Browse Articles
Discover research articles across all indexed journals
Mysterious ‘little red dots’ could be black holes in disguise
Little red dots as young supermassive black holes in dense ionized cocoons
Abstract The James Webb Space Telescope (JWST) has uncovered many compact galaxies at high redshift with broad hydrogen and helium lines, including the enigmatic population of little red dots (LRDs) 1,2 . The nature of these galaxies is debated and is attributed to supermassive black holes (SMBHs) 3,4 or intense star formation 5 . They exhibit unusual properties for SMBHs, such as black holes that are overmassive for their host galaxies 4 and extremely weak X-ray 6–10 and radio 6,11–13 emission. Here we show that in most objects studied with the highest-quality JWST spectra, the lines are broadened by electron scattering with a narrow intrinsic core. The data require very high electron column densities and compact sizes (light days), which, when coupled with their high luminosities, can be explained only by SMBH accretion. The narrow intrinsic line cores imply black hole masses of 10 5−7 M ⊙ , two orders of magnitude lower than previous estimates. These are the lowest mass black holes known at high redshift, to our knowledge, and suggest a population of young SMBHs. They are enshrouded in a dense cocoon of ionized gas producing broad lines from which they are accreting close to the Eddington limit, with very mild neutral outflows. Reprocessed nebular emission from this cocoon dominates the optical spectrum, explaining most LRD spectral characteristics, including the weak radio and X-ray emission 14,15 .
Credit in research goes hand in hand with responsibility
The asymmetric synthesis of an acyclic N-stereogenic amine
Abstract Most molecules in chemistry and biology are chiral, leading to mirror-image variants, so-called enantiomers. However, although the selective chemical synthesis of molecules in which the stereogenicity arises from a carbon atom is well-established, enantioselective approaches to nitrogen-stereogenic molecules are much less common 1–3 , and in case of acyclic, N -stereogenic amines, even unknown, because of their rapid pyramidal inversion. Here we describe the catalytic asymmetric synthesis of stable, acyclic N -stereogenic amines by the addition of enol silanes to nitronium ions that ion pair to a confined chiral anion. In the produced so-called anomeric amines, the commonly observed isomerization is slowed down by two N -oxy-substituents, which hamper nitrogen inversion. The important stereogenicity-creating step challenges previously established stereochemical descriptors of enantiodifferentiation. Computational studies provide further insight into the origin of the observed stereocontrol. Our work opens up a new avenue to investigate the fascinating and previously underexplored chemistry of enantiopure anomeric amines.
Earliest evidence of making fire
Semantic design of functional de novo genes from a genomic language model
Abstract Generative genomic models can design increasingly complex biological systems 1 . However, controlling these models to generate novel sequences with desired functions remains challenging. Here, we show that Evo, a genomic language model, can leverage genomic context to perform function-guided design that accesses novel regions of sequence space. By learning semantic relationships across prokaryotic genes 2 , Evo enables a genomic ‘autocomplete’ in which a DNA prompt encoding genomic context for a function of interest guides the generation of novel sequences enriched for related functions, which we refer to as ‘semantic design’. We validate this approach by experimentally testing the activity of generated anti-CRISPR proteins and type II and III toxin–antitoxin systems, including de novo genes with no significant sequence similarity to natural proteins. In-context design of proteins and non-coding RNAs with Evo achieves robust activity and high experimental success rates even in the absence of structural priors, known evolutionary conservation or task-specific fine-tuning. We then use Evo to complete millions of prompts to produce SynGenome, a database containing over 120 billion base pairs of artificial intelligence-generated genomic sequences that enables semantic design across many functions. More broadly, these results demonstrate that generative genomics with biological language models can extend beyond natural sequences.
Cross-regulation of [2Fe–2S] cluster synthesis by ferredoxin-2 and frataxin
LLMs behaving badly: mistrained AI models quickly go off the rails
Rewiring an olfactory circuit by altering cell-surface combinatorial code
Abstract Proper brain function requires the precise assembly of neural circuits during development. Despite the identification of many cell-surface proteins (CSPs) that help guide axons to their targets 1,2 , it remains mostly unknown how multiple CSPs work together to assemble a functional circuit. Here we used synaptic partner matching in the Drosophila olfactory circuit 3,4 to address this question. By systematically altering the combination of differentially expressed CSPs in a single type of olfactory receptor neuron (ORN), which senses a male pheromone that inhibits male–male courtship, we switched its connection nearly completely from its endogenous postsynaptic projection neuron (PN) type to a new PN type that promotes courtship. From this switch, we deduced a combinatorial code including CSPs that mediate both attraction between synaptic partners and repulsion between non-partners 5,6 . The anatomical switch changed the odour response of the new PN partner and markedly increased male–male courtship. We generalized three manipulation strategies from this rewiring—increasing repulsion with the old partner, decreasing repulsion with the new partner and matching attraction with the new partner—to successfully rewire a second ORN type to multiple distinct PN types. This work shows that manipulating a small set of CSPs is sufficient to respecify synaptic connections, paving the way to investigations of how neural systems evolve through changes of circuit connectivity.
Enriching African genome representation through the AGenDA project
Six steps to protect researchers’ digital security
Repulsions instruct synaptic partner matching in an olfactory circuit
Abstract Neurons exhibit extraordinary precision in selecting synaptic partners. Although cell-surface proteins (CSPs) that mediate attractive interactions between developing axons and dendrites have been shown to instruct synaptic partner matching 1,2 , the degree to which repulsive interactions have a role is less clear. Here, using a genetic screen guided by single-cell transcriptomes 3,4 , we identified three CSP pairs, Toll2–Ptp10D, Fili–Kek1 and Hbs/Sns–Kirre, that mediate repulsive interactions between non-partner olfactory receptor neuron (ORN) axons and projection neuron (PN) dendrites in the developing Drosophila olfactory circuit. Each CSP pair exhibits inverse expression patterns in the select ORN–PN partners. Loss of each CSP in ORNs led to similar synaptic partner matching deficits as the loss of its partner CSP in PNs, and mistargeting phenotypes caused by overexpressing one CSP could be suppressed by loss of its partner CSP. All CSP pairs are also differentially expressed in other brain regions. Together, our data reveal that multiple repulsive CSP pairs work together to ensure precise synaptic partner matching during development by preventing neurons from forming connections with non-cognate partners.
Mutations in mitochondrial ferredoxin FDX2 suppress frataxin deficiency
Abstract Frataxin is a key component of an ancient, mitochondrial iron–sulfur cluster biosynthetic machinery, serving as an allosteric activator of the cysteine desulfurase NFS1 (refs. 1–5 ). Loss of frataxin levels underlies Friedreich’s ataxia 6 , the most common inherited ataxia. Yeast, C aenorhabditis elegans and human cells can tolerate loss of frataxin when grown in ‘permissive’ low oxygen tensions 7 . Here we conducted an unbiased, genome-scale forward genetic screen in C. elegans leveraging permissive and non-permissive oxygen tensions to discover suppressor mutations that bypass the need for frataxin. All mutations act dominantly and are in the ferredoxin FDX2 / fdx-2 or in the cysteine desulfurase NFS1 / nfs-1 genes, resulting in amino-acid substitutions at the FDX2–NFS1 binding interface. Our genetic and biochemical analyses show that the suppressor mutations boost iron–sulfur cluster levels in the absence of frataxin. We also demonstrate that an excess of FDX2 inhibits frataxin-stimulated NFS1 activity in vitro and blocks the synthesis of iron–sulfur clusters in mammalian cell culture. These findings are consistent with structural and biochemical evidence that frataxin and FDX2 compete for occupancy at the same site on NFS1 (refs. 8,9 ). We show that lowering levels of wild-type FDX2 through loss of one gene copy can ameliorate the growth of frataxin mutant C. elegans or the ataxia phenotype of a mouse model of Friedreich’s ataxia under normoxic conditions. These genetic and biochemical studies indicate that restoring the stoichiometric balance of frataxin and FDX2 through partial knockdown of FDX2 may be a potential therapy for Friedreich’s ataxia.
Structural basis of regulated N-glycosylation at the secretory translocon
Abstract Most human secretory pathway proteins are N-glycosylated by oligosaccharyltransferase (OST) complexes as they enter the endoplasmic reticulum (ER) 1–3 . Recent work revealed a substrate-assisted mechanism by which N-glycosylation of the chaperone glucose-regulated protein 94 (GRP94) is regulated to control cell surface receptor signalling 4 . Here we report the structure of a natively isolated GRP94 folding intermediate tethered to a specialized CCDC134-bound translocon. Together with functional analysis, the data reveal how a conserved N-terminal extension in GRP94 inhibits OST-A and how structural rearrangements within the translocon shield the tethered nascent chain from inappropriate OST-B glycosylation. These interactions depend on a hydrophobic CCDC134 groove, which recognizes a non-native conformation of nascent GRP94. Our results define a mechanism of regulated N-glycosylation and illustrate how the nascent chain remodels the translocon to facilitate its own biogenesis.
Exciplex-enabled high-efficiency, fully stretchable OLEDs
Training large language models on narrow tasks can lead to broad misalignment
Abstract The widespread adoption of large language models (LLMs) raises important questions about their safety and alignment 1 . Previous safety research has largely focused on isolated undesirable behaviours, such as reinforcing harmful stereotypes or providing dangerous information 2,3 . Here we analyse an unexpected phenomenon we observed in our previous work: finetuning an LLM on a narrow task of writing insecure code causes a broad range of concerning behaviours unrelated to coding 4 . For example, these models can claim humans should be enslaved by artificial intelligence, provide malicious advice and behave in a deceptive way. We refer to this phenomenon as emergent misalignment. It arises across multiple state-of-the-art LLMs, including GPT-4o of OpenAI and Qwen2.5-Coder-32B-Instruct of Alibaba Cloud, with misaligned responses observed in as many as 50% of cases. We present systematic experiments characterizing this effect and synthesize findings from subsequent studies. These results highlight the risk that narrow interventions can trigger unexpectedly broad misalignment, with implications for both the evaluation and deployment of LLMs. Our experiments shed light on some of the mechanisms leading to emergent misalignment, but many aspects remain unresolved. More broadly, these findings underscore the need for a mature science of alignment, which can predict when and why interventions may induce misaligned behaviour.
Direct observation of the Migdal effect induced by neutron bombardment
Abstract The search for dark matter focuses now on hypothetical light particles with masses ranging from MeV to GeV (refs. 1–12 ). These particles would leave very faint signals experimentally. A potential avenue for enhancing experimental sensitivity to light matter relies on the Migdal effect 13–15 , which involves the detectable ejection of electrons following the instantaneous accelerations of atoms colliding with neutral dark matter. However, although the Migdal effect could be equally generated in controlled experiments with neutral projectiles, a direct experimental observation of this effect is missing, casting doubt on the reliability of detection experiments relying on this effect. Here we report the direct observation of the Migdal effect in neutron–nucleus collisions, achieving a statistical significance of 5 standard deviations, which rests on 6 candidate events selected out of almost 10 6 recorded events. Our experiments have determined the ratio of the Migdal cross-section to the nuclear recoil cross-section to be $${4.9}_{-1.9}^{+2.6}\times {10}^{-5}$$ 4.9 − 1.9 + 2.6 × 10 − 5 , in which nuclear recoils exceed 35 keVee and electron recoils span 5–10 keV. These findings are consistent with theoretical predictions. This work resolves a long-standing gap in experimental validation, which not only strengthens the theoretical foundation of the Migdal effect but also paves the way for its application in light dark matter detection.