Cell systems最新文献

Society needs alternatives to painful, expensive, and cumbersome injections for diseases like diabetes. Franko et al. developed cells that sense melatonin to deliver glucagon-like peptide-1 (GLP-1) therapy during sleep. This circadian-synchronized approach restored normal blood sugar in diabetic mice, advancing the field of smart cells for patient-centered circadian medicine.

Single-cell RNA sequencing provides detailed insights into cellular heterogeneity and responses to external stimuli. However, distinguishing inherent cellular variation from extrinsic effects induced by external stimuli remains a major analytical challenge. Here, we present scCausalVI, a causality-aware generative model designed to disentangle these sources of variation. scCausalVI decouples intrinsic cellular states from treatment effects through a deep structural causal network that explicitly models the causal mechanisms governing cell-state-specific responses to external perturbations while accounting for technical variations. Our model integrates structural causal modeling with cross-condition in silico prediction to infer gene expression profiles under hypothetical scenarios. Comprehensive benchmarking demonstrates that scCausalVI outperforms existing methods in disentangling causal relationships, quantifying treatment effects, generalizing to unseen cell types, and separating biological signals from technical variation in multi-source data integration. Applied to COVID-19 datasets, scCausalVI effectively identifies treatment-responsive populations and delineates molecular signatures of cellular susceptibility.

High-content image-based phenotypic profiling combines automated microscopy and analysis to identify phenotypic alterations in cell morphology and provide insight into the cell's physiological state. Classical representations of the phenotypic profile cannot capture the full underlying complexity in cell organization, while recent weakly supervised machine-learning-based representation-learning methods are hard to biologically interpret. We used the abundance of control wells to learn the in-distribution of control experiments and used it to formulate a self-supervised reconstruction anomaly-based representation that encodes the intricate morphological inter-feature dependencies while preserving the representation interpretability. The performance of our anomaly-based representations was evaluated for downstream tasks with respect to two classical representations across four public Cell Painting datasets. Anomaly-based representations improved reproducibility and mechanism of action classification and complemented classical representations. Unsupervised explainability of autoencoder-based anomalies identified specific inter-feature dependencies causing anomalies. The general concept of anomaly-based representations can be adapted to other applications in cell biology. A record of this paper's transparent peer review process is included in the supplemental information.

To begin deciphering the hierarchical structure of the cell, we need to integrate multiple types of data of different scales on subcellular organization. To this end, we developed MIRAGE, a multi-modal generative model for integrating protein sequence, protein-protein interaction, and protein localization data. Our adversarial approach successfully learns a joint embedding space that captures the complex relationships among these diverse modalities and allows us to generate missing modalities. We evaluate our model's performance against existing methods, obtaining superior performance in protein function prediction and protein complex detection. We apply MIRAGE to construct a hierarchical map of subcellular organization in HEK293T cells, recovering known protein assemblies across multiple scales.

AI-enabled functional annotation reveals hidden enzyme diversity and distribution in fermented food microbiomes, shedding light on their ecological roles and biotechnological potential.

Microbial bioactivity is essential for the flavor, appearance, quality, and safety of fermented foods. However, the diversity and distribution of enzymatic resources in fermentation remain poorly understood. This study explored 10,202 metagenome-assembled genomes from global fermented foods using machine learning, identifying over 5 million enzyme sequences grouped into 98,693 homologous clusters, representing over 3,000 enzyme types. Functional analysis revealed that 84.4% of these clusters were unannotated in current databases, with high novelty in terpenoid and polyketide metabolism enzymes. Peptide hydrolases exhibited broad environmental adaptability based on predicted optimal temperatures and pH, and niche breadth calculations indicated 31.3% of enzyme clusters displayed food-type specificity. Additionally, we developed a machine learning model to classify fermented food sources by enzyme clusters, highlighting key enzymes differentiating habitats. Our findings emphasize the untapped potential of fermented food environments for enzyme resource exploration, offering valuable insights into microbial functions for future food research. A record of this paper's transparent peer review process is included in the supplemental information.

Single-molecule RNA fluorescence in situ hybridization (RNA FISH)-based spatial transcriptomics methods have enabled the accurate quantification of gene expression at single-cell resolution by visualizing transcripts as diffraction-limited spots. Although these methods generally scale to large samples, image analysis remains challenging, often requiring manual parameter tuning. We present Piscis, a fully automatic deep learning algorithm for spot detection trained using a loss function, the SmoothF1 loss, that approximates the F1 score to directly penalize false positives and false negatives but remains differentiable and hence usable for training by deep learning approaches. Piscis was trained and tested on a diverse dataset composed of 358 manually annotated experimental RNA FISH images representing multiple cell types and 240 additional synthetic images. Piscis outperforms other state-of-the-art spot detection methods, enabling accurate, high-throughput analysis of RNA FISH-derived imaging data without the need for manual parameter tuning. A record of this paper's transparent peer review process is included in the supplemental information.

Understanding bacterial gene function remains a major challenge. Double-mutant genetic interaction analysis addresses this challenge by uncovering the functional partners of targeted genes, enabling association of genes of unknown function with known pathways and unraveling of connections among well-studied pathways, but such approaches are difficult to implement at the genome scale. Here, we use double-CRISPR interference (CRISPRi) to systematically quantify genetic interactions at scale for the Bacillus subtilis cell envelope, including essential genes. We discover >1,000 genetic interactions, some known and others novel. Our analysis pipeline and experimental follow-ups reveal the shared and distinct roles of paralogous genes such as mreB and mbl in peptidoglycan and teichoic acid synthesis and identify additional genes involved in the well-studied process of cell division. Overall, our study provides valuable insights into gene function and demonstrates the utility of double-CRISPRi for high-throughput dissection of bacterial gene networks, providing a blueprint for future studies in diverse species. A record of this paper's transparent peer review process is included in the supplemental information.

Crohn's disease (CD) is an inflammatory gastrointestinal disease affecting approximately 1 in 1,000 people in North America. Incidence of pediatric CD has been rising in recent decades, and this group is especially at risk of more severe disease development because of the association of CD with developmental deficits. Genome-scale metabolic models (GEMs) present an opportunity to investigate systems-level changes in metabolism in specific contexts, such as pediatric CD. In this work, we utilized pediatric and adult omics data to create an ileum-specific GEM, Ileum1. We also developed reaction inclusion analysis (RIA) to quantify broad metabolic differences of several clinical cohorts and used this method to compare hundreds of subject-specific GEMs. RIA predicted altered cholesterol metabolism in males with CD, and in vitro testing found that cholesterol synthesis inhibition prevented an increase of inflammatory cytokines. We used transcriptomics from adult subjects and found that metabolism is uniquely altered in adult CD. A record of this paper's transparent peer review process is included in the supplemental information.

Cancer patients frequently suffer from anemia and cancer-related pain, which can be treated by non-opioid analgesics such as diclofenac (DCF) and acetaminophen (APAP) attenuating inflammatory responses. The pro-inflammatory cytokine interleukin (IL)-6 triggers the expression of acute-phase proteins, including the iron regulator hepcidin. Using proteomics and dynamic pathway modeling, we show that DCF and APAP directly impact IL-6 signaling by enhancing the induction of the feedback-inhibitor suppressor of cytokine signaling 3 (SOCS3), reducing signal transducer and activator of transcription (STAT)3 phosphorylation, and decreasing the expression of most acute-phase proteins except for hepcidin. In primary human hepatocytes (PHHs), the impact depends on the patient-specific extent of SOCS3 induction, which is anti-correlated with hepcidin expression. Whereas, in liver cancer cells, DCF and APAP stabilize the interaction of autocrine secreted bone morphogenic protein (BMP) with its receptor, resulting in strongly amplified hepcidin expression. Our studies suggest that co-inhibition of the BMP receptor counteracts excessive hepcidin production upon treatment with pain-relieving drugs and could prevent iron-deficiency-caused anemia in liver cancer. A record of this paper's transparent peer review process is included in the supplemental information.