellular polymerases. Additionally, as L1s rely on several components to form an active RNP, there exists the exciting possibility of developing small molecule inhibitors to target these assemblies more specifically. Novel strategies to combat inflammation triggered by cytoplasmic L1s, such as more efficient degradation of cytoplasmic L1 DNAs or inhibition of cGAS/STING-sensing pathways, may be equally effective. As SIRT6 plays a key role in silencing L1s, SIRT6 activators may prove to be effective at counteracting age-related loss of L1 silencing. The development of less toxic NRTIs in recent years may provide solutions in the interim.

      Several studies have demonstrated that cytosolic DNAs can trigger innate immune responses that contribute to cellular senescence and aging (Dou et al., 2017; Gl € uck et al., 2017; Thomas et al., 2017; Yang et al., 2017). The identity of the cytosolic DNAs, however, has been a mystery. Recently, it has been reported that accumulation of L1 cDNA in the cytosol of TREX1- deficient neurons drives type I interferon production and leads to apoptosis (Thomas et al., 2017). In this study, we demonstrate that these repetitive elements significantly contribute to the cellular and physiological pathologies experienced by SIRT6- deficient animals including type I interferon response and other inflammation markers. Many of these pathologies are mirrored in both mouse and human physiological aging, including L1 activation and sterile inflammation. Importantly, L1 inhibition in the WT aged mice leads to reduced inflammation and decrease in aging biomarkers including p16 expression and methylation age. Here we propose that the L1 activation and the resulting cytosolic L1 cDNAs, along with L1-mediated genomic instability, contribute to age-related inflammation and other aging pathologies (Figure 7K). Thus, L1 inhibition may be a viable strategy for treatment of age-related diseases. Interventions aimed at inhibiting L1 activity and the related inflammation may hold the potential to supplement other treatments or serve as a new form of therapy for age-related pathologies. Future work on effective dosages, late life animal treatments, and developing specific and less toxic anti-L1 inhibitors will pave the way for the future translational applications.

Limitations of Study

     While the work presented here has demonstrated the role of L1s in aging and SIRT6 deficiency-related pathologies, several aspects of this process remain to be clarified. We do not know the mechanism leading to formation of cytoplasmic cDNA copies. Such copies can potentially arise from reverse transcription in the cytoplasm primed by random RNA fragments, or L1 DNAs can escape from the nucleus.  Additionally, our analysiswas limited to active families of L1 transposons. As such, the activity of other transposable elements in aged tissues remains to be examined.

STAR+METHODS

Detailed methods are provided in the online version of this paper

and include the following:

● KEY RESOURCES TABLE

● CONTACT FOR REAGENT AND RESOURCE SHARING

● EXPERIMENTAL MODEL AND SUBJECT DETAILS

   Mice and Cell Lines

● METHOD DETAILS

  NRTI Treatments

  Analysis of Cytokines in Mouse Plasma Following d4T Treatment

 Cell Culture

 Transfections

 Quantitative RT-PCR

 L1 DNA Content

 Cytoplasmic DNA Extraction

 Mouse Activity Assays

 Immunofluorescence and Apoptosis

 Western Blotting

 ORF1p Immunostaining

 Histology

 Synthesis of 50 -O-Methyl d4T

 Comet Assays

 Blood Glucose Analysis

 Bone Marrow Stem Cell Counting

 LINE1 RNAi Vector Systems

 DNA FISH

 Immunoprecipitation of cGAS

 p16 Luciferase Measurement in d4T Treated Mice

 Methylation Clock Analysis

● QUANTIFICATION AND STATISTICAL ANALYSIS

SUPPLEMENTAL INFORMATION

Supplemental Information can be found with this article online at https://doi. org/10.1016/j.cmet.2019.02.014.

ACKNOWLEDGMENTS

We thank Jef Boeke and Emily Adney for the ORF1p antibody. This work was supported by grants from the NIH, USA to S.L.H., J.M.S., A.S., A.V.G., and V.G.; Roswell Park Alliance Foundation to A.V.G.; and Life Extension Foundation to A.S. and V.G.

AUTHOR CONTRIBUTIONS

M.S., A.S., and V.G. designed research and wrote the manuscript with input from all authors. M.S. performed all experiments related to L1 activity, DNA damage, and mouse phenotypic characterization. M.V.M. performed initial experiments described in Figure 1; J.A. performed mouse NRTI treatment and mouse pathology analysis; Z.K. performed analysis of stem cells; R.S.G. performed pathology evaluation of tissues; T.T. synthesized Methyl D4T; M.D.C. contributed to qRT-PCR analysis and performed in situ staining for ORF2; K.I.L. performed analysis of cytokine arrays; V.K. and N.N. contributed to bioinformatics analysis; S.L.H. contributed to data analysis; A.R. and H.Y.C. provided tissues from MOSES mice; M.V.M. and V.N.G. performed the analysis of methylation clock; and M.P.A., A.V.G., and J.M.S. performed NRTI treatment of aged mice, and analysis of cytokines and p16 expression (M.P.A. and A.V.G.), and contributed to study design, discussion, manuscript writing, and data analysis.

DECLARATION OF INTERESTS

The authors declare no competing interests.

Received: July 31, 2018

Revised: November 2, 2018

Accepted: February 22, 2019

Published: March 7, 2019

This article is excerpted from the Cell Metabolism 30, 871–885, April 2, 2019 by Wound World.