Eukaryotes are theorized to have an origin associated with the archaeal phylum Promethearchaeota (formerly Asgard archaea). The first cultured representatives revealed valuable insights4,5 but are still distantly related to eukaryotes, leaving many unknowns regarding eukaryogenesis. Here we report the isolation of two strains, HC1 and SC1, that belong to the archaeal lineage that branches closest to eukaryotes (Heimdallarchaeia);6–8 these isolates are members of a pure co-culture and a tri-culture with methanogenic partners. Both are obligately anaerobic, syntrophic, peptide-degrading and mesophilic archaea that produce protrusions and vesicles but lack persistent internal membrane structures, like Promethearchaeales4,5. All Promethearchaeota isolates demonstrate behaviour focused on cell construction and membrane structures rather than division, in contrast to typical prokaryotes9. Strain HC1 possesses genes associated with aerobic lifestyles (such as cytochrome c oxidase (Cox)) but co-cultures cannot grow under (micro)oxic conditions. Meanwhile, the strain demonstrates growth under microoxic conditions when in the presence of aerobes, suggesting that Cox supports oxygen detoxification rather than respiration. Comparative genomics provides evidence that, like HC1, Cox-possessing Heimdallarchaeia studied to date probably do not respire oxygen. The wide conservation of anaerobiosis, cell structure without persistent intracellular membranes and membrane protrusions across Promethearchaeota suggest that the archaeal ancestor of eukaryotes may have had such physiological features. In this cultivation-informed scenario for eukaryogenesis, the archaea–eukaryote transition would involve a steep transition in cell structure and physiological shift from anaerobiosis to (facultative) aerobiosis10, with early interactions between the archaeal host and mitochondrial ancestor probably occurring under low-oxygen conditions and at least partially mediated by oxygen detoxification. Cultures reveal conservation of cell structure, physical symbioses and anaerobiosis across Heimdallarchaeia (the archaeal lineage that branches closest to eukaryotes) and other Promethearchaeota.

Root-associated fungi and bacteria play pivotal roles in terrestrial ecosystem processes. However, identifying keystone taxa driving below-ground microbial assembly remains a major challenge. Here, we present a statistical framework for identifying microbes that reorganize microbiome assembly. Using 1,270 root samples collected from a cool-temperate forest in Japan, we reconstruct assembly landscapes describing the probabilistic structure of community states. We then find that several endophytic fungi (e.g., Phialocephala and Oidiodendron) are strongly associated with the reorganization of assembly landscape topography along abundance gradients, coinciding with shifts in mycorrhizal symbiosis. Independent datasets from Europe confirm that the highlighted genera are significantly associated with changes in mycorrhizal colonization patterns. Prokaryotic communities exhibit transitions among states characterized by core taxa, Pseudomonadota, or Actinomycetota, strongly associated with abundance gradients of Candidatus Udaeobacter and several endophytic fungi. These findings provide a scalable framework for prioritizing candidate keystone taxa and linking microbiome assembly to ecosystem-scale processes. Root microbiomes shape terrestrial ecosystems, yet identifying keystone microbes remains difficult. By statistically inferring assembly rules, the authors reveal endophytic fungi as hidden drivers reorganizing below-ground symbioses.

Adenosine-to-inosine RNA editing mediated by endogenous ADAR enzymes represents a promising therapeutic strategy because it can correct mutations at the transcript level without inducing permanent changes to the genome. However, existing systems face challenges, particularly the limited editing efficiency, which limits their practical applications. Here we introduce MARRS (multivalent ADAR1-recruiting RNA editing system)—an efficient and specific RNA editing platform that leverages engineered multivalent RNA aptamers to robustly recruit endogenous ADAR1 enzymes. By optimizing aptamers targeting the Zα domain of ADAR1 p150 and embedding them in a circular RNA scaffold, MARRS enables near-complete adenosine-to-inosine editing of RNA targets in cultured cells and mouse models, achieving efficiencies up to 100% in vitro and 84% in vivo. MARRS also exhibits remarkable versatility across a range of transcript targets while maintaining high specificity with minimal off-target effects. Operating without the need for exogenous enzymes, the MARRS system offers a safe, efficient and programmable platform. The multivalent ADAR1-recruiting RNA editing system (MARRS) is an RNA editing platform that efficiently recruits endogenous ADAR1 to achieve precise adenosine-to-inosine editing. MARRS enables up to 100% editing in vitro and 84% in vivo with high specificity across diverse RNA targets.

Panoptes illustrates the sophistication of bacterial immunity and the evolutionary plasticity of host-pathogen interactions. By inverting phage evasion into a self-destruct signal, bacteria have gained a decisive, if temporary, advantage. As phages respond with new countermeasures, the cycle continues, promising more surprises. The discovery of Panoptes reminds us that the most effective biological solutions often come from understanding the rules of the game and then subverting them.

Whole-cell models serve as powerful tools to predict cellular phenotype from genotype by integrating genome-scale processes within a single computational framework. Single-gene knockout (KO) growth is a stringent benchmark, but comparisons to experimental resources require close alignment of assay conditions, especially temperature and growth medium. We benchmark an E. coli whole-cell model by simulating multi-generation single-gene KOs in glucose minimal and amino-acid–supplemented media at 37 °C and by analyzing two independent eight-generation KO cohorts. For each KO we estimated generation-specific growth rates and harmonic-mean summaries across early generations. Experimental KO growth rates measured at 30°C were converted to s−1 and scaled to 37°C using an explicit Q10 correction Essential versus nonessential effects were tested using Welch’s t comparisons with 95% confidence intervals based on the Welch–Satterthwaite t approximation, and false discovery rate control across generations. Cross-medium generalization was evaluated using correlations and stratified regression for further prediction. Temperature adjustment shifted the experimental growth-rate distribution toward the simulated distribution. Essential KOs showed lower growth in the earliest generations, while the two eight-generation cohorts exhibited an overall decline in mean growth with substantial heterogeneity. Cross-medium correlations were weak, and in the held-out predictive comparison the global linear model achieved lower RMSE than the stratified SVR.

Achieving coordinated CRISPR-based transactivation (CRISPRa) regulation in vivo remains challenging, as the requisite genetic elements frequently exceed the packaging capacity of AAV vectors. Here we report the identification and engineering of the Cas12m protein family, a naturally nuclease-dead CRISPR system that possesses intrinsic self-processing activity for its precursor crRNA. We develop an optimized variant, denCas12m, and leverage it in the smCRISPRa platform. We show that a single transcriptional unit encoding a multi-gRNA array can be processed into individual functional guides within a target cell, enabling efficient synergistic activation of a single gene through multiple sites or simultaneous activation of distinct endogenous genes from a single all-in-one AAV vector. We demonstrate phenotypic rescue in mouse models of Duchenne muscular dystrophy and Scn1a haploinsufficiency. Furthermore, a single all-in-one AAV vector supports the simultaneous activation of multiple distinct endogenous genes in mice, with a single vector delivering up to eight crRNAs for activation of six different genes, demonstrating the platform’s high-order multiplexing. smCRISPRa is a compact platform for high-efficiency multiplexed gene activation in mouse models.

Built upon a multi-agent architecture comprising a leader agent, a kernel agent, a research agent and an enhancement agent (Fig. 1a and Supplementary Note 1), CopilotJ integrates task comprehension, software control, domain knowledge retrieval and advanced computation to support complex analysis...

The self-assembly of peptides into well-defined nanostructures represents a versatile strategy for developing biomedical agents. However, cellular internalization of pre-assembled peptide nanostructures is often hampered by insufficient uptake and low selectivity toward target cells, leading to off-target concerns and limited therapeutic efficacy. The past few years have witnessed the development of in situ assembly in living cells as a new endeavor for optimizing the performance of peptide assemblies. In situ assembly of peptides refers to intracellular assembling processes regulated by biocompatible stimuli, thereby enhancing peptide internalization and locally forming assemblies at target sites. Recently, we have developed strategies for establishing in situ assembly of peptides in living cells dictated by stimuli-responsive unnatural amino acids. Here we detail the procedure for developing in situ peptide assembly systems in living cells through incorporating stimuli-responsive unnatural amino acids. We present a thorough workflow encompassing peptide design and synthesis as well as the characterization of peptide responsiveness to glutathione or sulfatase. Self-assembly conditions are described for experiments in solution, in murine macrophage cells, human cervical carcinoma cells and murine breast cancer cells. The applications of in situ-formed peptide assemblies for subcellular targeting and cancer therapy in murine breast tumor-bearing BALB/c mice are also presented. The described methodology provides a robust platform for developing in situ-formed peptide assemblies and advancing the clinical translation of peptide-based supramolecular therapeutics. Performance of this Protocol requires expertise in peptide chemistry, materials engineering and biomedical engineering, taking an expected 92–115 d to complete. Peptide assemblies exhibit potential in biomedical applications due to their excellent biocompatibility and bioactivity. In this Protocol, a versatile strategy for the establishment of in situ peptide assembly systems in living cells is described.