Cas9d Ultra Editor
The compact Cas9d Ultra editor was placed in a single AAV viral vector and used to modify a gene in the livers of mice. On August 19, authors published a study on Cas9d Ultra, a compact genetic editor. One of its variants, 9dCBE, changes the letter C in DNA; it was packaged with a guide RNA in a single AAV9 viral vector and administered to newborn mice.
Through five weeks, the editor modified the target region of the Pcsk9 gene in the liver by an average of 15.7%, and the level of LDL cholesterol in the serum was lower than in mice after PBS injection. The base editor makes precise changes to one "letter" of DNA. The guide RNA sets the address in the genome, the CRISPR protein holds the editor in place, and the attached enzyme performs the chemical replacement of the base.
The AAV virus, which carries the genetic cargo into cells, typically holds around 4,700 nucleotides. Large editors are often split between two vectors. In this study, a single AAV9 delivered both the editor and its guide RNA. The Cas9d MG34-1 was chosen due to its size: this protein consists of 747 amino acids and works with a guide RNA of normal length for CRISPR. In a 2022 study, its early version achieved up to 22% base editing on three targets in human cells. The current authors modified four amino acids in the protein and the guide RNA scaffold. In a cellular test, where DNA cutting triggers a fluorescent signal, activity increased from 20.44% to 63.32%.
🔗 Read original →
The compact Cas9d Ultra editor was placed in a single AAV viral vector and used to modify a gene in the livers of mice. On August 19, authors published a study on Cas9d Ultra, a compact genetic editor. One of its variants, 9dCBE, changes the letter C in DNA; it was packaged with a guide RNA in a single AAV9 viral vector and administered to newborn mice.
Through five weeks, the editor modified the target region of the Pcsk9 gene in the liver by an average of 15.7%, and the level of LDL cholesterol in the serum was lower than in mice after PBS injection. The base editor makes precise changes to one "letter" of DNA. The guide RNA sets the address in the genome, the CRISPR protein holds the editor in place, and the attached enzyme performs the chemical replacement of the base.
The AAV virus, which carries the genetic cargo into cells, typically holds around 4,700 nucleotides. Large editors are often split between two vectors. In this study, a single AAV9 delivered both the editor and its guide RNA. The Cas9d MG34-1 was chosen due to its size: this protein consists of 747 amino acids and works with a guide RNA of normal length for CRISPR. In a 2022 study, its early version achieved up to 22% base editing on three targets in human cells. The current authors modified four amino acids in the protein and the guide RNA scaffold. In a cellular test, where DNA cutting triggers a fluorescent signal, activity increased from 20.44% to 63.32%.
🔗 Read original →
PubMed Central (PMC)
Compact 9dBEs Enable Efficient and Precise Genome Editing in Mammalian Cells and In Vivo
Despite the promise of DNA base editors for diverse genome editing applications, their utility remains constrained by off‐target effects, which are exacerbated by short spacers in miniature systems and the large size of Cas9‐derived editors, which ...
Jacek Hoffmann Proposal
Jacek Hoffmann proposed building systems from multiple AI agents with different data and verification methods. On August 20, Hoffmann published an essay on "heterogeneous cognitive ecology": humans and AI approach a task with different methods and data, then cross-check results. He calls the scenario where multiple agents share one mistake "Beryl Cage".
Hoffmann assumes a possible asymmetry: AI systems will increasingly create and transform information, while an individual will find it harder to reconstruct the solution path. He sees a sample verification device in science: one participant proposes an explanation, another looks for a counterexample, and a third repeats the analysis with different data or methods. This makes the error more noticeable where the results diverge.
Diverging data, methods, and criteria create different reasoning paths and different errors. The divergence can reveal a hidden assumption or condition that the first participant missed. Ten copies of one system can agree because they err in the same way. "The number of models is not equal to diversity," Hoffmann writes, citing an official report from Anthropic, August 13.
🔗 Read original →
Jacek Hoffmann proposed building systems from multiple AI agents with different data and verification methods. On August 20, Hoffmann published an essay on "heterogeneous cognitive ecology": humans and AI approach a task with different methods and data, then cross-check results. He calls the scenario where multiple agents share one mistake "Beryl Cage".
Hoffmann assumes a possible asymmetry: AI systems will increasingly create and transform information, while an individual will find it harder to reconstruct the solution path. He sees a sample verification device in science: one participant proposes an explanation, another looks for a counterexample, and a third repeats the analysis with different data or methods. This makes the error more noticeable where the results diverge.
Diverging data, methods, and criteria create different reasoning paths and different errors. The divergence can reveal a hidden assumption or condition that the first participant missed. Ten copies of one system can agree because they err in the same way. "The number of models is not equal to diversity," Hoffmann writes, citing an official report from Anthropic, August 13.
🔗 Read original →
Substack
Prisoners of the Stone Age in a World of Exabytes
From the Biological Limits of Mind to a Heterogeneous Cognitive Ecology
AC Immune Reports Progress
AC Immune reported that ACI-19764 reached the cerebrospinal fluid around the brain and dose-dependently suppressed the inflammatory signal IL-1β in blood analysis. On August 20, AC Immune published interim data from the first cohorts of the ACI-19764 study in healthy volunteers. The company checked if the drug reached the central nervous system environment, achieved the necessary concentration there, and was associated with a biochemical response in the blood.
The ACI-19764 is an orally administered small chemical compound. AC Immune is developing it as an NLRP3-inflammasome inhibitor - an intracellular complex of the immune system involved in the formation of inflammatory signals, including IL-1β. In the August 20 release, the company reported on cohorts with single and repeated doses. The detection of ACI-19764 in cerebrospinal fluid shows that the drug reached the central nervous system environment.
Daily doses of up to 10 mg gave concentrations above the IC90 level - the level of substance that in a laboratory test suppresses 90% of the chosen response. In whole blood samples, IL-1β release decreased with increasing dose. These measurements answer different questions: did the drug reach the central nervous system environment, is its concentration sufficient for the laboratory goal, and does IL-1β release in the blood decrease with dose. The original protocol in the NCT07463196 registry is designed for 78 healthy volunteers.
🔗 Read original →
AC Immune reported that ACI-19764 reached the cerebrospinal fluid around the brain and dose-dependently suppressed the inflammatory signal IL-1β in blood analysis. On August 20, AC Immune published interim data from the first cohorts of the ACI-19764 study in healthy volunteers. The company checked if the drug reached the central nervous system environment, achieved the necessary concentration there, and was associated with a biochemical response in the blood.
The ACI-19764 is an orally administered small chemical compound. AC Immune is developing it as an NLRP3-inflammasome inhibitor - an intracellular complex of the immune system involved in the formation of inflammatory signals, including IL-1β. In the August 20 release, the company reported on cohorts with single and repeated doses. The detection of ACI-19764 in cerebrospinal fluid shows that the drug reached the central nervous system environment.
Daily doses of up to 10 mg gave concentrations above the IC90 level - the level of substance that in a laboratory test suppresses 90% of the chosen response. In whole blood samples, IL-1β release decreased with increasing dose. These measurements answer different questions: did the drug reach the central nervous system environment, is its concentration sufficient for the laboratory goal, and does IL-1β release in the blood decrease with dose. The original protocol in the NCT07463196 registry is designed for 78 healthy volunteers.
🔗 Read original →
PubMed Central (PMC)
First‐in‐human safety, tolerability, and pharmacokinetic results of DFV890, an oral low‐molecular‐weight NLRP3 inhibitor
This first‐in‐human study evaluated the safety, tolerability, single‐ and multiple‐dose pharmacokinetic profiles with dietary influence, and pharmacodynamics (PD) of DFV890, an oral NLRP3 inhibitor, in healthy participants. In total, 122 ...
Perturb-ME Maps Cell Mechanisms
Researchers from Genentech and the Broad Institute presented the Perturb-ME method in a preprint on August 18, which combines genome-wide CRISPR knockout with cell sorting and single-cell analysis to build maps of cell mechanisms. The approach combines genome-wide gene knockout with cell sorting by target protein level and simultaneous reading of the knocked-out gene, RNA, and surface markers in each cell.
In an experiment on melanoma cells, the authors reconstructed a regulatory network of 221 regulator genes and 1998 response genes, and deposited the code and computational pipelines in an open repository. The main obstacle in studying complex cellular processes remains the gap between arrays of genetic correlations and testable causal links. When biologists want to determine how each individual gene controls a trait, they have to choose between two extremes.
The new Perturb-ME approach (Perturb-seq with Marker Enrichment) overcomes this limitation with a two-step design. First, genes are knocked out across the genome in a pool of cells using the CRISPR system. Then, the population is passed through a flow cytometer, selecting only the top and bottom 5% of cells with the minimum and maximum levels of the protein of interest. This step filters out cells without an expressed response and concentrates the effective mutations.
Only after this cell sorting step are the cells sent for multimodal single-cell profiling. Within each cell, the guiding RNA, transcriptome, and level of surface proteins are read simultaneously using DNA-barcoded antibodies. The work of the method was verified using the example of the major histocompatibility complex I (MHC-I) in human melanoma cells, and the results were published in Nature Aging, July 2026.
🔗 Read original →
Researchers from Genentech and the Broad Institute presented the Perturb-ME method in a preprint on August 18, which combines genome-wide CRISPR knockout with cell sorting and single-cell analysis to build maps of cell mechanisms. The approach combines genome-wide gene knockout with cell sorting by target protein level and simultaneous reading of the knocked-out gene, RNA, and surface markers in each cell.
In an experiment on melanoma cells, the authors reconstructed a regulatory network of 221 regulator genes and 1998 response genes, and deposited the code and computational pipelines in an open repository. The main obstacle in studying complex cellular processes remains the gap between arrays of genetic correlations and testable causal links. When biologists want to determine how each individual gene controls a trait, they have to choose between two extremes.
The new Perturb-ME approach (Perturb-seq with Marker Enrichment) overcomes this limitation with a two-step design. First, genes are knocked out across the genome in a pool of cells using the CRISPR system. Then, the population is passed through a flow cytometer, selecting only the top and bottom 5% of cells with the minimum and maximum levels of the protein of interest. This step filters out cells without an expressed response and concentrates the effective mutations.
Only after this cell sorting step are the cells sent for multimodal single-cell profiling. Within each cell, the guiding RNA, transcriptome, and level of surface proteins are read simultaneously using DNA-barcoded antibodies. The work of the method was verified using the example of the major histocompatibility complex I (MHC-I) in human melanoma cells, and the results were published in Nature Aging, July 2026.
🔗 Read original →
bioRxiv
Perturb-ME: Scalable mechanism discovery from phenotype-enriched genome-wide screens
Perturb-seq enables pooled genetic screens with rich single-cell profiling readouts, but genome-scale profiling remains costly and may not be associated with other established functional characteristics. Moreover, as screens grow in size and complexity, interpreting…
UK Clinical Trials
The UK has published clinical trial metrics, with 82% of already open studies recruiting participants on schedule, but launch after approval is lagging. On August 19, the UK's Department of Health and Social Care published July metrics for the UKCRD program, which tracks clinical trial timelines.
The program measures the path from application to first participant separately, including the stages of regulatory and ethics committee approval, research center opening, and first participant enrollment. The 82% metric describes recruitment in already open studies, while launch after approval is measured by separate metrics.
In the CPMS registry, a database of the UK's National Institute for Health and Care Research network, there were 4,347 open studies in July, with 82% of them recruiting on schedule and reaching target participant numbers, exceeding the 80% goal. The UKCRD has been publishing these metrics since 2024.
For commercial drug trials, a 150-day metric covers the entire path from application to first participant, with all nine January-submitted studies meeting this timeline. After approval, the report tracks two more stages: 56% of studies opened recruitment within 60 days, and 61% enrolled their first participant within 30 days of opening recruitment, with a 90% goal for both stages, as reported in the UKCRD July metrics.
🔗 Read original →
The UK has published clinical trial metrics, with 82% of already open studies recruiting participants on schedule, but launch after approval is lagging. On August 19, the UK's Department of Health and Social Care published July metrics for the UKCRD program, which tracks clinical trial timelines.
The program measures the path from application to first participant separately, including the stages of regulatory and ethics committee approval, research center opening, and first participant enrollment. The 82% metric describes recruitment in already open studies, while launch after approval is measured by separate metrics.
In the CPMS registry, a database of the UK's National Institute for Health and Care Research network, there were 4,347 open studies in July, with 82% of them recruiting on schedule and reaching target participant numbers, exceeding the 80% goal. The UKCRD has been publishing these metrics since 2024.
For commercial drug trials, a 150-day metric covers the entire path from application to first participant, with all nine January-submitted studies meeting this timeline. After approval, the report tracks two more stages: 56% of studies opened recruitment within 60 days, and 61% enrolled their first participant within 30 days of opening recruitment, with a 90% goal for both stages, as reported in the UKCRD July metrics.
🔗 Read original →
GOV.UK
UK Clinical Research Delivery key performance indicators: data to July 2026
The UK Clinical Research Delivery (UKCRD) key performance indicators (KPIs) report brings together data from the National Institute for Health and Care Research (NIHR) and the Medicines and Healthcare products Regulatory Agency (MHRA) to monitor the delivery…
Blood Age Shift
A recent analysis of 48 DNA segments found that changes in immune cells explained most of the decade-long shift in blood methylation. On August 19, a preprint was published with paired blood samples from 86 elderly participants over ten years. After accounting for the estimated change in immune cell composition, 32.2% of the original rate of shift in the 48-segment metric remained; this residue was not statistically different from zero.
Methylation refers to chemical marks on DNA, which are used to build epigenetic age indicators. However, a blood test always contains a mixture of immune cells, each with their own methylation patterns. If the proportions of these cells change over the years, the average blood signal shifts, even without a similar shift within each cell type. The preprint author took a metric from 48 preselected DNA segments and tracked it in two longitudinal cohorts.
For the Danish cohort of 86 participants, the author estimated changes in the proportions of seven types of immune cells using a separate set of DNA markers and included these changes in the calculation. The rate of shift in the metric dropped to 32.2% of the original, and the model with changing cell composition explained 36.4% of the variance in metric change between participants. Comparing a person to themselves removes their constant features, including their usual blood composition, which can change over ten years.
In a 2024 study, the IntrinClock model was calibrated so that its readings did not change between ten validated types of immune cells. In purified naive CD8+ T cells, its metric still increased with age. Reanalysis of blood should show the shift in immune composition and what remains after such correction separately. Then, one figure does not mix the rearrangement of cells with the change in marks within them, as reported in Nature Aging, July 2026.
🔗 Read original →
A recent analysis of 48 DNA segments found that changes in immune cells explained most of the decade-long shift in blood methylation. On August 19, a preprint was published with paired blood samples from 86 elderly participants over ten years. After accounting for the estimated change in immune cell composition, 32.2% of the original rate of shift in the 48-segment metric remained; this residue was not statistically different from zero.
Methylation refers to chemical marks on DNA, which are used to build epigenetic age indicators. However, a blood test always contains a mixture of immune cells, each with their own methylation patterns. If the proportions of these cells change over the years, the average blood signal shifts, even without a similar shift within each cell type. The preprint author took a metric from 48 preselected DNA segments and tracked it in two longitudinal cohorts.
For the Danish cohort of 86 participants, the author estimated changes in the proportions of seven types of immune cells using a separate set of DNA markers and included these changes in the calculation. The rate of shift in the metric dropped to 32.2% of the original, and the model with changing cell composition explained 36.4% of the variance in metric change between participants. Comparing a person to themselves removes their constant features, including their usual blood composition, which can change over ten years.
In a 2024 study, the IntrinClock model was calibrated so that its readings did not change between ten validated types of immune cells. In purified naive CD8+ T cells, its metric still increased with age. Reanalysis of blood should show the shift in immune composition and what remains after such correction separately. Then, one figure does not mix the rearrangement of cells with the change in marks within them, as reported in Nature Aging, July 2026.
🔗 Read original →
Researchsquare
An error has occurred
Research Square is a preprint platform that makes research communication faster, fairer, and more useful.
Brain Ultrasound
The SonoMod module attaches to a mouse's head, focusing ultrasound on a brain area while simultaneously recording its response with a miniature camera. On August 18, authors published a preprint on SonoMod, releasing files for assembly, optics, analysis software, and data. Focused ultrasound concentrates sound wave energy in a small brain point.
The device allows for simultaneous stimulation and recording, overcoming previous setup limitations that required head fixation or separate sessions for stimulation and recording. According to the authors, "in neuroscience, tools for intervening in neural circuits and reading their response have traditionally relied on incompatible physics." In SonoMod, light and sound pass through a single module, with a transparent niobate disk converting an electrical signal into ultrasound.
The UCLA Miniscope v4, an open miniature microscope for experiments with freely moving animals, is used with the SonoMod module. Authors tested the device in a series of experiments, first verifying that the camera can see the brain through the module, and then directing ultrasound to the secondary motor cortex (M2). The results were published in a preprint repository and are awaiting review in a neuroscience journal.
🔗 Read original →
The SonoMod module attaches to a mouse's head, focusing ultrasound on a brain area while simultaneously recording its response with a miniature camera. On August 18, authors published a preprint on SonoMod, releasing files for assembly, optics, analysis software, and data. Focused ultrasound concentrates sound wave energy in a small brain point.
The device allows for simultaneous stimulation and recording, overcoming previous setup limitations that required head fixation or separate sessions for stimulation and recording. According to the authors, "in neuroscience, tools for intervening in neural circuits and reading their response have traditionally relied on incompatible physics." In SonoMod, light and sound pass through a single module, with a transparent niobate disk converting an electrical signal into ultrasound.
The UCLA Miniscope v4, an open miniature microscope for experiments with freely moving animals, is used with the SonoMod module. Authors tested the device in a series of experiments, first verifying that the camera can see the brain through the module, and then directing ultrasound to the secondary motor cortex (M2). The results were published in a preprint repository and are awaiting review in a neuroscience journal.
🔗 Read original →
PubMed Central (PMC)
General‐Purpose Ultrasound Neuromodulation System for Chronic, Closed‐Loop Preclinical Studies in Freely Behaving Rodents
Transcranial focused ultrasound stimulation (tFUS) is an effective noninvasive treatment modality for brain disorders with high clinical potential. However, the therapeutic effects of ultrasound neuromodulation are not widely explored due to ...
Virtual Gene Knockout
The authors of a preprint compared eight methods that attempt to predict the consequences of gene knockout based on single-cell RNA data. In a test on K562 leukemia cells, the direction predicted by the linear version of CellOracle matched the experiment in 18 out of 44 "transcription factor — gene" pairs. The data on single-cell RNA shows which genes are usually active together, allowing researchers to build a hypothesis about the regulatory network and choose a gene for the next experiment.
The actual gene knockout answers a different question: how will the work of each specific gene change after intervention. The authors tested four transcription factors — proteins that control the work of other genes — and 11 glycolysis genes, the first stage of a cell obtaining energy from glucose. In Perturb-seq, researchers use CRISPRi to suppress the work of a selected gene and then read the RNA of individual cells. The average activity of these 11 genes decreased in all four factors.
The measurements were then compared to the signs of coefficients in the linear version of CellOracle. The sign of the coefficient describes the relationship between genes in the RNA data, and CRISPRi shows their response to intervention. The directions matched in 18 out of 44 cases, as reported in the preprint of August 19.
🔗 Read original →
The authors of a preprint compared eight methods that attempt to predict the consequences of gene knockout based on single-cell RNA data. In a test on K562 leukemia cells, the direction predicted by the linear version of CellOracle matched the experiment in 18 out of 44 "transcription factor — gene" pairs. The data on single-cell RNA shows which genes are usually active together, allowing researchers to build a hypothesis about the regulatory network and choose a gene for the next experiment.
The actual gene knockout answers a different question: how will the work of each specific gene change after intervention. The authors tested four transcription factors — proteins that control the work of other genes — and 11 glycolysis genes, the first stage of a cell obtaining energy from glucose. In Perturb-seq, researchers use CRISPRi to suppress the work of a selected gene and then read the RNA of individual cells. The average activity of these 11 genes decreased in all four factors.
The measurements were then compared to the signs of coefficients in the linear version of CellOracle. The sign of the coefficient describes the relationship between genes in the RNA data, and CRISPRi shows their response to intervention. The directions matched in 18 out of 44 cases, as reported in the preprint of August 19.
🔗 Read original →
AI Model Helps Find Gene
Researchers combined an AI model with data on gene activity in individual blood stem cells, genetic regulator testing, and cell transplantation in mice. They found that the Pbx1 gene is involved in the age-related shift: after transplantation, old cells are less effective at restoring erythrocytes and more likely to produce thrombocyte precursors.
The study, published on August 21 in Science Advances, used the Geneformer AI model to identify the Pbx1 gene as a key regulator of this shift. The model was trained on ten datasets of young and old mouse HSC cells and used to predict which genes, when activated, would make young cells resemble old cells.
The team then tested the prediction in cells by activating 143 genetic regulators associated with blood formation in young HSC cells and tracking the CD48 marker. They found that increased Pbx1 activity slowed the appearance of CD48, and that its activity was higher in old HSC cells.
🔗 Read original →
Researchers combined an AI model with data on gene activity in individual blood stem cells, genetic regulator testing, and cell transplantation in mice. They found that the Pbx1 gene is involved in the age-related shift: after transplantation, old cells are less effective at restoring erythrocytes and more likely to produce thrombocyte precursors.
The study, published on August 21 in Science Advances, used the Geneformer AI model to identify the Pbx1 gene as a key regulator of this shift. The model was trained on ten datasets of young and old mouse HSC cells and used to predict which genes, when activated, would make young cells resemble old cells.
The team then tested the prediction in cells by activating 143 genetic regulators associated with blood formation in young HSC cells and tracking the CD48 marker. They found that increased Pbx1 activity slowed the appearance of CD48, and that its activity was higher in old HSC cells.
🔗 Read original →
PubMed Central (PMC)
Geneformer-guided multiomics integration identifies Pbx1 as a network hub of hematopoietic stem cell aging
Hematopoietic stem cells (HSCs) constitute an organized hematopoietic system that undergoes age-related alterations, including increased platelet production and decreased erythropoiesis. The fundamental mechanisms driving these shifts remain ...
Mambbot Project
Molecular biologist Майкл Левин proposes using a series of AI-driven experiments to search for the desired form of living tissue. On 21 августа, he released a lecture on "free lunches," where he describes Mambbot, a collaborative project that uses AI to suggest light, vibration, temperature, or chemical signals to apply to cells to search for a biobot with a specific form and function.
The concept of a "free lunch" refers to the gap between what a system provides and the effort explicitly invested in it through design, selection, or training. According to Levin, this gap defines the next experiment: what property of the system produced the result, and how can it be induced again. This idea grows out of his laboratory's long-standing work at Университете Тафтса on morphogenesis, or how cellular collectives assemble, repair, and change body shape.
In a 2021 article, the authors described ksenobots, mobile constructs made from frog embryo cells. These clusters assembled free cells into new clusters, and an algorithm selected forms that reproduced better. In work on anthrobots, adult human airway cells self-organized into mobile constructs, and in neural cultures, they accelerated the closure of damaged areas. These results provided researchers with a material in which to observe the form and function of cellular collectives after changing conditions.
🔗 Read original →
Molecular biologist Майкл Левин proposes using a series of AI-driven experiments to search for the desired form of living tissue. On 21 августа, he released a lecture on "free lunches," where he describes Mambbot, a collaborative project that uses AI to suggest light, vibration, temperature, or chemical signals to apply to cells to search for a biobot with a specific form and function.
The concept of a "free lunch" refers to the gap between what a system provides and the effort explicitly invested in it through design, selection, or training. According to Levin, this gap defines the next experiment: what property of the system produced the result, and how can it be induced again. This idea grows out of his laboratory's long-standing work at Университете Тафтса on morphogenesis, or how cellular collectives assemble, repair, and change body shape.
In a 2021 article, the authors described ksenobots, mobile constructs made from frog embryo cells. These clusters assembled free cells into new clusters, and an algorithm selected forms that reproduced better. In work on anthrobots, adult human airway cells self-organized into mobile constructs, and in neural cultures, they accelerated the closure of damaged areas. These results provided researchers with a material in which to observe the form and function of cellular collectives after changing conditions.
🔗 Read original →
PubMed Central (PMC)
Kinematic self-replication in reconfigurable organisms
Almost all organisms replicate by growing and then shedding offspring. Some molecules also replicate, but by moving rather than growing: They find and combine building blocks into self-copies. Here we show that clusters of cells, if freed from a ...
Gene Regulator Found
Researchers have discovered that the NHR-49 gene regulator in worms prevents egg cells from maturing without sperm and preserves the organism's reserve. On August 22, the journal Nature Communications published an article on how nutrition is linked to the decision to expend resources on reproduction in the roundworm C. elegans.
When the nhr-49 gene was disrupted, egg cells matured and were released without sperm, causing the organism to lose its yolk and fat reserves, and its lifespan was reduced. In C. elegans, sperm serve as a signal to initiate reproduction, and without this signal, egg cells remain in the oviduct.
A 2025 study by the same group found that sperm-free lines at 25 °C accumulated more fat and lived longer. The current article seeks to identify the molecular mechanism that maintains this waiting state, which was found to be NHR-49, a protein that regulates genes related to nutrition and fat metabolism.
🔗 Read original →
Researchers have discovered that the NHR-49 gene regulator in worms prevents egg cells from maturing without sperm and preserves the organism's reserve. On August 22, the journal Nature Communications published an article on how nutrition is linked to the decision to expend resources on reproduction in the roundworm C. elegans.
When the nhr-49 gene was disrupted, egg cells matured and were released without sperm, causing the organism to lose its yolk and fat reserves, and its lifespan was reduced. In C. elegans, sperm serve as a signal to initiate reproduction, and without this signal, egg cells remain in the oviduct.
A 2025 study by the same group found that sperm-free lines at 25 °C accumulated more fat and lived longer. The current article seeks to identify the molecular mechanism that maintains this waiting state, which was found to be NHR-49, a protein that regulates genes related to nutrition and fat metabolism.
🔗 Read original →
PubMed Central (PMC)
Nuclear hormone receptor NHR-49 is an essential regulator of stress resilience and healthy aging in Caenorhabditis elegans
The genome of Caenorhabditis elegans encodes 284 nuclear hormone receptor, which perform diverse functions in development and physiology. One of the best characterized of these is NHR-49, related in sequence and function to mammalian hepatocyte ...
AI Research Evaluation
Authors propose evaluating AI researchers based on a trail of research decisions. On August 18, nine authors posted a framework for assessing scientific AI agents on ChemRxiv, a preprint repository. The framework's unit is the "discovery episode": a preserved sequence of hypotheses, actions, data, and revisions.
The authors suggest evaluating the entire sequence, including episodes of discovery, which record what was known before the next step, the action chosen by the agent, what was observed, and how the plan changed afterwards. For each step, the episode records the code version, instrument settings, human involvement, and safety rules.
The framework breaks down research work into three connected parts: hypothesis, execution, and interpretation. The authors advise starting with limited tasks, which have a clear goal, can be automatically evaluated, and fit within acceptable time and cost. The separate stages can then be connected into episodes and tested through independent repetition, allowing the evaluator to trace the path from decision to data and the next question, as published in ChemRxiv.
🔗 Read original →
Authors propose evaluating AI researchers based on a trail of research decisions. On August 18, nine authors posted a framework for assessing scientific AI agents on ChemRxiv, a preprint repository. The framework's unit is the "discovery episode": a preserved sequence of hypotheses, actions, data, and revisions.
The authors suggest evaluating the entire sequence, including episodes of discovery, which record what was known before the next step, the action chosen by the agent, what was observed, and how the plan changed afterwards. For each step, the episode records the code version, instrument settings, human involvement, and safety rules.
The framework breaks down research work into three connected parts: hypothesis, execution, and interpretation. The authors advise starting with limited tasks, which have a clear goal, can be automatically evaluated, and fit within acceptable time and cost. The separate stages can then be connected into episodes and tested through independent repetition, allowing the evaluator to trace the path from decision to data and the next question, as published in ChemRxiv.
🔗 Read original →
ChemRxiv
Measuring AI Scientists: From Exams to Discovery | ChemRxiv
Large language models and agentic systems are increasingly embedded across the scientific
work-flow, from literature synthesis and hypothesis generation to code execution,
data analysis and writing. This broadening of use exposes a mismatch in evaluation:…
work-flow, from literature synthesis and hypothesis generation to code execution,
data analysis and writing. This broadening of use exposes a mismatch in evaluation:…
Brain Immunity Hub
Researchers have discovered structures in the skull's bone marrow where the immune system recognizes antigens, molecular markers from the brain. On August 19, a study in Nature revealed that in mice, these immune cell clusters responded to antigens from the brain. In a glioma brain tumor model, suppressing these clusters weakened the anti-tumor immune response and reduced animal survival.
The brain is surrounded by cerebrospinal fluid, which bathes the brain and spinal cord. Channels between the brain's hard shell and the skull's bone marrow allow this fluid and its dissolved substances to reach neighboring tissue. A 2022 study traced this path and its effect on innate immune cells. The current article's authors investigated whether immune cells in the skull's bone marrow can recognize antigens from the brain.
In the back of the skull's bone marrow, the authors found clusters of B cells, which produce antibodies, T helper cells, and cells that show antigens to T cells. The B cells showed signs of germinal centers, where they are selected and mature to produce antibodies. Comparison with bone marrow from the breastbone and thigh bone, microscopy, and single-cell analysis revealed that such clusters are characteristic of the back of the skull's bone marrow.
In another experiment, mouse neurons were engineered to produce a model protein antigen to track its path. It was detected in the cerebrospinal fluid, brain membranes, and skull bone marrow; immune T and B cells recognizing this protein were activated there. The authors also injected glioma cells with the same antigen either into the brain or under the skin on the mouse's side. When the tumor was inside the brain, the reaction occurred in the skull's bone marrow, while a tumor on the side triggered a response in the inguinal lymph nodes.
🔗 Read original →
Researchers have discovered structures in the skull's bone marrow where the immune system recognizes antigens, molecular markers from the brain. On August 19, a study in Nature revealed that in mice, these immune cell clusters responded to antigens from the brain. In a glioma brain tumor model, suppressing these clusters weakened the anti-tumor immune response and reduced animal survival.
The brain is surrounded by cerebrospinal fluid, which bathes the brain and spinal cord. Channels between the brain's hard shell and the skull's bone marrow allow this fluid and its dissolved substances to reach neighboring tissue. A 2022 study traced this path and its effect on innate immune cells. The current article's authors investigated whether immune cells in the skull's bone marrow can recognize antigens from the brain.
In the back of the skull's bone marrow, the authors found clusters of B cells, which produce antibodies, T helper cells, and cells that show antigens to T cells. The B cells showed signs of germinal centers, where they are selected and mature to produce antibodies. Comparison with bone marrow from the breastbone and thigh bone, microscopy, and single-cell analysis revealed that such clusters are characteristic of the back of the skull's bone marrow.
In another experiment, mouse neurons were engineered to produce a model protein antigen to track its path. It was detected in the cerebrospinal fluid, brain membranes, and skull bone marrow; immune T and B cells recognizing this protein were activated there. The authors also injected glioma cells with the same antigen either into the brain or under the skin on the mouse's side. When the tumor was inside the brain, the reaction occurred in the skull's bone marrow, while a tumor on the side triggered a response in the inguinal lymph nodes.
🔗 Read original →
Nature
Functional role of skull lymphoid structures in CNS immunosurveillance
Nature - Functional lymphoid structures within the skull bone marrow enable central nervous system immunosurveillance and shape immune responses to brain disease.
Mouse Lung Repair
Researchers found that disabling genes that transmit tension to the nucleus of connective tissue cells in mice improved lung recovery after injury. On August 18, authors published a preprint on bioRxiv about damaged mouse lungs. When they genetically disabled two genes in fibroblasts, by day 28, this group had less scarring and severely damaged tissue.
Alveoli, the air sacs where blood receives oxygen, are restored by AT2 cells after injury. Nearby alveolar fibroblasts send signals to the epithelium needed for repair. After severe damage, tissue stretching can alter their function for an extended period. To separate the action of stretching from toxic damage, authors partially removed the lung, causing the remaining tissue to stretch more.
In another experiment, they tied off a bronchus of one lobe, reducing stretching in that area. With increased stretching, fibroblasts temporarily lost signs of their normal alveolar state, while reduced stretching preserved them. The researchers had previously shown in a 2024 study that the connection between fibroblasts and epithelium after injury could lead to pathological tissue remodeling; now they checked how physical signals maintained this change. The LINC complex, including Sun1 and Sun2, transmits nuclear tension from intracellular fibers.
🔗 Read original →
Researchers found that disabling genes that transmit tension to the nucleus of connective tissue cells in mice improved lung recovery after injury. On August 18, authors published a preprint on bioRxiv about damaged mouse lungs. When they genetically disabled two genes in fibroblasts, by day 28, this group had less scarring and severely damaged tissue.
Alveoli, the air sacs where blood receives oxygen, are restored by AT2 cells after injury. Nearby alveolar fibroblasts send signals to the epithelium needed for repair. After severe damage, tissue stretching can alter their function for an extended period. To separate the action of stretching from toxic damage, authors partially removed the lung, causing the remaining tissue to stretch more.
In another experiment, they tied off a bronchus of one lobe, reducing stretching in that area. With increased stretching, fibroblasts temporarily lost signs of their normal alveolar state, while reduced stretching preserved them. The researchers had previously shown in a 2024 study that the connection between fibroblasts and epithelium after injury could lead to pathological tissue remodeling; now they checked how physical signals maintained this change. The LINC complex, including Sun1 and Sun2, transmits nuclear tension from intracellular fibers.
🔗 Read original →
Longevity Support
Aleksey Strigin called on participants in the longevity movement to regularly support strong texts from colleagues. On August 22, the author of the "Economy of Life Extension" channel asked people to spend 10-20 seconds liking or reposting when colleagues ask for help spreading their work, and to ask for such support themselves.
Strigin suggests making mutual recommendations a common practice, so that texts about aging can reach people beyond the usual circle of readers. A strong text can remain in a small channel, where it will be seen mainly by already interested subscribers. A repost shows the text to a different audience: readers trust the person sharing the link and decide whether to read further.
In Facebook, likes, comments, and reposts become signals for the feed that the service selects for each user. The Meta company, which owns Facebook, explains that one of its predictions estimates the likelihood of a repost; this prediction participates in selecting the order of posts. Strigin formulates his stake as: Attention. It is more important than money. It attracts money, talent, and other resources.
A repost associates a person's name with someone else's text. Strigin recalls that he used to be shy about making such requests. The entire community receives a new audience, and each distributor decides whether they are ready to recommend specific material. Mutual support is based on selection. A person first reads the text, then shares what they are willing to be responsible for in front of their subscribers. Repeated recommendations give strong material new audiences, if people continue to choose what they are willing to support with their name.
🔗 Read original →
Aleksey Strigin called on participants in the longevity movement to regularly support strong texts from colleagues. On August 22, the author of the "Economy of Life Extension" channel asked people to spend 10-20 seconds liking or reposting when colleagues ask for help spreading their work, and to ask for such support themselves.
Strigin suggests making mutual recommendations a common practice, so that texts about aging can reach people beyond the usual circle of readers. A strong text can remain in a small channel, where it will be seen mainly by already interested subscribers. A repost shows the text to a different audience: readers trust the person sharing the link and decide whether to read further.
In Facebook, likes, comments, and reposts become signals for the feed that the service selects for each user. The Meta company, which owns Facebook, explains that one of its predictions estimates the likelihood of a repost; this prediction participates in selecting the order of posts. Strigin formulates his stake as: Attention. It is more important than money. It attracts money, talent, and other resources.
A repost associates a person's name with someone else's text. Strigin recalls that he used to be shy about making such requests. The entire community receives a new audience, and each distributor decides whether they are ready to recommend specific material. Mutual support is based on selection. A person first reads the text, then shares what they are willing to be responsible for in front of their subscribers. Repeated recommendations give strong material new audiences, if people continue to choose what they are willing to support with their name.
🔗 Read original →
Meta
Our Approach to Facebook Feed Ranking| Transparency Center
Meta regularly publishes reports to give our community visibility into community standards enforcement, government requests and internet disruptions
Muscle Strength Boost
Researchers found that disabling the P311 gene helped damaged muscles in old mice develop 19% more strength. On August 22, in an article in npj Aging, authors described an experiment on 24-month-old mice: they disabled the P311 gene throughout the body and chemically damaged the tibialis anterior muscle.
Through 28 days after injury, this muscle developed 19% more strength than in similar old mice with P311 enabled. As muscles age, damaged areas often heal with excess collagen, forming scar tissue that hinders fiber function. The authors chose P311 because previous work linked the protein it codes to TGF-β production, a signaling molecule that promotes such tissue formation.
After injury, P311 and TGF-β levels in old muscle increased more than in young muscle. The authors first checked the tissue part of this chain: 14 days after injury, mice without P311 in muscle had less collagen and lower fibrosis-related gene activity. By 28 days, the average cross-sectional area of recovering fibers was 25% larger, and the same tibialis anterior muscle developed 19% more strength.
🔗 Read original →
Researchers found that disabling the P311 gene helped damaged muscles in old mice develop 19% more strength. On August 22, in an article in npj Aging, authors described an experiment on 24-month-old mice: they disabled the P311 gene throughout the body and chemically damaged the tibialis anterior muscle.
Through 28 days after injury, this muscle developed 19% more strength than in similar old mice with P311 enabled. As muscles age, damaged areas often heal with excess collagen, forming scar tissue that hinders fiber function. The authors chose P311 because previous work linked the protein it codes to TGF-β production, a signaling molecule that promotes such tissue formation.
After injury, P311 and TGF-β levels in old muscle increased more than in young muscle. The authors first checked the tissue part of this chain: 14 days after injury, mice without P311 in muscle had less collagen and lower fibrosis-related gene activity. By 28 days, the average cross-sectional area of recovering fibers was 25% larger, and the same tibialis anterior muscle developed 19% more strength.
🔗 Read original →
Nature
Deletion of P311 enhances regeneration in aged skeletal muscle
npj Aging - Deletion of P311 enhances regeneration in aged skeletal muscle
Longevity Research Funding
Biogerontologist Matt Kieberlain wrote that exaggerated claims about life extension may undermine trust in data and complicate funding for aging research. Investor Carl Pfleger suggested testing this connection using historical examples. In a detailed post, Kieberlain links gerontology - the study of aging biology - to two conditions: funding for work and data that colleagues are willing to trust.
To move faster, both resources and quality science are needed, he writes. This year, Kieberlain visited the US Congress offices four times, and in three cases, his interlocutors, who had already heard about aging science, associated it with hype and "snake oil". Before discussing new research, he had to return the conversation to the question of whether the data could be trusted.
Kieberlain sees the historical cause of this concern in Sirtris, a biomedicine company that GSK announced it would acquire in 2008 for $720 million. According to Kieberlain, exaggerated expectations around Sirtris long complicated the flow of resources to aging research. Carl Pfleger, an investor in rejuvenation startups, suggests testing this connection using historical examples, citing the 1970s cancer research as a comparable case.
🔗 Read original →
Biogerontologist Matt Kieberlain wrote that exaggerated claims about life extension may undermine trust in data and complicate funding for aging research. Investor Carl Pfleger suggested testing this connection using historical examples. In a detailed post, Kieberlain links gerontology - the study of aging biology - to two conditions: funding for work and data that colleagues are willing to trust.
To move faster, both resources and quality science are needed, he writes. This year, Kieberlain visited the US Congress offices four times, and in three cases, his interlocutors, who had already heard about aging science, associated it with hype and "snake oil". Before discussing new research, he had to return the conversation to the question of whether the data could be trusted.
Kieberlain sees the historical cause of this concern in Sirtris, a biomedicine company that GSK announced it would acquire in 2008 for $720 million. According to Kieberlain, exaggerated expectations around Sirtris long complicated the flow of resources to aging research. Carl Pfleger, an investor in rejuvenation startups, suggests testing this connection using historical examples, citing the 1970s cancer research as a comparable case.
🔗 Read original →
Nature
GSK moves on Sirtris
Nature Biotechnology - GSK moves on Sirtris
💯1
Epigenetic Clocks
Мартин Йенсен proposed checking epigenetic clocks against health outcomes before measuring therapy effects. He responded to TranslAGE, a new database on epigenetic clock responses to interventions, on August 22. The database, introduced in Nature Medicine on August 21, contains 3,128 samples from 51 longitudinal studies.
The authors calculated 16 epigenetic clocks for each dataset, which estimate age-related changes or mortality risk based on DNA chemical marks. The study allows comparison of how different clocks change after medications, diets, and other interventions. Йенсен suggests comparing these shifts with patient-important outcomes, such as organ function, disease, or mortality.
He uses the COSMOS study as an example, where daily multivitamins did not significantly reduce overall cardiovascular or mortality outcomes over a median of 3.6 years in 21,442 elderly participants. Йенсен proposes an independent test to validate the clocks, where developers make predictions for a set of interventions without knowing the outcomes, and then compare the predictions with the actual data.
🔗 Read original →
Мартин Йенсен proposed checking epigenetic clocks against health outcomes before measuring therapy effects. He responded to TranslAGE, a new database on epigenetic clock responses to interventions, on August 22. The database, introduced in Nature Medicine on August 21, contains 3,128 samples from 51 longitudinal studies.
The authors calculated 16 epigenetic clocks for each dataset, which estimate age-related changes or mortality risk based on DNA chemical marks. The study allows comparison of how different clocks change after medications, diets, and other interventions. Йенсен suggests comparing these shifts with patient-important outcomes, such as organ function, disease, or mortality.
He uses the COSMOS study as an example, where daily multivitamins did not significantly reduce overall cardiovascular or mortality outcomes over a median of 3.6 years in 21,442 elderly participants. Йенсен proposes an independent test to validate the clocks, where developers make predictions for a set of interventions without knowing the outcomes, and then compare the predictions with the actual data.
🔗 Read original →
PubMed Central (PMC)
Validation of biomarkers of aging
The search for biomarkers that quantify biological aging (particularly ‘omics’-based biomarkers) has intensified in recent years. Such biomarkers could predict aging-related outcomes and could serve as surrogate endpoints for the evaluation of ...
Oliver Barton Releases mTOR Atlas
The mTOR Atlas is a navigator for 322 works on the cellular system affected by rapamycin. Version 1.0.0 was released on August 22 and contains 45 topics, each with a model, intervention, and measured outcome to relate the publication to the question it answers.
The mTOR protein and signaling system, named after it, influences cell growth, protein synthesis, and autophagy through nutrient availability. Experiences in cell culture clarify the mechanism, experiences in animals test it in an organism, and human studies measure the outcome in humans. The Atlas index for each work preserves a link to the original publication, model, intervention, and outcome.
Two markings perform different functions: the pyramid shows how close the data stands to the outcome measured in humans, and levels A–D distinguish types of evidence. In the open upload, 29 level B works were conducted on humans, 84 level C works were conducted on animals, and 205 level D entries comprise mechanistic works and reviews. Such marking helps match the result with the conditions of the experience, as seen in fly experiments where the same dose of rapamycin on different diets changed the sign of the effect on lifespan.
One of the ten pages with open questions is dedicated to mTORC1 and mTORC2, two protein complexes of this system, and separates known results from a "justified hypothesis" about the rapamycin regimen that suppresses mTORC1 and spares mTORC2. To verify this, the Atlas suggests an experiment on mice: comparing lifespan, insulin sensitivity, and mTORC2 activity under different rapamycin regimens, as described in Nature Aging, July 2026.
🔗 Read original →
The mTOR Atlas is a navigator for 322 works on the cellular system affected by rapamycin. Version 1.0.0 was released on August 22 and contains 45 topics, each with a model, intervention, and measured outcome to relate the publication to the question it answers.
The mTOR protein and signaling system, named after it, influences cell growth, protein synthesis, and autophagy through nutrient availability. Experiences in cell culture clarify the mechanism, experiences in animals test it in an organism, and human studies measure the outcome in humans. The Atlas index for each work preserves a link to the original publication, model, intervention, and outcome.
Two markings perform different functions: the pyramid shows how close the data stands to the outcome measured in humans, and levels A–D distinguish types of evidence. In the open upload, 29 level B works were conducted on humans, 84 level C works were conducted on animals, and 205 level D entries comprise mechanistic works and reviews. Such marking helps match the result with the conditions of the experience, as seen in fly experiments where the same dose of rapamycin on different diets changed the sign of the effect on lifespan.
One of the ten pages with open questions is dedicated to mTORC1 and mTORC2, two protein complexes of this system, and separates known results from a "justified hypothesis" about the rapamycin regimen that suppresses mTORC1 and spares mTORC2. To verify this, the Atlas suggests an experiment on mice: comparing lifespan, insulin sensitivity, and mTORC2 activity under different rapamycin regimens, as described in Nature Aging, July 2026.
🔗 Read original →
Liver Age Tool Released
The authors have released LivAge, an open tool for assessing the age of mouse livers based on gene activity. On August 21, an article about LivAge was published in Aging Cell. The tool receives RNA-seq results and provides an estimate of the mouse liver age in months. The authors have also made the calculation code available.
In aging experiments, it can be difficult to determine whether a diet, medication, or genetic modification affects tissue condition, as the passport age of the control and experimental groups may be the same, and differences in lifespan may take a long time to become apparent. LivAge reduces a large table of gene activity to a single indicator that can be used to compare groups.
The authors trained the model on 432 liver samples from healthy C57BL/6 mice from 23 studies. The age of the animals ranged from one to 30 months, and only control groups without genetic modifications or interventions were included. The algorithm selected 268 genes whose joint activity determines the liver age estimate. The final test was conducted on 134 samples from four other studies that were not used for training and model tuning.
🔗 Read original →
The authors have released LivAge, an open tool for assessing the age of mouse livers based on gene activity. On August 21, an article about LivAge was published in Aging Cell. The tool receives RNA-seq results and provides an estimate of the mouse liver age in months. The authors have also made the calculation code available.
In aging experiments, it can be difficult to determine whether a diet, medication, or genetic modification affects tissue condition, as the passport age of the control and experimental groups may be the same, and differences in lifespan may take a long time to become apparent. LivAge reduces a large table of gene activity to a single indicator that can be used to compare groups.
The authors trained the model on 432 liver samples from healthy C57BL/6 mice from 23 studies. The age of the animals ranged from one to 30 months, and only control groups without genetic modifications or interventions were included. The algorithm selected 268 genes whose joint activity determines the liver age estimate. The final test was conducted on 134 samples from four other studies that were not used for training and model tuning.
🔗 Read original →
PubMed Central (PMC)
LivAge: An Online Aging Clock for Murine Transcriptomic Age Estimation
The increase in life expectancy over the past century has been accompanied by the recognition of age as the primary risk factor for a wide range of pathologies, including cardiovascular and neurodegenerative diseases, as well as cancer. Thus, the ...
Brain Reads Thoughts
Researchers have made a breakthrough in non-invasive EEG-neurointerfaces, discovering that they can read specific words, including rare ones, from an open dictionary. A massive dataset was collected, consisting of around 240,000 words read by one person over 49 hours in 393 separate sessions. The study utilized a 19-channel dry EEG, eliminating the need for gel, surgery, or invasive sensors.
The words were displayed in a rapid sequential presentation, with the font changing each time to prevent the brain from "guessing" the answer based on visual form. The model consisted of two parts: a convolutional EEG encoder and a causal transformer, trained using a contrastive scheme similar to CLIP. The system learned to associate brain activity with semantic and lexical features of words, as described in Nature Aging, July 2026.
The accuracy was measured as the top-10 hit rate and was consistently above the random level, including words with medium and low frequency. The quality improved log-linearly with the amount of data and did not reach saturation, meaning that the more data, the better the decoding. Removing occipital and parietal electrodes reduced accuracy by about a third but did not affect the model's ability to track text context. Control experiments showed that the model actually recognizes words, rather than just guessing based on position or context.
🔗 Read original →
Researchers have made a breakthrough in non-invasive EEG-neurointerfaces, discovering that they can read specific words, including rare ones, from an open dictionary. A massive dataset was collected, consisting of around 240,000 words read by one person over 49 hours in 393 separate sessions. The study utilized a 19-channel dry EEG, eliminating the need for gel, surgery, or invasive sensors.
The words were displayed in a rapid sequential presentation, with the font changing each time to prevent the brain from "guessing" the answer based on visual form. The model consisted of two parts: a convolutional EEG encoder and a causal transformer, trained using a contrastive scheme similar to CLIP. The system learned to associate brain activity with semantic and lexical features of words, as described in Nature Aging, July 2026.
The accuracy was measured as the top-10 hit rate and was consistently above the random level, including words with medium and low frequency. The quality improved log-linearly with the amount of data and did not reach saturation, meaning that the more data, the better the decoding. Removing occipital and parietal electrodes reduced accuracy by about a third but did not affect the model's ability to track text context. Control experiments showed that the model actually recognizes words, rather than just guessing based on position or context.
🔗 Read original →
arXiv.org
Decoding silent reading from non-invasive EEG
Non-invasive decoding of inner speech faces a fundamental data problem: a corpus pairing brain activity with a person's spontaneous inner monologue cannot be collected, and the available proxy...