🐟 Adaptive cellular evolution in the intestine of hyperdiverse cichlid fishes

Animals, Female, Male, Adaptation, Physiological, Biological Evolution, Cichlids, Diet, Enterocytes, Feeding Behavior, Intestinal Mucosa, Intestines, Lakes, Phylogeny, Single-Cell Gene Expression Analysis, Transcriptome

Researchers investigated how cichlid fishes from Lake Tanganyika adapted their digestive systems to radically different diets, using single-cell transcriptomics across 24 species with varied feeding habits. By combining this cellular-level gene expression data with ecological, morphological, and genomic information, they found that dietary adaptation in the intestine centers on a specific cell type: anterior enterocytes. Both the proportion of these cells and their gene expression patterns shifted in response to different diets, driven by rapidly evolving genes specific to that cell population. These findings reveal that ecological specialization operates simultaneously at multiple biological levels — from genome evolution to cell-type composition and molecular profiles within individual cell populations. The work broadens understanding of adaptive radiation beyond feeding structures like jaws and teeth to include the cellular architecture of the gut itself.

Read the original article →

🦴 Genetic diversity of late Neanderthals in northwestern Europe

Animals, Female, Humans, Male, Belgium, France, Gene Flow, Genetic Variation, History, Ancient, Neanderthals, Phylogeny

Researchers sequenced genetic material from 27 late Neanderthals (~52,500 years old or younger) recovered from ten sites across Belgium and France, including a high-coverage genome from a ~45,000-year-old individual. Most of these individuals were more closely related to each other than to other late Neanderthals elsewhere in Europe, suggesting a regional population. Unlike the famously isolated Altai Neanderthals, these western Europeans showed no signs of close-relative inbreeding, implying they lived in larger or better-connected social networks. Some individuals carried ancestry from an older Neanderthal lineage predating the divergence of late Neanderthals. Despite overlapping in time with early modern humans in the region, no evidence of recent interbreeding with those humans was found. Crucially, genetic load did not increase over time, undermining the hypothesis that progressive inbreeding-driven genetic decline caused Neanderthal extinction.

Read the original article →

🩸 Human haematopoietic stem cells remember inflammatory stress

Adult, Animals, Child, Female, Humans, Male, Mice, Aging, Anemia, Sickle Cell, Cell Differentiation, Clonal Hematopoiesis, COVID-19, Hematopoiesis, Hematopoietic Stem Cells, Heterografts, Inflammation, Single-Cell Analysis, Stress, Physiological

Researchers investigated how human haematopoietic stem cells (HSCs) respond to and remember inflammatory stress — a question with major implications for ageing and blood cancers. Using xenograft mouse models combined with single-cell multiomics, they identified two molecularly distinct HSC subsets, one of which, called HSC-iM, retains an epigenetic and transcriptional memory of past inflammation. HSC-iM cells tend toward quiescence and reduced blood cell output. This subset was detected in samples from patients recovering from COVID-19, living with sickle cell disease, ageing, and clonal haematopoiesis, validating its physiological relevance. Clonal haematopoiesis mutations partially counteracted HSC-iM's suppressive effects by promoting cell activation. The HSC-iM transcriptional signature was transmitted to downstream immune cells and, critically, its enrichment in circulating blood was linked to increased all-cause mortality risk in population cohorts, highlighting its potential as a biomarker of long-term health outcomes.

Read the original article →

🐭 Specific expansion of motor cortical projections in a singing mouse

Animals, Female, Male, Mice, Auditory Cortex, Motor Cortex, Neural Pathways, Species Specificity, Vocalization, Animal, Rodentia

Researchers investigated how differences in brain wiring might explain why Alston's singing mouse produces complex vocalizations that lab mice cannot. Using bulk neural tracing, serial two-photon tomography, and high-throughput DNA sequencing of over 76,000 individually barcoded neurons, they mapped motor cortical projections across the whole brain in both species. They found that orofacial motor cortex in the singing mouse has substantially expanded connections to an auditory cortical region and the midbrain periaqueductal grey — both areas tied to vocal control. Single-neuron projection analysis showed this expansion is driven largely by neurons projecting exclusively to the auditory cortex, rather than by neurons that broadcast to multiple targets. The findings suggest that selective strengthening of pre-existing cortical pathways can produce major behavioral differences over short evolutionary timescales, with implications for understanding how enhanced cortical vocal control may have paved the way for human language.

Read the original article →

🤖 A multi-agent system for automating scientific discovery

Animals, Humans, ATP Binding Cassette Transporter 1, Automation, Drug Discovery, Intelligent Systems, Macular Degeneration, Phagocytosis, Retinal Pigment Epithelium, rho-Associated Kinases, Sequence Analysis, RNA, Sulfonamides

Researchers built Robin, a multi-agent AI system that automates the full scientific discovery cycle—literature search, hypothesis generation, experimental design, and data analysis—for experimental biology. Applied to dry age-related macular degeneration (AMD), Robin proposed boosting retinal pigment epithelium phagocytosis as a therapeutic strategy, then identified two candidate compounds. One, ripasudil (a Rho kinase inhibitor already approved for glaucoma), had never previously been suggested for dry AMD; both compounds showed confirmed in vitro efficacy. Robin then designed and analyzed an RNA sequencing follow-up experiment, uncovering upregulation of ABCA1, a lipid efflux pump gene, as a potential novel mechanistic target. Notably, all hypotheses, analyses, and figures in the paper were generated by Robin itself. The system represents one of the first AI frameworks to autonomously complete iterative experimental discovery cycles, offering a new template for accelerating biomedical research.

Read the original article →

🤖 Accelerating scientific discovery with Co-Scientist

Humans, Artificial Intelligence, Drug Discovery, Reproducibility of Results, Time Factors, Generative Artificial Intelligence

Researchers built Co-Scientist, a multi-agent AI system using Google's Gemini models, designed to help scientists generate and refine research hypotheses. The system uses an asynchronous agent framework where multiple agents continuously produce, criticize, and improve hypotheses, with a tournament-style process that self-improves hypothesis quality as more compute is applied at inference time. To validate it beyond automated benchmarks, the team focused on three biomedical areas: drug repurposing, novel target discovery, and antimicrobial resistance mechanisms. Notably, Co-Scientist identified new drug-repurposing candidates and synergistic combination therapies for acute myeloid leukemia that were subsequently confirmed through in vitro laboratory experiments. The results suggest that scaling test-time compute meaningfully improves hypothesis quality, and that AI systems of this kind could meaningfully accelerate the front end of scientific discovery by generating experimentally testable, genuinely novel ideas.

Read the original article →

🔬 Ubiquitination of glycogen and metabolites in cells and tissues

Animals, Ubiquitination, Humans, Glycogen, Mice, Liver, Muscle, Skeletal, Male, Lysosomes, Ubiquitin, Female, Glycogen Storage Disease, Mass Spectrometry

Ubiquitin is well known for tagging proteins, but whether it also modifies non-protein molecules at physiologically relevant levels has been difficult to assess because standard proteomics methods miss such targets. Researchers developed a mass spectrometry technique called NoPro-clipping, which pairs ubiquitin-cleaving enzymes with sortase-based labeling to detect ubiquitinated non-protein substrates. Applying this approach to mammalian cells and mouse and human tissues, they discovered that glycogen is ubiquitinated, with the highest levels in liver and skeletal muscle. This glycogen ubiquitination appears to route glycogen to lysosomes for degradation, is altered in glycogen storage diseases, and rises during liver fasting—implying ubiquitin plays a previously unrecognized role in normal glycogen breakdown. The method also uncovered ubiquitination of glycerol and spermine. Collectively, the findings reframe ubiquitin as a broad biomolecule modifier rather than an exclusively protein-targeting signal.

Read the original article →

🧬 An X-linked long non-coding RNA, PTCHD1-AS, and the core features of autism

Animals, Female, Humans, Male, Mice, Autism Spectrum Disorder, Autistic Disorder, Chromosome Deletion, Disease Models, Animal, Exons, Genes, X-Linked, Genetic Predisposition to Disease, Hippocampus, Mice, Knockout, RNA, Long Noncoding, Social Behavior, Membrane Proteins

Researchers analyzed whole-genome sequencing data from over 17,000 individuals and found that small X-chromosome deletions affecting the long non-coding RNA gene PTCHD1-AS are associated with autism spectrum disorder (ASD) in males. Two knockout mouse models confirmed that disrupting the conserved exon 3 of this gene produces core autism-like traits—repetitive behaviors, impaired social interaction and communication—without the cognitive problems or ADHD-like symptoms often seen with other ASD-linked genes. The gene is expressed predominantly in the dorsal striatum, a GABAergic brain region tied to ASD. Multi-omics analyses revealed disrupted myelination and synaptic plasticity pathways in striatal cells, along with reduced protein kinase C isoforms and altered phosphorylation of SRC and GSK-3α/β, ultimately enhancing striatal long-term potentiation and depression. These results point to PTCHD1-AS as a rare, non-coding ASD susceptibility gene that drives core social and repetitive symptoms through striatal circuit dysregulation.

Read the original article →

🫀 Lineage and organ signals sequentially build organ intrinsic nervous systems

Animals, Cell Lineage, Extracellular Matrix, Mice, Neural Crest, Neurogenesis, Pancreas, Neurons, Nervous System, Cell Movement, Organ Specificity, Heart, Single-Cell Analysis, Intestines, Lung, Signal Transduction, Integrins, Coculture Techniques, Cell Differentiation

Organ intrinsic nervous systems (OINSs) — the local neural networks embedded in the heart, gut, pancreas, and lungs — are essential for linking organ function to whole-body physiology, yet how they develop from shared neural crest cell origins into distinct architectures was poorly understood. Researchers combined lineage tracing, 3D imaging, single-cell transcriptomics, and genetic experiments across all four organs to dissect this process. They found that the migration routes of neural crest cells establish the spatial layout of each OINS, while molecular identity is shaped later by local organ environments. Co-culture experiments showed that organ-derived signals reprogram neurons toward organ-specific gene expression profiles, with extracellular matrix (ECM) contact playing a central instructive role. In the heart specifically, ECM-integrin signaling drives neurogenesis, and ECM crosslinking stabilizes the characteristic ganglionic arrangement. The findings reveal a two-stage logic: lineage determines spatial scaffolding, while organ microenvironments specify molecular identity.

Read the original article →