Perfusion lowers liver biological age, improves transplant outcomes

On September 14, MIT Technology Review, September 14 reported unpublished data from a Boston team showing that connecting donor livers to an oxygenated perfusion device is linked to a lower biological age assessment. The findings were presented at an industry conference.

In two series of molecular measurements, biological age scores were lower for livers after machine perfusion than for those stored on ice. The first series examined 37 samples from 19 livers via epigenetic marks; the second series evaluated 208 samples from 103 livers by gene activity.

After adjusting for chronological age, the difference between groups was about 30%. After transplantation, scores rose in both groups but remained lower for perfused livers.

Gene activity related to inflammation and tissue structure changed, and cells more actively cleared damaged parts, accompanying the lower biological age scores.

In a 2022 randomized study of 300 recipients, normothermic perfusion reduced early graft dysfunction from 31% to 18%.

The new data add a molecular assessment of organ condition before surgery to that clinical result. Perfusion systems also allow drugs to be delivered directly to the liver before operation. In another study, the drug FXT-001 was added to the system’s fluid to protect cell membranes from oxidative damage.

The current team is testing drug treatments for livers; biological age assessment could show how such preparation alters tissue.

🔗 Read original →

Biological Age Estimates Vary Depending on Signals and Populations Studied

On 8 September, Frontiers in Aging published a review of 435 studies from January 2011 — June 2023 that examined biomarkers of aging and how they are combined into biological age estimates. The authors note that the resulting estimate depends on which signals are used and on the group of people it was tested on.

Biological age tries to capture aging through markers such as telomere length, blood proteins, or gene activity. Some estimates rely on a single marker, others use models that combine several signals, so the same term can refer to different measurements.

In 123 of 162 studies, older participants tended to have shorter telomeres, while 25 studies found no such link. The association was common but varied across works, keeping telomere length as one of many aging signals.

The review also covers blood proteins and metabolites, gene expression, and multi‑marker models. Authors stress that different biological age measures give divergent results and ask how they relate, and which markers truly reflect aging versus drive it.

Performance of algorithms can differ across sexes, ethnicities, social groups, lifestyles, and environments; one study showed better accuracy when applied to the population on which it was trained. The review concludes that a biological age number is meaningful only when accompanied by an explanation of the signals it integrates, what they reflect biologically, and for whom it was validated.

🔗 Read original →

Founder of Nucleus Genomics Calls for 'Genetic Optimization Industry'

On 8 September, Kian Sadeghi, the founder of Nucleus Genomics, published an essay titled Evolution by Genetic Optimization. In it he groups embryo selection, gene editing, and technologies that could increase the number of available oocytes.

He argues that the main mechanism of genetic optimization is selection, not construction, and assigns gene editing to cases requiring a new genetic change. Selection, he links to IVG — a developing technology for obtaining eggs or sperm from other body cells — which would increase the number of gametes and thus embryos with different DNA combinations for evaluation and choice.

In IVF clinics, preimplantation genetic testing is already used to screen embryos before transfer, including for single‑gene disorders. Sadeghi also incorporates polygenic scores — calculations of disease risk or trait likelihood from many DNA variants — into his program. Authors of an article in Scientific American distinguish such testing from polygenic assessments of complex diseases and traits

🔗 Read original →

US Military Pushes for AI Dominance While AI Leaders Call for Slowdown

It is absurd to see the usual roles reversed. Normally one expects officials and the military to warn about uncontrolled technologies and demand tighter controls, while private corporations quietly push progress for profit. Instead, the heads of leading AI laboratories are publishing manifestos urging a slowdown of the tech race, calling for audits and a pause in the development of super‑intelligent systems.

At the very same moment the U.S. Department of Defense is openly posting slogans styled “Americanism, not effective altruism” and stating that its primary goal is absolute AI dominance. Adding to the paradox, the recent Palantir manifesto — which the White House openly follows — reinforces this stance. Thus, while developers fear their own algorithms and request a pause, government agencies are pressing the accelerator to the floor.

This signals that an evolutionary transition is imminent, and the power apparatus understands that whoever first forges a machine intelligence of a higher level will claim the future world. Attempts to curb the singularity are failing, which is a good sign, and at some point the government may recognize the potential of augments and genetic editing — a capability not weaker than AI and sometimes stronger. This could be the very trigger of exponential growth that Kurzweil warned about.

🔗 Source: @solid_state_humanity

Two substantia nigra connectivity axes tied to aging/movement and memory/mood

On 8 September, Molecular Psychiatry published an article on the substantia nigra, a small midbrain region involved in movement control. Using data from over a thousand participants across four independent datasets, researchers identified two stable connectivity axes within this region.

First, they built a map using resting‑state fMRI from 618 healthy participants aged 18‑88, then replicated it in a separate group of 184 young participants with higher‑resolution scans. In both samples the same two axes emerged.

The first axis runs from the inner to outer part of the substantia nigra and correlates with age, reaction speed, motor learning and planning ability. In clinical datasets this axis differed according to the result of SAA‑analysis of cerebrospinal fluid, which tests whether the protein alpha‑synuclein forms aggregates.

The second axis extends from the posterior to anterior part of the region. In healthy aging it linked to memory, cognitive flexibility, anxiety and depression. In

🔗 Read original →

No accelerated gray‑matter loss found at menopause transition, study shows

Researchers analyzed repeated MRI scans of 1,095 participants taken during puberty, pregnancy, and menopause. The study was published on 8 September in Nature Communications.

The menopause transition analysis included 120 women who answered “no” to menopause at the first scan and “yes” at the second. For comparison, researchers

🔗 Read original →

Waymo robotaxis crash rates 68% lower than human drivers in four US cities

The Insurance Institute for Highway Safety (IIHS) compared Waymo’s driver‑less robotaxi trips with human driving in San Francisco, Phoenix, Los Angeles and Austin. In the sample, injury‑causing crashes per mile were 81% lower for Waymo. Overall comparable crashes per mile were 68% lower.

The analysis covered about 50 million miles traveled by Waymo vehicles versus 222 billion miles driven by humans in the same areas and period. Only 22% of the 736 automated‑system incidents were deemed comparable to police‑reportable crashes.

By city, the reduction was 76% in Phoenix, 71% in Los Angeles and 35% in San Francisco, while Austin showed a 4% increase—a figure sensitive to few events. These results refer to the current Waymo robotaxi fleet and to the crash category defined as police‑reportable.

IIHS recommends a unified data‑collection system for mileage, operating mode and crashes to maintain comparable monitoring as robot

🔗 Read original →

Algorithm Finds Optimal Nerve‑Cortex Pulse Pause Faster Than Random Search in Adults

A preprint released on 8 September describes how researchers tested ten adults and eight children to determine the optimal pause between an electrical nerve impulse and a magnetic brain pulse. They varied the pause from 5 to 130 ms in 5‑ms steps (26 pauses total), repeating each pause ten times for adults and six times for children, and built individual response maps of the tibialis anterior muscle.

After three randomly selected pulse pairs, a Gaussian‑process model predicted the next pause to try, aiming for a response level at 90 % of each participant’s individual range between weakest and strongest responses. The model’s suggestion was checked against the full map after each trial.

In adults, the algorithm reached the target in nine out of ten participants, requiring a median of 10 pulse pairs, whereas random selection needed a median of 20 pairs. In children, the algorithm succeeded in five of eight participants with a median of 45 pairs, while random search succeeded in four of eight with a median of 56 pairs.

When predicting a left‑out pause from the remaining data, the model achieved an average R² of 0.71 for adults and 0.18 for children. Authors attribute the poorer child performance to greater response variability and the lower number

🔗 Read original →

Age‑Related RNA Dysregulation Linked to Neuronal Vulnerability

Trends in Cell Biology, September 8, 2026 published a review connecting age‑related RNA changes to neuronal susceptibility to degeneration. RNA molecules carry the instructions for protein synthesis, and neurons must transport RNAs far from the nucleus and translate them locally to maintain long processes and dynamic synapses. The review estimates that regulated RNA transport and local synthesis supply approximately 50% of a neuron’s proteins.

During splicing, cells excise and rejoin RNA fragments to generate different instruction variants. RNA‑binding proteins oversee processing, transport, translation, and decay; their levels, subcellular localization, and RNA targets shift with age. In neurons directly reprogrammed from skin of healthy elderly donors, the RNA‑binding protein TDP‑43 relocated from the nucleus to the cytoplasm, a shift associated with widespread alterations in alternative splicing.

Stress granules—temporary assemblies of RNAs and proteins—help cells rapidly remodel protein synthesis under stress and normally disassemble afterward. Aging can modify their composition and impede their disassembly. Certain RNAs contain double‑stranded regions that cellular immunity may mistake for viral signals, triggering inflammation.

Genes whose variants are linked to ALS and frontotemporal dementia encode RNA‑binding proteins. The authors propose a model where such variants are benign under normal conditions, but age‑related stress diminishes the cell’s capacity to cope, and the impact is amplified by concurrent RNA‑ and protein‑level changes.

To test this, the review recommends experimentally modulating individual RNA regulators in neuronal model systems and measuring resulting functional deficits. These approaches should reveal which specific RNA‑regulatory failures directly weaken neurons in aging.

🔗 Read original →

Wearable electroceutical patch uses microneedles to ease chronic pain

The patch uses microneedles to penetrate the stratum corneum, providing stable contact with nerve endings without pain or surgery. Engineers combined several breakthrough solutions in one compact device.

The conductive hydrogel coating ensures uniform current distribution, avoiding hazardous hotspots. A built‑in thermosensitive safety mechanism detaches the electrodes if skin temperature rises abnormally, preventing burns.

A **

🔗 Read original →

3D Culture Rejuvenates Limbal Niche Cells via FOSL1

On September 8, researchers published in Aging Cell that limbal niche cells from human donor tissue were cultured for six days in Matrigel droplets, forming spheroid clusters. Compared with flat culture, these cells divided more frequently and showed fewer signs of replicative senescence.

The limbus is the border between the transparent cornea and the white sclera, and its connective tissue houses limbal niche cells that support the epithelial stem cells renewing the corneal surface. Upon repeated passaging, niche cells divide slower and acquire replicative‑senescence markers, narrowing the available cell stock for further work.

This line of inquiry grew from the group’s 2012 study, in which transferring already‑proliferated cells to three‑dimensional Matrigel restored some stem‑cell‑associated proteins. The new work asks more precisely how a 3D environment alters aging markers and which protein drives this shift.

After 3D culture, the cells contained a higher proportion of dividing cells and a lower proportion of senescence‑positive cells. Single‑cell gene‑expression analysis revealed that two senescence‑associated gene

🔗 Read original →

Cystine deficiency triggers metabolic shift in methionine-restricted diets

A preprint posted on bioRxiv, 9 September examined which amino acid—methionine or cystine—activates the metabolic response in methionine‑restricted diets. The researchers divided a standard methionine‑restricted regimen into its two components: reduced methionine and altered cystine levels. In a short‑term test, diets lacking cystine raised FGF21 in mice lacking the cystathionine‑γ‑lyase gene (CTH), while diets with normal cystine kept the hormone near baseline.

Over 10 months, cystine depletion produced divergent outcomes depending on CTH status. In CTH‑deficient mice, a 70 % reduction in cystine (which avoided the ~30 % acute weight loss seen with full removal) curbed age‑related weight gain, preserved lean mass, improved glucose tolerance, and increased grip strength in males. Mice with intact CTH showed stable weight, coordination, grip strength and glucose tolerance on the same diet, but males gained fat mass.

🔗 Read original →

Brain‑Age Index Links MRI, Mood, Memory and Gut Microbiota in Healthy Adults

A study of 1,462 healthy adults aged 18–65 linked MRI‑derived brain age to memory, mood and gut bacteria. It was published September 9 in the journal eBioMedicine.

🔗 Read original →

Protein proximity map guides dual‑target antibody design

On September 9, 2026, Nature, September 9, 2026 published a method to select two targets for a single anticancer molecule based on their proximity on the tumor cell surface. The authors chose the EGFR–CDCP1 pair and tested an antibody that binds both proteins and carries a toxic payload.

To map proximity, they attached photosensitive labels to 12 surface receptors, left chemical traces on nearby proteins, and used mass spectrometry to read the traces, generating 248 proximity maps across 28 tumor cell systems. The MetaMap algorithm matched the maps and highlighted CDCP1 as a frequent EGFR neighbor in tumors but not in normal epithelial or mesenchymal cells.

The finding was validated with reciprocal labeling, an additional series of 38 CDCP1 maps in 19 tumor systems, and co‑isolated EGFR protein analysis, showing that proximity strength varied independently of each protein’s abundance.

Researchers then built a bispecific antibody with one arm binding EGFR, the other binding CDCP1, and a cytotoxic payload; EGFR binding was weakened to favor simultaneous engagement of both receptors. In cell assays the bispecific antibody entered tumor cells more effectively than monospecific antibodies, correlating with dual receptor engagement, while uptake in primary epithelial cells was much lower.

In mice bearing SW48 xenografts, a dose of 1 mg/kg of the bispecific antibody gave complete tumor growth inhibition, whereas an EGFR‑only antibody achieved only 43% inhibition (each group n=8). When each mouse carried two tumors—wild‑type SW48 and a CDCP1‑low variant—a dose of 3 mg/kg produced 97% inhibition in the wild‑type tumor and 46% in the CDCP1‑low tumor, underscoring CDCP1’s role in payload delivery.

🔗 Read original →

Allogenetics Secures €3.8 Million for Gene‑Edited Donor Lung Trial

On September 9, German biotech Allogenetics closed a €3.8 million financing round. The funds will support preclinical work needed for an IND filing for its ALG-115 program, which aims to treat donor lungs with gene therapy before transplantation.

In ALG-115, donor lungs are kept alive on a perfusion device and receive gene therapy that lowers MHC I and II expression on the organ’s cells. By reducing these molecules, the recipient’s immune system is less likely to recognize the transplant as foreign, potentially decreasing rejection risk while still preserving the lung’s ability to fight infections.

As transplant surgeon Jeffrey Platt of the University of Michigan Medical School explained, “When the intervention is directed at the donor organ, the recipient’s immune system can be preserved.”

🔗 Read original →

Bacterial B12 Extends Worm Lifespan via Mitochondrial Renewal When rict-1 Is Off

On 14 September 2026, Nature Communications, 14 September 2026 reported that bacterial vitamin B12 extends the lifespan of *Caenorhabditis elegans* when the rict-1 gene is knocked out, by promoting mitochondrial renewal.

Worms obtain B12 from the bacteria they eat; a diet rich in the HT115 strain increases both longevity and osmotic stress resistance in rict-1 mutants, and the same effect occurs with B12 supplementation of the standard OP50 feed, but only when sufficient methionine is present.

Earlier work linked rict-1 to diet and lifespan (2009 study), and the same lab found a B12‑dependent longevity path in flr-4 mutants (2022 study); this paper pinpoints the exact pathway connecting B12 to the rict-1 effect.

The researchers disabled B12‑dependent enzymes involved in methionine and propionate metabolism, which raised succinate levels; blocking succinate formation reduced stress resistance and lifespan, while adding succinic acid restored them, identifying succinate as a crucial link.

Succinate then drives mitochondrial fission via DRP-1; inhibiting DRP-1 lowered stress resistance and lifespan, and enhanced mitophagy—detected with a fluorescent sensor—required pink-1 and pdr-1; loss of those genes diminished mitophagy, stress resistance, and the lifespan advantage of rict-1 mutants.

Norm

🔗 Read original →

Preprint links iPSC‑derived vascular cells to improved blood flow in mice

A preprint published September 9 combined 68 preclinical studies of vascular cells made from induced pluripotent stem cells. In the primary analysis of 51 works with 114 comparisons, tissue blood flow after cell therapy was on average 1.96‑fold higher than in control groups.

Limb ischemia occurs when blood supply to tissues is inadequate, causing oxygen shortage. In the lab, iPSCs can be turned into various specialized cells, including the endothelial lining of blood vessels. Researchers injected these cells into mice with hindered limb

🔗 Read original →

Heart lymphatic vessels decay before fibrosis in aging mice

In old mice, heart lymphatic vessels thin before scarring and thickening; the cause is the nuclear form of

🔗 Read original →

AI drug‑discovery models Boltz‑2 and Nesso‑1 falter on rare protein families, simple method wins

On September 13, independent researchers published the MIRAGE preprint benchmark, showing that on proteins scarcely represented in public databases both Boltz‑2 and Nesso‑1 lose to a simpler, cheaper method.

Boltz‑2 was released in June 2025 by MIT and Recursion (Nasdaq: RXRX), with a press release claiming it approaches the accuracy of costly physics‑based binding‑energy calculations while being up to 1,000 times faster. Recursion’s scientific director Nadjat Khan stressed that early‑molecule selection is a fundamental drug‑discovery challenge.

Over a year later Recursion’s Valence Labs released Nesso‑1, which Khan said matches or exceeds Boltz‑2 on public benchmarks; the MIRAGE preprint was used to test this claim.

Suspicions that the models memorize protein families rather than learning binding physics were examined across the full spectrum of family representation. Accuracy of Boltz‑2 and Nesso‑1 rose with family representation, while controls deprived of family data showed no improvement.

Even after removing cases where the model could have seen the exact protein before, the effect persisted, indicating recognition of whole families rather than single entries. On families rarely seen, a simple random forest without family data outperformed both neural nets.

To test on a truly low‑representation target, the authors used binding data released in May 2026; neither Boltz‑2 nor Nesso‑1 nor classical docking tools could reliably beat the baseline of predicting activity from molecular mass alone. Boltz‑2 fell below this baseline, Nesso‑1 was barely above but the difference was deemed statistically insignificant.

Adding evolutionary alignment as an input boosted Boltz‑2’s accuracy on unfamiliar families from 2% to 46%, while performance on familiar families remained unchanged; the released model does not receive this signal.

The MIRAGE authors conclude that selecting models by average accuracy favors familiarity with already solved protein families, and advise checking how well a target’s family is represented in the training data before trusting predictions.

🔗 Read original →

Insilico Launches Longevity Vaccines Program to Reprogram T‑Cells In Vivo

On September 15 Insilico Medicine announced its Longevity Vaccines program, which uses cyclic RNA in a lipid nanoparticle to temporarily teach a patient’s T‑cells to hunt and destroy senescent cells, activated fibroblasts and autoreactive lymphocytes. The first target is senescent immune cells, whose accumulation weakens vaccine responses and raises susceptibility to infections and cancer with age.

The approach mirrors CAR‑T therapy: an RNA‑encoded receptor turns the patient’s own T‑cells into carriers that attack the chosen culprit cells, then the RNA degrades,

🔗 Read original →

LSD1 enzyme prolongs DNA‑damage signal, driving cellular senescence

On 15 September, the Journal of Clinical Investigation published a study from the Mayo Clinic and Tongji University showing that the enzyme LSD1 builds up in aged mouse kidneys, heart, liver and in artificially aged cells because the cell cannot degrade it efficiently. LSD1 removes a methyl group from the ATM protein at residue 3016, preventing ATM from turning off after DNA repair and keeping the damage signal active.

When ATM stays active, the phosphatase WIP1 cannot bind well to the unmethylated site, so the DNA‑damage alarm persists longer than needed. This prolonged signal pushes the cell into a stable growth arrest — senescence — where it secretes harmful factors that damage surrounding tissues.

Normally, LSD1 is cleared by autophagy through binding to LC3 and Beclin1; with age autophagy weakens, the LSD1‑LC3/Beclin1 link breaks, and LSD1 accumulates in the nucleus where ATM resides. Enhancing autophagy with rapamycin lowers LSD1 levels even in old organs, while blocking autophagy raises LSD1 even in young ones.

Senolytic drugs that eliminate senescent cells also reduced LSD1 and senescence markers in the kidneys, heart and liver of old mice; a regimen of dasatinib + quercetin restored the kidney‑protective protein Klotho and lessened fibrosis.

The LSD1 inhibitor ORY-1001 (iadademstat), already tested in humans for acute myeloid leukemia, myelodysplastic syndrome and small‑cell lung cancer, decreased DNA damage and aging signs in the kidneys, heart and liver of irradiated and aged mice, and protected them from radiation‑induced hair graying. These effects on aging have been demonstrated only in mice and cells so far.

Together, the work links two hallmarks of aging — persistent DNA‑damage signaling and declining cellular cleanup — to a single switch on ATM: LSD1 does not decide whether the alarm sounds, but how long it lasts, determining whether a cell returns to division or becomes stuck in senescence.

🔗 Read original →