The clocks get a report card — and senescence's oldest assay gets audited
This week the field turned its instruments on its own instruments. Nature Medicine published the first systematic test of whether DNA-methylation clocks behave like surrogate endpoints should — 16 clocks across 51 longitudinal interventional studies — and the answer is conditional: mortality- and pace-trained clocks respond, first-generation chronological predictors largely do not, and study design dominates the signal. In the same journal, a deep-learning histological clock built on 25,712 GTEx slides argues that tissue architecture is itself a readable aging substrate. Against that, a research-integrity investigation found at least 54 cell-aging papers that appear to have used an E. coli β-galactosidase antibody to detect mammalian SA-β-gal — a reminder that the senescence literature's most-used marker has a validation problem underneath the biology. Mechanistically, CDK4/6, CK2α–STING and RANKL all gained credible geroscience cases this week, all still preclinical.
of circulating T cells are CD4 cytotoxic lymphocytes in supercentenarians, vs 4% at ages 70–90 (Cell Reports, n = 28)
Where the human evidence is
Count by statusEvidence-stage map
Evidence stage × investigator promise (1–10)Aging clocks on trial
First systematic surrogacy test of epigenetic clocks: 16 clocks across 51 longitudinal interventional studies (TranslAGE)
Sehgal, Borrus and Higgins-Chen curated TranslAGE, a harmonized database of 51 public and private longitudinal interventional studies, and computed a consistent set of 16 prominent epigenetic clocks plus 94 additional DNAm biomarkers for each. The pattern is directional rather than uniform: clocks trained on mortality or pace of aging show the strongest and most mutually consistent responses across interventions, pharmacological and lifestyle interventions drive the largest biomarker movement, and study population and duration are decisive determinants of whether any response is detectable at all. Multi-subscore 'explainable' clocks gave more mechanistic specificity than single-score clocks. In the accompanying News & Views, Steve Horvath frames this as a report card on surrogacy — and discloses GrimAge/PhenoAge patent interests and Altos Labs equity, which is worth holding in view. This is the most rigorous argument yet that clock choice is not interchangeable, and that a single consumer biological-age number is the wrong unit of inference.
Read the paperDeep-learning histological 'tissue clocks' from 25,712 GTEx slides predict organ-specific aging — and can be inferred from blood
Using 25,712 whole-slide histopathology images spanning 40 tissue types from 983 GTEx donors, the authors trained tissue-level biological-age predictors reported at roughly 4.9 years mean prediction error. Age gaps tracked telomere attrition, subclinical pathology and comorbidity, and organ-specific acceleration mapped onto demographic and lifestyle factors. By integrating paired histology and transcriptomics they then predicted tissue-specific age gaps from blood, validating against eight prevalent diseases including Alzheimer's, stroke and Crohn's in independent cohorts. Two caveats for clinical readers: GTEx is a post-mortem donor cohort, and the blood-inferred predictions are a derived layer, not a directly measured tissue readout. Still, this positions tissue architecture as a genuinely orthogonal aging substrate to methylation and proteomics.
Read the paperSenescence: a real drug target and a real measurement problem
Cyclin D1–CDK6 controls the inflammatory senescence secretome; palbociclib improves motor function and frailty in aged mice
Rajesh, Havas and Adams show that the cell-cycle regulators cyclin D1 and CDK6 govern the inflammatory arm of the senescent phenotype via DNA damage and cGAS–STING activation, and that blocking this axis with palbociclib improves motor function and reduces frailty in aged mice. This is a mechanistic complement to — not a repeat of — the abemaciclib senomorphic work covered on Aug 9; the two papers appeared together with a joint News & Views arguing CDK4/6 inhibitors could be repurposed as senomorphics. The translational appeal is that palbociclib and abemaciclib are approved drugs with known human PK and toxicity. The translational brake is equally clear: these are myelosuppressive agents used in oncology, with no aging-indication human data, and the mouse benefit is functional rather than survival.
Read the paperAt least 54 cell-aging papers appear to have used an E. coli β-galactosidase antibody to detect mammalian SA-β-gal
Independent molecular biologist Sholto David identified at least 54 papers — spanning Cell and Nature Aging down to low-impact titles — whose methods list an antibody raised against Escherichia coli β-galactosidase, used to detect β-galactosidase in mammalian cells as a senescence marker. Cross-kingdom reactivity is, as David puts it, far-fetched and unsupported. The trail began with a 2016 Cell paper on in vivo partial reprogramming from the Izpisua Belmonte lab (now at Altos), which listed an E. coli β-gal antibody used on mouse liver; Cell says it is looking into it and Springer Nature says it will assess. Nothing here is proof that any specific conclusion is wrong — several affected experiments were ancillary — but it is a direct hit on the reliability of the single most widely used senescence readout, and it compounds the pre-existing dispute over whether SA-β-gal staining identifies senescence at all. Treat SA-β-gal-only senescence claims, including in supplement and clinic marketing, as unverified.
Read the paperInflammaging: two new upstream controllers
Lamin A/C scaffolds STING and CK2α at the nuclear periphery to drive STING turnover; losing the axis produces systemic aging
Zhang, Jin and Ao report that lamin A/C acts as a scaffold bringing STING and the kinase CK2α together at the nuclear periphery, where CK2α-mediated phosphorylation promotes STING turnover. Loss of this LMNA–CK2–STING brake allows STING to accumulate, driving cGAS–STING-dependent inflammaging, myeloid senescence and systemic aging phenotypes. This is a satisfying mechanistic bridge between two hallmarks that have been discussed in parallel — nuclear envelope decline and DNA-sensing inflammation — and it offers a rationale for why laminopathies phenocopy accelerated aging. Entirely mouse and cell-based; no compound, no human exposure.
Read the paperNoncircadian BMAL1–YAP enhancer activity amplifies persistent inflammation in aged epidermis
Bonjoch, Solá and Solanas show that BMAL1 and YAP cooperate at enhancers to maintain epidermal homeostasis, and that with age this cooperation is redirected — driven by microenvironmental change — toward inflammatory enhancers, sustaining chronic skin inflammation. The notable point is that the effect is explicitly noncircadian: a core clock protein is repurposed away from its rhythmic role. This complicates the common framing that circadian realignment interventions act through BMAL1 rhythmicity per se. Mouse epidermis; no intervention tested.
Read the paperTwo druggable youth signals: brain and bone
TIMP2, a youth-associated plasma factor, reverses aged microglial state and restores phagocytosis in aged mice
Building on TIMP2's established role as a young-plasma factor acting on the extracellular matrix to regulate synaptic plasticity, this work shows TIMP2 deletion exacerbates aging-associated microglial phenotypes — activation transcriptomics, altered lysosomal markers and phagocytosis, and elevated stress and inflammatory proteins in brain extracellular space measured by in vivo microdialysis. Treating aged mice with TIMP2 reversed several of these: reduced microglial activation, a lower proportion of proinflammatory microglia, and enhanced phagocytosis of physiological substrates. TIMP2 is high in very young human and mouse plasma and declines steeply into adulthood. Mouse-only; TIMP2 is a matrix metalloproteinase inhibitor with plausible off-target consequences in tumor biology that would need addressing before any human protocol.
Read the paperRANKL neutralization restores bone mass, reduces muscle fibrosis and extends lifespan in progeroid mice
In Zmpste24-null mice — a Hutchinson-Gilford progeria model — osteocyte-specific RANKL deletion reversed bone loss in long bones and vertebrae, increased grip strength, improved endurance and increased survival. A translational arm using a neutralizing anti-RANKL antibody reproduced the bone effect, reduced muscle fibrosis and extended lifespan. The extra-osseous benefit is the interesting part: it argues RANKL has a role in aging beyond bone remodeling, consistent with earlier signals for RANKL in muscle aging and with the survival data reported for bisphosphonate users. The obvious clinical hook is denosumab, already approved. The obvious limitation is that this is a progeroid model, not normal aging — a naturally aged cohort has not been reported — and denosumab carries a well-characterized rebound bone-loss risk on discontinuation.
Read the paperInterventions testing & translation infrastructure
Five mechanistically distinct lifespan-extending interventions converge on a shared metabolomic signature — with ergothioneine rising in brain, plasma and muscle
Male UM-HET3 mice treated from 4 to 12 months with rapamycin, acarbose, 17α-estradiol, canagliflozin or caloric restriction were profiled metabolomically across seven tissues. A feature-stabilized XGBoost pipeline identified treated animals as recipients of a lifespan-extending intervention well before survival curves separate, and — the striking part — a leave-one-intervention-out procedure showed models trained on any four interventions correctly classified mice receiving the fifth, unseen intervention. The dietary antioxidant ergothioneine rose consistently in brain, plasma and muscle across interventions, alongside shared lipid remodeling. If this holds under peer review it is a candidate short-horizon efficacy readout for geroprotector screening. It is not evidence that supplemental ergothioneine extends lifespan — the convergence is correlative and the direction of causality is untested.
Read the paperXPRIZE Healthspan names 20 finalists; 10 receive $1M Milestone 2 awards, clinical testing runs 2026–2029
XPRIZE announced 20 finalist teams in the $101M Healthspan competition at a ceremony in Salt Lake City, with 10 teams from the US, South Korea, Japan and China receiving $1M Milestone 2 awards. Finalists will run trials of up to one year each between 2026 and 2029, benchmarked against restoration of muscle, cognitive and immune function by at least 10 years (target 20) in adults aged 50–90; the grand prize resolves in 2030. The approaches span biologics, stem-cell and gene therapy, extracellular vesicles and AI-guided precision protocols. The structural significance is that this is a head-to-head, prespecified-functional-endpoint comparison — an explicit rejection of clock-based scoring — which is the right instinct given this week's clock-surrogacy findings. No efficacy data exist yet for any finalist.
Read the paper