Deep soils store over half of terrestrial organic carbon, yet the viral mechanisms governing carbon persistence in subsurface contexts remain unclear. Here, by conducting a soil profile survey (0–700 cm) across four agroecosystems and a global synthesis of subsurface metagenomes, we established the Deep Soil Viral Resources Database, comprising 50,949 viral operational taxonomic units and 3,902 auxiliary metabolic genes (AMGs). Viral richness increased significantly with depth, accompanied by a shift towards lysogeny driven by declining carbon availability. The deep-soil virome carried AMGs and lysozymes involved in polymer degradation and cell endolysis. Viral endolysin significantly promoted necromass carbon accumulation, which contributed to increases in mineral-associated organic carbon. A synthesis of 138 global soil profiles further revealed a positive effect of viral abundance on soil organic carbon. Our findings establish a framework where viruses strengthen the microbial carbon pump in deep soils by facilitating host carbon scavenging via AMGs and promoting necromass formation via endolysin activity. Combining a metagenomic analysis of agroecosystem soils along a 7-metre depth profile with microcosm incubations and a global synthesis, the authors find increases in viral diversity, lysogeny and necromass carbon accumulation in deeper soils.

Programming biological function—designing bespoke proteins to perform specified molecular tasks—is a foundational goal of molecular engineering. However, current design models remain fundamentally limited: they typically require either natural proteins as starting points for optimization or manual reformulation of functional goals as geometric and sequence-level constraints to guide candidate generation. Here we introduce Pinal, a 16-billion-parameter foundation model that designs candidate proteins from natural-language descriptions of desired function. Trained on 1.7 billion synthetically annotated protein–text pairs, Pinal links functional intent to protein sequence and structure. In computational evaluations, generated candidates combined high predicted foldability with functional-description alignment and sequence diversity, providing a basis for prioritizing experimentally testable designs. We applied Pinal to four distinct design tasks—a fluorescent protein, a polyethylene terephthalate hydrolase, an alcohol dehydrogenase and a metabolic H-protein—and experimentally observed the intended function in each case, including catalytic activity for both designed enzymes. Without iterative experimental optimization, a Pinal-designed H-protein increased product titre by 1.7-fold relative to the corresponding native Escherichia coli H-protein in a multi-enzyme CO2 fixation pathway. These findings support natural language as a high-level interface for candidate generation in protein design, enabling programmable exploration with reduced reliance on manually specified structural or sequence constraints. Pinal, a 16-billion-parameter artificial intelligence model, designs functional proteins from natural-language descriptions, demonstrating experimentally validated activities and improved CO2-fixation product yields without iterative optimization.

Mitochondria are essential organelles containing their own genomes, encoding a few proteins essential for energy production. Most of the mitochondrial proteins are nucleus-encoded, translated as precursors in the cytosol, with a large fraction of these precursors properly addressed by an N-terminal mitochondrial targeting sequence (MTS). These MTS share common features but no consensus sequence can explain their functionality nor the precursors-specific determinants of mitochondrial import. To help understand this mechanism, we created a simple computational model to generate functional synthetic MTS while maintaining a tight control on the design parameters. Using the budding yeast, we demonstrated the presence of precursors-specific import capabilities in addressing artificially nucleus-relocated OXPHOS proteins. Our research confirms the existence of cargo-dependent functional compatibilities and takes us one step closer towards improving gene therapy-based treatment of mitochondrial diseases.

The rise of Klebsiella pneumoniae combining antimicrobial resistance and virulence genes poses a major health threat, but the evolutionary routes and phenotypic consequences of this convergence are poorly understood. Here, phylogenetics of 1468 isolates and population analysis of 7520 plasmids, from >50 countries through the last two decades, reveal that convergence follows preferred, constrained evolutionary paths. The dominant route involves multidrug-resistant classical K. pneumoniae acquiring conjugative IncFIB(Mar)/IncHI1B(Mar) plasmids carrying an incomplete set of virulence biomarkers. Across 25 independent convergence events, the acquisition of the aerobactin siderophore locus was the only universal feature. These convergent isolates exhibit enhanced siderophore production but consistently lack the hypervirulent phenotype in vivo. In contrast, genuine hypervirulent strains that gain resistance remain rare. We conclude that enhanced siderophore production, not hypervirulence, is the primary adaptive trait driving the success of globally emerging convergent lineages, representing a distinct evolutionary state optimized for transmission rather than systemic invasion. Authors present a global study revealing that drug-resistant Klebsiella pneumoniae repeatedly acquires conjugative plasmids with aerobactin, boosting iron-scavenging ability without becoming hypervirulent. This may help emerging strains spread in healthcare settings.

Non-ribosomal peptides play important roles in microbial ecology and medicine, for example as toxins and antibiotics such as penicillin. Like ordinary proteins, non-ribosomal peptides are subject to evolutionary change, which is driven by rampant recombinations of their synthetases. The evolutionary dynamics and causes of non-ribosomal peptide diversification are not well understood, because recombinations complicate phylogenetic analysis and statistical tests for natural selection and drift. Here we study the non-ribosomal GameXPeptides, which entomopathogenic bacteria use to kill their insect prey. We use recombination-aware phylogenetics, ancestral sequence reconstruction and statistical analysis of historical recombination patterns to retrace the evolution of GameXPeptides. This uncovers dozens of new GameXPeptides, including striking patterns of structural convergence. We provide statistical and experimental evidence that this diversity is shaped by selection, but also that GameXPeptides experience substantial genetic drift and biased recombination rates that limit the space of accessible peptide variants. Our work illuminates the complex interplay of adaptive and non-adaptive evolutionary forces acting on some of the most versatile chemical weapons in nature. The structural diversity of non-ribosomal peptides is largely determined by recombination. This study shows that selection, genetic drift and biased recombination rates contribute to the diversity of GameXPeptides, non-ribosomal peptides used by entomopathogenic bacteria to kill their insect prey.

Tools for manipulating RNA that rely on nonhuman proteins, such as ISC-based and Cas-based scaffolds, can lead to unintended and little-understood effects on human cells and present challenges in therapeutic translation. Replacing them with human proteins engineered for expanded, programmable functions could bypass the limitations of these foreign proteins. Here we demonstrate that human Argonaute 2 (AGO2) can be repurposed as a modular RNA targeting domain. AGO2 fused to a C-to-U deaminase enables AGO-led targeted editing of RNA (ALTER). Using guide RNAs that remodel target RNA structure for selective editing and reduced nuclease activity, we show that ALTER achieves, on average, 1.9-fold higher C-to-U editing on mRNA targets in HEK293T cells compared with Cas13 editors CURE and RESCUE. Transcriptome-wide RNA-seq revealed similar or lower levels of off-target editing compared to the same Cas13-based editors. Machine learning models predict that far fewer immunogenic peptides will result from ALTER and we show that ALTER is amenable to a validated workflow for artificial intelligence-directed deimmunization. Human Argonaute is repurposed for RNA editing.

G-quadruplexes (G4s) are non-canonical nucleic acid structures with critical roles in gene regulation, genomic stability, and disease, making them prime targets for therapeutic and biotechnological applications. However, the absence of a curated, centralized, experimentally validated repository of short (mostly synthetic) G4-forming sequences, paired with appropriate single-stranded or hairpin controls, has limited reproducibility and hindered progress in the field. Here, we introduce G4All, a comprehensive, curated database of G4-forming short DNA and RNA sequences, complemented by rigorously selected non-G4 controls, all studied under roughly similar experimental conditions (generally around 100 mM potassium and near neutral pH). G4All consolidates sequences validated by diverse experimental methods, including circular dichroism, NMR, UV spectroscopy, with annotations for topology, thermal stability, and experimental conditions. By providing both positive and negative datasets, G4All enables rigorous comparative analyses, assay benchmarking, and the development of predictive models. The database supports a broad range of applications, from fundamental studies of G4 biophysics to the rational design of aptamers, as well as benchmarking prediction algorithms. Future developments will expand G4All to include user-submitted datasets and more RNA sequences. Freely accessible at https://g4-all.streamlit.app/, G4All offers a searchable interface and downloadable datasets, establishing a community-driven hub to accelerate discovery and standardization in G4 research.

Transcriptomic measurement technologies provide powerful approaches to understand biology but are limited to providing either a high-resolution snapshot at a moment in time or a low-resolution history of transcription constrained to a few genes over a narrow time range. Here, we introduce the ‘recordanelle’, a single-component synthetic organelle capable of recording the transcriptome in human cells. Recordanelles are composed of a mitochondrial membrane anchor, a condensate assembler and a poly(A)-binding protein. We demonstrate that the recordanelle can be engineered to recruit engineered RNAs or a sampling of up to 90% of all endogenous mRNAs and protect them over several days. We further show that transient and dynamic transcriptional responses resulting from inducible promoters or interferon signaling are preserved throughout time within the recordanelle at the single-cell level. This work provides a nondisruptive and unbiased transcriptome-wide single-cell RNA recording technology to enable a deeper understanding of biology. Recordanelles continuously and unbiasedly record the transcriptome throughout time in living mammalian cells.