Antimicrobial resistance poses a significant challenge to conventional antibiotics, underscoring the urgent need for alternative therapeutic strategies. Antimicrobial peptides (AMP) have emerged as promising candidates due to their broad-spectrum antibacterial activity and distinct mechanisms of action. This study presents ANIA, a deep learning framework developed to predict the minimum inhibitory concentration (MIC) values of AMPs against three clinically significant bacteria: Staphylococcus aureus, E, coli, and Pseudomonas aeruginosa. ANIA leverages Chaos Game Representation (CGR) to transform AMP sequences into frequency-based image features, which are subsequently processed through a hybrid architecture comprising stacked Inception modules, a Transformer encoder, and a regression head. This integrative architecture enables ANIA to capture both local motif-based features and global contextual patterns embedded within AMP sequences. In benchmarking experiments, ANIA achieved notably superior performance compared to existing tools, including ESKAPEE-Pred, AMPActiPred, and esAMPMIC, achieving higher correlation coefficients and lower predictive errors across all bacteria targets, with the most pronounced improvement observed for P. aeruginosa, a pathogen renowned for its multidrug resistance. Specifically, ANIA achieved PCCs of 0.75–0.79 and MSEs of 0.23–0.26 across all species. Furthermore, motif-based interpretability analyses combining Grad-CAM visualizations, correlation heatmaps, motif frequency distributions, and hydrophobicity profiling revealed biologically meaningful subregions within the CGR matrix that are plausibly associated with antimicrobial efficacy. In conclusion, this study develops ANIA as a robust predictive tool for MIC estimation, offering valuable insights into the design of effective antimicrobial agents and contributing to the fight against antimicrobial resistance. A user-friendly web server for ANIA is available at https://biomics.lab.nycu.edu.tw/ANIA/.

Rhodospirillum rubrum owns a dynamic poly(3-hydroxybutyrate) (PHB) cycle: During growth PHB is accumulated and subsequently degraded under carbon starvation. Interestingly, this cycle is typically found for acetate grown R. rubrum but not for fructose grown cells, where no PHB accumulation has been observed. This study aimed to determine whether expression of PHB cycle genes correlates with the phases of PHB accumulation and degradation on acetate in comparison to absence of PHB synthesis during growth on fructose and ΔphaC1ΔphaC2 mutant unable to polymerize PHB on acetate. Surprisingly, transcriptomic analyses of the wild-type strain demonstrated that PHB cycle genes were not only expressed during growth on acetate but also for growth on fructose, regardless of PHB content. Substrate-specific expression patterns were identified: The PHB depolymerase gene phaZ1 was predominantly expressed on acetate, while phaZ2 and the depolymerase regulator apdA were upregulated on fructose. Interestingly, phaC3 and phaZ3 showed distinct expression patterns compared with other PHB cycle genes, particularly in mutant strains. Despite the absence of PHB granules in the ΔphaC1ΔphaC2 strain, several PHB cycle genes remained expressed, and volatile fatty acid assimilation pathways were transcriptionally impacted. These findings highlight the complexity of the PHB cycle and suggest that PHB participates in other physiological processes, such as substrate assimilation, potentially via regulatory actions of PHB granule bound regulator PhaR.

Plant root growth accounts for a major part of the net primary production in grassland and forest ecosystems and influences the global carbon and nutrient cycles. Measuring the production of roots is inherently difficult, prone to inconsistencies and time-consuming. Notably, there are currently no methods yet to automate this task.  We have developed GINGER, a new method for automated estimation of the fine root production from a time series of mini-rhizotron images. It compares pairs of consecutive images with each other, separating new root growth from standing crop.  The method was evaluated on four datasets from grassland, drained fen peatland and forest ecosystems. It exhibits performance on a similar level to that of human annotators while substantially reducing the time required for the data analysis. Human annotators showed a significant degree of variability among each other, confirming that the task is subjective and error-prone.  For demonstration, this pipeline was applied on two real-world image datasets, spanning 2 and 3 years, to compute the total annual root production. End-to-end, including annotation and model training, GINGER reduced the required human workload from several thousand to less than 40 work hours. It could allow to scale up monitoring efforts and enable full automation in the future.

The development of micro- and nano-robots has amplified the demand for intelligent multifunctional machines in biomedical applications, but most microrobotic systems struggle to achieve the attributes needed for those applications. Here we introduce enzymatic microbubble robots that exhibit steerable motion, enhanced biodegradability, high in vivo imaging contrast, and effective targeting and penetration of disease sites. These microrobots feature natural protein shells modified with urease to decompose bioavailable urea for autonomous propulsion, whereas an internal microbubble serves as an ultrasound imaging contrast agent for deep tissue imaging and navigation. Magnetic nanoparticle integration enables imaging-guided magnetically controlled motion and catalase functionalization facilitates chemotactic movement towards hydrogen peroxide gradients, directing robots to tumour sites. Focused ultrasound triggers robot shell collapse and inertial cavitation of the released microbubbles, creating mechanical forces that enhance therapeutic payload penetration. In vivo studies validate the tumour-targeting and therapeutic efficacy of these robots, demonstrating enhanced antitumour effects. This multifunctional microbubble robotic platform has the potential to transform medical interventions and precision therapies. Biodegradable enzymatic microbubble robots self-propel in urea, are magnetically or chemotactically guided, provide ultrasound imaging and enhance intratumoural drug delivery with focused ultrasound.

Bacterial contractile injection systems (CISs) are multiprotein complexes that facilitate the bacterial response to environmental factors or interactions with other organisms. Multiple novel CISs have been characterised in laboratory bacterial cultures recently; however, studying CISs in the context of the native microbial community remains challenging. Here, we present an approach to characterise a bioinformatically predicted CIS by directly analysing bacterial cells from their natural environment. Using cryo-focused ion beam milling and cryo-electron tomography (cryoET) imaging, guided by 16S rRNA gene amplicon sequencing, we discovered that thermophilic Chloroflexota bacteria produce intracellular CIS particles in a natural hot spring microbial mat. We then found a niche-specific production of CIS in the structured microbial community using an approach combining metagenomics, proteomics, and immunogold staining. Bioinformatic analysis and imaging revealed CISs in other extremophilic Chloroflexota and Deinococcota. This Chloroflexota/Deinococcota CIS lineage shows phylogenetic and structural similarity to previously described cytoplasmic CIS from Streptomyces and probably shares the same cytoplasmic mode of action. Our integrated environmental cryoET approach is suitable for discovering and characterizing novel macromolecular complexes in environmental samples.

Retrons represent a novel class of bacterial defense systems that employ reverse transcriptase (RT), noncoding RNA, and effector proteins to counteract phage infections. In this study, we elucidate the molecular mechanism of a retron system, Retron–Eco8. Biochemical experiments reveal that the Retron–Eco8 holocomplex, rather than the effector alone, exhibits double-stranded DNA cleavage activity, triggering abortive infection and therefore effectively halting phage propagation. Cryo-electron microscopy (cryo-EM) analysis reveals a supramolecular assembly comprising four RT subunits, four multicopy single-stranded DNA molecules, and four overcoming lysogenization defect (OLD) nucleases—a configuration critical for antiphage defense. Notably, we examine the activation of Retron–Eco8 by diverse single-stranded DNA-binding (SSB) proteins, and phylogenetic analysis of these SSB proteins elucidates the phage resistance specificity. Collectively, our findings delineate the structural architecture of the Retron–Eco8 defense complex and provide mechanistic insights into retron-mediated bacterial immunity.

Intrinsically disordered regions (IDRs) in proteins drive phase separation (PS) to form biomolecular condensates, which organize cellular matter. While IDRs are recognized as critical drivers of PS, the systematic identification of sequence motifs governing this phenomenon and their compositional determinants remain a key challenge. Here we develop PhaSeMotif, a deep learning framework for interpretable and precise predictions of essential phase-separating motifs within IDRs. We experimentally validate PhaSeMotif, demonstrating that mutations of predicted motifs significantly reduce or eliminate the PS capabilities of IDRs. The identified motifs possess diverse amino acid compositional features that are critical for determining PS propensities and condensate partitioning. Furthermore, PhaSeMotif integrates generative models to create validation-ready motifs that preserve these critical compositional features, empowering direct experimental verification and deeper mechanistic investigation of PS-driving IDR motifs. Overall, by combining motifs prediction, generation, and validation, PhaSeMotif provides an open-access toolkit to facilitate more efficient IDR motifs investigation and provides insights into the molecular determinants of PS. Deciphering the rules of protein phase separation remains a challenge. Here, the authors develop PhaSeMotif, an interpretable and generative deep learning framework that identifies essential sequence motifs and designs functional synthetic variants.