🧬 Aging Wiki

Targeting senescent cells enhances adipogenesis and metabolic function in old age

13 min read

Xu 2015 (eLife) — Targeting senescent cells enhances adipogenesis and metabolic function in old age

Companion study note: This is the eLife 2015 paper on senescent fat progenitors, activin A, and adipogenesis (Xu, Palmer, Ding, Weivoda, Kirkland et al.). It is a distinct study from the PNAS 2015 paper by some of the same authors (Xu, Tchkonia, et al., DOI 10.1073/pnas.1515386112) on JAK inhibition of SASP and physical frailty. Both appeared in 2015 from the Kirkland lab; they are companion studies and should not be conflated. This page covers only the eLife paper.

A study from the Kirkland/Tchkonia laboratory at the Mayo Clinic establishing that senescent fat progenitor cells secrete activin A — a TGF-β superfamily ligand — as a key SASP factor that paracrinally suppresses adipogenesis in neighboring healthy progenitors. Clearance of senescent cells (via the INK-ATTAC genetic system) or suppression of their SASP (via JAK inhibition with ruxolitinib/INCB018424) in old mice partially rescued fat mass, adipogenic gene expression, systemic metabolic parameters, and insulin sensitivity. Published in eLife 2015 (open access).

Question

Why does fat mass paradoxically decline in very old age even as obesity is common in middle-aged adults, and what role do senescent cells in adipose tissue play in this decline? Specifically: do senescent cells in fat tissue inhibit new adipocyte formation (adipogenesis), and if so, what is the molecular mechanism?

Design

Human in vitro experiments

  • Donors: Primary human subcutaneous fat progenitors (stromal-vascular fraction) from healthy, lean kidney donors (n=6; age 39 ± 3.3 years; BMI 26.6 ± 0.9 kg/m²).
  • Senescence induction: Ionizing irradiation at 10 Gy; >70% SA-β-galactosidase (SABG)-positive cells achieved within 20 days post-irradiation.
  • Co-culture assay: Senescent vs. non-senescent “source” progenitors co-cultured with naĂŻve “target” progenitors under adipogenic induction conditions; lipid accumulation scored as readout of adipogenesis.
  • Conditioned medium (CM) experiments: CM from senescent progenitors applied to naĂŻve target progenitors ± neutralizing antibody against activin A or ± SB-431542 (activin A receptor inhibitor; the paper notes it also inhibits TGFβ signaling via ALK5).
  • JAK inhibitor in vitro: Ruxolitinib (INCB018424) at 0.6 ÎĽM for 72 hours applied to senescent progenitors; activin A mRNA and secreted protein measured.

Mouse experiments — INK-ATTAC genetic clearance (Fig 4)

  • Model: INK-ATTAC transgenic mice express an FKBP–caspase-8 fusion under the p16^Ink4a^ promoter; administration of AP20187 (a chemical inducer of dimerization) selectively triggers apoptosis in p16^Ink4a^-expressing senescent cells 1.
  • Ages: 18-month-old wild-type (WT) C57BL/6 and INK-ATTAC+/- mice.
  • AP20187 dosing: 10 mg/kg i.p., two courses of three consecutive days each, with 14 days between courses — 6 total injection days over ~3 weeks.
  • n: n=8 for body composition endpoints; n=7 for SABG+ cell quantification.
  • Endpoints: Fat mass (MRI), SABG+ cells in adipose tissue, circulating activin A, adipose tissue gene expression (C/EBPα, PPARÎł, Lipin-1, IL-6, p16^Ink4a^, p21^Cip1

Mouse experiments — JAK inhibitor in vivo (Figs 6–8)

  • Ages: 22-month-old (n=9 per group; results replicated in a second independent cohort with the same regimen) and 8-month-old C57BL/6 mice (young control arm). NIA supplied mice as 22–24 months old.
  • Drug: Ruxolitinib (INCB018424) mixed into food; each mouse fed ~0.5 g food containing 60 mg/kg drug daily for 8 weeks; vehicle control = DMSO in food.
  • Endpoints:
    • Body composition (MRI): fat mass, lean mass
    • Fat depot weights: inguinal, subscapular, brown adipose tissue
    • Circulating activin A (ELISA)
    • Adipose tissue gene expression: PPARÎł, C/EBPα, FABP4, adiponectin (Adipo-Q), Lipin-1, GPAT4, ATGL, HSL; activin A transcript in whole fat and isolated progenitors
    • Plasma free fatty acids (FFA), plasma triglycerides
    • Liver weight, hepatic triglycerides (TG; both absolute total and TG/protein ratio)
    • Oral glucose tolerance test (OGTT) with area-under-curve (AUC)
    • Insulin tolerance test (ITT) with AUC
    • Fasting plasma glucose
    • Glucose-stimulated insulin secretion (GSIS)
    • Fat tissue ex vivo p-AKT/total AKT ratio (insulin challenge, n=6)
    • Metabolic rate and food intake (metabolic cages; no difference detected)

Findings

1. Senescent fat progenitors inhibit adipogenesis via a paracrine mechanism (Figs 1–2)

Co-culture of naïve target progenitors with senescent source progenitors dramatically reduced lipid accumulation: only ~20% of target progenitors accumulated lipid when co-cultured with senescent source cells, compared to >50% when co-cultured with non-senescent source cells 2. This inhibitory effect was recapitulated by conditioned medium (CM) from senescent progenitors, confirming a secreted, paracrine mechanism rather than direct cell–cell contact.

ConditionLipid accumulation (% target progenitors)
Co-culture with non-senescent source>50%
Co-culture with senescent source~20%

2. Activin A is the key SASP mediator of adipogenesis inhibition (Fig 3)

Secretome analysis of senescent fat progenitors identified activin A — a homodimer of Inhibin β-A subunits (encoded by INHBA), belonging to the TGF-β superfamily — as the dominant paracrine inhibitor. Evidence:

  • Neutralizing antibody against activin A added to senescent CM substantially restored adipogenesis in target progenitors.
  • SB-431542 (activin A receptor inhibitor; also blocks TGFβ signaling, acting on ALK4/5/7-type receptors upstream of SMAD2/3) similarly restored adipogenesis when added to senescent CM.
  • Direct addition of recombinant activin A to non-senescent progenitors mimicked the senescent CM effect, inhibiting adipogenesis 3.
  • Activin A levels were elevated in conditioned medium from irradiation-senescent progenitors relative to non-senescent controls.

This positions activin A secretion — acting through its type I receptors (ALK4/5/7) and downstream SMAD2/3 signaling — as the primary SASP axis through which senescent fat progenitors suppress local adipogenesis. gap/needs-human-replication

3. JAK inhibition suppresses activin A secretion from senescent progenitors in vitro (Fig 5)

Because the JAK/STAT pathway is a major transcriptional driver of the SASP, the authors tested whether JAK inhibition reduces activin A production from senescent progenitors:

  • Ruxolitinib at 0.6 ÎĽM for 72 hours significantly reduced activin A mRNA transcript in senescent progenitors (n=7, p<0.05).
  • Ruxolitinib at 0.6 ÎĽM for 72 hours significantly reduced secreted activin A protein by ELISA (n=6, p<0.05).
  • Conditioned medium from ruxolitinib-treated senescent progenitors caused less inhibition of adipogenesis in naĂŻve target progenitors, confirming functional relevance of the activin A reduction.

4. INK-ATTAC genetic senescent cell clearance in old mice blunts fat loss and reduces activin A (Fig 4)

In 18-month-old C57BL/6 mice, AP20187-mediated clearance of p16^Ink4a^-positive senescent cells over 3 weeks:

  • Blunted progressive fat loss: WT mice lost more fat mass than INK-ATTAC+/- mice during the treatment period (p<0.05 at 3 weeks). Old mice are in a state of ongoing fat tissue deterioration; senescent cell clearance partially reversed this trend.
  • Reduced circulating activin A: INK-ATTAC+/- mice showed a >30% reduction in circulating activin A from baseline, while WT controls showed an ~10% increase.
  • Reduced SABG+ cells in adipose tissue (INK-ATTAC+/- vs WT, p<0.05).
  • Increased adipose tissue expression of pro-adipogenic transcription factors:
    • C/EBPα: significantly higher in INK-ATTAC+/- vs WT (p<0.05)
    • PPARÎł: significantly higher in INK-ATTAC+/- vs WT (p<0.05)
    • Lipin-1: increased
  • Decreased adipose tissue expression of senescence/SASP markers:
    • IL-6: reduced
    • p16^Ink4a^: reduced
    • p21^Cip1^: reduced

5. JAK inhibition in old mice rescues fat mass and metabolic parameters (Figs 6–8)

Ruxolitinib (60 mg/kg/day orally in food for 8 weeks) in 22-month-old mice produced a coordinated set of metabolic improvements:

Body composition:

  • Fat mass was preserved in ruxolitinib-treated mice; vehicle controls showed progressive fat mass loss (p<0.05).
  • Inguinal, subscapular, and brown adipose depots were reduced in vehicle controls but preserved in ruxolitinib-treated mice.
  • Lean mass was unchanged in both groups.
  • Body weight did not differ significantly between groups.

Adipose tissue gene expression (ruxolitinib vs. vehicle in 22-month mice; all p<0.05 unless noted):

  • PPARÎł: increased
  • C/EBPα: increased
  • FABP4: increased
  • Adiponectin (Adipo-Q): increased
  • Lipin-1: increased
  • GPAT4 (triglyceride synthesis): increased
  • ATGL (adipose triglyceride lipase): induced
  • HSL (hormone-sensitive lipase): induced
  • Activin A (whole fat and isolated progenitors): suppressed

Systemic activin A:

  • Circulating activin A was significantly reduced in ruxolitinib-treated 22-month-old mice vs. vehicle controls (p<0.05).
  • Young (8-month-old) mice showed minimal/non-significant change in circulating activin A with ruxolitinib.

Lipid metabolism:

  • Plasma free fatty acids (FFA): reduced in ruxolitinib-treated aged mice (p<0.05).
  • Plasma triglycerides: not significantly different.
  • Liver weight: decreased in ruxolitinib-treated mice (p<0.05).
  • Hepatic triglyceride/protein ratio: decreased (p<0.05).
  • Total hepatic triglycerides: decreased (p<0.05).

Glucose homeostasis and insulin sensitivity:

  • Oral glucose tolerance: improved (p<0.05).
  • Insulin tolerance: improved (area under curve, p<0.05).
  • Fasting plasma glucose: not significantly changed.
  • Glucose-stimulated insulin secretion: not altered.
  • Fat tissue ex vivo p-AKT/total AKT ratio (insulin challenge): significantly higher in ruxolitinib-treated mice (n=6, p<0.05) — indicating improved tissue-level insulin signaling.

Specificity — young mice:

  • 8-month-old mice treated with ruxolitinib showed minimal effects on fat mass, adipogenesis markers, and insulin tolerance, consistent with the mechanism requiring a substantial senescent cell burden as the driver.

No metabolic rate or food intake differences were detected in metabolic cages — ruling out effects mediated by altered energy expenditure or appetite.

Extrapolation table (mouse → human)

DimensionStatus
Pathway conserved in humans?Partial — activin A, ALK4/5/7, SMAD2/3 signaling are conserved; in vitro component uses primary human progenitors demonstrating activin A inhibits human adipogenesis; but aging-adipose senescence accumulation in humans not directly quantified in this paper
Phenotype conserved in humans?Partial — age-associated fat redistribution and loss occurs in humans; JAK/STAT-driven SASP is conserved; however, whether senescent fat progenitor activin A secretion accounts for a meaningful fraction of the human phenotype is unknown
Replicated in humans?No — all in vivo endpoints are mouse only; in vitro components use human cells; no clinical trial data gap/needs-human-replication

Why it matters

  1. Mechanistic specificity for a poorly understood aging phenotype. Age-associated fat redistribution (loss of subcutaneous fat, accumulation of visceral and ectopic fat) is a major contributor to metabolic dysfunction in older adults 4. This paper provides a tractable molecular mechanism: senescent fat progenitors → activin A → SMAD2/3-mediated suppression of adipogenesis in healthy neighboring progenitors.

  2. Dual-target validation. Both genetic (INK-ATTAC clearance) and pharmacological (JAK inhibition) approaches converged on the same phenotypic improvement — fat mass preservation, activin A reduction, adipogenic gene rescue — strengthening causal inference despite the absence of randomization or pre-registration.

  3. Ectopic lipid and insulin resistance connection. The reduction in hepatic triglycerides and improvement in systemic insulin sensitivity in ruxolitinib-treated aged mice links senescent-cell-driven fat dysfunction to the broader metabolic syndrome phenotype, including type-2-diabetes risk. The likely mechanism: without functional adipogenesis, non-esterified fatty acids released from dysfunctional old adipocytes are preferentially deposited ectopically in liver (and likely skeletal muscle).

  4. JAK inhibitors as senomorphics. This paper (alongside the companion PNAS paper on JAK inhibition and frailty) established the rationale for repurposing JAK inhibitors as senomorphics — agents that suppress SASP without killing senescent cells. Ruxolitinib is FDA-approved for myelofibrosis; the dose used here (60 mg/kg/day in mice) substantially exceeds the human therapeutic range, limiting direct translatability.

  5. Activin A as aging biomarker candidate. The observation that circulating activin A was reduced >30% by senescent cell clearance (INK-ATTAC) and suppressed by ruxolitinib positions it as a pharmacodynamic biomarker for senomorphic/senolytic treatments in future clinical trials. gap/needs-human-replication

Connections to the wiki

  • sasp — activin A added as a novel SASP effector from fat progenitors with a specific paracrine adipogenesis-inhibition function; extends the SASP canon beyond canonical IL-6/IL-8/MMP axes
  • cellular-senescence — canonical evidence that senescent fat progenitors accumulate in aged adipose and functionally impair the tissue through paracrine mechanisms
  • activin A — central mechanistic effector (protein page being seeded in parallel); this study is the primary in-vivo-validated citation for activin A as a SASP factor suppressing adipogenesis
  • tgf-beta — activin A is a TGF-β superfamily member; ALK4/5/7 → SMAD2/3 signaling axis
  • smad2-smad3 — downstream transcriptional effectors of activin A signaling; SB-431542 rescue experiment confirms this arm
  • jak-stat-pathway — upstream transcriptional driver of SASP activin A expression; ruxolitinib target
  • senolytics — INK-ATTAC clearance arm establishes that elimination of senescent cells rescues adipogenesis; conceptually relevant to senolytic drug strategies
  • senomorphics — ruxolitinib arm is one of the foundational studies establishing JAK inhibition as a senomorphic strategy (cited alongside the companion Xu 2015 PNAS paper on the senomorphics page)
  • type-2-diabetes — insulin sensitivity improvement and hepatic steatosis reduction mechanistically links senescent-fat-progenitor burden to metabolic disease
  • obesity — fat mass paradox in aging (loss of subcutaneous adipose in very old age) is the phenotype studied; ectopic fat redistribution may underlie visceral obesity risk

Limitations

  • All in vivo results are mouse-only. The human components are limited to isolated fat progenitor cell culture experiments. Whether the observed effects translate to aged humans is unknown. gap/needs-human-replication
  • INK-ATTAC model selects for p16^Ink4a^-expressing cells, which may not represent all senescent cells in aged fat tissue. Fat progenitors may differ in their p16^Ink4a^ expression trajectory from other cell types.
  • No randomization or power analysis reported. Authors explicitly note this in their methods: statistical analysis used two-tailed Student’s t-tests, and no formal pre-registration was conducted.
  • Ruxolitinib dose (60 mg/kg/day) is supra-therapeutic in humans. The clinical dose for myelofibrosis is ~15–25 mg twice daily in adults (~0.4–0.7 mg/kg/day), roughly 100-fold below the mouse dose on a mg/kg basis. Allometric scaling reduces (but does not eliminate) this gap; whether effective SASP suppression occurs at clinically achievable human doses is unresolved. gap/dose-response-unclear
  • Young mouse (8-month) comparator arm was not treated with ruxolitinib in the full metabolic battery. The specificity to old mice is inferred from differential response, not a powered age-stratified design.
  • Activin A levels not reported in absolute units (pg/mL) in the primary text — changes expressed as percent from baseline or as relative comparisons to controls. Exact fold-changes and absolute concentrations would strengthen pharmacodynamic biomarker utility.
  • Fat mass preservation vs. restoration. The intervention blunted ongoing fat loss; it did not dramatically restore fat mass to young-mouse levels. Magnitude of rescue in humans is highly uncertain.
  • The companion PNAS paper (Xu et al. 2015 — JAK inhibition, frailty, and SASP) used a different treatment protocol and endpoints; the two papers together strengthen the JAK-inhibitor-as-senomorphic concept but should not be cited interchangeably. See senomorphics for the frailty endpoints from the PNAS paper.

Footnotes

Footnotes

  1. doi:10.1038/nature10600 · Baker DJ et al. · Nature 2011 · “Clearance of p16Ink4a-positive senescent cells delays ageing-associated disorders” · the foundational INK-ATTAC model paper; established that AP20187-mediated clearance of p16^Ink4a^+ cells in BubR1-progeroid mice delayed lordokyphosis, cataracts, and loss of adipose tissue · n=20–25/group · model: BubR1^H/H^ progeroid + INK-ATTAC mice ↩

  2. doi:10.7554/eLife.12997 · Xu M et al. · eLife 2015 · n=6 human donors (in vitro); n=8 mice (INK-ATTAC arm); n=9/group (ruxolitinib arm) · model: primary human fat progenitors + aged C57BL/6 mice · in-vitro + in-vivo · open access (PMC4758946) ↩

  3. doi:10.2337/db10-0013 · PMID 20530742 · Zaragosi LE et al. · Diabetes 2010;59(10):2513–2521 · “Activin A plays a critical role in proliferation and differentiation of human adipose progenitors” · demonstrated that direct addition of recombinant activin A to non-senescent human fat progenitors inhibits adipogenesis; Xu 2015 builds on this to show senescent cells are an endogenous source of activin A in aged fat ↩

  4. doi:10.1111/j.1474-9726.2010.00608.x · PMID 20701600 · PMC2941545 · Tchkonia T et al. · Aging Cell 2010;9(5):667–684 · “Fat tissue, aging, and cellular senescence” · foundational review establishing that senescent cell accumulation in adipose tissue increases with age and is associated with impaired adipogenic capacity and pro-inflammatory changes in fat tissue · gold OA (PMC) ↩

Graph View

Backlinks