Phages are viruses that infect bacteria and play essential roles in shaping microbial communities. Identifying phage-host interactions (PHIs) is crucial for understanding infection dynamics and developing phage-based therapeutic strategies. Recent deep learning approaches have shown great promise for PHI prediction; however, their performance remains constrained by the limited number of experimentally validated positive pairs and the overwhelming abundance of unlabeled or non-validated samples. Moreover, most existing models overlook higher-level phylogenetic relationships among hosts, which could provide valuable structural priors for guiding representation learning. To address these challenges, we propose a phylogenetic tree-aware positive-unlabeled deep metric learning framework for phage-host interaction (PHI) identification. Unlike traditional approaches that train classification models to strictly separate positive and negative phage-host pairs, the proposed method learns representations under supervision from both confirmed positive PHIs and host phylogenetic tree constraints on non-positive samples. The proposed method can seamlessly formalize contrastive learning and deep metric learning within the same framework that explicitly optimizes PHI encoders with biological constraints in the learning functions. We show that this metric learning formulation outperforms conventional contrastive learning approaches that enforce separation between positive and negative samples without consistently aligning the learned representations with evolutionary distances. Experiments on the Cherry benchmark dataset and metagenome Hi-C multi-host dataset demonstrate that our approach enhances species-level prediction accuracy, improves cross-host generalization, and yields more interpretable representations of phage-host relationships.

Enzymes do not operate as static structures, but continuously fluctuate between different conformations. Enzymes therefore dynamically sample conformations with varying catalytic activity. However, it remains largely unexplored whether evolution can exploit the conformational dynamics between sub-states to improve activity. Here, we dissect the evolutionary trajectory of the β-lactamase OXA-48 toward improved ceftazidime hydrolysis. Evolution relieved conformational bottlenecks by promoting alternate functional sub-states, gradually shifting the rate-limiting step from substrate binding to sub-state interconversion, and finally to the chemical step. Reorganization of the conformational landscape enhanced OXA-48's ability to hydrolyze ceftazidime and introduced a trade-off in its native activity against meropenem. This trade-off stemmed from catalytic incompatibility between the native and the evolved sub-state populations. Our findings highlight the transitions between functional sub-states as a mechanism of natural selection, shaping functional divergence and offering new strategies for enzyme and antibiotic engineering.

Droplet microfluidics is a core technology that powers high-throughput single cell sequencing. However, the current generation of single-cell microfluidics faces notable limitations, including cell aggregation, suboptimal on-chip reactions that compromise experimental outcomes and elevate background noise, as well as a dependence on costly commercial barcode beads. To address these challenges, we present MusTer, an integrated next-generation platform with multi-parametric singlet droplet sorting and triple-droplet merging capability. MusTer's multi-parametric singlet sorting module enables in-line droplet analysis of intrinsic fluorescence peak amplitude, width and interval from single-nucleus- (singlet) or multiple-nuclei (multiplet)-encapsulating droplets, subsequently allowing an effective separation of the singlet droplets from multiplet droplets and empty droplets. MusTer's triple-droplet merging module enables precise multi-step reactions, with each step performed under its own optimal conditions, thereby significantly enhancing experimental flexibility and efficiency. We validated MusTer's performance by performing single-cell ATAC-seq on maize leaves. The results demonstrate that MusTer significantly reduces the doublet rate, enhances the signal-to-noise ratio, and yields improved cell clustering compared with traditional methods. These results validate MusTer's capability to overcome key limitations in droplet-based single-cell analysis, effectively enhancing data quality and reliability, and also paves the way for its use in other challenging sample types and multi-step single-cell assays.

Temperature offers a simple yet powerful signal to program cellular behavior. Here, we engineered and characterized a set of temperature-dependent genetic circuits that integrate RNA thermometers with site-specific DNA recombinases to achieve precise, irreversible control of gene expression. Using the serine recombinase Bxb1 placed under the control of the Salmonella FourU RNA thermometer, we demonstrate how promoter strength critically defines thermal sensitivity: weak promoters’ activity clears ON/OFF transitions, while strong promoters lead to continuous, quasi-temperature-independent recombination. Furthermore, temperature pulse duration and growth phase of cell culture were found to modulate recombination efficiency, providing additional layers of control. We illustrate the potential of this framework through proof-of-concept applications, including (i) the generation of spatial expression patterns on 2D surfaces via localized heating, (ii) a paper-based device capable of recording temperature gradients as stable genetic outputs, and (iii) a temperature-triggered lysis system for controlled cellular release. Together, these results establish temperature-regulated recombinase circuits as versatile and robust tools for programmable, spatially resolved, and irreversible control of gene expression, paving the way for new applications in synthetic biology, biosensing, and bioproduction.

Design of experiments (DOE) principles are increasingly applied to biological assays, yet it remains unclear whether their foundational assumption, orthogonal decomposition, holds in nonlinear biological systems. We addressed this question using Perturb-seq as a case study. By benchmarking a design commonly used in Perturb-seq and related experiments against the orthogonal Plackett-Burman (PB) design via simulations, we uncovered a counter-intuitive phenomenon: while orthogonal designs generally excel, the multicollinearity inherent to the common design is functionally advantageous in systems with significant signal amplification. This challenges the blind application of DOE to biology. Based on these findings, we developed the PB suitability index (PBSI), a simple, parameter-free metric that predicts the optimal design solely from network structure. Our work not only provides practical guidelines for Perturb-seq but also establishes a "biology-oriented DOE" framework, bridging the gap between statistical rigor and biological complexity.

Transcriptional control arises from the specific recognition of promoter DNA by transcription factors (TFs), forming the basis of cellular information processing and gene regulation. In synthetic biology, TF-promoter interactions are assembled into gene circuits to program cellular behaviors. To ensure reliable circuit performance, most synthetic gene circuits rely on well-characterized and orthogonal regulatory parts. This reliance minimizes crosstalk but constrains circuit complexity and information integration. Creating hybrid TFs that combine or interpolate promoter specificities could therefore expand the design space of synthetic regulatory systems. However, it remains unclear whether hybrid functions can be created by mixing amino acid sequences, and how such functional integration could be achieved in a principled manner. Here we show that a variational autoencoder (VAE) trained on LuxR-family DNA-binding domains can generate transcription factors with hybrid and partially novel promoter recognition properties. By sampling intermediate regions of the VAE-learned latent space, we designed hybrid TFs that activate both the lux and las promoters. High-throughput sort-seq assays together with individual in vivo assays revealed that a subset of functional variants exhibited dual-responsive behavior while maintaining sequence-selective DNA recognition. Together, these results provide a data-driven strategy for exploring functional intermediate sequences between closely related proteins.

3'-5' translation (termed "backward translation") mediated by eukaryotic circular RNAs has been reported recently. Whether linear mRNA can produce proteins through backward translation remained elusive. Here, we demonstrate that linear mRNAs can mediate backward translation in vitro. Backward translation was detected in human cells transfected with expression vectors used to produce linear mRNAs. Importantly, synthetic linear mRNAs could produce backward translation signal in multiple in vitro translation systems, with a level comparable to that of conventional 5' to 3' forward translation. Furthermore, we demonstrated that a single translation initiating sequence (TIS) was able to drive backward and forward translation in an in vitro experimental setup, with the efficiency of backward translation found to be significant lower than that of forward translation. In summary, this study further reinforces the notion of translation flexibility, demonstrates a broader applicability of backward translation and heralds a new strategy in synthetic biology.

MicroRNAs (miRNAs) are key posttranscriptional regulators of gene expression that influence cancer initiation, progression, and therapeutic response. While most studies have focused on endogenous miRNAs, emerging evidence has highlighted the role of plant-derived miRNAs as exogenous dietary regulators capable of cross-kingdom gene modulation. This review summarises current knowledge regarding plant-derived miRNAs and their ability to regulate human cancer-related genes. Experimental findings indicate that plant miRNAs can withstand gastrointestinal digestion, enter the circulation, and regulate the expression of oncogenes, tumor suppressors, long noncoding RNAs, and immune checkpoint molecules via canonical RNA-induced silencing mechanisms. Specific examples include miR-156a, miR-159a-3p, miR-166a, miR-167e-5p, miR-171, miR-395e, miR-2911, miR-4995 and miR-5754, which exhibit anticancer activities across various cancer types and modulate key signalling pathways in mammalian cells, highlighting their potential as cross-kingdom regulators with therapeutic relevance. In addition to these characterised miRNAs, certain plant groups, which are rich in bioactive compounds, remain unexplored as sources of functional miRNAs, representing a promising avenue for future research. Collectively, these studies underscore the ability of plant-derived miRNAs to modulate mammalian gene expression and suggest their potential as diet-based or synthetic therapeutic agents. Further investigations into their bioavailability, target specificity, and functional relevance could inform innovative strategies for cancer prevention, integrating nutritional, molecular biological, and therapeutic approaches.

Solanoeclepin A (SolA) is a triterpenoid exuded from the roots of Solanaceae plants, originally identified as hatching stimulant for plant parasitic cyst nematodes. Assuming that evolution would have selected against the production of such a fitness lowering molecule, we postulated that SolA must serve another, beneficial, role for the plant. In this study, we demonstrate that nitrogen (N) deficiency strongly increases the SolA concentration in tomato root exudate. Moreover, SolA is produced only under non-sterile conditions, indicating that soil microbiota are involved in its production. Time-resolved RNAseq analysis revealed several candidate genes for SolA biosynthesis, which were all upregulated under N deficiency. Transient silencing of two SolA biosynthetic genes (e.g., CYP749A19 and CYP749A20) significantly reduced SolA production. Microbiome analysis on the rhizosphere of these plants demonstrated that the recruitment of beneficial Massilia spp. was inhibited in the transiently silenced plants. Isolation of a Massilia strain, identified as Massilia cellulosiltytica, allowed us to show that it has growth-promoting activity under N deficiency, likely via indole-3-acetic acid production and enhanced N acquisition. Root exudate of N-starved tomato displayed strong chemotactic activity towards this strain. Together, these findings demonstrate that SolA production under N deficiency relies on the interaction between tomato and soil microbiota that convert a plant-produced precursor that is a recruitment signal for beneficial, growth-promoting microbes to SolA, which was hijacked by cyst nematodes as a reliable host presence cue. This dual functionality resembles the microbial transformation of primary into secondary bile acids, important signalling molecules, in the gut of animals, and suggests convergent evolution in host microbe co-metabolism. Overall, our study positions SolA as a multifunctional signaling molecule in rhizosphere interactions, with potential application for enhancing crop resilience and sustainable pest management.