Targeting senescent cells alleviates obesity-induced metabolic dysfunction

Abstract

Adipose tissue inflammation and dysfunction are associated with obesity-related insulin resistance and diabetes, but mechanisms underlying this relationship are unclear. Although senescent cells accumulate in adipose tissue of obese humans and rodents, a direct pathogenic role for these cells in the development of diabetes remains to be demonstrated. Here, we show that reducing senescent cell burden in obese mice, either by activating drug-inducible "suicide" genes driven by the p16^Ink4a promoter or by treatment with senolytic agents, alleviates metabolic and adipose tissue dysfunction. These senolytic interventions improved glucose tolerance, enhanced insulin sensitivity, lowered circulating inflammatory mediators, and promoted adipogenesis in obese mice. Elimination of senescent cells also prevented the migration of transplanted monocytes into intra-abdominal adipose tissue and reduced the number of macrophages in this tissue. In addition, microalbuminuria, renal podocyte function, and cardiac diastolic function improved with senolytic therapy. Our results implicate cellular senescence as a causal factor in obesity-related inflammation and metabolic derangements and show that emerging senolytic agents hold promise for treating obesity-related metabolic dysfunction and its complications.

Keywords: adipogenesis; aging; cellular senescence; dasatinib; quercetin; senolytics; type 2 diabetes.

Figure 1

Removal of obesity‐induced senescent cells from adipose tissue. (a, b) Renilla luciferase activity in DIO p16‐3MR mice (a, representative image), quantified in b (n = 6–9 per group). (c) Renilla luciferase activity localization in DIO p16‐3MR dissected tissues. (d, e) Senescence‐associated beta‐galactosidase (SA‐β‐gal) activity as whole tissue activity (d; representative image) and % positive cells of total DAPI⁺ cells in p16‐3MR VAT (e; chow n = 3, DIO n = 7–9 per group). (f) Expression of p16‐3MR transgene (eGFP) components (Renilla luciferase and mRFP) and p16^Ink4a (chow n = 3, DIO n = 11 per group) in p16‐3MR VAT. (g, h) Senescence‐associated beta‐galactosidase (SA‐β‐gal) activity as whole tissue activity (g; representative image) and % positive cells of total DAPI⁺ cells in p16‐3MR VAT (h; n = 3–4 per group). (i) p16^Ink4a mRNA levels (c; chow n = 8, DIO n = 19–21 per group) in VAT of D + Q‐treated DIO mice. (j) Percent of VAT stromal vascular fraction (SVF) cells highly expressing FLAG (a component of the p16^Ink4a promoter‐driven ATTAC fusion protein), CENP‐B, and p21^Cip1 after a single course of D + Q (o, n = 6 per group) in DIO INK‐ATTAC mice. Means ± SEM are shown. Box and whisker plot show minimum, mean, maximum, 25th and 75th percentiles. *p < 0.05, **p < 0.005, ***p < 0.0005; one‐way ANOVA with Bonferroni correction or two‐tailed Student's t test when comparing two groups

Figure 2

Eliminating senescent cells enhances glucose homeostasis and insulin sensitivity. (a, b) Intraperitoneal glucose tolerance test in DIO p16‐3MR (a; chow n = 4, DIO n = 6–7 per group) and DIO wild‐type mice treated with D + Q (b, chow n = 4, DIO n = 11 per group) following senescent cell clearance. (c) Hemoglobin A1c in DIO p16‐3MR (chow n = 4, DIO n = 15–18 per group) and DIO wild‐type mice treated with D + Q (chow n = 6, DIO n = 11–12 per group). (d, e) ITT following ganciclovir treatment in p16‐3MR mice (d, n = 4 chow, n = 18–19 DIO groups), or D + Q treatment in DIO wild‐type mice (e, chow n = 6, DIO n = 11–12 per group). (f) Fold change in AKT serine‐473 phosphorylation after 5‐min ex vivo 5 nM insulin stimulation in freshly isolated p16‐3MR VAT (n = 3 per group). For each mouse, p‐AKT was normalized to total AKT and expressed as a ratio of p‐AKT in insulin‐exposed tissue to p‐AKT in noninsulin‐exposed tissue. (g) Glucose infusion rate (GIR) during hyperinsulinemic–euglycemic clamp in DIO mice treated with vehicle or D + Q (n = 8 per group). (h) Plasma insulin concentration at baseline and during hyperglycemic clamping (HGC) in DIO mice treated with D + Q (n = 8 per group). (i) Glucose appearance rate (Ra) during basal, hyperglycemic, and hyperinsulinemic–euglycemic clamping in DIO mice treated with D + Q (n = 8 per group). Means ± SEM are shown. *p < 0.05, **p < 0.005, ***p < 0.0005; one‐way ANOVA with Bonferroni correction for multiple comparisons. ^# p < 0.05, two‐tailed Student's t test comparing DIO vehicle‐treated group to DIO ganciclovir‐treated, or D + Q‐treated group. Groups of interest were compared at each time point for GTTs and ITTs

Figure 3

Adipogenesis is enhanced by senescent cell reduction. (a, b) Plasma activin A in p16‐3MR mice (a; chow n = 3, DIO n = 9–11) and D + Q‐treated DIO mice (b, chow n = 3, DIO n = 11–12). (c, d) Adipogenic gene expression in cells isolated from subcutaneous adipose tissue stromal vascular fraction (SVF) of p16‐3MR (c; n = 5–7 per group) and D + Q‐treated mice (d, n = 3 per group). (e) Representative images of lipid droplet formation during differentiation of adipocyte progenitors isolated from vehicle‐ or ganciclovir‐treated p16‐3MR mice after 5‐day exposure to differentiation medium (scale bars indicate 50 μm). (f) VAT cell size in p16‐3MR mice and D + Q‐treated mice (n = 3–7 per group). (g, h) Subcutaneous:intra‐abdominal adipose ratio in DIO p16‐3MR mice (g, n = 5–6 per group) and D + Q‐treated mice (h, n = 18–21 per group). Means ± SEM are shown. *p < 0.05; one‐way ANOVA with Bonferroni correction or two‐tailed Student's t test when comparing two groups

Figure 4

Senescent cells promote macrophage infiltration. (a) Correlation of FLAG⁺ cells with F4/80⁺/Cd11b⁺ macrophages in the SVF of DIO INK‐ATTAC VAT as analyzed by CyTOF (n = 30). (b) Plasma cytokines in ganciclovir‐treated DIO p16‐3MR mice (n = 19–22 per group). (c) F4/80 mRNA in VAT of ganciclovir‐treated p16‐3MR DIO mice (chow n = 4, DIO n = 17–22 per group). (d, e) F4/80⁺ crown‐like structures in VAT of ganciclovir‐treated p16‐3MR DIO mice (representative photographs in d, quantified in e; chow n = 4; DIO n = 5–8 per group). (f) THP‐1 cell migration into a Transwell containing either nonirradiated or irradiated (senescent) adipocyte progenitors, with no addition of antibody (control), addition of nonspecific antibody (IgG), or addition of antibodies against MCP‐1, MIP1‐β, or M‐CSF (n = 4 per condition). (g, h) Representative image of luminescence signal in AP20187‐treated INK‐ATTAC;db/db mice 24 hr following i.v. injection of 1 × 10⁶ luciferase⁺ monocytes (g), normalized to vehicle‐treated mice and quantified in h (n = 6 per group). (i) Quantification of luminescence in D + Q‐treated db/db mice 24 hr following i.v. injection of 1 × 10⁶ luciferase⁺ monocytes isolated from CAG‐luc mice, normalized to vehicle‐treated mice (n = 5 per group). Means ± SEM are shown. *p < 0.05, **p < 0.005, ***p < 0.0005; one‐way ANOVA with Bonferroni correction or two‐tailed Student's t test when comparing two groups

Figure 5

Complications of diabetes are alleviated by senescent cell clearance. (a–c) Parameters of diastolic function in p16‐3MR db/db mice treated with ganciclovir determined by echocardiography–Doppler measurements (n = 5–8 per group): (a) Early diastolic velocity from medial mitral annulus (e′); (b) isovolumic relaxation time (IVRT) c, left ventricular filling pressure (E/e′) (n = 5–8 per group); (d) left ventricular ejection fraction (EF), an indicator of systolic function, in p16‐3MR lean or db/db mice treated with ganciclovir (n = 5–8 per group). (e) Number of cells per glomerulus staining positively for Wilms tumor protein in D + Q‐treated DIO mice (WT‐1; e, n = 4 per group). (f) Average number of p16^Ink4a‐positive cells per field in the renal cortex of D + Q‐treated DIO mice (n = 4 per group). (g, h) Urine albumin/creatinine ratio (ACR) in DIO p16‐3MR mice treated with ganciclovir (g, chow n = 4, DIO n = 9–11 per group) or DIO wild‐type mice treated with D + Q (h, n = 6–9 per group). Means ± SEM are shown. *p < 0.05, ***p < 0.0005, ****p < 0.0001; one‐way ANOVA with Bonferroni correction or two‐tailed Student's t test when comparing two groups