The probiotics field, a historically popular yet scientifically debated discipline, is moving beyond a decades-long promotion of ‘first-generation’ food-derived strains towards the development of ‘next-generation probiotics’ (NGP) or ‘precision probiotics’, natural and engineered strains featuring improved human colonization, clinical efficacy and safety profiles. In this Review, we outline the evolution of NGP and means by which their development is designed to tackle challenges of live bacterial therapy related to colonization resistance, in-host evolution, long-term safety and insufficient understanding of therapeutic and off-target mechanisms of activity. We showcase how a variety of emerging strategies enable the identification of NGP strains and define consortia featuring therapeutic potentials in metabolic, immune and oncological diseases. Finally, we discuss how computational and artificial intelligence (AI) advances can reshape NGP development, including AI-based discovery of strains and bioactive compounds; computational-driven design of engineered microorganisms and multi-kingdom consortia; and AI-assisted structural and metabolic network-based modelling predicting personalized NGP function, interactions and therapeutic impacts. In this Review, Kern, Tofield, Frame and Elinav discuss recent advances in the design of next-generation probiotics, from identification of candidates to therapeutic applications across diverse disease contexts, and highlight major challenges and the potential of artificial intelligence to develop effective, personalized probiotics with therapeutic functions.

mRNA-based therapeutics have revolutionized the treatment and prevention of infectious, neurological, and cancer diseases. However, their linear topology makes them susceptible to rapid degradation in vivo, which limits their therapeutic efficacy. Engineered circular RNAs (circRNAs) due to their closed ends and high stability are emerging as a promising alternative to linear RNA therapies. Engineered circRNAs are also increasingly used to mimic naturally occurring circRNAs in functional studies. Both applications, however, depend on production of precise circRNAs with homogenous sequences to enable accurate interpretation of biological outcomes. To address this, we developed and optimized methods for generating precise circRNAs. We employed enzymatic ligation of linear RNAs rather than autocatalytic splicing to produce circRNAs to minimize extraneous nucleotides remaining from the ribozymes. We carefully designed the DNA transcription template to maintain sequence and structural integrity. A permuted DNA template leveraging three internal guanosines (Gs) was synthesized and amplified using a reverse primer containing two 2′-O-methyl groups. This approach optimally produced the linear precursor RNA with correct 5′ and 3′ ends. After testing multiple workflows, we found that GMP-primed in vitro transcription, T4 RNA ligase 2-mediated circularization, and urea–polyacrylamide gel electrophoresis (PAGE) gel extraction produced the highest fidelity circRNAs.

Natural products, as metabolites from microorganisms, animals or plants, exhibit diverse biological activities, making them crucial for drug discovery. Nowadays, existing deep-learning methods for natural products research primarily rely on supervised learning approaches designed for specific downstream tasks. However, such one-model-for-a-task paradigm often lacks generalizability and leaves substantial room for performance improvement. Additionally, existing molecular characterization methods are not well-suited for the unique tasks associated with natural products. Here, to address these limitations, we pretrained a foundation model for natural products (NaFM) based on their unique properties. Our approach employs a pretraining strategy specifically tailored to natural products. By incorporating contrastive learning and masked graph learning objectives, we emphasize evolutional information from molecular scaffolds while capturing side-chain information. NaFM achieves state-of-the-art results in various downstream tasks related to natural product mining and drug discovery. We first compare taxonomy classification with synthetic molecule-focused baselines to demonstrate that current models are inadequate for understanding natural synthesis. Furthermore, by diving into a fine-grained analysis at both the gene and microbial levels, NaFM demonstrates the ability to capture evolutionary information. Eventually, our method is applied to virtual screening, illustrating informative natural product representations that can lead to more effective identification of potential drug candidates. Ding et al. present a scaffold-aware foundation model for small-molecule natural products leveraging masked objectives and contrastive learning to enhance taxonomy classification, genome mining and virtual screening in drug discovery.

Milk is one of the most vital foods worldwide, valued not only for its nutrient-rich composition but also for its diverse range of bioactive compounds. In addition to its nutritional importance, milk contains a variety of proteins that serve significant biological functions. Among these are lactic acid bacteria (LABs), which can produce antimicrobial peptides and organic acids. The antimicrobial effects of milk are primarily attributed to its bioactive proteins, including whey proteins and caseins. Whey proteins, such as lactoferrin, lysozyme, and immunoglobulins, as well as peptides derived from these proteins, exhibit significant antimicrobial activity, particularly against Gram-positive and Gram-negative bacteria. These peptides are released during proteolysis, either through enzymatic digestion or fermentation, and can interact with bacterial membranes, destabilising them and preventing microbial growth. The concentration of antimicrobial proteins varies across mammalian milks, with higher levels often observed in species such as sheep and goats, reflecting adaptations to specific environmental and immune challenges. Despite the reduction in antimicrobial efficacy following heat treatments like ultra-high temperature (UHT) or pasteurisation, fermented dairy products such as yoghurt and cheese retain significant antimicrobial properties, mainly due to the presence of bioactive peptides and increased acidity. These antimicrobial activities underscore the potential of milk-derived compounds as natural alternatives to antibiotics, particularly in food safety and therapeutic applications. Further research into milk’s bioactive peptides could expand their use in the prevention and treatment of microbial infections.

Many microbiological outcomes are shaped by the determinants of community composition, including the factors that allow pathogens to invade healthy microbiota and the processes that maintain the diversity that underpins soil function. Community ecology provides a rich conceptual toolset to investigate patterns of coexistence in ecosystems that can be adapted to explain and manage these outcomes. However, these concepts have complex histories of controversy and debate that must be considered when applying them to the microbial context. Microorganisms also have distinctive characteristics that must be accounted for when applying ideas that were originally developed to describe macroscopic ecosystems. Here, we provide a concise overview for microbiologists to five key frameworks from community ecology: Niche theory, Trophic levels, Keystone species, Succession, and Metacommunities. We discuss the historical context and controversies surrounding each framework and outline existing and potential applications to microbial systems. This work therefore provides a practical guide for microbiologists who wish to apply community ecology for understanding and manipulating microbial community composition.

Analyte detection through aptamer-induced signal generation by CRISPR-Cas enzymes has rapidly emerged as a popular biosensing approach. Here, we investigated the implementability and analytical performance of this setup for the detection of diverse small molecule analytes. We selected nine aptamers from the literature targeting seven analytes and tested a commonly used assay design whereby analyte binding by the aptamer liberates a short complementary ssDNA strand, which in turn activates Cas12a to generate a fluorescence signal. After extensive optimization, the assay functioned for only two of the seven analytes, and several previously reported results could not be reproduced. While Cas12a fluorescence ssDNA detection was robust, the low success rate is likely due to aptamers not functioning reliably, underscoring the need for careful aptamer validation. Overall, our study provides a critical assessment of aptamer-Cas12a assay performances and discusses potential strengths, limitations, and pitfalls of this biosensing strategy.

Developing engineered genetic switches that can respond to specific stimuli in mammalian systems has great potential to advance next-generation therapies. This study introduces a novel tetracycline-inducible artificial riboswitch that regulates gene expression at the translational level via a modified −1 programmed ribosomal frameshifting (−1 PRF) platform. Our dose-response analysis reveals an EC50 of 0.41 μM, indicating exceptional sensitivity compared to other tetracycline-dependent artificial riboswitches developed for mammalian applications. By varying the precise spacing and composition of the tetracycline aptamer as the stimulatory structure on the −1 PRF platform, a series of switches has been constructed that enable robust induction of gene expression upon ligand addition. To attain a translational gene control platform with broader applicability, we developed a concept that allows it to be integrated upstream of a gene of interest (GOI), enabling precise control without requiring extensive re-engineering and expressing the GOI as a fusion protein. We validate the platform’s effectiveness in several genetic contexts, confirming its modularity. As optimization efforts continue to develop artificial riboswitches with optimized properties in mammalian systems, our findings contribute to the development of highly sensitive and efficient regulatory devices with beneficial characteristics, such as reversibility, modularity, and compatibility with RNA-based delivery therapies.

PockFlex is a web server designed to analyze pockets across protein structural ensembles and support the reconstruction, characterization, and prioritization of recurrent binding site organizations. Applicable to ensembles derived from molecular dynamics simulations, multiple experimental structures, or protein structure predictions, PockFlex detects pockets independently in each conformation, retains those overlapping a user-defined region of interest, and groups them across the ensemble by residue-level similarity. This residue-centred clustering framework identifies recurrent binding site clusters, quantifies residue recurrence and variability, and distinguishes persistent from transient binding site regions across the ensemble. Pocket-level druggability, predicted using the PockDrug workflow, is summarized at the cluster level to support binding site prioritisation under conformational variability while preserving access to individual pocket scores. The web application provides interactive, residue-level insights into pocket organisation, variability, and druggability in structural ensembles. The web server is free and open to all users, without login requirement, at https://pockflex.rpbs.univ-paris-diderot.fr/.

Hyperuricemia has emerged as the fourth most prevalent metabolic disorder, necessitating the development of safer and more effective therapeutic strategies. In this study, we constructed a recombinant probiotic strain expressing the PucL and PucM enzymes, which demonstrated a uric acid degradation rate of 65% in vitro. To enhance this activity, we performed modular optimization by employing three ribosome binding sites (RBSs) of different strengths—RBS 29, RBS 31, and RBS T7—to tune the expression levels of pucL and pucM, resulting in highly efficient uric acid degradation. Further improvement was achieved by overexpressing the uric acid transporter gene ygfU and the hydrogen peroxide–degrading catalase gene katG, leading to significant uric acid degradation. Furthermore, the engineered E. coli Nissle 1917 strain was evaluated in a mouse model of hyperuricemia; treatment with the optimized probiotic reduced serum uric acid levels to 39.11 mg/L, representing a 15.98% decrease compared with the control group. Further analysis revealed that this engineered bacterium ameliorates hyperuricemia by modulating the Firmicutes-to-Bacteroidetes ratio, increasing microbial diversity, and promoting the growth of beneficial genera. Collectively, this study establishes an engineered probiotic cell factory for uric acid degradation and demonstrates a proof-of-concept for the microbial remediation of hyperuricemia.