Loss of sirtuin 3 disrupts cellular senescence signaling pathways

Abstract

Cellular senescence is a multifaceted stress response marked by stable proliferative arrest and the secretion of diverse biologically active factors, collectively known as the senescence-associated secretory phenotype (SASP). The senescent phenotype is remarkably variable and subject to various regulatory influences. We previously demonstrated that mitochondrial dysfunction induced by diverse stimuli, including the loss of sirtuin 3 (SIRT3), leads to the hyperactivation of AMPK and p53, culminating in senescence while concurrently suppressing much of the proinflammatory SASP. Here, we extend our findings by revealing that the absence of SIRT3 can suppress segments of the SASP even in the absence of p53. Intriguingly, SIRT3 deficiency renders cells resistant to stimulation by exogenous cytokines, such as interleukin-1. Fibroblasts derived from Sirt3 knockout mice exhibit a diminished SASP, including reduced levels of Pdgfa, and these mice display impaired wound healing and a more expansive granulation area. Furthermore, aged Sirt3 knockout mice show disrupted patterns of senescence relative to wild type controls, including increases in senescence markers in adipose tissue, but surprisingly also decreases in liver and heart. Collectively, these data underscore a role for SIRT3 in orchestrating cellular senescence phenotypes, shedding light on its regulatory influence beyond the p53-dependent pathway.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-026-02284-x.

Introduction

Cellular senescence is a stress or damage response characterized by an essentially irreversible proliferative arrest accompanied by the secretion of various bioactive molecules including, but not limited to, cytokines, chemokines, growth factors, proteases, and signaling lipids, collectively known as the senescence-associated secretory phenotype (SASP) [1, 2]. Senescent cells accumulate with age and have been associated with age-related pathologies such as neurodegenerative diseases, atherosclerosis, and cancer [3–7]. Clearance of senescent cells in mice has been linked to both health span and lifespan extension [3, 8].

Sirtuins (SIRTs) are a family of nicotinamide adenine dinucleotide (NAD⁺)-dependent enzymes comprised of seven homologs, three of which (SIRT3, 4, and 5) are localized to mitochondria [9]. Sirtuin 3 (SIRT3), the primary mitochondrial protein deacetylase, has a complicated relationship with senescence and aging. For example, disease-accelerating conditions, such as metabolic disease [10], neuro-inflammation [11], and pulmonary arterial hypertension [12], are exacerbated in Sirt3 knockout (Sirt3-KO) mice. However, lifespans of Sirt3-KO mice are comparable to wild type mice on a fixed calorie diet and are extended upon caloric restriction [13]. Adding to this complexity, a separate study reported that Sirt3-KO mice under ad libitum conditions were shorter lived than their WT counterparts, citing cardiac abnormalities as the major pathological difference between the two [14]. Thus, the role of Sirt3 in murine aging remains unclear, as are the mechanisms that mediate its effects.

Loss of Sirt3 activity in human or murine fibroblasts or human mesenchymal stem cells results in adoption of cellular senescence, but these cells lack many of the proinflammatory features of the SASP that are found in response to other inducers of senescence [15], so it is unclear if the net result of Sirt3 loss is beneficial or detrimental. Additionally, Sirt3-KO mice are viable, and MEFs isolated from these animals do not show increased levels of senescence [15], suggesting that germline transmission is sufficient to spare Sirt3-KO mice from accumulation of senescent cells during early life. Other drivers of mitochondrial dysfunction, including loss of mtHSP70 (HSPA9), mitochondrial DNA depletion, and electron transport chain inhibition, also result in this mitochondrial dysfunction-associated senescence (MiDAS). MiDAS is distinct from other inducers of senescence, as these cells show a loss of many proinflammatory SASP factors at least in part due to activation of an AMPK-p53 signaling cascade. Despite showing a weaker metabolic phenotype than other forms of MiDAS (e.g., higher ATP:AMP ratios and less AMPK activation), depletion of SIRT3 resulted in stronger SASP suppression [15], but the mechanisms behind this remained unexplored.

We show here that loss of SIRT3 is required for many additional aspects of the SASP, independent of p53 activity or cell cycle arrest. Rather, SIRT3 is required for robust activation of the IL-1 signaling pathway, rendering SIRT3-depleted cells resistant to a major autocrine and paracrine driver of the proinflammatory SASP. Sirt3 knockout also lowers Pdgfa levels in senescent cells, and Sirt3-KO mice display commensurate alterations in wound healing kinetics. Finally, aged Sirt3-KO mice have tissue-specific senescence responses relative to age-matched WT controls. Together, our data may help explain some of the complexity observed in phenotypes associated with Sirt3 depletion and age-related conditions via disrupted regulation of the senescent phenotype.

Results

Loss of Sirt3 suppresses proinflammatory segments of the SASP in a p53-independent manner

We previously reported that mitochondrial dysfunction can both drive senescence and suppress proinflammatory aspects of the SASP via AMPK-dependent activation of p53 [15]. However, we observed a notable difference between other inducers of mitochondrial dysfunction and loss of SIRT3. For example, while depletion of p53 allows SIRT3-depleted cells to continue to divide and lowers senescence-associated beta-galactosidase (Supplemental Fig. 1a-c) [15], we did not observe a commensurate restoration of IL-6 secretion in IMR-90 (Fig. 1a) or BJ 1(Supplemental Fig. 1d) fibroblasts, or RNA levels of many proinflammatory cytokines and other SASP factors (Fig. 1b) following loss of both SIRT3 and p53, including in response to ionizing radiation (IR)—a robust inducer of senescence and proinflammatory aspects of the SASP. Thus, SIRT3 has effects that extend beyond our previously described AMPK-p53 axis. Irradiated shP53 cells were analyzed at 3 days after irradiation to avoid loss of viability inherent to the effects of irradiation in the absence of p53, consistent with previous studies [16–18].

Fig. 1.

p53-independent suppression of proinflammatory SASP factors by SIRT3 depletion. a, b IMR-90 fibroblasts were depleted of SIRT3 (shSIRT3), p53 (shp53), or controls (Scramble [shSCR] shRNA for SIRT3, shGFP for p53) by lentiviral shRNAs. Cells were then either treated with 10 Gy of ionizing radiation (IR) or mock-irradiated (Mock). Ten days later (or 3 days later for shP53), a conditioned media from cells were analyzed by ELISA for IL-6 secretion, or b RNA was extracted and analyzed by quantitative PCR for RNA levels of the indicated SASP factors. c Scramble control or shSIRT3 cells were either treated with 10 Gy of ionizing radiation (IR) or mock-irradiated (Mock). Ten days later, a NF-ΚΒ luciferase reporter assay was performed. d. Cells were treated as in (c) and analyzed by western blot for phosphorylated (activated) p38-MAPK, ERK1/2, or JNK. Beta-actin served as a loading control. e Quantitation of western band intensity across 3 independent experiments (*p < 0.05, **p < 0.01, *** p < 0.001, **** p < 0.0001 by 2-way ANOVA for all panels)

Nuclear factor kappa-beta (NF-κB) is a major mediator of the proinflammatory arm of the SASP [19, 20]. We therefore measured NF-κB activity by luciferase reporter assay. NF-κB activity was elevated in control cells when induced to senesce by IR, contrasted by a strongly reduced NF-κB activity in SIRT3 knockdown cells, regardless of IR treatment (Fig. 1c, Supplemental Fig. 1e). NF-κB is activated downstream of the MAPK signaling pathway during senescence [21, 22]. We therefore measured the phosphorylated (activated) forms of three major MAPK signaling proteins that can activate NF-κB: p38MAPK, ERK1/2, and JNK and observed a decrease in each (Fig. 1d, e). Thus, SIRT3 is required for key stress and inflammatory signaling pathways in senescent cells, and this is independent of p53 status.

SIRT3 is required for activation of the IL-1 signaling pathway

The SASP is subject to both cell-autonomous and cell non-autonomous regulation [23]; we therefore sought to determine if SIRT3 was required for cell non-autonomous promotion of the SASP. To test this, we co-cultured either mCherry-labeled scramble or shSIRT3 cells with IR-induced senescent cells for 3 days and then stained them for interleukin 6 (IL-6) by immunofluorescence (Fig. 2a). IR-induced senescent cells promoted IL-6 expression in scramble control cells, as in previous studies [23]. However, shSIRT3 cells did not elevate IL-6 levels in response to co-culture (Fig. 2b, c), suggesting that SIRT3 is required for cell non-autonomous activation of the SASP. Because interleukin-1 (IL-1) signaling can drive the SASP by autocrine, juxtacrine, and paracrine mechanisms [23, 24], we treated scramble or shSIRT3 cells with recombinant IL-1α or IL-1β (rIL-1A or rIL-1B; 100 pg/mL) and analyzed conditioned media for release of IL-6, a SASP factor induced by the IL-1 signaling pathway. Depletion of SIRT3 significantly lowered IL-6 secretion in response to IL-1 treatment (Fig. 2d).

Fig. 2.

SIRT3 is required for activation of IL-1 signaling during senescence. a Experimental scheme. Scramble or shSIRT3 cells (mCherry positive, red) were cultured in the presence of IR-induced senescent cells at a ratio of 1:3 for 72 h. Cells were then fixed and stained for IL-6 (green in secretory vesicles). b Representative images of co-cultured cells. c Quantitation of mCherry positive cells that stain positive for IL6. d Scramble or shSIRT3 cells were cultured in the presence of 100 pg/mL recombinant IL-1α, IL-1β, or carrier (BSA) for 24 h, followed by generation of conditioned media, which was then analyzed for IL-6 by ELISA, normalized to cell counts. e IMR-90 fibroblasts were treated or not for 2 months with ethidium bromide to deplete mitochondrial DNA (Rho0), followed by treatment with 100 pg/mL of recombinant IL-1α for 24 h. Conditioned media was generated for 24 h and analyzed for IL-6 by ELISA, normalized to cell number. f Scramble or shSIRT3 cells were cultured in the presence of 1 ng/mL lipopolysaccharides (LPS) for 24 h, followed by condition media generation and IL-6 ELISA. g Scramble or shSIRT3 cells were treated with 100 pg/mL of recombinant IL-1α + calyculin A for 0, 30, or 60 min, and lysates were analyzed for the indicated proteins by western blot. h IL-1A coculture experimental scheme: IMR-90 fibroblasts were transduced with shRNAs to GFP or IL1A and irradiated. Ten days later, cells were added to either pSICOR- or shSIRT3-transduced (mCherry positive) cells in a 3:1 ratio. After 72 h, cells were fixed and stained for IL-6 (green), and DAPI was used as a nuclear stain. i Example of mCherry/IL-6/DAPI stained cells. j). Quantitation of percent of mCherry-positive cells that also stained positive for IL-6 under each condition. All panels use ANOVA except for c, which uses a 2-tailed unpaired t-test with Welch’s correction. (***p < 0.001, ****p < 0.0001 for all panels). Error bars indicate SEM for all panels

We previously showed that depletion of p53 increases SASP expression during mitochondrial DNA depletion [15], unlike the observed SASP suppression observed during loss of SIRT3. We therefore sought to determine if blunted IL-1 responses are a common feature of multiple inducers of mitochondrial dysfunction, or if it is idiosyncratic to SIRT3 depletion. We therefore depleted mitochondrial DNA (Rho0) and then removed pyruvate from the media to induce MiDAS, followed by treatment with rIL-1A (Fig. 2e). Mitochondrial DNA depletion had no effect on IL-6 secretion in response to IL-1, indicating that the observed effects were SIRT3-specific.

The toll-like receptor 4 (TLR4) signaling pathway uses a similar signal transduction pathway to the IL-1 receptor. We therefore measured IL-6 secretion following treatment with lipopolysaccharides (LPS), a TLR4 agonist (Fig. 2f). LPS induced IL-6 secretion in both shSIRT3 and scramble controls, with shSIRT3 responding even more strongly than scramble controls, similar to previous reports [25]. Thus, this response seems to be a feature of the IL-1 signaling pathway and not all related signaling pathways.

IL-1 signaling can occur from both secreted forms of IL-1α and IL1β and also from juxtacrine signaling from membrane-bound IL-1α [24]. To further test whether loss of SIRT3 disrupts the IL-1 signaling pathway, we treated scramble or shSIRT3 cells with rIL-1A (plus the phosphatase inhibitor, calyculin A) for 0, 30, or 60 min—followed by protein extraction and analysis by western blot (Fig. 2g). Treatment with rIL-1A resulted in increased phosphorylation of IRAK4, IRAK1, TAK1, MEKK1, and IKK—along with lowered total IRAK1 and IkBα (which are ubiquitinated and degraded as part of IL-1 signaling)—all consistent with activation of IL-1 signaling [26], and each of these markers was blunted by loss of SIRT3. Notably, loss of SIRT3 did not lower levels of the IL-1 receptor (gene: IL1R1) at the RNA level (Supplemental Fig. 2a), did not change IL-1R staining on the surface of the cell (Supplemental Fig. 2b), and levels of the IL-1 receptor accessory protein (IL1RAP) were increased, rather than decreased, by SIRT3 depletion (Supplemental Fig. 2c). Further, levels of the toll inhibitory interacting protein (TOLLIP) were unchanged (Supplemental Fig. 2d). Because the action of the IL-1 receptor can be countered by the activity of IL-1 receptor antagonist (ILIRA; gene: IL1RN), we also considered the possibility that loss of SIRT3 might drive IL1RA. However, if anything, IL1RN RNA levels were reduced by SIRT3 depletion (Supplemental Fig. 2e), and IL1RA levels did not change by western blot (Supplemental Fig. 2f). To assess paracrine activity, we co-cultured shSIRT3 or empty vector controls (pSICOR) with IR-induced senescent cells transduced with lentiviral shRNAs to either IL1A or GFP (Control) (Fig. 2h). IR-induced senescent cells failed to induce IL-6 expression in co-cultured shSIRT3 cells, and shIL1A prevented IR-induced cells from promoting IL-6 positivity, regardless of shSIRT3 status (Fig. 2i, j). Together, these data indicate that SIRT3 is required for robust activation of the IL-1 signaling pathway, but the mechanism behind this remains unclear.

Sirt3 is required for proper wound healing in the skin

We previously showed that cells from germline knockout (Sirt3-KO) mice proliferate normally in culture, whereas depletion of Sirt3 in WT cells resulted in senescence [15]. However, we did not measure the SASP in these cells at the time. We therefore irradiated MEFs from either WT or Sirt3-KO mice and measured expression of multiple SASP factors for qPCR. Sirt3 was absent, but other sirtuins were unchanged (Supplemental Fig. 2a), confirming the specificity of the knockout and lack of compensatory upregulation of other sirtuins. RNA levels of multiple SASP factors including Il1a, Il6, Mmp3, and Ccl2 were diminished in senescent Sirt3-KO MEFs (Fig. 3a). Additionally, senescent Sirt3-KO cells also showed reduced RNA levels of platelet-derived growth factor AA (Pdgfa) (Fig. 3b), a factor elevated early in the senescence program [27]. Because senescence-derived PDGF-AA is beneficial in healing at the site of cutaneous wounds [28], we conducted wound healing assays following punch biopsy and measured wound closure over time (Fig. 3c). Sirt3-KO mice displayed reduced wound healing kinetics (Fig. 3d), in agreement with previous studies [29, 30]. These early kinetics also correlated with the time frame in which elimination of p16-positive cells slows wound healing [28]. Furthermore, Sirt3-KO mice also displayed larger granulation thickness and area compared to WT mice 14 days after wounding (Fig. 3e–g). These data support a role of Sirt3 in wound healing, potentially through disruption in senescence responses.

Fig. 3.

Loss of Sirt3 disrupts the SASP and slows wound healing in mice. a, b Individual embryos (3–6) were harvested from pregnant WR Sirt3-KO mice (in 129S1 background) and used to derive individual WT or Sirt3-KO MEF lines. Cells were then irradiated with 10 Gy IR and analyzed for a Il6, Il1a, Il1b, Il18, Mmp3, Ccl2; or b Pdgfa. c Experimental scheme for wound healing assays: WT (n = 8) or Sirt3-KO (n = 9) mice were given 6 mm cutaneous wounds by punch biopsy, and wound area was measured every 2 days by caliper. d Would area measurements over time. e–g Mice from (c) were euthanized 14 days after injury and histological sections were analyzed. e Representative images of picrosirius red stained wounds from WT and Sirt3-KO mice. f, g Four histological sections were evaluated for granulation thickness (f) and area (g). h–n Skin RNA from 18-mo old WT or Sirt3-KO mice was evaluated by quantitative PCR for h p21^Waf1, i p16^Ink4a, j p15^Ink4b, k SASP factors, l oxylipin synthases, m Acta2, and n Col1a2 expression, normalized to beta-actin. *p < 0.05, **p < 0.01, ****p < 0.0001; two-tailed unpaired t-tests with Welch’s correction for all panels

Senescent cells that accumulate with age have recently been shown to slow wound healing [31]. We therefore sought to determine if skin from aged Sirt3-KO mice might show increased markers of senescence relative to WT mice. While sample availability limited the power of our analyses (n = 3 WT, 5 Sirt3-KO), we did not observe changes in the senescence arrest-promoting cyclin-dependent kinase inhibitors (CDKNs) p21 or p16 in skin from aged Sirt3-KO mice (Fig. 3h, i), but surprisingly observed a robust reduction in levels of another senescence-inducing CDKN, p15^INK4b (Fig. 3j). RNA levels of many common SASP factors were altered in the skin of aged Sirt3-KO mice but did not rise to the level of statistical significance (Fig. 3k). In addition, because we previously identified multiple oxylipins as a lipid component of the SASP and mediator of senescence phenotypes and their effects of fibrosis [32, 33], we measured levels of the enzymes that synthesize senescence-associated oxylipins. RNA levels of senescence-associated oxylipin synthases were decreased in skin from aged Sirt3-KO mice (Fig. 3l). We also considered the possibility that loss of Sirt3 might increase fibrotic signaling in aged skin, much like in wound healing, but observed no changes in smooth muscle actin (Acta2, myofibroblasts marker) and a decrease in collagen (Col1A2) expression (Fig. 3m, n). Together, these data suggest that Sirt3 is required for proper wound healing, possibly by altering the homeostatic properties of senescent cells—though other cellular responses, such as macrophage activation, remain possible.

Aged Sirt3-KO mice show tissue-specific changes in expression patterns of senescence markers

Because loss of Sirt3 both promotes senescence and antagonizes elements of the SASP in culture, we sought to determine if it plays a role in the development of senescence or the SASP during chronological aging. We therefore crossed Sirt3-KO mice with p16-3MR mice, which allow detection of p16-positive cells by luminescence [28], and isolated tissues for subsequent gene expression analyses. Using these mice, we observed that whole body luminescence was increased at 18 months of age in Sirt3-KO mice relative to WT mice (Supplemental Fig. 3b), suggesting that more senescent cells were present, though mice were not shaved, and thus could instead reflect differences in hair density [34]. Despite increases in whole body luminescence, we did not observe elevated p16 levels in any tissue from Sirt3-KO mice and in fact observed lower p16 levels in liver and heart (Fig. 4a). Conversely, adipose tissue from Sirt3-KO mice instead showed higher levels of the senescence marker p21 (Fig. 4b), whereas p21 levels were lower in heart. Similarly, p15 levels were increased in adipose tissue of Sirt3-KO mice (Fig. 4c), though results were not statistically significant for any tissue. No statistically significant changes in any CDKN were observed in the kidney. Consistent with our findings for p21 and p15, adipose tissue from aged Sirt3-KO mice also had significantly higher levels of senescence-associated beta-galactosidase (SA-Bgal) staining relative to WT mice (Fig. 4d, e). Thus, effects of Sirt3 on age-related CDKNs are dependent on tissue type.

Fig. 4.

Tissue-specific effects of Sirt3 loss on senescent cell accumulation in aged mice. Sirt3-KO mice were crossed with p16-3MR mice (C57Bl/6 background), and homozygous WT and KO F2 mice containing the p16-3MR transgene were aged for 18 months. Gonadal white adipose tissue, liver, heart, and kidney tissues were collected. RNA was then extracted and analyzed for the following: a p16^Ink4a, b p21^Waf1, and c p15^Ink4b. d Gonadal white adipose tissue from mice was analyzed for senescence-associated beta-galactosidase (blue). e Quantification of beta-galactosidase intensity for gonadal adipose tissue from 3 WT and 3 Sirt3-KO mice. All p-values reflect two-tailed unpaired t-tests with Welch’s correction, *p < 0.05 if not otherwise listed. Adipose n = 6 WT, 8 SIRT3-KO; all others n = 3 WT, 6 SIRT3-KO

We also assessed the SASP in aged WT and Sirt3-KO mice. Adipose tissue showed variegated effects, with many SASP factors elevated, and others decreased (Fig. 5a, left). Because we have also shown senescence-dependent, age-related increases in oxylipin synthase expression in aged mice [32], we measured oxylipin synthase levels in both sets of aged mice. Consistent with the CDKNs and many SASP factors, oxylipin synthase levels were elevated in adipose tissue from Sirt3-KO mice (Fig. 5c, d, left). Consistent with p16 and p21, levels of many SASP factors and oxylipin synthases were decreased in liver and heart (Fig. 5a, b, middle), and kidneys did not have strong changes in most SASP markers (Fig. 5a–d, right). Finally, because many of the changes were associated with profibrotic elements of the SASP (e.g., Thbs1, Serpine1, Plaur), we assessed fibrosis markers and found that the smooth muscle actin (Acta2, marker of stellate cell and myofibroblasts activation) was lower in livers from aged Sirt3-KO mice, but not in any other tissues (Supplemental Fig. 3c), and this had no effect of collagen (Col1a2) gene expression (Supplemental Fig. 3d), suggesting little to change in age-related fibrosis responses. Thus, Sirt3-deficient animals have altered senescence responses depending on tissue, though it remains unclear what effects these alterations might have on longevity and healthspan. Together, our data elucidate a complex relationship between SIRT3, cellular senescence, and outcomes generated by senescent cells.

Fig. 5.

Effects of Sirt3 loss on the SASP in aged mice. Sirt3-KO mice were crossed with p16-3MR mice (C57Bl/6 background), and homozygous WT and KO F2 mice containing the p16-3MR transgene were aged for 18 months. Gonadal adipose tissue, liver, heart, and kidney tissues were collected. RNA was then extracted and analyzed for the following: a RNA levels of SASP factors and b aggregate SASP scores, as well as c oxylipin synthase levels and d aggregate oxylipin synthase scores. Adipose n = 6 WT, 8 SIRT3-KO; all others n = 3 WT, 6 SIRT3-KO. All p-values reflect two-tailed unpaired t-tests with Welch’s correction, *p < 0.05 if not otherwise listed

Discussion

Deficiency in SIRT3, a mitochondrially localized sirtuin, has been linked to senescence and to regulation of age-related phenotypes, but how does SIRT3 play this role? Here, we describe new mechanistic insights on how these relationships might occur—including both loss of proinflammatory IL-1 signaling and other secretory elements, as wells as alterations in age-related senescent cell accumulation and secretion.

While we show that SIRT3 is required for IL-1 signaling during senescence, it remains unclear how loss of SIRT3 blunts IL-1 signaling. We initially considered the possibility that the IL-1 signaling pathway connects with the mitochondrion through the ECSIT adapter protein [35, 36]. However, ECSIT is also required for TLR4 signaling, which was enhanced—rather than blunted—by loss of SIRT3. Additionally, ECSIT is an intermediate protein in the IL-1 signaling pathway [35] and therefore would not be expected to lower IRAK4 and IRAK1 phosphorylation, which are immediately downstream of IL-1 receptor activation and upstream of ECSIT in IL-1 signaling [26]. Rather, it seems likely that SIRT3 is required for IL-1 signaling at or near the plasma membrane by an as-yet unidentified mechanism. However, no changes in the IL-1R complex or its inhibitors were observed. Because TLR4 signaling was enhanced by loss of SIRT3 and uses many of the same proteins to drive signaling [37], it is possible that SIRT3 plays an unknown role in favoring signaling complex assembly at TLR4 at the expense of IL-1R, but this remains speculative.

Sirt3-KO mice have previously been shown to heal wounds more slowly [29, 30], but links to senescent cells have not been explored in this context. Notably, Sirt3 is decreased in macrophages during diabetes, and this slows wound healing [38]. Similarly, Sirt3 deficiency is linked to increased necroptosis during skin wound healing—a process we and others have shown can be driven by senescent cells during premature and chronological aging [39, 40]. Thus, it is tempting to speculate that these phenotypes might be either directly or indirectly by aberrant senescence responses in Sirt3-KO mice.

Our findings in aged Sirt3-KO mice suggest a perturbation in age-related senescence responses that are not clearly pro- or anti-geronic. For example, most senescence markers assayed, including cyclin-dependent kinase inhibitors and multiple SASP factors, were increased in the adipose tissue of aged Sirt3-KO mice relative to their wild type littermates. Conversely, skin, heart, and liver from aged Sirt3-KO mice show reduced levels of many of the same senescence markers. Thus, it appears that the role of Sirt3 in age-related senescence under ad libitum conditions is more complex than directional. Importantly, senescent cells are not the only source of many inflammatory cytokines and other related molecules, and this remains a limitation of this study. Reports that Sirt3-KO mice have roughly equivalent lifespans to wild type mice would seem to support this idea [13]. However, Sirt3 is also a key player in lifespan and disease following metabolic challenges, as calorically restricted Sirt3-KO mice live longer than their wild type counterparts [13], and disease-promoting features of high fat diets and diabetes are exacerbated in these mice [10, 41, 42]. Thus, it may be that Sirt3 is more of a key integrator of metabolic responses during aging, and these in turn might be mediated at least in part through control of cellular senescence.

Methods

Cell culture

IMR−90 and BJ1 (ATCC) human fibroblasts were mycoplasma-free, cultured in DMEM supplemented with 10% FBS and penicillin/streptomycin at 3% O₂ and 5% CO₂ as described previously [15], and used under 40 population doublings. Cells were mock irradiated or irradiated with 10 Gy X-rays, as described in [43]. To deplete mitochondrial DNA for rho0 cells, cells were cultured in 100 ng/ml ethidium bromide, 100 mM sodium pyruvate, and 50 mg/ml uridine for 2 months, followed by pyruvate removal to induce MiDAS.

Signaling ligands

Recombinant IL-1α or IL-1β (R&D Systems, 100 pg/mL), or LPS (1 μg/mL) were administered to cells in culture media for 24 h, followed by wash out and replacement with serum-free media for 24 h and analysis by IL-6 ELISA.

Co-culture experiments

“Donor” cells were transduced with shRNAs to GFP or IL1A, were selected using puromycin, and, 4 days later, received 10 Gy of ionizing radiation. Donor cells were used 7 days later. “Acceptor” cells were transduced with shSIRT3 or the empty pSICOR vector and selected using puromycin, and cells were confirmed to be > 90% mCherry-positive before use. Acceptor cells were then plated 1:3 with donor cells and co-cultured for 3 days before fixation, immunofluorescence, and analysis.

Lentiviral transduction

IMR-90 fibroblasts were transduced with lentiviral vectors that constitutively express shRNAs to either p53 [43] or SIRT3 [15] in polybrene as previously described. One day after transduction, cells were selected using puromycin (2 μg/mL) for 48 h. Resultant cells were confirmed SIRT3 or P53 depleted by western blot and used for experiments.

IL-6 ELISA

3 × 10⁴ cells were seeded in 12-well plates, were treated as indicated, and cultured in 0.5 ml serum-free DMEM for 24 h. CM were collected and clarified at 2000 × g for 10 min. Supernatants were transferred to a microcentrifuge tube, and cells were trypsinized and counted using a Coulter counter. Conditioned media (2.5 µl) were analyzed by bead-based ELISAs (AlphaLISA, Perkin-Elmer, Catalog #AL223F) in half-well, 96-well plates as instructed by the manufacturer on an Agilent BioTek microplate reader, and normalized to cell number.

Quantitative real-time RT-PCR

RNA was isolated from cells using a kit (Bioline) and from IAT using a Lipid Tissue RNA Isolation Kit (Qiagen), according to the manufacturer instructions. RNA (250 ng/ml) was used to synthesize cDNA, which was analyzed by qPCR using the universal probe library (Roche) or using TaqMan assays (Thermo Fisher). mRNA levels were normalized to beta actin, unless noted otherwise. Primer sequences and probes were described previously [15]. SASP scores were calculated as geometric means of all SASP factors and normalized to WT controls.

NF-κB reporter assays

IMR-90 fibroblasts were transduced with Cignal NF-κB Reporter Assay and Cignal renilla control lentiviral particles (Qiagen), followed by selection and the indicated treatment regimens. Cells were then lysed and analyzed via a dual luciferase reporter assays kit (Promega). NF-kB activity was normalized to renilla control.

Western blots

Cells were lysed in 5% SDS in 10 mM Tris, pH 7.4, with protease and phosphatase inhibitors. After BCA assay quantitation, 20 mg protein was electrophoresed and transferred to PVDF membranes. Membranes were blocked in TBST + 5% BSA, incubated overnight with primary antibody, washed 3 × in TBST, incubated with HRP-conjugated secondary antibody for 30 min, and visualized by chemiluminescence. Antibodies were from Cell Signaling Technology (p38 - #9212, phospho-JNK – #4668, phospho-ERK1/2 - #4370, phospho-IRAK4 - #7652, phospho-TAK1 - # 4531, phospho-IRAK1 - #4365, phospho-IKK -#2697), PhosphoSolutions (phospho-p38 - p190-1802), Santa Cruz (IRAK1 - sc-5288, phospho-MEKK1 - sc-130202, MEKK1 - sc-252, IkB-a - sc-371, SIRT3 - sc-49744), and Abcam (beta actin - ab8226, alpha-tubulin - ab4074).

Immunofluorescence

Immunofluorescence was performed as described previously [1]. Briefly, cells were treated as described, formalin fixed for 10 min, washed, and permeabilized in triton-X-100 for 30 min. Cells were washed again and incubated overnight in IL6 antibody (R&D Systems AF-206-NA), washed, and incubated in donkey-anti-Goat secondary antibody conjugated to ALEXA-488. Cells were counterstained with DAPI and photographed using fluorescence microscopy.

Animal models

All mice were maintained in the AALAC-accredited Buck Institute for Research on Aging animal facility. All procedures were approved by the Buck Institute Institutional Animal Care and Use Committee. Mice were bred in house and fed an ad libitum diet. Sirt3-KO mice in the 129S1 and C57Bl/6 backgrounds were provided by the Eric Verdin lab. P16-3MR were provided by Judith Campisi and crossed to Sirt3-KO mice in the C57Bl/6 background and F2 hybrids were genotyped for homozygosity. WT and Sirt3-KO x p16-3MR were aged for 18 months before analysis.

Wound healing assays

Sirt3-KO mice were in the 129S1 background with 129S1 serving as WT controls. Mice were used at 3 months of age. Two equal-sized wounds were created on dorsal skin by a 6 mm punch, and wounds were measured every 2 days by caliper and averaged across both wounds for each animal. At 14 days, all animals were euthanized and dorsal skin surrounding the wound site was collected.

In vivo luminescence

Homozygous WT and SIRT3-KO mice in the p16-3MR C57Bl/6 background were intraperitoneally injected with 15 μg of Xenolight RediJect Coelentarazine and quantified 25 min after injection on an IVIS Lumina S5 In Vivo Imaging system. Mice were not shaved.

Senescence-associated beta galactosidase

Staining for senescence-associated beta-galactosidase was accomplished as previously described [44] using a commercial kit (Biovision-Abcam). Briefly, cells were fixed 500 µl of Fixation solution for 10 min, washed in ice cold PBS 3x, and incubated at 37 °C for 24 h until analysis. For adipose tissue staining, 50 µg gonadal adipose tissue samples was fixed in 200 µl of Fixation solution for 15 min, followed by 3 washes using ice cold PBS. Tissues were then placed in 200 µl of Staining Solution and incubated at 37 °C with observation every 3 h until color (blue) developed. Stained tissues were photographed together and analyzed for beta-galactosidase intensity using the blue channel in ImageJ.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

CW: experimental and conceptual design and execution, wrote manuscript. NK, BM, and STJ: execution of experiments and writing. MCV: wound healing assays.

Funding

National Institute on Aging,AG051729,Agricultural Research Service,USDA-ARS 58-8050-9-004,Christopher Wiley,8050-10700-003,Christopher Wiley,National Institute of Diabetes and Digestive and Kidney Diseases,DK124170,Samantha T Jezak,SENS Foundation

Data availability

All data are presented in the manuscript.

Declarations

Competing interests

CW holds patents related to the detection, modification, and elimination of senescent cells and declares no non-financial interests. NK, BM, SJ, and MV declare no competing financial or non-financial interests.