Restoration of energy homeostasis by SIRT6 extends healthy lifespan

Abstract

Aging leads to a gradual decline in physical activity and disrupted energy homeostasis. The NAD⁺-dependent SIRT6 deacylase regulates aging and metabolism through mechanisms that largely remain unknown. Here, we show that SIRT6 overexpression leads to a reduction in frailty and lifespan extension in both male and female B6 mice. A combination of physiological assays, in vivo multi-omics analyses and ¹³C lactate tracing identified an age-dependent decline in glucose homeostasis and hepatic glucose output in wild type mice. In contrast, aged SIRT6-transgenic mice preserve hepatic glucose output and glucose homeostasis through an improvement in the utilization of two major gluconeogenic precursors, lactate and glycerol. To mediate these changes, mechanistically, SIRT6 increases hepatic gluconeogenic gene expression, de novo NAD⁺ synthesis, and systemically enhances glycerol release from adipose tissue. These findings show that SIRT6 optimizes energy homeostasis in old age to delay frailty and preserve healthy aging.

Fig. 1. SIRT6 regulates lifespan and healthspan of C57BL/6JOlaHsd mice of both sexes.

a Kaplan–Meier survival curves for male, female, and combined sexes of WT (n = 52 males, n = 50 females), SIRT1-tg (n = 47 males, n = 30 females), SIRT6-tg (n = 51 males, n = 41 females), and SIRT1 + 6-tg (n = 47 males, n = 44 females) mice. p-values for SIRT6-tg vs. WT mice are shown and were derived from two-tailed log-rank calculations. Median lifespan for each genotype is shown in parentheses. b–d Neoplasia (b), cancer (c), and gastrointestinal adenoma (d) incidence at the age of 25 months. For (b–d), n = 16 mice per genotype, a mixture of both sexes. Two-tailed Fisher’s exact test; *p < 0.05 vs. WT. Exact p-values are reported in Supplementary Table 1. e Respiratory exchange ratio (RER) averaged over 3 days in young and old WT and SIRT6-tg mice from either sex measured every half-hour. Three-way ANOVA with repeated measures with age, genotype, and time as variables. Sidak’s post hoc; young WT (* in blue) or old SIRT6 (* in green) vs. old WT. n = 8 mice per group, except for young SIRT6 where n = 7 mice. f Spontaneous wheel running activity of males at the ages of 8 months (n = 8 mice per genotype) and 15 months (n = 9 mice for WT, SIRT1, SIRT6, n = 7 mice for SIRT1 + 6). Two-way ANOVA with time and genotype as variables. g Exercise ability on treadmill in young and old males. Two-way ANOVA with Sidak’s post hoc. n = 6 mice per group, except for young WT where n = 5 mice. *p < 0.05, **p < 0.01, ***p < 0.001. In (e) and (f), values are shown as mean ± SEM. In (g), box extends from the 25th to 75th percentiles, line in the middle of the box is the median and whiskers go down to the smallest value and up to the largest. For (e–g), exact p-values are reported in the Source Data file.

Fig. 2. Serum metabolites associated with fasting and sirtuin overexpression.

a Heatmap showing serum metabolites from key metabolic pathways that changed significantly either with dietary state (fed, and 4 h and 16 h of fasting), genotype, or both as measured in two-way ANOVA. Each square represents average metabolite abundance of 7 WT mice, 7 SIRT1-tg mice, 6 SIRT6-tg mice, and 5 SIRT1 + 6-tg mice. b–d Scatterplots of data from (a) showing the levels of BCAAs (b), glucose and the GNG precursors glycerol and lactate (c), and TCA cycle metabolites (d). For (b–d), n = 7 WT mice, n = 7 SIRT1-tg mice, n = 6 SIRT6-tg mice, and n = 5 SIRT1 + 6-tg mice. Box extends from the 25th to 75th percentiles, line in the middle of the box is the median and whiskers go down to the smallest value and up to the largest. Statistical analysis was performed using two-way ANOVA (time, genotype) with Dunnett’s post hoc correction. p-value for genotype or fasting effect is indicated in graphs, exact p-values are reported in Supplementary Table 4. *p < 0.05, **p < 0.01.

Fig. 3. SIRT6 overexpression blocks age-dependent normoglycemia and deterioration in gluconeogenesis capacity.

a Blood glucose levels in young and old WT (blue) and SIRT6-tg (green) male mice at fed and fasted time points, as indicated in the graph. n = 10 young WT mice, n = 8 young SIRT6 mice, n = 7 old WT and old SIRT6 mice. b Same as (a), but in females. n = 8 young WT mice, n = 7 young SIRT6, old WT, and old SIRT6 mice. c Lactate tolerance test in 6 h-fasted male mice; n = 6 young WT, old WT, and old SIRT6 mice, n = 5 young SIRT6 mice. d Glycerol tolerance test in 6 h-fasted male mice; n = 6 mice per group. e Lactate tolerance test in 6 h-fasted female mice; n = 6 old WT, young SIRT6, and old SIRT6 mice, n = 7 young WT mice. f Glycerol tolerance test in 6 h-fasted female mice; n = 6 young WT and young SIRT6 mice, n = 8 old WT mice, n = 9 old SIRT6 mice. The area under the curve (AUC) for each tolerance test is shown on the right. For all panels, ages were 4–6 (young) and 22–24 (old) months. Statistical significance was calculated using three-way ANOVA with Sidak’s post hoc for line graphs, and two-way ANOVA with Fisher’s LSD method for bar graphs. Young WT (* in blue) or old SIRT6 (*in green) vs. old WT. In all panels, values are mean ± SEM. Exact p-values are reported in the Source Data file. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 4. SIRT6 activates hepatic mitochondrial catabolic and anti-inflammatory pathways.

RNA-seq was carried out on the liver of 6 months old WT and SIRT6-tg mice from either sex. a Principal component analysis was performed on all expressed genes. Each data point represents an individual mouse. b Gene Set Enrichment Analysis showing the top pathways up- or down-regulated in SIRT6-tg mice, sorted by normalized enrichment scores (NES). FDR q-values are shown in orange. Pathways related to catabolism and to inflammation are presented in blue and red, respectively. c Four metabolic pathways up-regulated in livers from male SIRT6-tg mice. In the upper part, each vertical black bar represents a gene in the gene set and its corresponding location in the sorted gene list. In the lower part, corresponding heatmaps show the expression values of the top genes in each pathway, which contribute most to the NES. High expression is colored in red and low expression in blue. Bold text indicates genes validated by qPCR in (e). d Ingenuity upstream regulator analysis common to males and females, based on genes differentially expressed between WT and SIRT6-tg mice. p-values for PPARα are shown, calculated by Fisher’s exact test. e Real-time PCR validation of gene expression from the indicated pathways, and mtDNA content in the livers of young and old WT and SIRT6-tg males. Two-way ANOVA with Fisher’s LSD method, asterisks indicate values significantly different between WT and SIRT6-tg at the same age. n = 8 mice per group, except for old SIRT6 for Coq10a (n = 6) and G6pc (n = 7), and for young WT and young SIRT6 for mtDNA content (n = 6). Mice ages are 6 and 25 months. RQ, relative quantity. Bars represent mean ± SEM. Exact p-values are reported in the Source Data file. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 5. Youthful-like hepatic TCA cycle, GNG, and redox metabolite levels in old SIRT6-tg.

a Lists of statistically significant metabolites from each of the four comparisons shown were used to create Venn diagram. The number of significant metabolites in each comparison is shown in parentheses. b PCA of significantly changed metabolites. Each dot represents one biological replicate. c Heatmap of significantly changed metabolites. Each square represents average metabolite abundance of n = 5 mice per genotype. TSP, transsulfuration pathway; Y, young; O, old; WT, wild type; TG, transgenic. d–f Scatterplots showing metabolite levels of hepatic glycolysis/GNG (d), TCA cycle (e), and redox metabolism (f) pathways. n = 5 mice per genotype, each dot represents one mouse. g Expression of key NAD⁺ de novo synthesis genes in the liver of young and old WT and SIRT6-tg mice. Asterisks indicate values significantly different between WT and SIRT6-tg at the same age. n = 8 mice for young WT, young SIRT6, and old SIRT6 (for Haao in young SIRT6, n = 7); and n = 7 mice for old WT (for Nmnat1, n = 8), males. RQ, relative quantity. h Protein levels of NMNAT1 (n = 5 mice) and TDO2 (n = 7 mice) in livers of old WT and SIRT6-tg littermates. ImageJ quantification of NMNAT1 normalized to α-tubulin and TDO2 normalized to ponceau is shown in the right. For all panels, mice were at the ages of 5–7 months (young) and 20–24 months (old). Bars represent mean ± SEM. For (d–g) data were analyzed using two-way ANOVA, for (h) using two-tailed Student’s t-test. Exact p-values are reported in the Source Data file. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 6. SIRT6 restores the significant reduction in hepatic lactate oxidation and contribution to GNG at old age.

a Schematic showing fates of lactate carbons following injection of [U-¹³C]-lactate, after one round of the TCA cycle or flux through pyruvate carboxylase (PC, Pyr→OAA). White balls, ¹²C carbons. Colored balls, ¹³C carbons. Red, carbon flux from lactate to glucose via the TCA cycle. Yellow, carbon flux from lactate to glucose via PC. b Hepatic glucose isotopologues, 15 min after [U-¹³C]-lactate injection. c–e, Relative abundance of hepatic glucose [M + 2] and glucose [M + 3] (c), the TCA cycle metabolites αKG [M + 2], succinate [M + 2] and malate [M + 2] (d), and total pyruvate and pyruvate [M + 3] (e) 15 min after [U-¹³C]-lactate injection. Metabolite peak area values were normalized to total ion count. f, g Hepatic lactate [M + 3] / pyruvate [M + 3] peak area ratio (f), and ratio of total liver metabolite levels of lactate/pyruvate (g) 15 min after [U-¹³C]-lactate injection. For (b–g), mice were at the ages of 5–7 months (young) and 20–24 months (old). n = 5 mice; bars represent mean ± SEM. One-way ANOVA with Fisher’s LSD method. Exact p-values are reported in the Source Data file. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. h Pyruvate tolerance test in 20 months old 6 h-fasted WT (n = 7 mice) and SIRT6-tg (n = 6 mice) male mice.

Fig. 7. Enhanced adipose tissue-derived glycerol contributes to maintenance of normoglycemia in aged SIRT6-tg mice.

a Lactate tolerance test in 22–24 months old liver-specific SIRT6-tg and appropriate littermate control male mice. Values are mean ± SEM of n = 6 WT mice, n = 5 lox mice, n = 9 Alb-cre mice and n = 5 liver SIRT6-tg mice. Two-way ANOVA showed no difference between liver SIRT6-tg and controls. Bars represent mean ± SEM. b Heatmap of GNG-related hepatic metabolites in 23–25 months old liver SIRT6-tg male mice and control littermates. Each square represents average metabolite abundance of n = 5 mice per genotype. See Supplementary Table 9 for source data. c Levels of the GNG precursors glycerol, lactate and pyruvate in the plasma of 6 h-fasted young (5–7 months) and old (20–24 months) mice, n = 5, males. Box extends from the 25th to 75th percentiles, line in the middle of the box is the median and whiskers go down to the smallest value and up to the largest. d Western blot and ImageJ quantification of HSL phosphorylation on Ser563 in white adipose tissue of 25 months old WT and SIRT6-tg littermates. n = 4 mice per genotype. Bars represent mean ± SEM, two-tailed Student’s t-test. For (c, d), exact p-values are reported in the Source Data file, *p < 0.05. e Schematic model depicting how SIRT6 improves longevity and healthspan by preserving NAD⁺ metabolism and energy production pathways in old age. Hepatic TCA cycle and GNG deteriorates during aging, leading to perturbed energy homeostasis and fasting glycemia in old age. SIRT6 overexpression increases hepatic energy production from fatty acids and AAs in the liver. This potentially spares GNG substrates, such as lactate and glycerol, whose shuttling during fasting from skeletal muscle and adipose tissue to the liver is increased by SIRT6. This cascade of events leads to preserved hepatic glucose production in old age. TG, triglyceride; HSL, hormone-sensitive lipase; DHAP, dihydroxyacetone phosphate; AAs, amino acids; Oxphos, oxidative phosphorylation.