Most bacterial pathogens are polylysogens, harbouring multiple vertically transmitted prophages1-3. These prophages enhance bacterial pathogenicity and survival by encoding virulence factors and anti-phage defence systems while retaining the capacity for horizontal transfer. Thus, prophage-encoded anti-phage defences must block propagation of external phages without inhibiting the spread of the prophages that encode them. Here we identify HepS-an abortive infection system encoded on the Gifsy-1 prophage constituted of a single HEPN domain protein-which restricts phages of the Siphoviridae family. We demonstrate that in its native host context of Salmonella enterica serovar Typhimurium, HepS both senses phage infection and enacts abortive infection. Structures of HepS reveal a tetrameric nuclease complex that undergoes allosteric activation upon recognition of Siphoviridae tail tip proteins during production of new phage particles. Once activated, HepS cleaves specific transfer RNA anticodon loops and arrests phage replication. Gifsy-1, a Siphoviridae itself, evades self-targeting by expressing a tail tip variant that does not trigger HepS, as do co-resident Siphoviridae prophages Gifsy-2 and Gifsy-3. This evasion permits Gifsy-1 to spread despite encoding HepS. These findings reveal a mechanism by which a prophage defends the host while maintaining its propagation abilities.

Epstein-Barr virus (EBV) is an endemic herpesvirus implicated in autoimmunity, cancer and neurological disorders. Although primary infection is often subclinical, persistent EBV infection can drive immune dysregulation and long-term complications. Despite the ubiquity of infection, the determinants of EBV persistence following primary exposure remain poorly understood, although human genetic variation partially contributes to this phenotypic spectrum1-3. Here we demonstrate that existing whole genome sequencing (WGS) data of human populations can be used to quantify persistent EBV DNA. Using WGS and health record data from the UK Biobank (n = 490,560) and All of Us (n = 245,394), we uncover reproducible associations between blood-derived EBV DNA quantifications and respiratory, autoimmune, neurological and cardiovascular diseases. We evaluate genetic determinants of persistent EBV DNA via genome association studies, revealing heritability enrichment in immune-associated regulatory regions and protein-altering variants in 148 genes. Single-cell and pathway level analyses of these loci implicate variable antigen processing as a primary determinant of EBV DNA persistence. Further, relevant gene programs were enriched in B cells and antigen-presenting cells, consistent with their roles in viral reservoir and clearance. Human leukocyte antigen genotyping and predicted viral epitope presentation affinities implicate major histocompatibility complex class II variation as a key modulator of EBV persistence. Together, our analyses demonstrate how re-analysis of human population-scale WGS data can elucidate the genetic architecture of viral DNA persistence, a framework generalizable to the broader human virome4.

Time-dependent drives hold promise for realizing non-equilibrium many-body phenomena that are absent in undriven systems1-3. Yet, drive-induced heating normally destabilizes the systems4,5, which can be parametrically suppressed in the high-frequency regime by using periodic (Floquet) drives6,7. It remains largely unknown to what extent highly controllable quantum simulators can suppress heating in non-periodically driven systems. Here, using the 78-qubit superconducting quantum processor, Chuang-tzu 2.0, we report the experimental observation of long-lived prethermal phases in many-body systems with tunable heating rates, driven by structured random protocols, characterized by n-multipolar temporal correlations. By measuring both the particle imbalance and subsystem entanglement entropy, we monitor the entire heating process over 1,000 driving cycles and observe the existence of the prethermal plateau. The prethermal lifetime is 'doubly tunable': one way by driving frequency, the other way by multipolar order; it grows algebraically with the frequency with the universal scaling exponent 2n + 1. Using quantum-state tomography on different subsystems, we demonstrate a non-uniform spatial entanglement distribution and observe a crossover from area-law to volume-law entanglement scaling. With 78 qubits and 137 couplers in a two-dimensional configuration, the entire far-from-equilibrium heating dynamics are beyond the reach of simulation using tensor-network numerical techniques. Our work highlights superconducting quantum processors as a powerful platform for exploring universal scaling laws and non-equilibrium phases of matter in driven systems in regimes where classical simulation faces formidable challenges.

Early development across vertebrates and insects critically relies on robustly reorganizing the cytoplasm of fertilized eggs into individualized cells1,2. This intricate process is orchestrated by large microtubule structures that traverse the embryo, partitioning the cytoplasm into physically distinct and stable compartments3,4. Here, despite the robustness of embryonic development, we uncover an intrinsic instability in cytoplasmic partitioning driven by the microtubule cytoskeleton. By combining experiments in cytoplasmic extract and in vivo, we reveal that embryos circumvent this instability through two distinct mechanisms: either by matching the cell-cycle duration to the time needed for the instability to unfold or by limiting microtubule nucleation. These regulatory mechanisms give rise to two possible strategies to fill the cytoplasm, which we experimentally demonstrate in zebrafish and Drosophila embryos, respectively. In zebrafish embryos, unstable microtubule waves fill the geometry of the entire embryo from the first division. Conversely, in Drosophila embryos, stable microtubule asters resulting from reduced microtubule nucleation gradually fill the cytoplasm throughout multiple divisions. Our results indicate that the temporal control of microtubule dynamics could have driven the evolutionary emergence of species-specific mechanisms for effective cytoplasmic organization. Furthermore, our study unveils a fundamental synergy between physical instabilities and biological clocks, uncovering universal strategies for rapid, robust and efficient spatial ordering in biological systems.

Allergic diseases are caused by overexuberant type II immune responses mounted against environmental antigens1. The allergic state is typified by the presence of allergen-reactive immunoglobulin E (IgE), which triggers mast cell degranulation upon allergen encounter, manifesting in pruritis, oedema and, in severe cases, anaphylaxis. Over the past century, the prevalence of allergic diseases has increased markedly, suggesting that environmental rather than genetic factors are mediating this change2. Although many hypotheses connecting environment to allergy exist3-6, the biological mechanisms that underpin environmentally mediated protection from allergy are unknown. Here we show, using a mouse model of allergic disease, that exposure to immunostimulatory environments generated cross-reactive adaptive immune memory, which tracked with obstructed type II immune responses upon allergen exposure. We found that engagement of cross-reactive adaptive immunity protected against future allergic sensitization and suppressed established allergic responses. Cross-reactivity in a tolerogenic context also prevented allergy, with the effect extending across antigenically complex exposures even at low protein sequence similarity. Our findings demonstrate a mechanistic relationship between environment and allergy, with general implications for adaptive immune function in natural settings.

Cranial nerves densely innervate the heart and vasculature, with sensory neurons reporting on blood pressure, respiratory gases and tissue damage1. The roles of arterial baroreceptors in systemic physiology are well appreciated2, but the functions of vagal cardiac mechanoreceptors have been more difficult to parse, in part due to the closed-loop structure of the cardiovascular system. Here we use genetic tools in mice to identify a small group of neurons that are acutely sensitive to circulating blood volume and initiate a reflex that compensates for decreased filling of the heart in an upright posture and haemorrhage. Vagal PIEZO2 neurons form characteristic end-net endings in the heart, lower blood pressure in response to optogenetic stimulation and display blood-volume-dependent responses with every heartbeat that are time-locked to atrial and ventricular systole. Knockout of Piezo2 and/or ablation of PIEZO2 neurons in vagal ganglia eliminates this heartbeat-coupled nerve activity, causes orthostatic hypotension and compromises cardiovascular stability during trauma-induced blood loss. Together, these findings demonstrate that vagal mechanoreceptors monitor the cardiac cycle and initiate a blood-volume-dependent reflex that defends the constancy of circulation.

The invention of the laser transformed optics by providing intense, coherent light in the visible region, but extending this concept to X-rays has been hindered by a lack of suitable gain media and mirrors. Current hard X-ray free-electron laser (XFEL) facilities1-5 overcome this by amplifying shot noise from a high-peak-current electron bunch via self-amplified spontaneous emission6 in a single pass through long undulators, delivering very high brightness but with a noisy, multi-spiked temporal and spectral profile. Cavity-based XFELs (CBXFELs)7-9 were proposed to close this gap by recirculating spectrally filtered X-ray pulses in a Bragg-reflecting cavity synchronized to a high-repetition-rate electron beam. Here we show lasing with multi-pass gain at 6.952 keV in a 132.8-m round-trip diamond-based Bragg cavity10 at the European XFEL, matched to the 2.23-MHz bunch spacing of the superconducting accelerator5. Under stringent length and angular stability requirements, a ring-up in the cavity across successive bunches was observed, producing spectrally pure, microjoule-level pulses. This establishes the feasibility of CBXFELs in an accelerator environment and validates diamond Bragg optics for X-ray resonators. The demonstrated spectral purity opens a path to next-generation X-ray science, which demands highly coherent, stable sources.

Human genetic variation influences all aspects of our biology, including the oral cavity1-3, through which nutrients and microbes enter the body. Yet it is largely unknown which human genetic variants shape a person's oral microbiome and potentially promote its dysbiosis3-5. We characterized the oral microbiomes of 12,519 people by re-analysing whole-genome sequencing reads from previously sequenced saliva-derived DNA. Human genetic variation at 11 loci (10 new) associated with variation in oral microbiome composition. Several of these related to carbohydrate availability; the strongest association (P = 3.0 × 10-188) involved the common FUT2 W154X loss-of-function variant, which associated with the abundances of 58 bacterial species. Human host genetics also seemed to powerfully shape genetic variation in oral bacterial species: these 11 host genetic variants also associated with variation of gene dosages in 68 regions of bacterial genomes. Common, multi-allelic copy number variation of AMY1, which encodes salivary amylase, associated with oral microbiome composition (P = 1.5 × 10-53) and with dentures use in UK Biobank (P = 5.9 × 10-35, n = 418,039) but not with body mass index (P = 0.85), suggesting that salivary amylase abundance impacts health by influencing the oral microbiome. Two other microbiome composition-associated loci, FUT2 and PITX1, also significantly associated with dentures risk, collectively nominating numerous host-microbial interactions that contribute to tooth decay.

Eukaryotic genome replication is surveyed by the S-phase checkpoint, which coordinates sequential origin activation to prevent the exhaustion of poorly defined, rate-limiting replisome components1-3. Here we show that excessive origin firing saturates chromatin-bound proliferating cell nuclear antigen (PCNA)-a sliding clamp for DNA polymerase processivity and Okazaki fragment processing4-thereby restricting further PCNA loading and lagging-strand synthesis when checkpoint control is lost. PCNA-associated factor 15 (PAF15) emerges as a dosage-sensitive regulator of this process5-9. During unperturbed S phase, the entire soluble PAF15 pool binds to chromatin, leaving no reserve to stabilize PCNA under conditions of excessive origin activation. PAF15 binds to PCNA specifically on the lagging strand through a high-affinity PIP motif and occupies the DNA-encircling channel, protecting the clamp and associated enzymes from premature unloading by the ATAD5-RFC complex. Conversely, overexpression of PAF15 or forced redistribution to the leading strand disrupts replisome progression and induces cell death. These detrimental effects are mitigated by Timeless-Claspin, which blocks PAF15-PCNA binding on the leading strand. E2F4-mediated repression fine-tunes PAF15 expression to ensure optimal dosage and strand specificity. These findings reveal a previously unrecognized replisome constraint: when PAF15-PCNA assemblies are exhausted, the S-phase checkpoint globally restricts origin activation, linking a strand-specific rate-limiting mechanism to global replication dynamics.

Pesticides are widely distributed in soils1-3, yet their effects on soil biodiversity remain poorly understood4-7. Here we examined the effects of 63 pesticides on soil archaea, bacteria, fungi, protists, nematodes, arthropods and key functional gene groups across 373 sites spanning woodlands, grasslands and croplands in 26 European countries. Pesticide residues were detected in 70% of sites and emerged as the second strongest driver of soil biodiversity patterns after soil properties. Our analysis further revealed organism- and function-specific patterns, emphasizing complex and widespread non-target effects on soil biodiversity. Pesticides altered microbial functions, including phosphorus and nitrogen cycling, and suppressed beneficial taxa, including arbuscular mycorrhizal fungi and bacterivore nematodes. Our findings highlight the need to integrate functional and taxonomic characteristics into future risk assessment methodology to safeguard soil biodiversity, a cornerstone of ecosystem functioning.

Benchmarks are important tools for tracking the rapid advancements in large language model (LLM) capabilities. However, benchmarks are not keeping pace in difficulty: LLMs now achieve more than 90% accuracy on popular benchmarks such as Measuring Massive Multitask Language Understanding1, limiting informed measurement of state-of-the-art LLM capabilities. Here, in response, we introduce Humanity's Last Exam (HLE), a multi-modal benchmark at the frontier of human knowledge, designed to be an expert-level closed-ended academic benchmark with broad subject coverage. HLE consists of 2,500 questions across dozens of subjects, including mathematics, humanities and the natural sciences. HLE is developed globally by subject-matter experts and consists of multiple-choice and short-answer questions suitable for automated grading. Each question has a known solution that is unambiguous and easily verifiable but cannot be quickly answered by internet retrieval. State-of-the-art LLMs demonstrate low accuracy and calibration on HLE, highlighting a marked gap between current LLM capabilities and the expert human frontier on closed-ended academic questions. To inform research and policymaking upon a clear understanding of model capabilities, we publicly release HLE at https://lastexam.ai .

Deep learning models that predict functional genomic measurements from DNA sequences are powerful tools for deciphering the genetic regulatory code. Existing methods involve a trade-off between input sequence length and prediction resolution, thereby limiting their modality scope and performance1-5. We present AlphaGenome, a unified DNA sequence model, which takes as input 1 Mb of DNA sequence and predicts thousands of functional genomic tracks up to single-base-pair resolution across diverse modalities. The modalities include gene expression, transcription initiation, chromatin accessibility, histone modifications, transcription factor binding, chromatin contact maps, splice site usage and splice junction coordinates and strength. Trained on human and mouse genomes, AlphaGenome matches or exceeds the strongest available external models in 25 of 26 evaluations of variant effect prediction. The ability of AlphaGenome to simultaneously score variant effects across all modalities accurately recapitulates the mechanisms of clinically relevant variants near the TAL1 oncogene6. To facilitate broader use, we provide tools for making genome track and variant effect predictions from sequence.

Galaxy clusters are the most massive gravitationally bound structures in the universe and serve as tracers of the assembly of large-scale structure1. Studying their progenitors, protoclusters, sheds light on the earliest stages of cluster formation. However, detecting protoclusters is demanding: their member galaxies are loosely bound and the emerging hot intracluster medium (ICM) may only be in the initial stages of virialization2-4. Recent James Webb Space Telescope (JWST) observations located several protocluster candidates by identifying overdensities of z ≳ 5 galaxies5-9. However, none of these candidates was detected by X-ray observations, which offer a powerful way to unveil the hot ICM. Here we report the combined Chandra and JWST detection of a protocluster, JADES-ID1, at z ≈ 5.68, merely one billion years after the Big Bang. We measure a bolometric X-ray luminosity of Lbol=(1.5-0.6+0.5)×1044ergs-1 and infer a total gravitating mass of M500=(1.8-0.7+0.6)×1013M⊙ , making this system a progenitor of today's most massive galaxy clusters. The detection of extended, shock-heated gas indicates that substantial ICM heating can occur in massive halos as early as z ≈ 5.7. Also, given the limited survey volume, the discovery of such a massive cluster is statistically unlikely10, implying that the formation of the large-scale structure must have occurred more rapidly in some regions of the early universe than standard cosmological models predict.

The quantum superposition principle is a fundamental concept of physics1 and the basis of numerous quantum technologies2,3. Yet, it is still often regarded counterintuitive because we do not observe its key features on the macroscopic scales of our daily lives. It is, therefore, interesting to ask how quantum properties persist or change as we increase the size and complexity of objects4. A model test for this question can be realized by matter-wave interferometry, in which the motion of individual massive particles becomes delocalized and needs to be described by a wave function that spans regions far larger than the particle itself5. Over the years, this has been explored with a series of objects of increasing mass and complexity6-9 and a growing community aims at pushing this to ever larger limits. Here we present an experimental platform that extends matter-wave interference to large metal clusters, a qualitatively new material class for quantum experiments. We specifically demonstrate quantum interference of sodium nanoparticles, which can each contain more than 7,000 atoms at masses greater than 170,000 Da. They propagate in a Schrödinger cat state with a macroscopicity10 of μ = 15.5, surpassing previous experiments5,9,11 by an order of magnitude.

Plasticity-the ability of cells to undergo phenotypic transitions-drives cancer progression and therapy resistance1-3. Recent studies have suggested that plasticity in solid tumours is concentrated in a minority subset of cancer cells4-6, yet functional studies examining this high-plasticity cell state (HPCS) in situ are lacking. Here we develop mouse models enabling the detection, longitudinal lineage tracing and ablation of the HPCS in autochthonous lung tumours in vivo. Lineage tracing reveals that the HPCS cells possess a high capacity for cell state transitions, giving rise to both early neoplastic (differentiated) and progressed lung cancer cell states in situ. Longitudinal lineage tracing using secreted luciferases reveals that HPCS-derived cells have a high capacity for growth compared with bulk cancer cells or another cancer cell state with features of differentiated lung epithelium. Ablation of HPCS cells in early neoplasias abrogates benign-to-malignant transition, whereas ablation in established tumours by suicide gene or chimeric antigen receptor (CAR) T cells robustly reduces tumour burden. We further demonstrate that the HPCS gives rise to therapy-resistant cell states, whereas HPCS ablation suppresses resistance to chemotherapy and oncoprotein-targeted therapy. Notably, an HPCS-like state is ubiquitous in regenerating epithelia and in carcinomas of multiple other tissues, revealing a convergence of plasticity programs. Our work establishes the HPCS as a critical hub enabling reciprocal transitions between cancer cell states. Targeting the HPCS in lung cancer and in other carcinomas may suppress cancer progression and eradicate treatment resistance.

Sub-Neptunes and super-Earths, the most abundant types of planet in the galaxy, are unlike anything in the Solar System, with radii between those of Earth and Neptune1,2. Fundamental questions remain regarding their structure and origin. Although super-Earths have a rocky composition3, sub-Neptunes form a distinct population at larger radii and are thought to consist of a rocky core overlain by a hydrogen-rich envelope4,5. At the extreme conditions of the core-envelope interface (exceeding several gigapascals and several thousand kelvin4,6), reaction between core and envelope seems possible, but the nature and extent of these reactions are unknown. Here we use first-principles molecular dynamics driven by density functional theory to show that silicate and hydrogen are completely miscible over a wide range of plausible core-envelope pressure-temperature conditions. We find the origin of miscibility in extensive chemical reaction between hydrogen and silicate, producing silane, SiO and water species, which may be observable with ongoing or future missions. Core-envelope miscibility profoundly affects the evolution of sub-Neptunes and super-Earths, by dissolving a large fraction of the hydrogen of the planet in the core and driving exchange of hydrogen between core and envelope as the planet evolves.