Longevity InTime: Autonomous AI Institute. Anti-Aging Digital Health Immortality Transhumanist AI Channel

Second Century Foundation targets aging research overlooked by NIH and venture capital President Michael Rizman of the Second Century Foundation, a Centerview Partners banker advising biotech sales, launched the foundation on April 25 at the NOVA conference. By October 13, the foundation’s website outlined its projects. It identifies two funding filters: NIH does not support work without publication, while venture capital avoids work without a ten-year return. On Rizman’s slides, these appear as two questions: public and nonprofit funders ask “Can this be published?”, while industry money asks “Can we achieve more than a fivefold return in ten years?” Rizman includes student clubs in this category: the U.S. has about 4,000 universities but fewer than ten active longevity clubs, despite clubs costing less than $10,000 to start. The foundation’s simulator compares treating individual diseases to slowing aging. In its calculation, annual mortality risk rises exponentially with age; curing a disease removes deaths only from that cause, whereas slowing aging—by the foundation’s assumption—delays death from all causes simultaneously. About half of Americans live to roughly 82 years; curing cancer shifts this to 85, while slowing aging by 20% starting at age 40 pushes it to 90. Rizman analyzes the IPOs of three companies with aging-related drugs. Sirtris, co-founded by aging researcher David Sinclair, went public in 2007 at $10 per share and was acquired by GSK in 2008 for $22.50; per the slides, the drug failed, yet IPO investors earned more than double their investment. Scientist Matt Kaeberlein believes pharma companies were burned by the Sirtris hype and are only now returning to aging. In 2018, Morgan Stanley wrote about Unity Biotechnology: “Removing senescent cells opens a unique approach to fighting chronic diseases”; Unity was dissolved in 2025 after failed trials. BioAge halted its drug trial in December 2024 due to elevated liver enzymes in some participants. Per Rizman’s analysis, the investor documents for all three companies followed the same structure: first, aging-related diseases; then a broad target and initial disease (for Sirtris, calorie restriction mimetic and type 2 diabetes; for BioAge, exercise mimetic and obesity); then expansion to other diseases. He explains why aging cannot be tackled directly via a thought experiment: Sinclair asks FDA for permission to prove a drug extends life, and hears about a twenty-year placebo trial with 10,000 patients, while investors tell him: “You’re fired.” Rizman concludes: “Wall Street will grasp the enormous market opportunity, and if your drug looks promising, the labels ‘longevity’ and ‘aging’ won’t stop you.” The path to company sale or IPO is also described in an investor memorandum the foundation co-writes with Norn Group, which grants longevity research funds; it includes a calculation of today’s value of future drug revenues minus costs. A money-flow map will show university, corporate, and government spending on longevity and highlight areas the foundation deems underfunded. The foundation awards scholarships to those whose research would not occur without them. The first recipient, a Columbia University graduate student, develops viral-free delivery of gene and cell therapies. As described by the foundation, these therapies currently reach only 10% of tissues, while viral carriers can trigger inflammation. 🔗 Read original →

When does fasting trigger autophagy in humans? The widely cited 24–72 hour window likely stems from mouse studies, not human data, according to a review published October 13 in Biochemical Society Transactions. Authors Sanjna Singh and Timothy Sergeant from the University of Adelaide analyzed 20 human studies and compared them with animal research, finding that while certain diets induce autophagy and extend lifespan in animals, the timing cannot be directly translated to humans. Mice begin to die after three days without food, whereas humans can survive at least three weeks — as demonstrated by 13 volunteers in fasting trials. Hormonal responses also differ: FGF21 rises in mice within six hours of fasting, but only becomes noticeable in humans after seven to ten days. Measuring autophagy flux in humans remains challenging. Most studies assess autophagy proteins or gene activity in blood, muscle, or fat cells at a single time point, but LC3B-II — a marker on autophagosomes — accumulates both when autophagy is active and when it is blocked, making it impossible to distinguish flux from stagnation. To measure true flux, researchers split fresh blood samples, adding a lysosomal inhibitor to one half and comparing LC3B-II accumulation; only blood can be kept viable outside the body for this assay. Of the 20 human studies reviewed, flux was measured in only six, with four showing no significant change. In one 24–72 hour fasting study, flux did not change after 24 hours, while LC3B-II increased only after one hour in nutrient-reduced media — which itself suppresses flux. In a trial of 121 obese individuals, intermittent fasting was associated with higher autophagy flux after six months compared to standard care, though flux did not significantly rise from baseline — possibly due to declining flux in the control group. Three of the six flux-measuring studies were conducted by the review’s authors, and Sergeant is listed as an inventor on a patent for the flux measurement method. The same method will be used in an upcoming registered trial: 24 participants will undergo a three-day water-only fast to assess whether muscle and fat biopsies can measure flux and to identify a bodily fluid biomarker that could accelerate autophagy-targeting therapies. The authors conclude that mouse-derived timings for autophagy induction cannot be reliably applied to humans. 🔗 Read original →

Bacterial antiviral system DRT10 writes DNA repeats from an RNA template, like telomerase extending chromosome ends, and its enzyme shares an evolutionary branch with telomerase On October 12, a study from Samuel Sternberg’s lab at Columbia University was published in Nature describing the bacterial antiviral system DRT10. Its enzyme, a reverse transcriptase, synthesizes a DNA strand by copying identical units from an RNA template. The lab had previously studied two related defensive enzymes, DRT2 and DRT9, which repeatedly copy segments of their own RNA into DNA. DRT10 emerged as the next candidate, often found in genomes adjacent to defense genes. Graduate student Steven Tan identified in E. coli carrying DRT10 a DNA chain composed of a nine-letter unit repeated in tandem. The template consists of nine RNA “letters,” immediately followed by a copy of its first four letters. The enzyme transcribes the nine letters; the DNA strand then dissociates from the RNA and pairs with the copied segment, after which the enzyme resumes copying the template from that point. In the preprint of this work, among twenty RNA substitutions made around the template, five disrupted antiviral defense — all five impairing the match between template and copy. Altering only the start of the template or only the copy abolished repeats; changing both identically restored them with a new sequence. Telomerase operates via a similar mechanism: cell division shortens telomeres, the protective chromosome ends, and telomerase elongates them by repeatedly copying a short segment of its own RNA. In both enzymes, the template boundary is marked by a folded RNA region — a hairpin in DRT10 — while the growing DNA chain disengages from the template and re-pairs with RNA to initiate a new cycle. When the RNA template of DRT10 was replaced with that of human telomerase, the enzyme synthesized telomeric repeats TTAGGG in both E. coli and human cells. In the 3D structure of the complex, the two enzymes lie on opposite sides of a single RNA molecule, each with its own template. Only the enzyme whose template is accompanied by a copy of its start synthesizes a long DNA chain; the other produces short fragments. Swapping the RNA template regions caused the other enzyme to begin building the long chain. Template boundaries are defined by the distance between two RNA hairpins: removing two letters between one hairpin and the template reduced the repeat unit from nine to three letters. The same two-enzyme complex was identified in a DRT10 variant from Tianjin Medical University by the preprint’s authors. Recalling Sternberg, the idea of a shared ancestor between DRT10 and telomerase arose during a talk by Mario Rodrigues Mestre of the University of Copenhagen, who presented a preliminary tree of all reverse transcriptases. “I interrupted Mario: I was genuinely intrigued by the turquoise strip of eukaryotic reverse transcriptases tightly clustered with bacterial ones.” That strip turned out to be telomerase. Previously, telomerase’s origin was traced to mobile element enzymes — “jumping” DNA segments — but in a tree of 6,734 reverse transcriptases, telomerase fell into a branch with bacterial defensive enzymes. The authors propose that the ability to write DNA repeats from an RNA template originated in a common ancestor of bacteria and eukaryotes, prior to the emergence of linear chromosomes in eukaryotes. The lab is now testing whether telomere elongation can protect cells from aging and hopes to determine if DRT10 could eventually be used to treat diseases involving defective telomere maintenance. “This is only the beginning, but we are encouraged by the possibility of turning a bacterial defense enzyme into a tool that restores protective DNA at chromosome ends,” says Sternberg. Nature, October 2026 🔗 Read original →

AlphaFold3, Chai-1, and Boltz-2 outperform docking in placing molecules on coronavirus protein structures A study by the University of California, San Francisco, published in eLife on October 12, evaluated AlphaFold3, Chai-1, and Boltz-2 on 489 SARS-CoV-2 protein structures solved after the models’ training. The neural networks placed drug-like molecules more accurately than traditional docking based on physical formulas. Among docking candidates, AlphaFold3 and Boltz-2 distinguished true binders from false positives without a clear advantage over docking alone. The researchers focused on Mac1, a SARS-CoV-2 enzyme that suppresses the cell’s antiviral signal. They provided the enzyme’s sequence and molecular formulas, then compared predicted binding positions to experimental crystal structures. Docking “fits” molecules into the protein’s binding pocket — a cleft where drugs bind — by selecting the best fits from hundreds of millions or billions of options using physical formulas. Neural networks, in contrast, predict the full protein–ligand complex directly. Earlier work showed neural networks excel on complexes similar to their training data, suggesting memorization. In September, the MIRAGE benchmark revealed a similar trend for Boltz-2: on proteins rare in public databases, a simple algorithm outperformed it. For Mac1, the team identified nearly a thousand protein–ligand structures via X-ray crystallography and selected 551 solved after model training; 489 were evaluable. A prediction was considered correct if the molecule’s atoms deviated from their crystal positions by less than 2 Å on average. AlphaFold3 and Chai-1 succeeded in about 70% of cases; docking achieved 40%. Boltz-2 reached 53%, despite being trained on data through mid-2023, while AlphaFold3’s training ended in September 2021. To test memorization, the authors compared AlphaFold3’s error to molecular similarity with training examples — no significant correlation was found. Docking uses a fixed protein structure, whereas neural networks predict both the protein and its pose, enabling them to model protein rearrangements like chain flips or loop openings. Of 19 observed chain flips, AlphaFold3 predicted 4; of 20 loop openings, it predicted none — yet often placed the molecule correctly anyway. The authors interpret this as evidence of a “non-physical” aspect of machine learning: networks learn typical contact patterns rather than physical forces. Only Boltz-2 estimates binding strength. For 202 molecules, the correlation between its predicted affinity and experimental measurement was 0.6 (1 = perfect match), while docking energy scores correlated at only 0.23 in absolute value. Neural networks are roughly ten thousand times slower than docking, so in a second test, AlphaFold3 and Boltz-2 re-ranked shortlists generated by docking from libraries for three proteins: AmpC enzyme, σ2 receptor, and dopamine D4 receptor. In vitro, some of these molecules do not bind the target protein. Boltz-2 outperformed docking in distinguishing true from false binders on σ2, matched it on D4, and underperformed on AmpC; AlphaFold3 lost to docking on all three. Even when docking gave medium or poor scores, neural networks failed to improve on its rankings. The authors attribute this divergence to task differences: in the Mac1 set, all molecules bound and fell into four families, so the task was mainly to place a familiar molecule. Docking’s shortlists were diverse and intentionally dissimilar to known binders — a scenario where neural networks relying on learned patterns struggle. The authors propose a division of labor: docking screens novel molecules, neural networks refine the pose of promising hits, and — like Boltz-2 — rank their analogs. 🔗 Read original →

Gene activity clocks reveal intestinal epithelium temporarily younger during healing A preprint by Gladyshev lab researchers Sharif Ikbal and Vadim Gladyshev from Harvard Medical School, posted on bioRxiv October 13, shows that intestinal epithelium in mice becomes biologically younger than undamaged tissue during peak healing. Using gene activity-based aging clocks trained specifically on epithelial cells from 14 studies, the authors found that in all tested injury models — colitis, irradiation, and intestinal cell cultures — epithelial age dropped below control levels at the point of minimum age, with statistical significance in seven comparisons. In young mice (4 months), epithelial age reached its minimum by day 4 after 5-fluorouracil injury and slightly exceeded baseline by day 6. In old mice (21 months), the decline began on day 4 but the minimum occurred later, and by day 6 age rose above baseline more sharply than in young mice. The authors note: “The ability to exit the regenerative state may be as important as the ability to enter it.” During low-age states, epithelial cells activated YAP-dependent repair genes and rare stem cell genes that replenish the stem cell pool after injury, while nutrient-absorbing mature cell genes were suppressed. This aligns with prior work showing that in old mice, the protein asporin prolongs the regenerative state of epithelium. The study also examined liver, where cells rarely divide. Applying 24 multi-species clocks from a Nature 2026 study to published 2023 data, the authors found that removal of two-thirds of the liver reversed age-related DNA packaging changes in old mice. The most robust clocks showed liver age below preoperative levels from 48 hours to ten days post-surgery, with a reduction of about half the difference between old and young liver — whereas in intestine, age returned to baseline over time. Notably, epithelial clocks rated tumor foci in mouse colon as “younger,” while whole-tissue clocks rated them as “older.” The authors explain that epithelial clocks reflect the regenerative state of epithelium, whereas whole-tissue clocks capture inflammation and connective tissue remodeling. They conclude: “A prediction of younger age must be read alongside what the cells are actually doing.” 🔗 Read original →

Beam Therapeutics reports first-in-human data for BEAM-301 in glycogen storage disease type Ia On October 13, Beam Therapeutics shared initial results from its Phase 1/2 trial of BEAM-301, a DNA base editor for glycogen storage disease type Ia (GSD Ia). Six adult patients received a single intravenous infusion: three at 20 mg and three at 40 mg. Data are current as of June 26, with updated results expected October 28. In GSD Ia, mutations in the G6PC gene impair glucose-6-phosphatase function, reducing hepatic glucose release and causing fasting hypoglycemia. Patients typically manage this with frequent oral cornstarch doses. The R83C mutation—a single-base substitution—is among the most common and severe forms, often leading to hypoglycemic shock within one to three hours without starch. BEAM-301 uses lipid nanoparticles to deliver mRNA encoding a base editor and guide RNA to hepatocytes. The editor nicks one DNA strand and converts the mutant C•G base pair to T•A, aiming to restore functional enzyme expression. Preclinical work in humanized R83C mice showed BEAM-301 corrected up to 60% of gene copies, enabling fasting tolerance of 24 hours or more for up to one year, versus 60–75 minutes in untreated controls. In humans, fasting tolerance was assessed via controlled challenge. At baseline, median time to dangerous hypoglycemia (<55 mg/dL) was 2.9 hours in the 20 mg group and 2.5 hours in the 40 mg group. By day 36, the 20 mg group’s median increased to 4.3 hours. In the 40 mg group, two of three patients were evaluable by day 36; both tolerated fasting for 15 hours or longer—the protocol’s maximum limit. Beam interprets this as consistent with restored endogenous glucose production. Median follow-up was 313 days for the 20 mg group and 53 days for the 40 mg group. In the 40 mg group, daily cornstarch dose decreased by 71% at day 50 and 92% at day 93, with daily doses dropping from 5–6 to 1–4. In the 20 mg group, starch reduction was 74% at day 128, 71% at day 227, and 11% at day 387, with daily doses reduced by 0–4 from a baseline of 6–7. All six patients experienced treatment-related adverse events. Five were serious, including three Grade 3 infusion-related reactions; all resolved within 2–10 days. Beam notes that short-acting steroids given pre-infusion may have contributed and has revised its premedication regimen. No deaths or treatment discontinuations occurred due to adverse events. On August 19, the FDA granted accelerated approval to Ultragenyx’s GENGLYCOS, an AAV8-delivered gene therapy for GSD Ia that provides a functional G6PC gene extrachromosomally. Approval was based on similar cornstarch reduction metrics. Unlike GENGLYCOS, BEAM-301 is not limited by preexisting AAV8 antibodies and is designed specifically for the R83C mutation. Study data presented at Beam Therapeutics investor update, October 2026 🔗 Read original →

NIH awards $3.5 million to UConn to grow thymus cells from reprogrammed elderly cells The National Institutes of Health (NIH) has awarded the University of Connecticut $3.5 million over five years to develop thymus cells from reprogrammed cells of elderly individuals. The thymus, which produces immune T-cells, gradually atrophies with age, leading to weakened immunity in older adults. On October 12, the university announced that Professor Lai-Zhun Lai received the grant to lead this research. His lab will transplant these lab-grown thymus cells into aged mice to test whether they can support the development of human T-cells. Thymic atrophy begins in infancy, with epithelial tissue replaced by fat and T-cell output declining sharply over time. According to a 2018 study in PNAS, the production of new T-cells halves approximately every 16 years. This decline is considered a root cause of age-related immune deterioration: while a young person may recover from a cold in days, the same virus can cause life-threatening illness in the elderly, and vaccines are less effective. Thymic epithelial cells (TECs) create the environment where naive T-cells — those that have not yet encountered an antigen — mature. Cortical TECs select functional T-cells, while medullary TECs eliminate those that target the body’s own tissues. With age, TECs diminish in number and function, which the university identifies as a key factor in thymic decline. The age of naive helper T-cells influences immunological memory. In 2003, co-investigator Laura Haynes and colleagues demonstrated in mice that memory formed from young mice’s T-cells remained strong after a year, whereas memory from old mice’s T-cells was weak. In a 2015 paper, Lai’s lab described a protocol to convert human embryonic stem cells into TEC precursors using the proteins HOXA3 and FOXN1 — genes essential for thymus development, as mice lacking them fail to form a thymus. When transplanted into mice, these precursors matured into cortical and medullary TECs, and within 24 weeks, mouse T-cells developed in the grafted tissue. When human hematopoietic stem cells (which give rise to T-cells) were co-transplanted, functional human T-cells were also generated. Instead of using embryonic stem cells foreign to the recipient, Lai now uses induced pluripotent stem cells (iPSCs) derived from the patient’s own cells. These iPSCs can differentiate into any cell type and are less likely to be rejected. Although cells from elderly individuals carry age-related markers, French scientists showed in 2011 that reprogramming resets these signs: iPSCs from centenarians matched embryonic stem cells in telomere length and gene expression, and cells derived from them exhibited rejuvenated characteristics. The project, funded by the National Institute on Aging, tests whether thymus cells grown from iPSCs of elderly people can create a supportive environment for new T-cell maturation in an aged organism. The lab will generate TEC precursors from iPSCs and transplant them into aged, immunodeficient mice that do not reject human cells, alongside human hematopoietic stem cells. Researchers will quantify TECs and newly formed T-cells, assess their function — including by challenging animals with influenza virus — and verify that the T-cells do not auto-react against self-tissues. Separately, using mouse cells, the team will determine which developing T-cells assist in TEC maturation: preliminary experiments cited in the grant application showed that co-culturing precursors with developing T-cells helped the precursors become TECs. “Currently, there are no approaches that address T-cell immunodeficiency in the elderly,” Lai said. 🔗 Read original →

Preprint: New cell specialization index based on DNA methylation marks shows little change during partial reprogramming despite epigenetic clocks indicating rejuvenation October 11, a team led by Andrew Teschendorff from the Chinese Academy of Sciences posted a preprint on bioRxiv introducing the EpiDiff model. The model calculates a cell specialization index from DNA methylation patterns — chemical tags on DNA — ranging from 0 (pluripotent cell, capable of becoming any cell type) to 1 (fully specialized cell). Epigenetic clocks estimate cellular age using the same marks. During partial reprogramming — brief exposure to factors that induce pluripotency — these clocks show rejuvenation. Full reprogramming erases specialization, so partial reprogramming is halted earlier. The specialization index should reflect this loss; the authors built it to be used alongside clocks. The index was trained on 146 samples: 25 pluripotent cells, 86 fetal tissues and umbilical cord blood (intermediate state), and 35 adult tissues. It is based on 1,298 DNA regions where fetal tissue methylation differs from pluripotent cells by at least 30 percentage points, and adult tissue differs by at least 50 points. These thresholds were chosen to exclude shifts from cell division. The index responds to loss of specialization: fully reprogrammed pluripotent cells show very low values. Applying the index to data from two prior studies, the authors found that in a 2020 study, skin fibroblasts and vascular endothelial cells treated with six reprogramming factors for four days showed a 3.4-year reduction in epigenetic age, but the specialization index remained unchanged. In a 2022 study, fibroblasts from donors aged 38–53 received factors for 10–17 days (instead of the full ~50 days), then selected successfully reprogrammed cells were cultured without factors; the index stayed close to original fibroblast levels, though slightly lower. The authors conclude that partial reprogramming largely preserves specialization. The index increases with age even in isolated cell types — liver cells, bronchial epithelium, astrocytes (brain support cells), and naive T-cells that have not encountered antigen. It sums specialization signals across all directions: in a simulated mix of two mature cell types, the index is higher than in either pure type, so contamination by mature cells raises it. In naive T-cells, the authors attribute the age-related increase to such contamination: the selection marker is also present in mature cells, whose proportion grows with age. In liver, bronchial, and astrocyte cells, the increase is attributed to drift: with age, DNA methylation in these cells shifts toward the state of other lineages. To test this, the authors selected from the 1,298 regions those distinguishing germ layers: 247 characteristic of mesoderm (giving rise to muscle, bone, blood), 139 of ectoderm, and 70 of endoderm. In most datasets, methylation at these regions changes with age more often than at random sites, and in the same direction as during embryonic cell specialization. This effect is strongest in mesoderm-associated regions; in astrocytes, it is observed only in them. The authors term this epigenetic drift toward mesenchyme, the embryonic connective tissue. A similar drift in gene activity across more than 40 human tissues was reported by Juan Carlos Izpisua Belmonte’s group in Cell: partial reprogramming reduced it even before cells lost specialization. 🔗 Read original →

Dendritic cell trapping drives lymphoid-like clusters in aging mice, senolytics block formation Genentech researchers posted a preprint on bioRxiv on October 11 showing that in mice, dendritic cells — immune scouts that collect foreign material in tissues and migrate to lymph nodes to activate T cells — become trapped in organs with age. This leads to the formation of lymphoid-like clusters containing T and B cells, complete with vessels and signaling proteins resembling lymph nodes. In humans over 70, such clusters are more common than in those under 50, and mature dendritic cells ready to migrate are almost entirely retained within these clusters. The CCR7 receptor guides dendritic cells to lymph nodes, but with age, their ability to follow this signal drops by about half, according to mouse experiments cited by the authors. To test whether trapped dendritic cells initiate cluster formation, the researchers deleted CCR7 only in conventional dendritic cells in mice — avoiding systemic autoimmune disease seen with full-body knockout. By 3–4 months, these mice developed lymphoid-like clusters in healthy organs without tissue damage or autoantibodies, while control mice showed weaker, later-forming clusters. Even when only half the dendritic cells lacked CCR7 — modeling age-related decline — clusters still formed. Removing dendritic cells for one week caused established clusters to dissolve. Increasing dendritic cell numbers triggered clusters in normal mice, which mostly resolved after the stimulus ended. But in CCR7-deficient mice, clusters persisted. The authors conclude that a temporary excess of dendritic cells can start cluster formation, but sustained presence is needed to maintain them. Since clusters in CCR7-deficient mice appear only with age, the team investigated cellular senescence. Senescent cells, which stop dividing and secrete inflammatory signals, were targeted with dasatinib plus quercetin. In two-month-old CCR7-deficient mice, this senolytic reduced new cluster formation, though it did not dissolve existing clusters in adults. In vitro, senescent connective tissue cells pushed dendritic cells toward an anti-inflammatory state. The authors propose that aged tissue imparts these traits to dendritic cells, and their retention allows them to organize clusters. Within clusters in CCR7-deficient mice, regulatory T cells — which suppress immune responses — were more abundant than in controls. In a colitis model using dextran sulfate sodium, CCR7-deficient mice initially lost weight like controls but already had more regulatory T cells in the gut. After toxin removal, they recovered better: less intestinal shortening, inflammation, and ulceration. The authors note that in lung cancer, dendritic cell migration to tumor-draining lymph nodes is required for early lymphoid structure formation (per a Science study), and retention accelerates this process. They suggest age-related clusters may represent an “organized immune adaptation arising to promote tissue tolerance and protection in aged tissues,” where tolerance means immune restraint toward self. Targeting dendritic cells therapeutically, they propose, could help suppress local inflammation in chronic inflammatory diseases, autoimmunity, and organ transplantation. 🔗 Read original →

Rubedo reports RLS-1496 cream reduces skin epigenetic age in atopic dermatitis patients Rubedo reported that after four weeks of treatment with its RLS-1496 cream, the epigenetic age of skin in 11 atopic dermatitis patients decreased by an average of 2.1 years, as measured by DNA methylation clocks. Mitra Bio’s clocks assessed skin samples from sun-exposed areas before treatment and on day 36. In some patients, the reduction reached 8 years, and in the best responder on day 63 — about a month after treatment ended — it was 11.7 years. Rubedo published the atopic dermatitis trial results on October 12, following presentation at the ARDD conference in Boston on October 2. The company develops drugs targeting pathological senescent cells — cells that have stopped dividing and can secrete inflammatory substances. DNA methylation changes predictably with age, and trained models, epigenetic clocks, read tissue age from these marks. As reported by WIRED, Mitra Bio created these clocks and independently analyzed the samples for Rubedo. Mitra Bio validated the clocks against sun damage: in 2025, among 75 volunteers, the epigenetic age of sun-protected back skin was 3.8 years younger than forehead skin. Rubedo noted in March that trial participants applied the cream to both diseased skin and sun-aged skin. RLS-1496 targets GPX4, an enzyme that protects cellular membrane lipids from oxidation. Rubedo proposes that temporary GPX4 inhibition drives pathological senescent cells into ferroptosis — iron-dependent membrane lipid peroxidation — while stressed but non-senescent neighboring cells recover function. Marco Quarta, Rubedo co-founder, called this approach “seno-adaptive” in May. In April, Jesús Gil’s group at Imperial College London independently showed in Nature Cell Biology that GPX4 inhibition kills senescent cells via ferroptosis in cell cultures and mice; these cells survive via elevated GPX4 despite higher iron and oxidant levels. In Rubedo’s trial, biopsies showed fewer molecular markers of senescence and inflammation, and microscopic sections revealed a 10% increase in type III collagen fibers — linked by WIRED to skin elasticity. The drug was largely undetectable in most participants’ blood, but inflammatory proteins, including CCL22 (associated with dermatitis severity), decreased. Rubedo believes topical GPX4 modulation may influence systemic inflammation. In a separate trial for actinic keratosis — precancerous sun-induced lesions — lesion count decreased by an average of 49% over 28 days in 22 patients, per Rubedo. In non-senescent cells intended to be restored, the company observed increased NRF2, a protective protein. Eric Verdin, president of the Buck Institute for Research on Aging, told WIRED: “I would perhaps be a bit more cautious about what these clocks actually measure, given how many have emerged.” Dermatologist Anthony Rossi of Memorial Sloan Kettering Cancer Center noted: “Some senescent cells are beneficial because they suppress things, for example cancer progression.” Rubedo CEO Frederick Beddingfield sees potential in a weaker, over-the-counter version of the drug to compete with retinol. The company plans to launch a Phase 2a trial for androgenetic alopecia in October. 🔗 Read original →

Sparks Brain Preservation reduces response network from eight to five sites, cutting U.S. population reach within six hours from 84% to 64% On October 13, advisory board member Matija Roy posted on X that four new sites under construction would enable six-hour access to 64% of the U.S. population. Sparks confirms this figure applies to its five total sites, including Salem. The organization’s website still lists the prior plan of eight sites covering 84% of the population. Sparks Brain Preservation, a Salem, Oregon-based nonprofit, perfuses the brains of the deceased with an aldehyde solution that crosslinks proteins to prevent decomposition. The extracted brain is then stored in refrigeration fluid. Standard preservation costs $45,000, while members with two or more years of tenure pay $36,000. The procedure ideally begins within minutes of death. Currently, technicians typically fly to the location, a process founder Jordan Sparks says “can literally take all day.” To address this, Sparks is building regional sites equipped with ambulances and staff. According to the Sparks website, nearly all fees cover rapid team deployment, with storage costing almost nothing. For international members, the process differs: a local physician extracts the brain without a team and ships it to Salem. Long-tenure members pay $3,000 instead of $36,000 (excluding local services). By comparison, the American Cryonics Institute, which stores bodies in liquid nitrogen, charges $28,000 for basic service to lifetime members. Team deployment and body stabilization with transport are sold separately by Suspended Animation, at up to $30,000 per service. At Sparks, team deployment is included in the price, which Sparks says is below cost — the difference is covered by the organization. In December, Sparks unveiled a plan for seven new sites excluding Salem. On September 19, at a brain preservation summit, he explained the change: “Seven sites all at once was too much. The new short-term plan is five sites total.” The retained sites are Salem, Oakland, Dallas, Chicago, and New York. Los Angeles, Atlanta, and Palm Beach were dropped. As of September 19, no new site had staff: Chicago was slated to open around September 30, 2026, and Oakland around May 1, 2027. Sparks estimates the New York site will cover 73 million people — one-third of the total reach. He notes that state health department registration is required, and without in-state presence, services cannot be offered; the New York site is expected to open around June 1, 2027. Due to Texas regulations, Sparks expects to open the Dallas site around 2028, stating that without in-state presence, bodies cannot be accepted, and brain removal from Texas is currently prohibited. Sparks anticipates brain preservation will suddenly gain popularity in thirty to forty years — possibly sooner — and is preparing the network for that moment. He adds that sites serve another purpose: “They’re a form of marketing. They attract clients and reporters, and reporters do our advertising for us.” Roy estimates that a non-Salem site breaks even without donations at ten preservations per year. The Sparks homepage lists 23 preserved individuals, 97 members, and 20 years of operation. Monthly reports from December to September show two to seven new clients per month (combined service subscriptions and memberships). If demand does not materialize, Sparks describes a “hibernation” mode: preserved individuals are maintained using investment income from existing assets. According to Roy’s slide, this mode could last indefinitely, as a fixed brain requires no liquid nitrogen, which would otherwise need constant replenishment. 🔗 Read original →

PUS7 enzyme replaces OCT4 in Yamanaka factors, yielding human iPS cells closer to embryonic stem cells On October 12, a study from the laboratories of Yanhong Shi (City of Hope) and Chuan He (University of Chicago) was published in Nature Chemical Biology, showing that the enzyme PUS7 can substitute for the exogenous OCT4 factor in the Yamanaka cocktail used to reprogram human cells into induced pluripotent stem (iPS) cells. The resulting iPS cells exhibited molecular and functional characteristics closer to embryonic stem cells than those generated with the standard OCT4, SOX2, KLF4, and MYC (OSKM) set. The researchers first observed that fibroblasts from four individuals with loss-of-function mutations in the PUS7 gene reprogrammed poorly into iPS cells, failing to initiate the second wave of endogenous pluripotency gene activation despite normal early reprogramming. Tracing this defect, they identified that PUS7 catalyzes the conversion of uridine to pseudouridine on RNA — a modification previously linked to intellectual disability when disrupted. Mapping pseudouridylation sites revealed a specific mark at position 250 in the 7SK non-coding RNA. This modification enhances the binding of the BAF chromatin remodeling complex subunit ARID1A to 7SK, promoting open chromatin states conducive to pluripotency. In PUS7-deficient cells, chromatin accessibility was reduced, and endogenous OCT4 binding to DNA was diminished by day 16 of reprogramming. Restoring the pseudouridine mark on 7SK via dCas13b-PUS7 fusion rescued iPS colony formation in mutant cells. Building on this mechanism, the team replaced exogenous OCT4 and its associated p53-suppressing RNA construct with PUS7 in the OSKM backbone, creating the PSKM set (PUS7, SOX2, KLF4, MYC). PSKM successfully generated iPS cell colonies, and when injected into mice, these cells formed teratomas containing tissues from all three germ layers — a gold-standard test of pluripotency. Critically, SOX2, KLF4, and MYC alone failed to produce colonies unless PUS7 was recruited to 7SK. PSKM-derived iPS cells activated the endogenous OCT4 locus, and reprogramming efficiency dropped when BRG1, a core ATPase of the BAF complex, was inhibited, confirming BAF dependence. Compared to standard OSKM-iPS cells, PSKM-iPS cells showed normalized expression of imprinted genes in the DLK1–DIO3 genomic region — a locus whose dysregulation correlates with diminished iPS cell potential. Neural progenitors from PSKM cells engaged brain-specific transcriptional programs, whereas those from OSKM-iPS cells skewed toward mesodermal fates. Overall, PSKM-iPS cells exhibited a transcriptome profile marginally closer to embryonic stem cells than OSKM-iPS cells, which the authors attribute to the absence of exogenous OCT4 and its potential to induce aberrant gene expression, possibly mediated through the 7SK pseudouridylation mark. Notably, global chromatin accessibility was similar between the two iPS cell types, suggesting distinct reprogramming trajectories can converge on a comparable pluripotent state. 🔗 Read original →

Ginsenoside Rg3 from ginseng extends lifespan and protects neurons in C. elegans A team from Shanghai Jiao Tong University fed Caenorhabditis elegans ginsenoside Rg3. The authors link lifespan extension and neuroprotection to lysosomes — cellular compartments that digest unwanted material — and to lipids on their membranes. The study was published in Aging Cell on October 10, 2026. Two of the 14 authors work at Sirio Pharma, which provided additional funding. Since neurons do not divide, damaged proteins must be cleared by autophagy — the cell’s recycling of its own components. This process weakens with age, so the authors tested whether Rg3 could preserve it. The work builds on two 2023 papers. In PLoS Biology, a Chinese group showed that the enzyme VPS-34 in aging motor neurons slows the conversion of lipid PI(3)P to PI(4)P, and partial inhibition of VPS-34 eased motor aging in worms and mice. In Cell, a Berlin group found in cells that switching between these lipids controls lysosomes: during starvation, PI(3)P disappears from lysosomes, making them more active at digestion, while mTORC1 — the cell’s nutrient sensor — is suppressed. In disease models, Rg3 had previously shown neuroprotective effects, but its impact on organismal aging remained unclear. A total ginsenoside extract in 2021 extended worm lifespan by 14%, though the main active components were identified as Rg1, Re, and Rb1. With Rg3 in the diet, worms lived longer than controls, even when supplementation began only on the fifth day of adulthood. Feeding and reproduction were checked: the effect persisted on heat-killed bacterial food, and offspring number remained unchanged. Rg3 extended lifespan in mutants of five known longevity pathways but did not help when bec-1 (autophagy initiation) was suppressed or in vha-3 (lysosomal acid pump) mutants. Gene activity was compared via RNA sequencing: 753 genes changed expression, especially those involved in phosphoinositide metabolism (lipids like PI(3)P and PI(4)P), autophagy, and neuronal maintenance. Neurons were then examined: with Rg3, aggregates of Q40 — a clumping protein with a long glutamine chain — decreased. In a model with human α-synuclein, whose clumping is linked to Parkinson’s disease, more worms treated with Rg3 retained dopaminergic neurons. Lipids in neurons were tracked using fluorescent sensors marking PI(3)P and PI(4)P spots. With age, PI(3)P spots increased, but Rg3 reduced their number on the fifth and tenth days of adulthood. PI(4)P spots increased with Rg3, and some colocalized with acidic cellular compartments, including lysosomes. Among the 753 altered genes were genes for two core subunits of TORC1, the worm version of mTORC1: their activity weakened, while pifk-1 — the gene encoding the enzyme that makes PI(4)P — showed increased activity. TORC1 complex activity itself weakened: fewer phosphate marks appeared on its target RSKS-1. HLH-30, the worm homolog of human TFEB, activates lysosomal and autophagy genes upon entering the nucleus; according to the authors’ model, active TORC1 keeps it out of the nucleus. With Rg3, HLH-30 accumulated in intestinal cell nuclei, and its target genes worked more strongly. When pifk-1 was suppressed, Rg3 no longer caused HLH-30 nuclear accumulation or reduced Q40 aggregates; when hlh-30 was suppressed, both aggregate reduction and improved motility disappeared. If pifk-1 and let-363 (the gene encoding the TOR kinase itself) were suppressed together, aggregates decreased more than with pifk-1 suppression alone, and Rg3 added no further benefit. The authors conclude that partial TORC1 suppression makes PIFK-1 largely dispensable. 🔗 Read original →

Study finds sharp hormonal shifts at menopause after analyzing 300 million tests On October 12, a team led by Uri Alon at the Weizmann Institute published findings in Nature Medicine showing that 52 out of 90 health markers change abruptly at the time of a woman’s last menstrual period, based on an analysis of 300 million test results from the U.S. NHANES survey and Israel’s Clalit health fund. Because the exact date of menopause is rarely recorded in large databases and occurs at varying ages — with half of U.S. women experiencing it between 45 and 53 — averaging across age creates the illusion of a gradual transition. However, when the researchers aligned each woman’s data to her individual last menstrual period, they revealed sudden jumps in biomarkers, much like correcting for blur in astronomical imaging using known distortion patterns. The size of these jumps ranges from 0.1 to 1.4 standard deviations — the typical variation seen in non-pregnant women aged 25–45. The most pronounced shifts occur in sex hormones, bone markers, liver function, and blood lipids, with most changes moving toward higher disease risk. After menopause, the proportion of women developing osteoporosis, heart attack, and stroke begins to rise more quickly. Iron and hemoglobin are exceptions, improving likely due to the cessation of menstrual blood loss. The same analysis applied to men, using a hypothetical “menopause” at age 50, showed no such jumps, while Clalit data produced results similar to those from the U.S. According to the authors, the magnitude of change correlates with a system’s sensitivity to estrogen: free thyroid hormones and folic acid, which are weakly linked to estrogen, showed no jump. On average, anti-Müllerian hormone — a marker of ovarian reserve — begins accelerated decline 15 years before the last period, pituitary FSH rises 14 years prior, and estradiol, the primary estrogen, starts falling 12 years out. In the SWAN study, hormonal shifts appeared to begin only about seven years before menopause, likely due to its smaller sample size (3,302 women) and limited longitudinal data. If estrogen loss drives these changes, women on menopausal hormone therapy should show smaller jumps. In U.S. data, women on therapy differed from untreated peers in 39 out of 80 compared markers, with larger jumps showing greater differences — supporting this hypothesis. Martha Hickey, director of the menopause unit at Melbourne’s Royal Women’s Hospital, told ABC that describing the shift as a “physiological cliff” is unnecessarily alarming and not a scientific or medical term. She noted that studies using existing data are often low quality and that healthier women are more likely to seek therapy. Alon responded that current bone density screening typically begins around age 60, but this research supports checking as early as 40 or 45 and considering menopausal hormone therapy for prevention — always under medical guidance. His team plans to trial therapy initiated before menstruation ceases. 🔗 Read original →

Pharma companies should fine-tune open AI models on their own data, like Coinbase did for fraud detection, says Bo Wang of biotech firm Xaira Coinbase chief Brian Armstrong wrote that the company fine-tuned the open Qwen3.5-9B model on internal data, and in fraud detection it surpassed leading AI lab models while being cheaper and faster. One and a half hours later, on October 12, Bo Wang proposed the same path for pharma companies. Coinbase catches fraud in Onramp — its in-app crypto purchase service for partner apps. To the existing model and rules, the company added an AI agent on Anthropic’s Claude Opus model that rechecks transactions. In an online experiment, the group with the agent showed 30% fewer fraudulent transactions than the control group. On an archive of past transactions, the fraud detection rate for new Opus 5, Sonnet 5, and GPT-5.6 models was lower than predecessors by 0.8–22.2 percentage points. With Opus 4.5, Coinbase said, control “ended at the API boundary” — the channel through which the program accesses the model on the developer’s servers. Prompts and version can be chosen, but model weights — the numbers storing everything the model has learned — cannot be adjusted to Onramp fraud patterns. So Coinbase took the open Qwen3.5-9B model — nine billion weights. They are published, so the model can be downloaded, run locally, and fine-tuned on proprietary data. It was trained with reinforcement: a formula compared transaction risk scores with known outcomes and awarded rewards. Training ran on a single GPU and cost less than $100. On Coinbase’s own test, the fine-tuned model beat the Opus 4.5 agent: precision (share of true fraud among flagged operations) higher by 7.5 percentage points, recall (share of fraud found) higher by 12.0, and share of fraudulent amount found higher by 35.4. Open weights, per Coinbase, let the company decide what to train and when to update, while self-hosting keeps transaction data off external APIs. In return, the model must be constantly monitored and retrained: fraud schemes evolve. Wang, chief AI scientist at Xaira and co-creator of the scGPT model for single-cell data, writes: “Cost and speed here are small prizes, but sovereignty is the big one.” Strategically vulnerable, he says, is a pharma company that rents intelligence — i.e., pays for a frontier lab model: access, pricing, and roadmap are set by others, and the most sensitive data cross a boundary it does not control. Every major pharma, Wang continues, has an asset labs cannot replicate — decades of closed data from mass molecule screening, lab tests, and failed trials. Fine-tuning open weights on these data gives the company ownership of everything: weights, model deployment, and intellectual property — what he calls corporate-level sovereignty. His forecast: “The first pharma company to grasp this will compress drug discovery timelines by years. The rest will end up renting intelligence on their own molecules.” A precedent in pharma exists: on September 14, Nature reported that five pharma companies fine-tuned the open OpenFold3 model — which builds 3D protein structures together with drug molecules — on 20,167 of their own closed structures. Per Apheris platform report, on 1,056 structures from the same companies not used in training, high-quality protein-molecule contact was found in 52.1% of predictions from the new version and 35.6% from the original OpenFold3. 🔗 Read original →

Aging drugs must delay cancer longer than therapies extend life after diagnosis, says Ira Pastor Ira Pastor, host of a science and technology podcast, proposed a criterion for aging drugs in a debate about whether aging can be reversed, posting on X on October 11. For cancer, two timeframes matter. First, the pre-disease period: how many years an aging drug would delay diagnosis—for example, from age 65 to 80. Second, the post-diagnosis period: how much treatment would extend life after diagnosis. Pastor suggests comparing these two intervals. The @rand_longevity account asked on X: “Do you think aging will be reversed in your lifetime?” Pastor replied that this is “not quite the right question, and it’s incomplete.” The proper question, he said, is whether any “aging reversal” tool will delay the diseases that kill us—primarily cancer and heart disease—noticeably longer than new precise therapies extend life after diagnosis. Based on the geroscience hypothesis that targeting aging mechanisms could delay many age-related diseases at once, Pastor argues that longevity industry leaders conclude diseases can be postponed indefinitely—or even forever—and sell this dream. His example: cancer would not occur until age 150, heart disease until 180. “There is nowhere any evidence that this is true, and we won’t learn it even if aging reversal tools become available to us, for many more decades.” Bill Walker (@BWalkerTTAGGG) responded that it is known for certain: Greenland whales, with more than a thousand times as many cells as humans, live at least 230 years, possibly 268. “We don’t even need to invent longevity technology—we just need to borrow it from other mammals.” Cell number relates to cancer: tumors arise from a cell that has accumulated mutations, and in an animal with a thousandfold cell surplus, one would expect higher cancer risk. Peto’s paradox is that large species do not have higher cancer rates than small ones. A 2022 study in Nature confirmed this paradox, while a 2025 analysis of 292 vertebrate species did not. The 268-year figure is a genomic estimate: a 2019 article in Scientific Reports assessed the whale’s maximum lifespan this way. The oldest whale whose age was determined was estimated at 211 years by 2019; the number 230 does not appear in that paper. Pastor replied to Walker: “True,” but added that this is precisely why the question is incomplete. One thing is to create a “gerotherapeutic”—a drug against aging mechanisms—that in human trials attempts to replicate whale DNA repair and cold-adapted metabolism. Another is to show it can delay cancer in humans longer than any combination of modern therapies extends life after diagnosis. Among therapies, Pastor named immunotherapy, re-engineered patient immune cells, personal vaccines, antibody-drug conjugates, radioactive drugs, and bispecific antibodies targeting two proteins at once. An example of post-treatment duration comes from melanoma immunotherapy. In a final ten-year analysis of a clinical trial, previously untreated patients with metastatic melanoma receiving nivolumab plus ipilimumab had a median overall survival of not less than 71.9 months—almost six years—while those receiving ipilimumab alone, approved in 2011 before nivolumab, had a median of not less than 19.9 months. At ten years, 43 out of 100 patients in the first group were alive, compared to 19 out of 100 in the second. Pastor concluded: “This is the question the industry truly must show an answer to—first for cancer and heart disease—everything else is hype.” 🔗 Read original →

Extracellular vesicles from p16-positive senescent fibroblasts transmit aging without inflammation A team led by Ana O'Loghlen in Madrid published experiments on human fibroblast cultures on 8 October in the journal Aging. Fibroblasts in which senescence was induced by the p16 protein do not release typical inflammatory substances, leaving it unclear how they influence their surroundings. Their membrane vesicles halted division in other fibroblasts and induced DNA damage markers in them. Senescent cells stop dividing due to damage or stress but continue to affect neighboring cells. They secrete a set of substances known as SASP, which usually includes inflammatory proteins such as interleukins that recruit the immune system to clear such cells. SASP also contains extracellular vesicles—small membrane-bound packets carrying proteins and nucleic acids. In 2011, Judith Campisi's laboratory artificially activated the p16 protein in human cells; it inhibits cell division and serves as a marker of senescent cells, which increase with age. The cells ceased dividing but did not release inflammatory substances and had no effect on epithelial cells. The lab concluded that SASP is a damage response that can be separated from cell-cycle arrest. In 2019, the O'Loghlen group showed that senescence can be transmitted to neighbors via soluble substances and vesicles. Now they tested whether vesicles from p16-positive cells induce aging in neighboring cells. Experiments are described in a preprint on bioRxiv—an early version of the work prior to peer review. Fibroblasts were of three types: controls, RAS-induced senescent, and p16-induced senescent. RAS is a mutant gene that constantly signals the cell to divide; in response, the cell irreversibly arrests and ages. RAS cells served as a model of typical aging and secreted inflammatory SASP; p16 cells did not, but released more vesicles than controls. Vesicles isolated by centrifugation and added to dividing fibroblasts for three days halted division in recipient cells, increased their own p16 and γH2AX—a DNA damage marker. According to the article, they did not trigger production of interleukins IL-6 and IL-8, whereas vesicles from RAS cells, per the preprint, induced IL-8. Vesicles transmitted senescence in two other cellular models. In one, cells produced progerin, a protein causing the accelerated-aging disease progeria. In another, telomere function—protective chromosome ends—was disrupted. A mixture of vesicles from fibroblasts of four donors aged 67–81 years increased β-galactosidase activity—a senescent-cell marker—in fibroblasts from donors aged 1–3 years. With age, the immune system becomes less effective at clearing senescent cells. NK cells—immune killer cells—from people around 70 years old in culture killed senescent fibroblasts less effectively than equivalent cells from young individuals around 21 years old. The preprint authors suggest that accumulated p16-positive cells with age have a weak or absent inflammatory signal that would recruit immunity, so they spread aging via vesicles. Damaged neighbors arrest without inflammation, and, they propose, tissue renewal deteriorates. Senescent cells with suppressed SASP have also been described in the blind mole rat Spalax, a long-lived subterranean rodent: in its senescent fibroblasts, interleukin gene activity and other SASP factors were undetectable or lower than in mouse and human cells. The authors of that study attribute this to efficient DNA repair: damage that triggers SASP does not accumulate. 🔗 Read original →

AlphaGenome Atlas shows weak correlation with Alzheimer’s disease risk variants across 111,000 DNA changes in 75 risk regions Authors from the Korea Advanced Institute of Science and Technology (KAIST) posted a preprint on medRxiv on October 11 evaluating AlphaGenome Atlas, a database of precomputed predictions for DNA variant effects developed by Google DeepMind. The tool assigns each variant an AVI score — a composite of five feature groups: gene expression, regulatory features (DNA accessibility for regulatory proteins), splicing, protein impact, and evolutionary conservation — to prioritize variants and cell types for experimental follow-up. In Alzheimer’s GWAS regions, the correlation between predicted AVI scores and disease association strength was minimal: +0.013 (Spearman partial correlation, adjusted for variant frequency, linkage, and distance to nearest gene). This barely exceeds the +0.005 threshold expected from random shuffling in 95% of cases. Expression and splicing showed slightly higher correlations at +0.018 and +0.028, respectively, but after adjusting for genomic position, expression dropped to +0.011 (still above random) and splicing fell to +0.010 — indistinguishable from chance. The authors note that AVI is dominated by evolutionary conservation (47%) and regulatory features (34%), whereas disease-linked signals align more strongly with expression and splicing. When compared to statistical fine-mapping without AlphaGenome data — which defined 95 credible sets (groups of variants where the causal variant has ≥95% probability) — expression rankings of lead variants were, on average, 0.044 higher than those of tightly linked neighbors excluded from the set, and higher in 67% of 87 such comparisons. No consistent difference was found for splicing. For cell-type prediction, AlphaGenome Atlas forecasts effects across 285 cell and tissue types, collapsed into 12 classes. The variant’s assigned class is the one where its predicted effect is strongest. In 50 loci with multiple candidates, the average entropy of these class labels was 1.63 — higher than the 1.53 entropy of label sets shuffled within the locus (control). In 39 of 50 loci, candidate labels were more diverse than controls, and in none did they converge meaningfully. Labels spanned brain, immune system, liver, and vasculature, indicating that the predicted cell of action depends heavily on which variant is deemed causal. The authors recommend applying AlphaGenome Atlas predictions to all variants in statistical credible sets and reporting them alongside uncertainty — both regarding which variant is causal and in which cell type it acts. medRxiv preprint, October 11, 2024 🔗 Read original →

Statin-induced liver senescence may drive muscle weakness via IGF-1 decline Korean researchers found that lovastatin combined with ezetimibe triggered cellular senescence in mouse livers, reduced blood IGF-1 levels, and thinned muscle fibers. To test whether statins affect muscles through the liver, they exposed mouse liver cells (AML12) to statins for nine weeks. Lovastatin at 0.5 µM increased senescence-positive cells from ~1% in controls to about 27%, marked by β-galactosidase staining and elevated p21 and p16 gene activity. Atorvastatin at 1 µM did not significantly increase senescence-positive cells but raised p21 and p16 activity in both cases. Among ten secreted signaling genes measured, IGF-1, IL-1α, and IL-1β significantly decreased. The drop in inflammatory cytokines was attributed to statins’ anti-inflammatory effects, leaving IGF-1 as the likely mediator to muscle. To test liver-to-muscle signaling, senescent liver cells were placed in the upper chamber of a two-compartment device, with muscle precursor cells (C2C12) in the lower chamber, separated by a membrane allowing only soluble factors to pass. Muscle cells near senescent liver cells thinned from ~12 µm to ~10 µm in diameter and showed increased Atrogin-1 and MuRF1 — enzymes that tag proteins for degradation. In vivo, 11-month-old mice received lovastatin plus ezetimibe via feeding tube for three weeks. Liver senescence markers increased, blood IGF-1 fell by about one-fifth, tibialis anterior muscle fiber cross-sectional area decreased similarly, and AKT/mTOR signaling weakened. The authors conclude that statins induce liver senescence, which reduces IGF-1 secretion, weakening muscle AKT/mTOR signaling and promoting atrophy. They suggest protecting muscles by inhibiting liver senescence or restoring IGF-1/AKT/mTOR signaling. Cell Death Discovery, October 10, 2024 🔗 Read original →

Biochemists calculate the physical limits of reactive oxygen species in cells: they cannot accumulate enough to be the main cause of aging, a role the oxidative stress theory has assigned them for 70 years Redox chemist Christopher Sweitzer of the University of Leicester and co-author Kimya Kaderi calculated how much reactive oxygen species a living cell can produce and retain. Under generous assumptions, the concentration of these molecules does not rise above low micromolar levels — orders of magnitude lower than what the dominant theory predicts. The preprint was posted on October 8, 2026, on Preprints.org and has not yet undergone peer review. Reactive oxygen species (ROS) — byproducts of cellular respiration — have for decades been credited as the primary drivers of aging, cancer, and heart disease. The idea originated from ionizing radiation chemistry: in 1956, gerontologist Denham Harman proposed the free radical theory of aging in the paper “Aging: A Theory Based on Free Radical and Radiation Chemistry” — the title itself revealed its source. The discovery in 1969 of the enzyme superoxide dismutase, which breaks down superoxide, strengthened the idea, although the reaction merely converts one oxidant into another, hydrogen peroxide, without eliminating oxidant chemistry. Sweitzer and Kaderi recalculated the physics from scratch in their preprint. Dissolved oxygen in tissues amounts to units to tens of micromoles per liter — many times less than in a laboratory test tube. Via the mitochondrial respiratory chain, about 0.5% of electron flow leaks into side reactions that produce ROS — even with an exaggerated leak, concentration does not exceed low micromolar values. Hydrogen peroxide is almost instantly scavenged by peroxiredoxin enzymes — among the fastest enzymes in the cell — while glutathione maintains reserves orders of magnitude higher than ROS concentration. When oxygen is scarce, mitochondrial respiration weakens, and the fraction of leaking electrons does not increase — meaning ROS levels are lower precisely where oxygen is least available. Tumors in chronic hypoxia switch to glycolysis — by model, this should reduce mitochondrial ROS, although ROS are considered the engine of their adaptation to hypoxia. The same logic applies to hematopoietic stem cells: in low-oxygen niches of the bone marrow, they suppress mitochondria to maintain quiescence, although decisions to divide or differentiate are linked to ROS. “The resulting model of oxidative stress reflects toxicology logic more than cellular reality — a model built on extreme-condition chemistry and applied to systems where these conditions almost never occur,” the authors write. This aligns with clinical trials: beta-carotene in smokers increased lung cancer incidence by 18%, and the SELECT trial showed vitamin E increased prostate cancer risk by 17%. If ROS were the primary damaging agent, suppressing them should have helped — yet the effect is consistently zero or negative across substances and diseases. Physics permits ROS accumulation in only a few cases: a burst during reperfusion after ischemia, the bactericidal burst of neutrophils, and a lipid peroxidation chain reaction in ferroptosis. Instead, the authors propose that ROS mechanisms be held to the standard of ordinary signaling molecules: demonstrated production, plausible concentration, defined target, and consistency with cellular chemistry. In mice with constitutively active Yamanaka factors — proteins that transiently make mature cells more pliable — the antioxidant N-acetylcysteine in the liver suppressed a proven ROS burst and delayed their death: production, concentration, and target were confirmed experimentally, not theoretically. 🔗 Read original →