Programming biological function—designing bespoke proteins to perform specified molecular tasks—is a foundational goal of molecular engineering. However, current design models remain fundamentally limited1–3: 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.
Hantaviruses are rodent-borne pathogens that cause severe human disease for which no approved therapeutics are available. New World hantaviruses, including Andes virus (ANDV), use the cell-surface receptor protocadherin-1 (PCDH1) to enter host cells1. However, the structural basis of this interaction remains unclear. Here we describe the structure of PCDH1 bound to the ANDV spike at near-atomic resolution. PCDH1 asymmetrically engages a deep central pocket formed by four Gn protomers in the spike, explaining why only intact viral particles or soluble tetramers recognize it. Mutagenesis identified three Gn loops that shape the receptor-binding site and afford the transfer of PCDH1 binding from New World to Old World glycoproteins. Structural and functional analyses of four neutralizing antibodies revealed that they block receptor binding through two distinct mechanisms—occluding access to the PCDH1-binding site or sterically preventing receptor engagement. Together, these findings define the structural determinants of hantavirus–receptor recognition, reveal how a virus can mask its receptor-binding site by recessing it in a surface depression and how antibodies can counter this strategy, and provide a framework for the design of antiviral therapeutics targeting the hantavirus spike. The structure of PCDH1 bound to the ANDV spike defines the structural determinants of hantavirus–receptor recognition.
Anaerobic degradation of complex organic compounds to methane and carbon dioxide is common in the deep subsurface biosphere and has long been considered to require syntrophic cooperation, with methanogens confined to the terminal use of simple C1 and C2 substrates. Thermodynamic considerations, multi-omics analyses and microcosm experiments suggest that archaeal lineages are capable of methanogenic degradation of complex substrates1–6. However, experimental validation in pure culture was lacking. Here we report the isolation of a thermophilic archaeon, Methanoglobus dolicholipophagus AA-75, affiliated with Archaeoglobi, from a Chinese petroleum reservoir. Strain AA-75 not only converts methanol into methane, carbon dioxide and acetate, but also uses saturated fatty acids varying from C7:0 to C18:0 and unsaturated fatty acids including C16:1, C18:1 and C18:2 to produce methane. We term the latter lipotrophic methanogenesis, which integrates β-oxidation of fatty acids with the Wood–Ljungdahl pathway and the canonical methyl-tetrahydromethanopterin:coenzyme M methyltransferase and methyl-coenzyme M reductase complexes, together with an intracellular electron transfer system. A formaldehyde-methanol branch probably functions as an auxiliary route for C1 metabolism during methanogenesis. Methane produced from stearate is moderately 13C-enriched (δ13C–CH4 = −40.0‰ to −36.3‰), distinct from the strong 13C depletion characteristic of canonical methanogenic pathways. These findings demonstrate that methanogens can function as primary degraders of complex organic matter, challenging the classical syntrophic paradigm and calling for refinement of current models of carbon cycling in the deep subsurface biosphere. Methanoglobus dolicholipophagus AA-75 is a thermophilic archaeon isolated from a Chinese petroleum reservoir that converts methanol into methane, carbon dioxide and acetate, and uses saturated and unsaturated fatty acids to produce methane.
Mitochondria are bound by a double membrane and are semiautonomous organelles, rendering them highly vulnerable to global and internal proteostatic stress1. Multiple surveillance mechanisms at proteostatic and inter-organellar levels have evolved to maintain mitochondrial integrity and cellular health2–7. However, the precise mechanisms by which proteostatic damage relays stress signals across organelles and how the quality surveillance machinery coordinates these functions remain unknown. Here we report that under proteostatic stress, damaged mitochondrial proteins are sequestered to form stress-induced mitochondrial protein aggregates (SIMPAs) that activate specific quality control mechanisms in a spatially confined manner without perturbing the mitochondrial network. Quantitative proteomic and correlative light–electron microscopy analysis revealed that SIMPA mediates inter-organellar communication via mitochondria–endoplasmic reticulum contact sites or mitochondrial-associated membranes. Mechanistically, GRP75 coordinates SIMPA biogenesis and focally activates endoplasmic reticulum unfolded protein response pathways via mitochondria–endoplasmic reticulum contact sites or mitochondrial-associated membranes for the targeted clearance of SIMPAs. Functionally, in both cellular and liver disease models, we found that a GRP75-dependent SIMPA signalling cascade attenuates mitochondrial stress and preserves mitochondrial integrity under proteotoxic stress. Overall, we show that SIMPAs, functioning as protein condensates, serve as a central cross-compartment communication hub for mitochondrial quality control, supporting cellular and organismal fitness. Proteotoxic stress drives formation of GRP75-dependent protein aggregates that signal through mitochondria–ER contact sites to activate targeted quality-control pathways, enabling selective clearance of damaged proteins and preserving mitochondrial integrity.
Astrocytes perform a diverse range of tasks that facilitate brain physiology, raising the possibility that these diverse astrocytic roles are enabled by functionally distinct subtypes1,2. Recent studies identified molecular diversity among astrocytes3; however, whether this molecular heterogeneity reflects specialized astrocyte subpopulations with distinct functions remains unclear. Here, on the basis of previous molecular and functional studies4, we generated a CD51-FLP mouse line that enables direct comparison of the functional properties between CD51+ and CD51− astrocyte subpopulations in neural circuits in vivo. We show that CD51+ and CD51− astrocytes have unique morphological, physiological and developmental properties in both the cortex and hippocampus. Functional studies demonstrate that CD51+ astrocytes selectively promote synapse formation in the developing cortex and circuit plasticity in the hippocampus. Mechanistically, we found that CD51+ astrocytes function through glypican 6 to promote synapse formation in the cortex, while CD51 itself drives circuit plasticity in the hippocampus. Applying these findings to brain injury, we found that CD51+ astrocytes promote functional recovery after stroke through enhanced synaptic remodelling. Taken together, our studies identify specialized astrocyte subpopulations that exhibit distinct functional properties in the developing, mature and injured brain. CD51+ astrocytes promote synapse formation in the developing cortex, circuit plasticity in the hippocampus and circuit remodelling after stroke.
