Browse Articles

Stabilized real-time Brillouin microscopy reveals fractal organization of protein condensates in living cells

Abstract Mechanical alterations of protein condensates are increasingly recognized in the etiology of several neurodegenerative diseases, yet their characterization remains technically challenging. Although Brillouin microscopy could offer a promising solution, its use is hindered by instrumental instabilities demanding frequent adjustments and manual calibrations with reference materials. Here, we present an enhanced Brillouin Microscope that incorporates an electro-optic modulator, serving simultaneously as frequency reference, spectrometer calibrator, and temporal stabilizer. This integration enables robust, real-time spectral stability over multiple days in a fully automated workflow. Using this system, we quantify Brillouin shifts of several protein condensates in living cells and validate our findings with FRAP. The correlation between techniques reveals a fractal internal architecture of the condensates, providing important insights into their physical nature while probing the mechanical behavior of entire compartments containing multiple protein species. Our method offers a unique framework for distinguishing physiological from pathological condensates, paving the way for long-term, user-independent, high-precision mechanical measurements in living cells.

Acute peritonitis-induced adipose CD127+ ILC1s express PD-L1 and ameliorate inflammation in mice

scLong: a billion-parameter foundation model for capturing long-range gene context in single-cell transcriptomics

Abstract Single-cell RNA sequencing (scRNA-seq) has revolutionized the study of cellular heterogeneity by providing gene expression data at single-cell resolution, uncovering insights into rare cell populations, cell-cell interactions, and gene regulation. Foundation models pretrained on large-scale scRNA-seq datasets have shown great promise in analyzing such data, but existing approaches are often limited to modeling a small subset of highly expressed genes and lack the integration of external gene-specific knowledge. To address these limitations, we present scLong, a billion-parameter foundation model pretrained on 48 million cells. scLong performs self-attention across the entire set of 28,000 genes in the human genome. This enables the model to capture long-range dependencies between all genes, including lowly expressed ones (containing unexpressed genes with zero expressions), which often play critical roles in cellular processes but are typically excluded by existing foundation models. Additionally, scLong integrates gene knowledge from the Gene Ontology using a graph convolutional network, enriching its contextual understanding of gene functions and relationships. In extensive evaluations, scLong surpasses both state-of-the-art scRNA-seq foundation models and task-specific models across diverse tasks, including predicting transcriptional responses to genetic and chemical perturbations, forecasting cancer drug responses, and inferring gene regulatory networks.

An mRNA-delivered consensus allergen induces a neutralizing IgG response against food and pollen allergens

Sclerostin deficiency sensitizes white adipocytes to thermogenic signals that induce beiging in mice

Strong long-term variability in active galactic nuclei affects virial black hole mass measurements

Utilizing the predictable binding kinetics of DNA-PAINT to denoise super-resolution images

Abstract DNA-Point Accumulation for Imaging in Nanoscale Topography (DNA-PAINT) is a versatile super-resolution technique that relies on the predictable binding kinetics between fluorescent imager strands and docking strands attached to target proteins. This makes DNA-PAINT particularly suitable for multiplexing and quantitative applications, but its performance is often limited by spurious signals from non-specific binding of imager strands. Here we describe a method to remove these non-specific binding events using a statistical test to distinguish between DNA-specific and non-specific interactions. To demonstrate the method, we imaged mosaic epithelial tissues in Drosophila melanogaster egg chambers and showed that > 90% of non-specific and otherwise indistinguishable signal in the super-resolved images can be removed. This denoising improves the quality of DNA-PAINT super-resolved images and is essential for accurate measurements of spatial relationships and protein quantification.

Closed-form feedback-free learning with forward projection

Abstract State-of-the-art backpropagation-free learning methods employ local error feedback to direct iterative optimisation via gradient descent. Here, we examine the more restrictive setting where retrograde communication from neuronal outputs is unavailable for pre-synaptic weight optimisation. We propose Forward Projection (FP), a randomised closed-form training method requiring only a single forward pass over the dataset without retrograde communication. FP generates target values for pre-activation membrane potentials through randomised nonlinear projections of pre-synaptic inputs and labels. Local loss functions are optimised using closed-form regression without feedback from downstream layers. A key advantage is interpretability: membrane potentials in FP-trained networks encode information interpretable layer-wise as label predictions. Across several biomedical datasets, FP achieves generalisation comparable to gradient descent-based local learning methods while requiring only a single forward propagation step, yielding significant training speedup. In few-shot learning tasks, FP produces more generalisable models than backpropagation-optimised alternatives, with local interpretation functions successfully identifying clinically salient diagnostic features.

Almost optimal geometrically local quantum LDPC codes in any dimension

Abstract Geometrically local quantum codes, comprised of qubits and checks embedded in $${{\mathbb{R}}}^{D}$$ R D with local check operators, have been a subject of significant interest. A key challenge is identifying the optimal code construction that maximizes both code dimension and distance under the geometric constraints. In this work, we introduce a construction that can transform any good quantum LDPC code into an almost optimal geometrically local quantum code. Our approach hinges on a novel yet simple procedure that extracts a two-dimensional structure from an arbitrary three-term chain complex, building a connection between geometric operations and code constructions. We expect that this procedure will find broader applications in areas such as weight reduction and the geometric realization of chain complexes.

Observation of a superfluid-to-insulator transition of bilayer excitons

Author Correction: Human cerebrospinal fluid net flow enhanced by respiration during the awake state

A stabilized MERS-CoV spike ferritin nanoparticle vaccine elicits robust and protective neutralizing antibody responses

Abstract Middle East respiratory syndrome coronavirus (MERS-CoV) was identified as a human pathogen in 2012 and causes ongoing sporadic infections and outbreak clusters. Despite case fatality rates (CFRs) of over 30% and considerable pandemic potential, a safe and efficacious vaccine has not been developed. Here we report the design, characterization, and preclinical evaluation of MERS-CoV antigens. Our lead candidate comprises a stabilized spike displayed on a self-assembling ferritin nanoparticle that can be produced from a high-expressing, stable cell pool. This vaccine elicits robust MERS-CoV pseudovirus and authentic virus neutralizing antibody titers in BALB/c mice. Immunization of male non-human primates (NHPs) with one dose of Alhydrogel-adjuvanted vaccine elicited a > 10 3 geometric mean titer of pseudovirus neutralizing antibodies that was boosted with a second dose. Sera from these NHPs exhibited cross-reactivity against spike-pseudotyped lentiviruses from MERS-CoV clades A, B, and C as well as a distant pangolin merbecovirus. In human DPP4 transgenic mice, immunization provided dose-dependent protection against MERS-CoV lethal challenge, and in an established alpaca challenge model using female alpacas, immunization fully protected against MERS-CoV infection. This MERS-CoV nanoparticle vaccine is a promising candidate for clinical advancement to protect at-risk individuals and for future use in a potential outbreak setting.

FLASH-MM: fast and scalable single-cell differential expression analysis using linear mixed-effects models

Abstract Single-cell RNA sequencing (scRNA-seq) enables detailed comparisons of gene expression across cells and conditions. Single-cell differential expression analysis faces challenges like sample correlation, individual variation, and scalability. We develop a fast and scalable linear mixed-effects model (LMM) estimation algorithm, FLASH-MM, to address these issues. We reformulate aspects of the linear mixed model estimation procedure to make it faster, by reducing computational complexity and memory usage. Simulation studies with scRNA-seq data show that FLASH-MM is accurate, computationally efficient, effectively controls false positive rates, and maintains high statistical power in differential expression analysis. Tests on tuberculosis immune and kidney single cell data demonstrate FLASH-MM’s utility in accelerating single-cell differential expression analysis across diverse biological contexts.

Dated gene duplications elucidate the evolutionary assembly of eukaryotes

Abstract The origin of eukaryotes was a formative but poorly understood event in the history of life. Current hypotheses of eukaryogenesis differ principally in the timing of mitochondrial endosymbiosis relative to the acquisition of other eukaryote novelties 1 . Discriminating among these hypotheses has been challenging, because there are no living lineages representative of intermediate steps within eukaryogenesis. However, many eukaryotic cell functions are contingent on genes that emerged from duplication events during eukaryogenesis 2,3 . Consequently, the timescale of these duplications can provide insights into the sequence of steps in the evolutionary assembly of the eukaryotic cell. Here we show, using a relaxed molecular clock 4 , that the process of eukaryogenesis spanned the Mesoarchaean to late Palaeoproterozoic eras. Within these constraints, we dated the timing of these gene duplications, revealing that the eukaryotic host cell already had complex cellular features before mitochondrial endosymbiosis, including an elaborated cytoskeleton, membrane trafficking, endomembrane, phagocytotic machinery and a nucleus, all between 3.0 and 2.25 billion years ago, after which mitochondrial endosymbiosis occurred. Our results enable us to reject mitochondrion-early scenarios of eukaryogenesis 5 , instead supporting a complexified-archaean, late-mitochondrion sequence for the assembly of eukaryote characteristics. Our inference of a complex archaeal host cell is compatible with hypotheses on the adaptive benefits of syntrophy 6,7 in oceans that would have remained largely anoxic for more than a billion years 8,9 .

A magnetically soft yet mechanically strong and ductile Ta free CoFeNi high entropy alloy with Al and Ti additions

Early diagnosis of pancreatic ductal adenocarcinoma by signal-enhanced lateral flow immunoassay: SELFI

Automated diagnostic of cervical spondylosis on multimodal medical images with a multi-task deep learning model

NEO-STIM advances personalized neoantigen-specific adoptive T cell therapy

Abstract Neoantigen-based adoptive T cell therapies (ACTs) represent a promising avenue in cancer immunotherapy due to their exquisite tumor specificity. The first cell-based immunotherapy for a solid tumor, comprising tumor-infiltrating lymphocytes, recently received FDA approval. Building on this, we designed a distinct ACT approach, where T cell responses against personalized neoantigens are systematically generated from autologous peripheral blood. Here we report the establishment of NEO-STIM, an ex vivo induction process to prime and expand pre-existing memory and de novo CD8 + and CD4 + T cell responses, thereby highlighting critical parameters for generating potent neoantigen-specific T cell responses. The drug products comprise mutant-reactive, polyfunctional, and cytotoxic CD8 + and CD4 + T cells, able to recognize autologous tumor material. Following infusion, T cell responses are detected in tumor and blood of a patient, and display activated/exhausted and cytotoxic phenotypes. A first-in-human clinical trial (NCT04625205) recently further validated proof-of-concept, supporting continued development of this ACT approach.

Molecular insights into the capsular polysaccharide transporter Wza-Wzc complex

Abstract Capsular polysaccharides (CPS) are key virulence determinants, constituting the protective capsule that surrounds bacterial pathogens. Here, we present the complete cryo-EM structure of Gram-negative bacterial CPS secretion machinery formed by the E. coli K12 Wza-Wzc complex. The structure reveals an elongated, continuous channel spanning the entire envelope that facilitates CPS secretion. Multiple structural snapshots of the ADP-bound Wza-Wzc complex capture intermediate conformations of the double membrane assembly, highlighting its remarkable intrinsic dynamics. In-depth analysis of the isolated Wza translocon and Wzc co-polymerase, reveals mechanistic details of both complex formation and CPS transport. We further uncover the jellyroll domain of Wzc as a CPS-binding module, likely guiding CPS repeat units into a proposed Wzc-Wzy polymerization platform. Collectively, this work provides structural and functional insights into CPS synthesis and transport, advancing our understanding of bacterial capsule formation and virulence mechanisms.

Electrolyte chemistry of adaptive hydrogen bonded domains for high voltage lithium metal batteries