Cell systems最新文献

Protein-function prediction is crucial for elucidating molecular mechanisms driving biological processes and therapeutics development. Despite numerous computational tools demonstrating promising performance, they fall short when predicting rare, uncharacterized functions or indirect activities. Here, we present COSMOS, a context-aware Gene Ontology (GO) subgraph mining system for protein-function prediction. By leveraging inductive subgraph foundation models and an enriched knowledge graph of protein-GO relationships, COSMOS performs zero-shot, few-shot, and low-homology protein-function prediction. Built on 7,923,952 functional semantic relationships, COSMOS demonstrates robust capabilities to (1) generate state-of-the-art predictions for GO classes with sparse or no experimental annotations, (2) provide interpretable functional subgraphs for transparent rationale analysis, and (3) deliver complementary benefits when integrated with existing embedding-based prediction methods. We anticipate that COSMOS will serve as a complementary approach to conventional protein annotation methods and an interpretable tool for predicting protein functions within underexplored GO classes, thereby advancing genomics and therapeutic research.

Resonance allows systems to amplify their response to periodic stimuli and is well established in physics but not yet described in gene regulatory networks. Here, we asked whether resonance exists in the dynamics of p53, a tumor suppressor that oscillates after DNA damage to activate growth-inhibitory pathways. We developed a mathematical framework predicting that p53 exhibits damped oscillations after a single stimulus and frequency-dependent amplitudes under periodic stimulation, both hallmarks of resonance. Using live single-cell imaging, we confirmed these predictions: a single drug pulse that stabilizes p53 produced damped oscillations, while periodic pulses triggered frequency-dependent responses with maximal amplitudes at the natural p53 oscillation frequency as well as minor peaks. Finally, theoretical analysis suggested that resonance may enhance transcriptional responses and selectively activate downstream targets. Together, our results identify resonance as a regulatory principle in gene networks, potentially linking oscillations of transcription factors with selective gene activation through signal amplification.

Genetic circuit engineering enables new cellular functions, yet most circuits are developed in the model host Escherichia coli, limiting their availability and performance in alternative organisms. To expand chassis options, we developed an experimental-theoretical pipeline to evaluate NOT logic circuits, or inverters, in the soil bacterium Pseudomonas protegens Pf-5, a species with valuable environmental traits and a newcomer to bioengineering. We characterized the inverter's input-output behavior and used mathematical modeling to infer key dynamic principles. The model quantified how parameters such as translation efficiency, repressor performance, and promoter activity shape circuit output and influence inter-host portability. Pf-5 displayed distinct properties, including steeper on/off transitions than the synthetic biology workhorse Pseudomonas putida. A model-guided design of two compatible inverters connected in series was validated, producing a YES logic response. This work provides DNA parts, circuits, and models that establish P. protegens Pf-5 as a promising chassis for environmental synthetic biology. A record of this paper's transparent peer review process is included in the supplemental information.

Auxotrophs are prevalent in microbial communities, enhancing their diversity and stability-a counterintuitive effect considering their dependence on essential resources from other species. To address the ecological roles of auxotrophs, our study introduced a consumer-resource model (CRM) to capture the complex higher-order interactions within these communities. We also developed an intuitive graphical and algebraic framework, which assesses the feasibility of auxotroph communities and their stability under resource fluctuations and biological invasions. Validated against experimental data from synthetic E. coli auxotroph communities, the model accurately predicted outcomes of community assembly. Our findings highlight the critical role of higher-order interactions and resource dependencies in maintaining the diversity and stability of microbial ecosystems dominated by auxotrophs. A record of this paper's transparent peer review process is included in the supplemental information.

Probiotic interventions are effective strategies to modulate the gut microbiome, but how exogenous probiotics compete with native gut microbiota remains elusive. Here, we use a mouse model and a well-documented probiotic, Bifidobacterium animalis subsp. lactis V9 (BV9), to mechanistically investigate its competitive strategies. We perform metagenomic and whole-genome sequencing of stool samples and isolated BV9, longitudinally collected from 24 mice orally administered with BV9 and different diets. Results show that a high-fiber diet most effectively supports the colonization of BV9, where BV9 selectively competes with Parabacteroides distasonis (P. distasonis), rather than extensively with other gut bacteria. By comparing the genomic structures of BV9 and P. distasonis isolated during the washout period, we infer their co-evolution mechanisms, highlighting their competition and compromise in utilizing inulin-derived glucose. Finally, our in vitro co-culture experiments validate such competitive dynamics. This study fills a critical gap in our understanding of niche competition in colonization.

CRISPR-mediated gene activation (CRISPRa) allows concurrent transcriptional control of many genes and is widely used in genome-wide screening, bioproduction, and therapeutics. Multi-gene control is possible due to the sequence specificity by which guide RNAs (gRNAs) recruit dCas9 and an activator protein to target genes. Still, the optimization of CRISPRa systems remains difficult. Here, we show that, despite sequence specificity, different gRNAs interfere with each other by competing for dCas9 and the activator protein. This competition breaks modularity and hinders CRISPRa. We also discover that gene activation is biphasic, wherein increased level of a gRNA leads to target repression instead of activation. We introduce a chemical reaction-network model that captures these effects and use it for improving the dynamic range of CRISPRa. Our results demonstrate that CRISPRa is not as modular or scalable as previously thought and establish a predictive modeling tool that enables systematic design and optimization of multi-gRNA CRISPRa systems. A record of this paper's transparent peer review process is included in the supplemental information.

Integral feedback enables perfect adaptation, but antithetic implementations can be slow and noisy. Filo et al. show that sensor-based antithetic integral feedback (sAIF) control yields effective proportional-integral (PI) behavior without adding a separate proportional module. Implemented in E. coli using split inteins, it improves disturbance rejection and noise in defined regimes.

Emerging research links the gut, skin, and oral microbiomes to allergies, with serine proteases (SPs) identified as potential allergens. This study leverages deep learning and pre-trained protein language models (pLMs) to uncover allergenic SPs in metagenomic data. First, we develop a model to identify the catalytic serine residue in serine hydrolases, demonstrating how pLMs capture structural information. Next, we create a deep learning framework to detect candidate SP allergens across gene catalogs, using the conserved catalytic triad to identify homologs in gut and oral sites despite low sequence identity. Our model predicts a putative SP allergen resembling V8 protease, a known trigger for protease-activated receptor 1. It also identifies a cysteine protease similar to Der f 1 from dust mites. Immunization with these proteases induced allergic responses, validating their allergenic potential experimentally. This approach uncovers candidate allergens beyond traditional methods, offering new targets for allergy research. A record of this paper's transparent peer review process is included in the supplemental information.

Effective cellular regulation relies on feedback control mechanisms to maintain homeostasis and mitigate environmental fluctuations. We develop and analyze a sensor-based antithetic integral feedback (sAIF) controller that achieves this by embedding proportional and integral actions within a minimal genetic architecture. Arising from a single modification to the classical antithetic control motif, this sAIF architecture intrinsically incorporates proportional feedback without requiring additional circuitry. Control-theoretic and stochastic analyses show that this proportional action speeds up the system's dynamic response and counteracts the noise amplification typical of pure integral feedback, enabling both improved speed and reduced cellular variability. Using intein-mediated splicing, we implement sAIF in E. coli and demonstrate robust perfect adaptation, strong disturbance rejection, and favorable noise properties. These findings establish a generalizable design principle for engineering high-performance biological controllers, with broad implications for synthetic biology, metabolic engineering, and cell-based therapies. A record of this paper's transparent peer review process is included in the supplemental information.