Curr Opin Biotechnol. 2024 Jul 17;88:103172. doi: 10.1016/j.copbio.2024.103172. Online ahead of print.

ABSTRACT

Microbes orchestrate nearly all major biogeochemical processes. The ability to program their influence on plant growth and development is attractive for sustainable agriculture. However, the complexity of microbial ecosystems and our limited understanding of the mechanisms by which plants and microbes interact with each other and the environment make it challenging to use microbiomes to influence plant growth. Novel technologies at the intersection of microbial ecology, systems biology, and bioengineering provide new tools to probe the role of plant microbiomes across environments. Here, we summarize recent studies on plant and microbe responses to abiotic stresses, showcasing key molecules and micro-organisms that are important for plant health. We highlight opportunities to use synthetic microbial communities to understand the complexity of plant-microbial interactions and discuss future avenues of programming ecology to improve plant and ecosystem health.

PMID:39029405 | DOI:10.1016/j.copbio.2024.103172

G3 (Bethesda). 2024 Jul 19:jkae158. doi: 10.1093/g3journal/jkae158. Online ahead of print.

ABSTRACT

To remodel their hosts and escape immune defenses, many pathogens rely on large arsenals of proteins (effectors) that are delivered to the host cell using dedicated translocation machinery. Effectors hold significant insight into the biology of both the pathogens that encode them and the host pathways that they manipulate. One of the most powerful systems biology tools for studying effectors is the model organism, Saccharomyces cerevisiae. For many pathogens, the heterologous expression of effectors in yeast is growth inhibitory at a frequency much higher than housekeeping genes, an observation ascribed to targeting conserved eukaryotic proteins. Abrogation of yeast growth inhibition has been used to identify bacterial suppressors of effector activity, host targets, and functional residues and domains within effector proteins. We present here a yeast-based method for enriching for informative, in-frame, missense mutations in a pool of random effector mutants. We benchmark this approach against three effectors from Legionella pneumophila, an intracellular bacterial pathogen that injects a staggering >330 effectors into the host cell. For each protein, we show how in silico protein modeling (AlphaFold2) and missense-directed mutagenesis can be combined to reveal important structural features within effectors. We identify known active site residues within the metalloprotease RavK, the putative active site in SdbB, and previously unidentified functional motifs within the C-terminal domain of SdbA. We show that this domain has structural similarity with glycosyltransferases and exhibits in vitro activity consistent with this predicted function.

PMID:39028840 | DOI:10.1093/g3journal/jkae158

Sci Adv. 2024 Jul 19;10(29):eadp6039. doi: 10.1126/sciadv.adp6039. Epub 2024 Jul 19.

ABSTRACT

The adult hippocampus generates new granule cells (aGCs) with functional capabilities that convey unique forms of plasticity to the preexisting circuits. While early differentiation of adult radial glia-like cells (RGLs) has been studied extensively, the molecular mechanisms guiding the maturation of postmitotic neurons remain unknown. Here, we used a precise birthdating strategy to study aGC differentiation using single-nuclei RNA sequencing. Transcriptional profiling revealed a continuous trajectory from RGLs to mature aGCs, with multiple immature stages bearing increasing levels of effector genes supporting growth, excitability, and synaptogenesis. Analysis of differential gene expression, pseudo-time trajectory, and transcription factors (TFs) revealed critical transitions defining four cellular states: quiescent RGLs, proliferative progenitors, immature aGCs, and mature aGCs. Becoming mature aGCs involved a transcriptional switch that shuts down pathways promoting cell growth, such SoxC TFs, to activate programs that likely control neuronal homeostasis. aGCs overexpressing Sox4 or Sox11 remained immature. Our results unveil precise molecular mechanisms driving adult RGLs through the pathway of neuronal differentiation.

PMID:39028813 | DOI:10.1126/sciadv.adp6039

Sci Adv. 2024 Jul 19;10(29):eadl6366. doi: 10.1126/sciadv.adl6366. Epub 2024 Jul 19.

ABSTRACT

Physical processes ultimately shape tissue during development. Two emerging proposals are that cells migrate toward stiffer tissue (durotaxis) and that the extent of cell rearrangements reflects tissue phase, but it is unclear whether and how these concepts are related. Here, we identify fibronectin-dependent tissue stiffness as a control variable that underlies and unifies these phenomena in vivo. In murine limb bud mesoderm, cells are either caged, move directionally, or intercalate as a function of their location along a stiffness gradient. A modified Landau phase equation that incorporates tissue stiffness accurately predicts cell diffusivity upon loss or gain of fibronectin. Fibronectin is regulated by WNT5A-YAP feedback that controls cell movements, tissue shape, and skeletal pattern. The results identify a key determinant of phase transition and show how fibronectin-dependent directional cell movement emerges in a mixed-phase environment in vivo.

PMID:39028807 | DOI:10.1126/sciadv.adl6366

STAR Protoc. 2024 Jul 18;5(3):103205. doi: 10.1016/j.xpro.2024.103205. Online ahead of print.

ABSTRACT

The electroencephalogram (EEG) is crucial for real-time brain physiology research in epilepsy. However, maternal care reliance limits its use in immature rodents. Our "pup-in-cup" setup overcomes this, enabling continuous, uninterrupted video-EEG/electromyogram (EMG) recordings in neonatal rats. This protocol details the steps for video-EEG/EMG system setup, EEG headmount implantation, and recording continuous video-EEG/EMG traces from postnatal days 4-12. For complete details on the use and execution of this protocol, please refer to Choudhary et al.1.

PMID:39028620 | DOI:10.1016/j.xpro.2024.103205

Neuro Oncol. 2024 Jul 19:noae135. doi: 10.1093/neuonc/noae135. Online ahead of print.

ABSTRACT

BACKGROUND: Glioblastoma is a highly aggressive brain cancer that is resistant to conventional immunotherapy strategies. Botensilimab, an Fc-enhanced anti-CTLA-4 antibody (FcE-aCTLA-4), has shown durable activity in "cold" and immunotherapy-refractory cancers.

METHOD: We evaluated the efficacy and immune microenvironment phenotype of a mouse analogue of FcE-aCTLA-4 in treatment-refractory preclinical models of glioblastoma, both as a monotherapy and in combination with doxorubicin delivered via low-intensity pulsed ultrasound and microbubbles (LIPU/MB). Additionally, we studied 4 glioblastoma patients treated with doxorubicin, anti-PD-1 with concomitant LIPU/MB to investigate the novel effect of doxorubicin modulating FcγR expressions in tumor associated macrophages/microglia (TAMs).

RESULTS: FcE-aCTLA-4 demonstrated high-affinity binding to FcγRIV, the mouse ortholog of human FcγRIIIA, which was highly expressed in TAMs in human glioblastoma, most robustly at diagnosis. Notably, FcE-aCTLA-4 mediated selective depletion of intra-tumoral regulatory T cells (Tregs) via TAM-mediated phagocytosis, while sparing peripheral Tregs. Doxorubicin, a chemotherapeutic drug with immunomodulatory functions, was found to upregulate FcγRIIIA on TAMs in glioblastoma patients who received doxorubicin and anti-PD-1 with concomitant LIPU/MB. In murine models of immunotherapy-resistant gliomas, a combinatorial regimen of FcE-aCTLA-4, anti-PD-1, and doxorubicin with LIPU/MB, achieved a 90% cure rate, that was associated robust infiltration of activated CD8+ T cells and establishment of immunological memory as evidenced by rejection upon tumor rechallenge.

CONCLUSION: Our findings demonstrate that FcE-aCTLA-4 promotes robust immunomodulatory and anti-tumor effects in murine gliomas and is significantly enhanced when combined with anti-PD-1, doxorubicin, and LIPU/MB. We are currently investigating this combinatory strategy in a clinical trial (clinicaltrials.gov NCT05864534).

PMID:39028616 | DOI:10.1093/neuonc/noae135

Elife. 2024 Jul 19;12:RP90964. doi: 10.7554/eLife.90964.

ABSTRACT

Normal aging leads to myelin alterations in the rhesus monkey dorsolateral prefrontal cortex (dlPFC), which are positively correlated with degree of cognitive impairment. It is hypothesized that remyelination with shorter and thinner myelin sheaths partially compensates for myelin degradation, but computational modeling has not yet explored these two phenomena together systematically. Here, we used a two-pronged modeling approach to determine how age-related myelin changes affect a core cognitive function: spatial working memory. First, we built a multicompartment pyramidal neuron model fit to monkey dlPFC empirical data, with an axon including myelinated segments having paranodes, juxtaparanodes, internodes, and tight junctions. This model was used to quantify conduction velocity (CV) changes and action potential (AP) failures after demyelination and subsequent remyelination. Next, we incorporated the single neuron results into a spiking neural network model of working memory. While complete remyelination nearly recovered axonal transmission and network function to unperturbed levels, our models predict that biologically plausible levels of myelin dystrophy, if uncompensated by other factors, can account for substantial working memory impairment with aging. The present computational study unites empirical data from ultrastructure up to behavior during normal aging, and has broader implications for many demyelinating conditions, such as multiple sclerosis or schizophrenia.

PMID:39028036 | DOI:10.7554/eLife.90964

bioRxiv [Preprint]. 2024 Jul 11:2024.07.11.603094. doi: 10.1101/2024.07.11.603094.

ABSTRACT

Gastrointestinal colonization by Clostridioides difficile is common in healthcare settings and ranges in clinical presentation from asymptomatic carriage to lethal C. difficile infection (CDI). We used a systems biology approach to investigate why patients colonized with C. difficile have a range of outcomes. Microbiota-humanization of germ-free mice with fecal samples from toxigenic C. difficile carriers revealed a spectrum of virulence among clade 1 lineages and identified commensal Blautia associated with markers of non-pathogenic colonization. Using gnotobiotic mice engrafted with defined human microbiota, we observed strain-specific CDI severity across clade 1 strains. Yet, mice engrafted with a higher diversity community were protected from severe disease across all strains without suppression of C. difficile colonization. These results indicate that when colonization resistance has been breached without overt infection, commensals can attenuate a diversity of virulent strains without inhibiting pathogen colonization, providing insight into determinants of stable C. difficile carriage.

PMID:39026847 | PMC:PMC11257545 | DOI:10.1101/2024.07.11.603094

bioRxiv [Preprint]. 2024 Jul 10:2024.07.07.602397. doi: 10.1101/2024.07.07.602397.

ABSTRACT

The Gillespie algorithm is commonly used to simulate and analyze complex chemical reaction networks. Here, we leverage recent breakthroughs in deep learning to develop a fully differentiable variant of the Gillespie algorithm. The differentiable Gillespie algorithm (DGA) approximates discontinuous operations in the exact Gillespie algorithm using smooth functions, allowing for the calculation of gradients using backpropagation. The DGA can be used to quickly and accurately learn kinetic parameters using gradient descent and design biochemical networks with desired properties. As an illustration, we apply the DGA to study stochastic models of gene promoters. We show that the DGA can be used to: (i) successfully learn kinetic parameters from experimental measurements of mRNA expression levels from two distinct E. coli promoters and (ii) design nonequilibrium promoter architectures with desired input-output relationships. These examples illustrate the utility of the DGA for analyzing stochastic chemical kinetics, including a wide variety of problems of interest to synthetic and systems biology.

PMID:39026759 | PMC:PMC11257475 | DOI:10.1101/2024.07.07.602397

bioRxiv [Preprint]. 2024 Jul 8:2024.07.04.602041. doi: 10.1101/2024.07.04.602041.

ABSTRACT

Cellular senescence is known to drive age-related pathology through the senescence-associated secretory phenotype (SASP). However, it also plays important physiological roles such as cancer suppression, embryogenesis and wound healing. Wound healing is a tightly regulated process which when disrupted results in conditions such as fibrosis and chronic wounds. Senescent cells appear during the proliferation phase of the healing process where the SASP is involved in maintaining tissue homeostasis after damage. Interestingly, SASP composition and functionality was recently found to be temporally regulated, with distinct SASP profiles involved: a fibrogenic, followed by a fibrolytic SASP, which could have important implications for the role of senescent cells in wound healing. Given the number of factors at play a full understanding requires addressing the multiple levels of complexity, pertaining to the various cell behaviours, individually followed by investigating the interactions and influence each of these elements have on each other and the system as a whole. Here, a systems biology approach was adopted whereby a multi-scale model of wound healing that includes the dynamics of senescent cell behaviour and corresponding SASP composition within the wound microenvironment was developed. The model was built using the software CompuCell3D, which is based on a Cellular Potts modelling framework. We used an existing body of data on healthy wound healing to calibrate the model and validation was done on known disease conditions. The model provides understanding of the spatiotemporal dynamics of different senescent cell phenotypes and the roles they play within the wound healing process. The model also shows how an overall disruption of tissue-level coordination due to age-related changes results in different disease states including fibrosis and chronic wounds. Further specific data to increase model confidence could be used to explore senolytic treatments in wound disorders.

PMID:39026713 | PMC:PMC11257496 | DOI:10.1101/2024.07.04.602041

Int J Infect Dis. 2024 Jul 16:107178. doi: 10.1016/j.ijid.2024.107178. Online ahead of print.

ABSTRACT

OBJECTIVES: Human babesiosis, an emerging potential fatal tick-borne disease caused by intraerythrocytic parasites of the Babesia genus. B. duncani is one of the Babesia species cause severe and life-threatening infections in humans. Detecting B. duncani infection is essential for accurate diagnosis and effective disease management. While molecular assays for detection in blood exist, there is still no reliable method to detect biomarkers of active infection.

METHODS: We developed the first B. duncani antigen capture assays, targeting two immunodominant antigens, BdV234 and BdV38. These assays were validated using established in vitro and in vivo B. duncani-infection levels, both before and after therapy.

RESULTS: The assay displayed no cross-reactivity with other species such as B. microti, B. divergens, Babesia MO1, or P. falciparum. It can detect as few as 115 infected erythrocytes/µl blood. Screening of 1,731 blood samples from various biorepositories, including samples previously identified as Lyme and/or B. microti positive, as well as new specimens from field mice, revealed no evidence of B. duncani infection and cross reactivity.

CONCLUSION: These assays have potential applications, such as point-of-care testing for early detection of B. duncani in patients, field tests for screening reservoir hosts, and high-throughput screening of blood samples collected for transfusion.

PMID:39025200 | DOI:10.1016/j.ijid.2024.107178

Int J Biol Macromol. 2024 Jul 17;276(Pt 1):133473. doi: 10.1016/j.ijbiomac.2024.133473. Online ahead of print.

NO ABSTRACT

PMID:39024924 | DOI:10.1016/j.ijbiomac.2024.133473

Genome Biol. 2024 Jul 18;25(1):189. doi: 10.1186/s13059-024-03334-3.

ABSTRACT

Single-cell RNA-sequencing enables testing for differential expression (DE) between conditions at a cell type level. While powerful, one of the limitations of such approaches is that the sensitivity of DE testing is dictated by the sensitivity of clustering, which is often suboptimal. To overcome this, we present miloDE-a cluster-free framework for DE testing (available as an open-source R package). We illustrate the performance of miloDE on both simulated and real data. Using miloDE, we identify a transient hemogenic endothelia-like state in mouse embryos lacking Tal1 and detect distinct programs during macrophage activation in idiopathic pulmonary fibrosis.

PMID:39026254 | DOI:10.1186/s13059-024-03334-3

Genome Biol. 2024 Jul 18;25(1):190. doi: 10.1186/s13059-024-03333-4.

ABSTRACT

BACKGROUND: Interactions among cis-regulatory elements (CREs) play a crucial role in gene regulation. Various approaches have been developed to map these interactions genome-wide, including those relying on interindividual epigenomic variation to identify groups of covariable regulatory elements, referred to as chromatin modules (CMs). While CM mapping allows to investigate the relationship between chromatin modularity and gene expression, the computational principles used for CM identification vary in their application and outcomes.

RESULTS: We comprehensively evaluate and streamline existing CM mapping tools and present guidelines for optimal utilization of epigenome data from a diverse population of individuals to assess regulatory coordination across the human genome. We showcase the effectiveness of our recommended practices by analyzing distinct cell types and demonstrate cell type specificity of CRE interactions in CMs and their relevance for gene expression. Integration of genotype information revealed that many non-coding disease-associated variants affect the activity of CMs in a cell type-specific manner by affecting the binding of cell type-specific transcription factors. We provide example cases that illustrate in detail how CMs can be used to deconstruct GWAS loci, assess variable expression of cell surface receptors in immune cells, and reveal how genetic variation can impact the expression of prognostic markers in chronic lymphocytic leukemia.

CONCLUSIONS: Our study presents an optimal strategy for CM mapping and reveals how CMs capture the coordination of CREs and its impact on gene expression. Non-coding genetic variants can disrupt this coordination, and we highlight how this may lead to disease predisposition in a cell type-specific manner.

PMID:39026229 | DOI:10.1186/s13059-024-03333-4

Nat Protoc. 2024 Jul 18. doi: 10.1038/s41596-024-01017-8. Online ahead of print.

ABSTRACT

Carbohydrates comprise the largest fraction of most diets and exert a profound impact on health. Components such as simple sugars and starch supply energy, while indigestible components, deemed dietary fiber, reach the colon to provide food for the tens of trillions of microbes that make up the gut microbiota. The interactions between dietary carbohydrates, our gastrointestinal tracts, the gut microbiome and host health are dictated by their structures. However, current methods for analysis of food glycans lack the sensitivity, specificity and throughput needed to quantify and elucidate these myriad structures. This protocol describes a multi-glycomic approach to food carbohydrate analysis in which the analyte might be any food item or biological material such as fecal and cecal samples. The carbohydrates are extracted by ethanol precipitation, and the resulting samples are subjected to rapid-throughput liquid chromatography (LC)-tandem mass spectrometry (LC-MS/MS) methods. Quantitative analyses of monosaccharides, glycosidic linkages, polysaccharides and alcohol-soluble carbohydrates are performed in 96-well plates at the milligram scale to reduce the biomass of sample required and enhance throughput. Detailed stepwise processes for sample preparation, LC-MS/MS and data analysis are provided. We illustrate the application of the protocol to a diverse set of foods as well as different apple cultivars and various fermented foods. Furthermore, we show the utility of these methods in elucidating glycan-microbe interactions in germ-free and colonized mice. These methods provide a framework for elucidating relationships between dietary fiber, the gut microbiome and human physiology. These structures will further guide nutritional and clinical feeding studies that enhance our understanding of the role of diet in nutrition and health.

PMID:39026121 | DOI:10.1038/s41596-024-01017-8

Br J Cancer. 2024 Jul 19. doi: 10.1038/s41416-024-02763-y. Online ahead of print.

ABSTRACT

BACKGROUND: Studies have shown that hepatitis B virus (HBV)-associated B-cell non-Hodgkin lymphoma (NHL) constitutes a unique subgroup with distinct clinical features. It still leaves open the question of whether the integration of HBV DNA into the B-cell genome is a causal mechanism in the development of lymphoma.

METHODS: Using the hybridisation capture-based next generation sequencing and RNA sequencing, we characterised the HBV integration pattern in 45 HBV-associated B-cell NHL tumour tissues.

RESULTS: A total of 354 HBV integration sites were identified in 13 (28.9%) samples, indicating the relatively low integration frequency in B-cell NHLs. High plasma HBV DNA loads were not associated with the existence of HBV integration. The insertion sites distributed randomly across all the lymphoma genome without any preferential hotspot neither at the chromosomal level nor at the genetic level. Intriguingly, most HBV integrations were nonclonal in B-cell NHLs, implying that they did not confer a survival advantage. Analysis of the paired diagnosis-relapse samples showed the unstable status of HBV integrations during disease progression. Furthermore, transcriptomic analysis revealed the limited biological impact of HBV integration.

CONCLUSION: Our study provides an unbiased HBV integration map in B-cell NHLs, revealing the insignificant role of HBV DNA integration in B-cell lymphomagenesis.

PMID:39026081 | DOI:10.1038/s41416-024-02763-y

Commun Biol. 2024 Jul 18;7(1):879. doi: 10.1038/s42003-024-06563-1.

ABSTRACT

In clinical situations, peripheral blood accessible CD3+CD4+CXCR5+ T-follicular helper (TFH) cells may have to serve as a surrogate indicator for dysregulated germinal center responses in tissues. To determine the heterogeneity of TFH cells in peripheral blood versus tonsils, CD3+CD4+CD45RA-CXCR5+ cells of both origins were sorted. Transcriptomes, TCR repertoires and cell-surface protein expression were analysed by single-cell RNA sequencing, flow cytometry and immunohistochemistry. Reassuringly, all blood-circulating CD3+CD4+CXCR5+ T-cell subpopulations also appear in tonsils, there with some supplementary TFH characteristics, while peripheral blood-derived TFH cells display markers of proliferation and migration. Three further subsets of TFH cells, however, with bona fide T-follicular gene expression patterns, are exclusively found in tonsils. One additional, distinct and oligoclonal CD4+CXCR5+ subpopulation presents pronounced cytotoxic properties. Those 'killer TFH (TFK) cells' can be discovered in peripheral blood as well as among tonsillar cells but are located predominantly outside of germinal centers. They appear terminally differentiated and can be distinguished from all other TFH subsets by expression of NKG7 (TIA-1), granzymes, perforin, CCL5, CCR5, EOMES, CRTAM and CX3CR1. All in all, this study provides data for detailed CD4+CXCR5+ T-cell assessment of clinically available blood samples and extrapolation possibilities to their tonsil counterparts.

PMID:39025930 | DOI:10.1038/s42003-024-06563-1

Nat Commun. 2024 Jul 18;15(1):6054. doi: 10.1038/s41467-024-50168-5.

ABSTRACT

The homeostatic regulation of sleep is characterized by rebound sleep after prolonged wakefulness, but the molecular and cellular mechanisms underlying this regulation are still unknown. In this study, we show that Ca2+/calmodulin-dependent protein kinase II (CaMKII)-dependent activity control of parvalbumin (PV)-expressing cortical neurons is involved in homeostatic regulation of sleep in male mice. Prolonged wakefulness enhances cortical PV-neuron activity. Chemogenetic suppression or activation of cortical PV neurons inhibits or induces rebound sleep, implying that rebound sleep is dependent on increased activity of cortical PV neurons. Furthermore, we discovered that CaMKII kinase activity boosts the activity of cortical PV neurons, and that kinase activity is important for homeostatic sleep rebound. Here, we propose that CaMKII-dependent PV-neuron activity represents negative feedback inhibition of cortical neural excitability, which serves as the distributive cortical circuits for sleep homeostatic regulation.

PMID:39025867 | DOI:10.1038/s41467-024-50168-5

Nat Commun. 2024 Jul 18;15(1):6066. doi: 10.1038/s41467-024-50365-2.

ABSTRACT

DNA N6-adenine methylation (6mA) has recently gained importance as an epigenetic modification in eukaryotes. Its function in lineages with high levels, such as early-diverging fungi (EDF), is of particular interest. Here, we investigated the biological significance and evolutionary implications of 6mA in EDF, which exhibit divergent evolutionary patterns in 6mA usage. The analysis of two Mucorales species displaying extreme 6mA usage reveals that species with high 6mA levels show symmetric methylation enriched in highly expressed genes. In contrast, species with low 6mA levels show mostly asymmetric 6mA. Interestingly, transcriptomic regulation throughout development and in response to environmental cues is associated with changes in the 6mA landscape. Furthermore, we identify an EDF-specific methyltransferase, likely originated from endosymbiotic bacteria, as responsible for asymmetric methylation, while an MTA-70 methylation complex performs symmetric methylation. The distinct phenotypes observed in the corresponding mutants reinforced the critical role of both types of 6mA in EDF.

PMID:39025853 | DOI:10.1038/s41467-024-50365-2

Nat Commun. 2024 Jul 18;15(1):6046. doi: 10.1038/s41467-024-50170-x.

ABSTRACT

Energy status and nutrients regulate photosynthetic protein expression. The unicellular green alga Chromochloris zofingiensis switches off photosynthesis in the presence of exogenous glucose (+Glc) in a process that depends on hexokinase (HXK1). Here, we show that this response requires that cells lack sufficient iron (-Fe). Cells grown in -Fe+Glc accumulate triacylglycerol (TAG) while losing photosynthesis and thylakoid membranes. However, cells with an iron supplement (+Fe+Glc) maintain photosynthesis and thylakoids while still accumulating TAG. Proteomic analysis shows that known photosynthetic proteins are most depleted in heterotrophy, alongside hundreds of uncharacterized, conserved proteins. Photosynthesis repression is associated with enzyme and transporter regulation that redirects iron resources to (a) respiratory instead of photosynthetic complexes and (b) a ferredoxin-dependent desaturase pathway supporting TAG accumulation rather than thylakoid lipid synthesis. Combining insights from diverse organisms from green algae to vascular plants, we show how iron and trophic constraints on metabolism aid gene discovery for photosynthesis and biofuel production.

PMID:39025848 | DOI:10.1038/s41467-024-50170-x